.EXAMINATION OF THE ENVIRONMENTAL IMPACT OF A GENETICALLY-MODIFIED PLANT GROWTH PROMOTING RHIZOBACTERIUM
Walter F. Mahaffeea, Ellen M. Bauskea, Joseph J. Shawb, and Joseph W. Kloeppera
aDepartment of Plant Pathology, and bDepartment of Botany and Microbiology, Biological Control Institute, Alabama Agriculture Experiment Station, Auburn University
SUMMARY
It has been hypothesized that the introduction of genetically-modified organisms poses an environmental risk that differs from the introduction of unaltered parental strains. A two-year field study was established to examine two measures of environmental impact; i. whether the survival and spatio-temperal popualtion dynamics of a genetically engineered plant growth-promoting rhizobacterium (PGPR) differed from those of the wild-type parent, and ii. whether the indigenous soil and root bacterial communities were affected differently by the bacterial inoculants. Treatments included a nontreated control, the wild-type PGPR strain (89B-27), and a bioluminescent derivative (Lux+; 89B-27::Tn4431). Plant and soil samples were taken 0, 7, 14, 21, 35, and 70 days after planting and processed to examine populations of the introduced strain and bacterial communities from both the rhizosphere and internal root tissues. Root populations of both 89B-27 and the Lux+ derivative declined during the season and were not detectable at harvest. In addition, they were not detected in any root-free soil sample taken during the growing season or in subsequent sampling of soil during the winter. In general, the populations of the Lux+ derivative were always lower than 89B-27 populations. Over 10,000 bacterial strains were isolated from soil, rhizosphere and internal root tissue samples. These strains have been identified using the Sherlock System (Microbial ID, Inc.) for fatty acid methyl ester analysis. Preliminary analysis of the bacterial community structure indicates that the rhizosphere community of each bacterial treatment differed from the noninoculated control, but did not differ from each other. There was no effect on the soil community by the bacterial treatments. The data indicate that the introduction of a genetically modified derivative of 89B-27 does not pose a greater environmental risk than its unaltered wild-type with respect to the measured parameters.
INTRODUCTION
With more and more bacterial inoculants for agriculture becoming commercially available and the probability that registration may be pursued for genetically engineered derivatives of these strains, the need exists for a knowledge base of how these microorganisms interact with and affect the indigenous biota and survive in the environment. In most cases, technologies and methodologies must be developed or adapted for this purpose. This information could then be used to assess if the introduction of an organism to the environment poses a potential environmental problem by examining the survival of the introduced organism, and assessing alterations in community structure and/or function of the indigenous biota associated with the introduction. The bacterial communities associated with soil and plant roots are extremely important in the development of plants and would most likely be the most affected by the introduction of organisms to plant roots, and thus would serve as models for developing these techniques and methods.
Assessment of the risks of using genetically engineered bacteria requires assessment of multiple ecological parameters, (e.g., colonization patterns, survival, spread from target area, etc.) under field conditions. The hypothesis which must be tested in each case is that the impact of the genetically engineered strain is similar to the wild-type non-engineered strain. While this is straightforward in concept, there is a major limitation with currently available techniques for bacteria. Monitoring introduced bacteria in the natural environment requires the ability to differentiate the specific introduced strain from a background of genetically similar and dissimilar indigenous bacteria (Kloepper and Beauchamp, 1992). Traditionally, this has been accomplished through the use of "marked" strains. One standard approach is to use spontaneous resistance to antibiotics, such as rifampicin or nalidixic acid, to which most environmental bacteria are sensitive. These marked strains are not truly wild-type strains, and yet they have been reported as the wild-type for comparison in many previous reports on the ecology of plant-associated bacteria genetically engineered with lux and lacZY genes (Cook et al., 1990; Beauchamp et al., 1992; Kluepfel et al., 1990, Shaw et al., 1992). Antibiotic marked strains frequently have pleiotropic mutations (Kloepper and Beauchamp, 1992; Kluepfel, 1993; Press et al., 1992) and may not express the resistance phenotype upon growth on selective media (McInroy et al., 1992). In one recent study (Beauchamp et al., 1992), lux genes were inserted via transposon mutagenesis into a Pseudomonas putida PGPR strain. Root colonization by the genetically engineered derivative was compared to that of its spontaneous rifampicin-resistant parent following seed treatments with each strain singly and with a 1:1 mixture of both strains. While colonization patterns for single-strain inoculations were similar for the genetically engineered and rifampicin-resistant mutant, the genetically engineered strain was displaced over time by the rifampicin-resistant mutant following inoculation of mixtures. It was concluded that the competitive fitness of the genetically engineered strain was reduced compared to nonengineered. However, such assessments of risks associated with environmental releases of genetically engineered microorganisms could reach false conclusions regarding ecological fitness and survival when an antibiotic-marked nonengineered strain is used for the "wild-type" control.
A possible solution to this dilemma is the use of immunological techniques, such as Immunofluorescence Colony Staining (IFC). IFC is based on combining isolation of viable bacteria with serological differentiation of target colonies (van Vuurde, 1987). The technique uses IgG antibodies, specific for a bacterial strain, conjugated with fluorescein isothiocyanate (FITC) to distinguish target colonies from nontarget colonies. Since the use of IFC does not require any alteration in the phenotype or genotype of the wild-type, this technique allows for the comparison of an unaltered wild-type to a genetically modified derivative of the wild-type. Thus direct assessment of effects on fitness of a genetically engineered bacterium due to the addition or genetic material is possible. Since bacteria identified by IFC remain viable, positive colonies may be subcultured for other uses such as identification by other methods (e.g., GC-FAME or biochemical reactions). IFC has been used by several investigators (van Vuurde and Roozen, 1990; Glandroff et al., 1992; Leeman et al., 1991; Underberg, 1992; Raaijmakers et al., 1994) to track and quantify bacterial strains introduced into soil or onto plants and to detect bacterial pathogens in soil and plant tissues with detection limits reported to be as low as 20 CFU/g soil (Leeman et al., 1991).
Another component to the assessment of the environmental impact associated with the introduction of a genecially engineered microorganism into an ecological habitat is the effect this introduction has on the indigenous fauna and flora. One approach to examining this impact is the assessment of community structure using various diversity indices and multivariate analyses. Gilbert et al. (1983) using cananical discriminant analysis, demonstrated that the physiological attributes of bacterial components of rhizosphere soil from plants grown from seeds coated with a bacterial strain differed from nontreated controls. Diversity indices were used by Workneh and van Bruggen (1994) to show that rhizosphere communities of actinomycetes on tomato roots were more diverse in organically than conventionally managed rhizosphere soil. Seidler (1992) noted that changes in microbial populations are valuable indicators for determining the environmental impact of introduced genetically modified organisms. Bolton et al. (1991) used this approach to monitor the effect of introducing a rifampicin-resistant Pseudomonas species in soil-core microcosms on wheat rhizoplane populations of heterotrophic bacteria. Species richness and evenness were used by Banerjee and Anderson (1992) to detect differences in rhizoplane microflora, which resulted from fumigation and organic nutrient addition. Differences in diversity of fungal communities lasted for at least 17 months, including two growing seasons.
A two-year field study has been established to test the hypothesis that the genetically modified bacterium does not differ in behavior from its unaltered parent, by examining two measures of environmental impact: i. whether the survival and spatio-temperal population dynamics of a genetically engineered plant growth-promoting rhizobacterium (PGPR) differs form those of the wild-type parent, and ii. whether the indigenous soil and root bacterial communities are affected differently by the bacterial inoculants. The results reported here are a preliminary analysis of the first year of field data.
MATERIAL AND METHODS
Basic Field Experiment. Each of the objectives was addressed by using different experimental approaches to elements of a common field experiment. A randomized complete block design, consisting of 3 treatments and 6 blocks with one replication of each treatment per block, was established in a sandy-loam field soil at the E.V. Smith substation of the Alabama Agricultural Experiment Station. Each experimental unit consisted of 2 rows with 10 cucumber plants per row spaced 90 cm apart in raised beds under plastic mulch with drip irrigation. The field was maintained according to recommendations for commercial cucumber production by the Alabama Cooperative Extension Service. Treatments were a nontreated control, wild-type strain 89B-27, and a lux transformant of 89B-27 (Lux+; 89B-27::Tn4431).
Seed were treated with a bacterial suspension in 2% methyl cellulose (72 µl/g seed) which resulted in inoculum density of approximately 1 × 106 cfu/seed. Suspensions were prepared by growing cultures, taken from -80C storage, on 5% tryptic soy broth agar (TSA) at 28C for 48 hrs. Bacteria were scraped from the surface of the agar after adding 10 ml of 0.1 M phosphate buffer, pH 7.0 (PB). This suspension was centrifuged (8000 × g, 10 min.), and the pellet was resuspended in 2% methyl cellulose, yielding a concentration of approximately 1 × 108 CFU/ml.
Samples were taken from field trials in the following manner: a 15 cm diameter soil core, centered on the plant stem, to a depth of 25 cm was removed from each experimental plot 0, 7, 14, 21, 35, and 70 days after planting. The root system was recovered and gently shaken to remove any loosely adhering nonrhizosphere soil and placed in sterile plastic bags. The remaining soil was mixed and a 500 g sample was removed for further analysis. Additional samples were taken during the winter to examine if the bacteria survived.
Habitat-Colonization Profile and Survival Studies. Seed, and root samples were weighed and processed as described below. Samples were sonicated in 10 ml PB to remove rhizosphere and rhizoplane or spermosphere and spermoplane microorganisms and then surface-disinfested with 1.05% NaClO with Tween 20 and rinsed three times in sterile PB (McInroy et al., 1995) before trituration in a Kleco Tissue Pulverizer (Kinetic Laboratory Equipment Company, Visalia, CA) with 1 ml PB to sample for bacterial endophytes. Sterility controls were performed on each sample by plating 1.0 ml of the final rinse buffer onto 5% TSA and incubating at 28C for 72 hrs. Any samples showing growth after 72 hrs incubation were removed from data analysis. The sonication washate and triturated samples were spiral-plated onto 5% TSA, 50% TSA and 50% TSA amended with 15 µg/ml tetracycline (5% TSAtc ). After 48 hrs incubation at 28C, all plates were counted by hand. In addition, bioluminescent colonies were counted by hand, in the dark, from the 50% TSA and 5% TSAtc plates. Bioluminescent and nonbioluminescent colonies from 50% TSA and 5% TSAtc plates were confirmed, based on light production, by placing a small portion of the colony (approx. 1 µg) in a monolight 2000 luminometer (Analytical Luminescence Laboratories, La Jolla, CA) (Shaw et al., 1987).
The sonication washate and triturated sample were processed for IFC as described below. Aliquats of 100 µl of the rhizosphere/rhizoplane or endophyte samples were placed in a well of a 24-well tissue culture plate and serially diluted using a FinnPipet Diluter (Labsystems Oy, Helsinki, Finland). Next, 300 µl of 5% TSA or 50% TSA (48C) was added to the sample while shaking at 250 rpm. Plates were allowed to harden before being inverted and incubated for 24 hrs at 28C. After overnight incubation, plates were air-dried to a thin film with warm blowing air (45C) for 4-6 hrs. The dried agar films were then stained with 300 µl of diluted FITC conjugated polyclonal antibodies specific for 89B-27 and incubated for 36 hrs at 4C. After staining, films were rinsed twice for 1 hr in 25% ringers with 0.01% tween 80. Stained preparations were viewed and counted microscopically with a 4x objective and 10x oculars under incident blue light at 490 nm. A random sample of IF-positive colonies was taken, purified and stored at -80C prior to GC-FAME analysis. GC-FAME analysis was used to check the specificity of the antiserum for 89B-27 by comparing samples to a library entry for 89B-27.
Microbial populations of the soil were also examined by shaking (350 rpm) 10 g soil, from the core samples, with 90 ml PB for 30 min. Samples were spiral-plated and processed as described above. 100 µl of each sample was also processed for IFC.
All population data were converted to log CFU/g fresh weight tissue or soil and analyzed by analysis of variance using PC-SAS (SAS Institute, Cary, NC). Significance differences were determined using single degree of freedom contrasts.
Community Structure. Random samples of 35, 25, and 35 bacterial colonies, as determined by rarefraction analysis (Hulbert 1971), were taken from the 5% TSA plates of the external and internal roots and soil samples, respectively. Each isolate was purified and stored at -80C until extraction. Fatty acid methyl esters were extracted from each isolate using standard and recommended procedures for GC-FAME analysis (Sasser, 1990). After extraction, samples were stored at -20C until analysis with a Hewlett-Packard gas chromatograph and the Sherlock Microbial Identification System software (MIDI, Newark, DE). After the isolates were identified, genera richness, evenness, and diversity were used to compare microbial communities of all three habitats among treatments over time. Genera richness was determined by direct counts of the numbers of genera identified in each sample. The total number of genera as well as number of isolates of each species were used to determined species evenness and diversity. The modified Hill's ratio (Ludwig and Reynolds, 1988) was calculated to determine evenness and Hill's N2 diversity, 1/ where = [ni (ni - 1)] / [n (n - 1)] (Hill, 1973) was used to calculate diversity.
RESULTS
Habitat-Colonization Profile and Survival Studies. Total bacterial rhizosphere-populations from 5% TSA plates ranged from 6.62 to 7.06 CFU/g fresh weight root and did not significantly differ among treatments at any sampling time; However, total seed populations varied significantly. The noninoculated control was 2 log units lower than the other treatments.
Populations as measured by IFC did not vary significantly between the bacterial treatments at any sampling time, while the populations declined from log 5.5 CFU/seed at planting to 0.80 CFU/g fresh root weight 70 days after planting (plant senescence). However, populations of Lux+ measured by number of bioluminescent colonies on 50% TSA, were significantly lower, generally more than 1 log unit, than populations of wild-type or populations of Lux+ measured by IFC. No IFC positive colonies (i.e., cross-reacting) were detected in any of the samples from the noninoculated controls.
Very few biolumminescent colonies were detected visually on 50% TSA plates and none were detected on 5% TSA plates. However, when colonies from 50% and 5% TSA plates with colony morphologies similar to the bioluminescent colonies were tested in luminometer, light emission was detected.
Community Structure. Genera richness, evenness and diversity indices of bacterial treatments were significantly different from the noninoculated control. However, these indices did not differ between the bacterial treatments. Bacterial treatments were more diverse and had more genera present than the noninoculated control. In addition, the genera in the noninoculated control were less evenly distributed than in the bacterial treatments as indicated by Hill's evenness index.
It was observed that different genera appeared to be more dominant (i.e., greater number of isolates) in the bacterial treatments than in the noninoculated control. However, the indices used in this preliminary analysis are not mathematically designed to detect this type of difference. Furthur analyses are being conducted using the index of biotic similarity (Pinkham and Pearson, 1976) and BIOSIM1 (Pearson and Pinkham, 1992) which will be able to test this observation.
DISCUSSION
Bacterial populations of Lux+, as determined by IFC, were greater than populations of Lux+ determined by counting biolouminescent colonies on 50% TSA and 50%TSAtc, indicating that this phenotypic marker may under-estimate the survival and colonization of the marked strain. One reason for this under estimation may be that the use of antibiotic amended media resulted in reduced culturability of the marked strain. Another could be that light production was reduced because of competition with other bacterial colonies on the plate. This is evident in that more bioluminescent bacteria were detected on 50% TSAtc, than on 50% TSA. There was a 2-3 log unit reduction in the number of total bacteria growing on 50% TSAtc verses 50% TSA. This under estimation could lead to false conclusions about the survival and possible environmental impact of an introduced bacterium. Regulatory polices based on such conclusions would be questionable. The population data, as measured by IFC, also indicate that there is not a loss of ecological fitness when a bacterial strain is genetically modified.
Since light production by luciferase is energy dependent, requires FNMH2 (Shaw et al., 1987), the lack of bioluminescence from Lux marked bacteria on 5% TSA and 50% TSAtc could be due to competition from other bacteria for nutrients. Competition could have reduced the availalbe nutrients and caused a reduction in the metabolic activity of the cells, which would reduce the available FNMH2 and thus light production.
Preliminary analysis of the bacterial community structure indicates that the rhizosphere community of each bacterial treatment differed from the noninoculated control. However, the community of wild-type and Lux+ treated plants did not differ from each other. These data indicate that the introduction of the luciferase operon and a gene for tetracycline resistance into a rhizobacterium does cause the genetically modidied bacterium to alter the community structure of the rhizosphere differently from the wild-type. This does not mean, however, that the introduction of less innocuous genetic material (e.g. genes for antibiotic or siderophore production) may not cause changes in the microbial structure. The data also demonstrates that the introduction of a rhizobacterial strain can cause a shift in the rhizosphere community structure, indicating that diversity indicies may be useful for assessing the environmental impact of rhizobacteria.
ACKNOWLEDGMENTS
We would like to thank the individuals that provided technical help with processing the tremendous number of bacterial samples that had to isolated, purified, frozen, extracted and analyzed. Aiyun Li, Brenda Conner, Patryce Curtis, Johannes Hallmann, Xho Hung, Natalia Martinez, John McInroy, Caroline Press, Andrea Quadt-Hallmann.
REFERENCES
Banerjee P and Anderson RC (1992) Long-term effects of soil fumigation and inorganic nutrient addition on the rhizoplane mycoflora of little bluestem (Schizachyrium scoparium). Mycologia 84:843-848.
Beauchamp CJ, Kloepper JW, Shaw JJ, and Lemke PA (1992) Bioluminescence as a marking system to monitor the fate of Pseudomonas putida strain 619A.3L, released in the soybean rhizosphere. (Abstr). Phytopathology 82:1179.
Bolton H Jr., Frederickson JK, Thomas JM, Li SW, Workman DW, Bentjen SA, and Smith JL (1991) Field calibration of soil-core microcosms: Ecosystem structural and functionalcomparisons. Microb Ecol 21:175-189.
Cook JR, Weller DM, Kovacevich P, Drahos D, Hemming B, Garnes G, and Pierson E (1990) Establishment, monitoring, and termination of field tests with genetically altered bactera applied to wheat for biological control of take-all. In: DR MacKenzie and SC Henry (eds) The Biosafety Results of Field Tests of Genetically Modified Plants and Microorganisms. USDA/ARS, Bethesda, MD. pp. 177-188.
Gilbert GS, Parke JL, Clayton MK, and Handelsman J (1993) Effects of an introduced bacterium on bacterial communities on roots. Ecology 74:840-854.
Glandorf DCM, Brand I and Bakker PAHM (1992) Stability of rifampicin resistance as a marker for root colonization studies of Pseudomonas putida in the field. Plant and Soil 147:135-142.
Hill MO (1973) Diversity and Evenness: A unifying notation and its consequences. Ecology 54:427-432.
Hulbert SH (1971) The non-concept of species diversity: A critique and alternative parameters. Ecology 52:577-586.
Kloepper JW and Beauchamp CJ (1992) A review of issues related to measuring colonization of plant roots by bacteria. Can J Microbiol 38:1212-1232.
Kluepfel DA (1993) The behavior and tracking of bacteria in the rhizosphere. Annu Rev Phytopathol 31:441-472.
Kluepfel DA, Kline EL, Hughes T, Skipper H, and Gooden D (1990) Field testing of a genetically engineered rhizosphere inhabiting pseudomonad: development of a model system. In: DR MacKenzie and SC Henry (eds) Biological monitoring of genetically engineered plants and microbes.. Agricultural Research Institute, Bethesda, Maryland. pp. 189-199.
Leeman M, Raaijmakers JM, Bakker PAHM and Schippers B (1991) Immunofluorescence colony staining for monitoring pseudomonads introduced into soil. In: ARB Beemster, GJ Bollen, M Gerlagh, MA Ruissen, B Schippers and A Tempel (eds) Biotic interactions and soil-borne diseases. Edited by . Elsevier, New York. pp. 374-380.
Ludwig JA and Reynolds JF (1988) Statistical Ecology. John Wiley and Sons, Inc, New York. 337 pp.
McInroy JA and Kloepper JW (1995) Survey of indigenous bacterial endophytes from cotton and sweet corn. Plant and Soil (In Press).
McInroy JA, Wei G, Musson G, and Kloepper JW (1992) Evidence for possible masking of rifampicin-resistance phenotype of marked bacteria in planta. (Abstr). Phytopathology 82:1117.
Pearson JG and Pinkham CFA (1992) Strategy for data analysis in environmental surveys emphasizing the index of Biotic similarity and BIOSIM1. Water Envrion Res 64:901-909.
Pinkham CFA and Pearson JG (1976) Application of a new coeffient fo similarity to pollution survevys. J. Water Pollut Control Fed 48:717-723.
Press CM, Kloepper JW, and McInroy JA (1992) Pleiotropic mutations associated with spontaneous antibiotic-resistant mutants of rhizobacteria. (Abstr). Phytopathology 82:1178.
Raaijmakers JM, Wilbert B. and Punte HLM (1994) Siderophore receptor PupA as a marker to monitor wild-type Pseudomonas putida WCS358 in natural environments. Appl Envrion Microbiol 60:1184-1190.
Sasser M (1990) Identification of bacteria through fatty acid analysis. In: Z. Klement, K. Rudolph, and D.C. Sands, eds. Methods in Phytobacteriology. Akadèmiai Kiadó, Budapest. pp. 199-204.
Seidler RJ (1992) Evaluation of methods for detecting ecological aspects from genetically engineered microorganisms and microbial pest control agents in terrestrial systems. Biotech Adv 10:149-178.
Shaw JJ, Dane F, Geiger D, and Kloepper JW (1992) Use of bioluminescence for detection of genetically engineered microorganisms released into the environment. Appl Environ Microbiol.
Shaw JJ, Rogowsky P, Close TJ and Kado CI (1987) Working with bacterial bioluminescence. Plant Molec Biol Rep 5:225-236.
Underberg HR (1992) Ecological model studies on the colonization patterns and population dynamics of Erwinia chrysanthemi Burkholder, McFadden and Dimock 1953 in the rhizosphere, Dissertation, Universität Tübingen.
van Vuurde JWL (1987) New approach in detecting phytopathogenic bacteria by combined immunoisolation and immunoidentification assays. EPPO Bulletin 17:139-148.
van Vuurde JWL, and Roozen NJM (1990) Comparison of immunofluorescence colony-staining in media, selective isolation on pectate mediu, ELISA, and immunofluorescence cell staining for detection of Erwinia carotovora subsp. atroseptica and E. chrysanthemi in cattle manure slurry. Neth J Plant Path 96:75-89.