INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

Genetic engineering tools make it possible to clone and manipulate almost any given piece of DNA and to transfer this material to a wide range of organisms. This procedure has many applications in medicine, including expression in heterologous hosts of a number of genes that code for proteins and hormones of medical and pharmaceutical interest and biosynthesis of new antibiotics and other chemicals with added value (4, 11, 14). In recent years a number of tools have been specifically designed to develop microorganisms as biopesticides and as enhancers of plant growth which have potential uses in agriculture (7). In the area of bioremediation, recombinant microorganisms with improved activities for amelioration of pollution have also been constructed (17, 19). Although genetically engineered microorganisms (GEMs) can be used for bioproduction of added-value pharmaceuticals and other chemicals in physically contained sites, GEMs that are used to biologically treat polluted sites or stimulate plant growth and plant protection will eventually be used in open environments. Although the benefits of cleaning a polluted site by biological means are obvious, the long-term behavior of GEMs released in open environments is an unexplored area. The consequences of the persistence of GEMs at sites and their effects on recolonization of the sites by indigenous microbes are also unknown. These issues raise serious concerns (15).

One way to decrease the persistence of GEMs in the environment is to provide them with active biological containment (ABC) systems. ABC systems are based on control of the expression of a lethal function (e.g., a porinlike protein or a nuclease) via sensory systems that recognize physical or chemical signals in the surrounding environment (15). Contreras et al. (5) designed the first circuit for biological containment of bacteria that degrade pollutants. This system was based on the well-characterized regulatory circuit for expression of the meta-cleavage pathway of the Pseudomonas putida TOL plasmid pWW0, the xylS gene and its cognate Pm promoter (21), and the gef gene of Escherichia coli, which encodes a membrane protein that collapses the cell membrane potential by generating pores (15). The model shown in Fig. 1 predicts that in the presence of XylS effectors (i.e., a wide range of alkyl-substituted, chloro-substituted, and other halosubstituted benzoates) (22), expression of the porinlike Gef protein is prevented and the strain survives and removes the pollutant. When the target compound is exhausted (or the microbe is spread to a nonpolluted site), the lack of induction of Pm results in the absence of the LacI protein in the host cell, which leads to expression of the lethal protein and cell death.

View larger version (15K): [in this window] [in a new window]   FIG. 1.   ABC model system. This model consists of the Pm promoter, which drives transcription of the meta-cleavage pathway of the TOL plasmid, and the xylS gene, which encodes the sensor protein () that interacts with alkylbenzoates and stimulates transcription from Pm (). In the containment system the lacI gene coding for the LacI repressor protein () was cloned downstream from Pm. The lethal element consists of the Plac promoter fused to the gef gene of E. coli, a member of the hok gene family (15), which encodes a pore-forming protein. The system performs as described in the text. 3MB, 3-methylbenzoate.

This model system was shown to be functional in E. coli (5) and in Pseudomonas putida (12, 26), but it exhibited a relatively high rate of killing escape, which was found to be associated with the fact that the regulatory element of the ABC system was plasmid borne (5, 12, 25). In this study we describe transfer of the elements of the ABC system to the chromosome of P. putida KT2440 (9). The resulting new biologically contained strain was shown to function under laboratory conditions in accordance with the model proposed in Fig. 1, and the frequency of killing escape was less than 1 in 10⁸ cells per cell and per generation. The strain with the ABC system was introduced into planted and unplanted pots with and without 3-methylbenzoate. Inoculated pots were kept under environmental conditions during the spring-summer and autumn-winter periods, and the survival of the contained strain and the survival of an uncontained control strain under these conditions were determined.

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Bacterial strains, plasmids, and growth conditions. The strains and plasmids used in this study and their relevant characteristics are shown in Table 1. Bacterial strains were grown with shaking at 30°C. E. coli strains were grown on Luria-Bertani (LB) medium at 37°C (27). P. putida strains were grown in modified M9 minimal medium (1) supplemented with 28 mM glucose, 5 mM benzoate, 5 to 15 mM 3-methylbenzoate, or 10 mM p-hydroxyphenylacetic acid as a carbon source. When appropriate, antibiotics were used at the following final concentrations: chloramphenicol, 30 µg/ml; kanamycin, 50 µg/ml; rifampin, 20 µg/ml; and tetracycline, 15 µg/ml.

Matings. Triparental matings were used to mobilize nonautotransmissible plasmids into P. putida strains. Equal numbers (about 10⁸ cells) of the recipient strain P. putida 2440 (benzoate positive, Cm^r), the donor strain E. coli CC118pir bearing the suicide pSM1350 plasmid, and the helper strain E. coli HB101 (pRK600) were mixed and deposited on a nitrocellulose filter placed on the surface of an LB agar plate containing 5 mM 3-methylbenzoate (6). Appropriate controls containing unmixed cells were included. P. putida transconjugants were selected on M9 minimal medium plates supplemented with kanamycin, chloramphenicol, 10 mM benzoate, and 5 mM 3-methylbenzoate. (Note that P. putida 2440 cannot use 3-methylbenzoate as a C source because it does not bear the TOL plasmid pWW0; in this case the aromatic carboxylic acid was used as a gratuitous inducer of the XylS protein to prevent cell killing.) Biparental matings were used to transfer the TOL plasmid pWW0 between P. putida strains under the conditions described previously with the controls described previously (23).

Transmission electron microscopy. P. putida cells were harvested by centrifugation (4,000 × g, 5 min) and then immediately fixed with 2% (vol/vol) glutaraldehyde-1% (vol/vol) formaldehyde in cacodylate buffer, postfixed with osmium tetroxide in the presence of 2% (wt/vol) potassium ferrocyanide, and embedded in Eponate 12. Thin sections were poststained with uranyl acetate and lead citrate and examined with a Zeiss transmission electron microscope at an accelerating voltage of 75 kV.

Light emission measurements. Soil leachates obtained from rhizosphere and bulk soils were prepared as follows: 10 g of soil was suspended in 90 ml of phosphate buffer, and after shaking for 1 h the soil was decanted and the liquid suspension was used for assays. To 1 ml of each soil leachate suspension we added 100 µl of a culture of P. putida CMC5 (Pm:'luxAB xylS) containing about 10⁸ CFU/ml and then incubated the resulting culture for 1 h at 30°C to allow induction of Pm. To determine luciferase activity, the turbidity of the culture at 660 nm was adjusted to 0.1, and to 1 ml of this P. putida CMC5 cell suspension we added 0.1 ml of 0.01% (vol/vol) n-decylaldehyde and then recorded the time course of light emission immediately thereafter for 1 min with an LKB model 1250 luminometer (16). Activity was expressed as the peak height in relative light units per turbidity unit.

Seed coating and soil inoculation. Cells in the mid-log phase in 1 liter of culture medium were harvested by centrifugation, washed twice in 50 mM phosphate-100 mM NaCl, and resuspended in the same buffer to a concentration of about 10⁸ CFU/ml. About 300 seeds of corn (Zea mays) or broad bean (Vicia faba) were soaked in 200 to 400 ml of the two cell suspensions used for 30 min with gentle shaking at 30°C. The number of bacteria attached per seed was estimated as follows. Two seeds coated with P. putida were air dried and then transferred to a test tube containing 1 to 5 ml of 1× M9 minimal medium without a C source. The preparations were vortexed for 2 min, and then serial dilutions were spread onto plates. Each seed was coated with about 10⁶ CFU (corn seeds) or 10⁷ CFU (broad bean seeds). Two seeds per pot were sown at a depth of 2 cm in pots that were 40 cm in diameter and contained 40 kg of soil. Twelve pots per treatment were used; one pot was used per sample, and each pot was analyzed in triplicate.

View larger version (21K): [in this window] [in a new window]   FIG. 2.   Distribution of subareas in the field release assay. In the spring-summer assay corn (Z. mays) plants were used, and in the autumn-winter assay the species used was broad bean (V. faba). Bacteria were introduced in the form of a biofilm on each seed or were homogeneously mixed with soil in pots without seeds. 3MB, 3-methylbenzoate.

Environmental release field design. The Spanish Ministry of Environment provided the necessary permits to carry out outdoor assays with P. putida EEZ32 (26) and CMC4 (this study). On the advice of the authorities, the soil was kept in pots, which were placed in different subareas in a controlled, fenced-in, 96-m² site located within the 2,000-m² experimental area at the Estación Experimental del Zaidín of the Consejo Superior de Investigaciones Científicas, Granada, Spain (Fig. 2). The granulometric composition of this soil was as follows: sand, 43%; silt, 41%; and clay, 16%. The pH of the soil was 7.8, and its CaCO₃ content was 6% (wt/wt). The experimental subareas in the agricultural experimental field site consisted of groups of pots filled with this soil. One group of pots contained seeds coated with the control bacteria or the contained bacteria, and other series of pots contained soil inoculated with the GEMs but no seeds. Some pots were supplemented with 0.01% (wt/wt) 3-methylbenzoate. To minimize edge effects and provide material to assess the dispersal of the GEMs, groups of pots were separated from each other by 1 m and the entire release area was completely surrounded by a buffer zone where GEMs were not introduced (Fig. 2). Z. mays (corn) seeds were used in the field trials performed during the spring and summer. The assay period lasted for 112 days from May to September 1996. During this period the daytime temperatures ranged from 25 to 45°C and the nighttime temperatures ranged from 10 to 20°C. V. faba (broad bean) seeds were used in the field trials performed during the autumn and winter. This assay was performed from December 1995 to March 1996. The temperature ranged from 3 to 13°C during the night and from 5 to 27°C during the day.

Monitoring bacteria in soil and the rhizosphere. After germination, the first samples were obtained from individual plants after the appearance of the first true leaf (11 days after sowing for corn and 21 days after sowing for broad bean), and samples were obtained at subsequent times. Whole plants were gently removed from the soil, and the bacteria in the soil attached to the roots (rhizosphere soil) and in the rest of the soil (bulk soil) were counted.

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

Construction of P. putida CMC4, a strain with the ABC system on the host chromosome. A mini-Tn5-Km transposon bearing the P_lac::gef fusion was transferred to the chromosome of P. putida 2440 via triparental mating as described in Materials and Methods. Twenty Km^r transconjugants, which appeared at a frequency of about 2 × 10⁸ transconjugant per recipient cell, were selected for further study. As expected, all transconjugant cells grew on minimal medium containing glucose as the sole C source only if 3-methylbenzoate was also present, which confirmed the functionality of the ABC system. When the clones were grown with glucose as the C source and with 3-methylbenzoate in the absence of kanamycin for about 50 generations, we always found that 100% of the cells retained the Km^r marker and died when they were spread onto plates without 3-methylbenzoate. This suggested that the mini-Tn5 element bearing the ABC system was stably maintained in the host.

1

2

6

8

8

8

View larger version (14K): [in this window] [in a new window]   FIG. 3.   Effect of removal of 3-methylbenzoate from the culture medium of P. putida EEZ32 and the contained strain P. putida CMC4. Cells of P. putida EEZ32 ( and ) and P. putida CMC4 ( and ) growing exponentially on M9 minimal medium containing glucose and 15 mM 3-methylbenzoate were filtered, thoroughly washed with 50 mM phosphate buffer, and resuspended in M9 minimal medium containing glucose but not 3-methylbenzoate ( and ) or in M9 minimal medium containing glucose and 15 mM 3-methylbenzoate ( and ). At the times indicated the numbers of viable cells (CFU per milliliter) were determined in triplicate in LB medium supplemented with 5 mM 3-methylbenzoate and appropriate antibiotics.

Autumn-winter outdoor assay. We examined the behavior of the two strains in pots in which broad beans were planted. After germination, the first sample was obtained when the first leaf appeared, 21 days after sowing. The results are presented in Fig. 4. Both in the absence and in the presence of 3-methylbenzoate the control strain tended to become established at a concentration of about 10⁵ CFU per g of rhizosphere soil. During the assay the number of organisms ranged from 10³ to 10⁶ CFU/g of rhizosphere soil. Control strain P. putida EEZ32 became established in pots containing 3-methylbenzoate faster than it became established in the absence of this aromatic compound, but at the end of the study the total counts were similar to the total counts obtained in the absence of 3-methylbenzoate (Fig. 4A). The number of CFU of the contained strain per gram of rhizosphere soil tended to decrease with time, and this phenomenon was more pronounced in the absence of 3-methylbenzoate, so that after 100 days the concentration of the strain was below our detection limit. In the presence of 3-methylbenzoate the number of CFU per gram of rhizosphere soil was 1 to 2 orders of magnitude higher in any given sample, and at the end of the study we found almost 10³ CFU/g of soil. Confirmation that the CFU in the pots into which the contained strain was introduced indeed represented contained strain cells was obtained from the fact that none of the bacteria selected on plates containing 3-methylbenzoate was able to grow on minimal medium supplemented with glucose, whereas all of the control bacteria survived in this medium.

View larger version (15K): [in this window] [in a new window]   FIG. 4.   Survival of P. putida CMC4 and EEZ32 under outdoor conditions in pots. Broad bean seeds were coated with P. putida EEZ32 (A) or P. putida CMC4 (B) and sown in pots in which the soil was not supplemented ( and ) or was supplemented ( and ) with 3-methylbenzoate. The numbers of CFU per gram of rhizosphere soil were determined at the times indicated. Data are the averages of values from three independent counts, and the standard deviations were less than 10%.

Spring-summer outdoor assay. In the spring-summer outdoor assay corn seeds were coated with the control or contained strain, and the survival of the bacteria was investigated. In soil without 3-methylbenzoate, the contained strain P. putida CMC4 became unrecoverable from the rhizosphere and bulk soils 19 and 11 days, respectively, after the appearance of the first true leaves of corn plants (Fig. 5 and Table 3). In contrast, the control strain was recovered at a level of about 10³ CFU/g of rhizosphere soil for about 28 days (Fig. 5); however, it was recovered from bulk soil only during the first 11 days after seed germination (Table 3). In soils containing 3-methylbenzoate the control strain was detected at levels of 10³ to 10⁶ CFU/g of rhizosphere soil during the first 70 days, whereas the contained strain was recovered at a level higher than 100 CFU/g of soil only during the first 40 days of the assay (Fig. 5). The contained and control strains became established poorly (if at all) in the bulk soil, and their levels never were greater than 10³ to 10⁴ CFU per g of soil (Table 3). The control strain could be recovered from bulk soil up to 28 days after sowing but not thereafter (Table 3), whereas the contained strain was recovered from bulk soil only up to day 11. 

View larger version (18K): [in this window] [in a new window]   FIG. 5.   Survival of P. putida CMC4 and EEZ32 under outdoor conditions in pots. The conditions were the same as those described in the legend to Fig. 4 except that corn seeds were used.

Lack of dispersal of GEMs during the field releases. During the outdoor trials reported in this study, five locations in the experimental field into which no GEMs had been introduced were used to monitor undesired dispersion of GEMs. Samples were taken every 14 days. Control or contained bacteria were never detected outside the pots. At the conclusion of the assay soil from the site was checked every fortnight for 6 months for the presence of the control or contained strain. Neither strain was detected in these tests.

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

Xenobiotic compounds contain structures or substituents rarely found in natural products. These compounds are usually not metabolized by microbes, and as a consequence they accumulate in the biosphere and contribute to the burden of pollution (17). Many microbes are able to evolve new catabolic activities against some of these recalcitrant compounds (3, 20, 28, 29), but most xenobiotic compounds remain in the biosphere unattacked (2). New catabolic pathways have been constructed for some of these pollutants, and in recent years it has become possible with molecular biology techniques to engineer bacteria that are able to degrade toxic compounds (17, 19). In the future recombinant microbes bearing genetically engineered catabolic pathways will probably be used in open environments, particularly in sites polluted with toxic chemicals (15). These GEMs should survive and perform for as long as the pollutant is present, but it is desirable that they should commit suicide once the compound is consumed (5, 26) to reduce concerns regarding unanticipated consequences. This behavior can be achieved by providing recombinant microbes with ABC systems.

In all genetic systems mutations appear at a certain frequency, and in the case of containment systems mutations that lead to killing escape have been reported (5, 12, 13). To decrease the rate of killing escape when cell death was induced, the control and killing elements of the ABC system were incorporated into the host chromosome of P. putida CMC4, where they were stably maintained even in the absence of selective pressure. When 3-methylbenzoate was exhausted, the killing genes were induced and the cells died (Fig. 3). Interestingly, the rate of killing escape under laboratory conditions was around 10⁸ mutants per cell and per generation.

The effectively contained strain P. putida CMC4 and the control strain P. putida EEZ32 were released under controlled conditions. Survival of P. putida EEZ32 and the contained strain P. putida CMC4 was better in the rhizospheres of plants than in bulk soil and unplanted soils. In the rhizospheres of corn or broad bean plants the control strain became established at similar levels regardless of the presence of the target compound. In contrast, the contained strain tended to die, although survival in the presence of the aromatic compound was better than survival in its absence (Fig. 4 and 5). It should nonetheless be noted that even in the presence of 3-methylbenzoate the contained strain survived poorly compared to the control. This may have been due to the reduced availability of the 3-methylbenzoate added to the soil. The ability to colonize the soil rhizosphere is a characteristic of P. putida 2440, which can become established around the root systems of many herbaceous plants (21, 24). Therefore, the marked decrease in the number of CFU of the contained strain per gram of rhizosphere soil can be attributed unequivocally to the containment system that this strain bears. However, under outdoor conditions a decrease in the number of CFU of the contained bacteria per gram of soil required long periods of time (weeks), whereas in the laboratory the counts decreased within hours (25). This difference may reflect the following two facts: (i) when the contained cells were introduced into the soil, they were full of LacI protein, and the killing gene was not expressed until degradation of the LacI protein occurred; and (ii) functioning of the Gef protein requires cells in an active metabolic state (25), as reported previously for cell lysis mediated by other porins (8). Once added to the soil, the cells were probably metabolically less active.

The contained strain colonized bulk soil less efficiently than the uncontained control strain. This reflects the situation described above for unplanted soils, and it follows that dispersal of the contained strain may be severely limited in a number of environments.

To monitor the possible effects of the release of P. putida CMC4 and EEZ32 on the natural bacterial populations, we selected p-hydroxyphenylacetic acid utilizers as the target population to monitor because the numbers of these organisms in the soil remained stable at around 10⁵ to 10⁶ CFU per g of soil throughout the year (26). The presence of the control strain or the contained strain had no significant effect on survival of this natural population.

Our results show that an ABC system based on killing genes can be developed in the laboratory and that such a system can work under outdoor conditions.