Prior to the release of a recombinant bacterium into the terrestrial environment, the safety and efficacy of the organism must first be assessed. Much of this information is obtained with the use of microcosms. The design of the microcosm significantly affects the ability of the recombinant microorganism to survive in soil and, thus, complicates the risk assessment process. To standardize microcosms for future use, we evaluated the survival of Pseudomonas aureofaciens 3732 RN-L11 (lac ZY, Rif^r, Nal^r) in intact soil cores (5.0 x 15 cm PVC core) and disturbed soil microcosms (50 g fresh, sieved soil). Survival was compared to data obtained during a field release. The intact soil core microcosm was shown to closely simulate results obtained in the field. The intact soil core microcosm closely predicts survival in bulk soil and in the rhizosphere of wheat. Data obtained with the microcosm was also similar when evaluated in separate studies in two different years. In 1993, P. aureofaciens survived for approximately 63 days in bulk soil and 96 days in the rhizosphere. The disturbed soil microcosm exhibited a much more rapid decline (34 days zero) in population compared to the intact core microcosm. These results demonstrate that a small, inexpensive and simple intact soil core microcosm may be appropriate for risk assessment purposes.

Key Words: Microcosms, risk assessment, survival, GEMs

Regulators are required to assess the risks associated witht he release of recombinant organisms into the environment. Microcosms are often used to gather the requisite risk assessment data. Microcosms, although uncoupled from nature (Heath, 1979; Schindler et al., 1980), represent a complex system containing the network from which it was obtained. As early as 1976, the EPA recognized the need for "effective, reliable, rapid and relatively inexpensive tools and techniques to screen and identify those chemicals having potentially harmful health and ecological effects" (Sanders, 1976).

Many different designs of microcosms are available. Designs range from as simple as a few grams of soil in a bottle (Jaffee et al., 1992) to a large chamber where many of the environmental variables are controlled (Gile et al., 1982; Gillett and Gile, 1976). Many variations in design were summarized in an EPA Microcosms Workshop (Cripe et al., 1992). Fourteen different microcosms designs were identified in what the authors concede is an incomplete list. Unfortunately, the design of the microcosm may significantly affect the results obtained.

Because the design of the microcosm affects the survival of a GEM in soil and so many microcosm designs are currently utilized, it is nearly impossible to compare data from one study to another. This has greatly complicated the decisions made by risk assessors. To begin the process of developing a standard microcosm design, the following study was initiated. Of the many microcosm designs in use, two were selected that were low cost, simple to use and most effectively adhered to basic principals of microbial ecology.

Microorganisms. Pseudomonas aureofaciens 3732 RN-L11 (lac ZY) was used for all studies. The construction of this microorganism, isolated from the rhizosphere, was described by Barry (1986; 1988). The lac ZY genes, located on the chromosome, are a highly stable marker for tracking bacteria in soil (Angle et al., 1994), and have been reviewed in detail by Drahos et al. (1992). Pseudomonas aureofaciens strain 3732 RN-L11 is also resistant to rifamycin SV (100 g/ml) and nalidixic acid (100 g/ml). The microorganism was maintained in Pseudomonas F broth (Difco, Detroit, MI) with 50% glycerol at -55⁰C.

Inoculum for addition to soil was prepared by adding a single loopful of P. aureofaciens RN-L11 to 10 ml Pseudomonas F broth containing rifamycin SV and nalidixic acid. The culture was grown to an approximate density of 10⁹ colony forming units (CFU) ml^-1. A small amount of broth was then transferred to 1.5 L broth and shaken for approximately 20 hrs, resulting in a density of 10⁹ CFU ml^-1. Prior to inoculation into soil, the culture was centrifuged at 5000xg for 15 min to pellet cells and to remove antibiotics. The pellet was then resuspended in distilled water to the desired concentration.

Soil. Soil used for all studies was a sandy loam collected from the University of Maryland Central Maryland Research and Education Center located in Calverton, Maryland. All soil was obtained from the top 15 cm of the soil profile. Chemical, physical and biological properties of the soil are presented in Table 1. Notably, the soil contained no background colonies for organisms that were resistant to the two antibiotics, grew on Pseudomonas F Agar, and expressed lac ZY.

Microcosms. PVC cores (15 cm long and 5 cm ID) were driven into soil and gently removed by hand. Soil filled all but the top 2 cm of the core. These cores are subsequently referred to as intact soil core microcosms. A polyester mesh membrane was placed over the bottom of the core to hold in soil yet allow for exchange of gasses and movement of solutes. Cores were immediately placed into a cooler and returned to the laboratory where they were placed into a refrigerator at 4⁰C. Microcosms were inoculated within 24 hrs after collection. One half of the intact soil core microcosms were seeded with three seeds of winter wheat (Triticum aestivum). Following germination, seedlings were thinned to one plant per core.

Disturbed soil microcosms consisted of 50 g soil placed into 160 ml milk dilution bottles. Immediately after the soil was collected from the field, and before it was placed into bottles, it was air dried and gently passed through a 0.5 cm sieve. Disturbed soil microcosms were not planted with wheat. Microcosm studies were repeated in both 1992 and 1993.

Following inoculation of microcosms, the microcosms were placed into a growth chamber and maintained at 28⁰C day and night temperatures and 70% relative humidity. The day photoperiod was 12 hrs with >400 E m^-2 sec^-1 photosynthetically active radiation from a combination of cool-white fluorescent and incandescent lamps.

Field plots. EPA approval was required and granted for release of P. aureofaciens 3732 RN-L11 into the field. One square meter plots were established six months prior to inoculation by killing all vegetation on the area. No pesticides were subsequently applied and occasional weeds were removed by hand. A 15 cm high plastic berm was buried around each of the plots to control runoff. Plot design was completely randomized; each plot was then split into two. One half of each plot was seeded to wheat at 10 cm row spacings.

In addition, uninoculated microcosms and plots were examined, at each sampling time, to ensure that background microorganisms were not present. At all times, no indigenous microorganisms were observed that were resistant to the antibiotics and able to express lacZY.

Inoculation. Microcosms without wheat were inoculated with P. aureofaciens 3732 RN-L11 by dribbling the broth culture evenly onto the soil surface. Soils were inoculated to a concentration of 1 x 10⁷ CFU per g soil. Field plots without wheat were sprayed with the inoculum to achieve a concentration of 1 x 10⁷ CFU per cm^-2 soil surface. Wheat plants in the field and microcosms were inoculated by pipetting 1 ml of the inoculum directly onto the crown area of the plant.

Enumeration. All soil within each microcosm was removed, screened to remove rocks and other debris and thoroughly mixed. Ten grams of soil was then added to 95 ml dilution buffer (1% CaCl₂) and shaken for 10 min. on a rotary shaker (Wollum, 1982). Soil was allowed to settle from solution for a period of 10 min. At this time, appropriate dilutions were spread on Pseudomonas F agar containing nalidixic acid (100 g/g), rifamycin SV (100 g/g), cycloheximide (100 g/g) and X-gal (80 g/ml). X-Gal is an analog of lactose and colonies expressing lac Zy appear blue. Plates were incubated at 28⁰C for two days at which time colonies were counted. Numbers were expressed as CFU g^-1 fresh soil.

Where a root system was present in the microcosms, the rhizosphere was analyzed. Roots were gently separated from the bulk soil and placed into 99 ml 1% CaCl₂. The mixture was shaken for two min. and the soil allowed to settle for 10 min. from solution. The solution was then assayed as above. Roots were removed from the dilution bottle, dried at 80⁰C for 12 hrs and weighed. Numbers were expressed as CFU per plant. This method of expression has previously been shown by Angle et al. to be the most appropriate for describing risk assessment data (unpublished data).

Soil was removed from the top 15 cm of the field with a 2.5 cm wide soil core. Five cores per plot were randomly collected and composited into a single sample. The sample was immediately returned to the laboratory and sieved, mixed and enumerated as before. Where wheat was present, the root system was removed with a hand trowel, placed in a plastic bag, returned to the laboratory and enumerated as above.

Statistics. Laboratory and field experiments were conducted with four replications in a completely randomized design. Analysis of variance was conducted on all data for each day using log transformed CFUs to determine significant differences among treatments. Analysis of variance was also conducted for data combined over experiments and days to test for the significance of the linear effect of days and whether the linear effect of days was significantly different in the field and lab microcosms. This analysis was used to compare the decline in CFUs between all treatments. Linear regression analysis was used to estimate the rate of decrease in the log CFUs over time. Linear regression equations were used to compare treatments and to predict the time for CFUs to decline to zero.

Since the number of cells inoculated into soil may vary from experiment to experiment, it is difficult to compare treatment means at each point in time. An alternative procedure and one that was highly effective was to statistically compare the slopes of the predicted lines (Table 2). This also allowed us to predict the time in which the population approached zero.

The population of P. aureofaciens in the 1993 disturbed soil microcosm declined at a significantly greater rate compared to the decline observed in the field or intact core microcosm (Figure 1). In the disturbed soil microcosm, the population was no longer detectable in soil after 34 days while the population in the intact core was detectable up to 63 days.

The slope of the population decline in the intact soil core microcosm was nearly identical to the slope observed in the field. There was an approximately one half log difference in the population at every point in time, including the predicted days to zero. This difference was due to the difference in the amount of inoculum added at time zero. Although we were careful to add similar amounts of inoculum to both the microcosms and field plots, the inoculation methods made this difficult. None-the-less, slopes were not significantly different, indicating that the intact core was an ideal tool for predicting survival in the field.

The population of P. aureofaciens in intact soil cores was compared between 1992 and 1993 (Fig. 2). Slopes of decline were not significantly different from one another and the days to zero were very similar.

Similar studies were repeated in core microcosms and the field by examing the population in the rhizosphere. The specific population within the rhizosphere was not measured until 7 days after the original inoculation. Before this time, the root system was too small to be accurately sampled. At all sampling times throughout the duration of the study, populations in the rhizosphere of wheat in the intact core microcosm and the field were statistically identical (Fig. 3). Further, the slopes over time were not statistically different (Table 2).

Pseudomonas aureofaciens was originally isolated from the rhizosphere and is therefore well adapted for survival in this region. The GEM survived approximately twice as long in the rhizosphere compared to the bulk soil (Figs. 1 and 3). In 1993, P. aureofaciens could not be isolated after approximately 63 days in the intact core microcosm and after 96 days in the rhizosphere. To some extent, this difference was related to the continual development and colonization of new roots. Since the population in the rhizosphere is expressed as CFU per plant, as opposed to CFU per g root as is the case with many other studies, we are measuring the total population within the soil as opposed to the concentration within this area.

Results obtained in core microcosms were also compared between 1992 and 1993 (Fig. 4). As noted, when expressed as CFUs per plant, P. aureofaciens survived for a significantly greater period of time in 1992 compared to 1993. The slopes of decline for each year were significantly different from one another. The lack of similarity can be partially explained by the poor growth of the wheat in 1993. In order to ensure uniformity between years, the same lot of seed was used, and between years, the seed was stored at 4^oC. Following storage, the germination of the seed and early growth in 1993 was significantly poorer than in 1992. Since results are expressed as CFU per plant, the size of the root system is an important variable controlling the population in the rhizosphere. When results were normalized to remove root size as a variable and the population was expressed as CFU per g root [as is typically used in many studies (DeFreitas and Germida, 1992; Dileep-Kumar and Dube, 1992; Kluepful and Tonkyn, 1992)], we found that the rhizosphere populations were more similar for each year (data not shown).

Drying of soil prior to addition to microcosms causes many chemical properties of the soil to be altered (Wolf et al., 1989). Specifically, drying releases a number of inorganic ions such as manganese that may be toxic to the bacterium. Several studies have recognized the inherent problems associated with the use of dried and sieved soil and have only utilized soil that has been maintained moist following collection and was passed through a relatively coarse sieve to remove only large debris (Iwasaki et al., 1994; Parmelee et al., 1993; Rocorbet et al., 1992).

The advantage to using dried and sieved soil is that much of the spatial variability associated with intact soil is removed. Coefficients of variation associated with the disturbed soil microcosm were lower than that associated with the field or core microcosms (data not presented). For predictive purposes, however, the variation with the intact core microcosm was not excessive and the small increase in variation was outweighed by other advantages with the intact core microcosm. Microcosms using small quantities of dried and sieved soil may also be suitable for examining relative differences between organisms or soils. In such studies, similarity to survival in the field is not of paramount importance.

Other studies have used extremely complicated microcosms where many of the environmental variables are controlled and much of the inherent soil structure is maintained. These microcosms may realistically simulate conditions in the field; however, their high cost and the difficulty in manipulating soil and treatments may limit their usefulness. Further, the inability to replicate large microcosms reduces the number of variables that can be examined in any single study.

The current study has shown that a simple and cost effective microcosm is adequate for providing the necessary information required by risk assessors considering a potential field release. The intact core microcosm is extremely inexpensive and simple to utilize. Cost and complexity will be a limiting factor affecting acceptance. These factors are important since tests may be required to utilize many soils under a variety of conditions.

Although it has been shown that the intact core microcosm is effective for predicting the survival of P. aureofaciens in a single Maryland soil, much work is required before it can be accepted as a standard risk assessment tool. Validation in additional soils and with other organisms is necessary prior to acceptance.

This work was supported by grants from the U.S. Environmental Protection Agency and Environment Canada. Intellectual and financial assistance is greatly appreciated.

Angle JS, Levin MA, Gagliardi JV, McIntosh MS, Glew JG (1994) Pseudomonas aureofaciens in soil: survival and recovery efficiency. Microbial Rel 2: 247-254.

Barry GF (1988) A broad-host-range shuttle system for gene insertion into the chromosome of Gram-negative bacteria. Gene 71: 75-84.

Barry GF (1986) Permanent insertion of foreign genes into the chromosomes of soil bacteria. Biotechnol 4: 446-449.

Dileep-Kumar BS, Dube HC (1992) Seed bacterization with a fluorescent pseudomonad for enhanced plant growth, yield and disease control. Soil Biol Biochem 24: 539-542.

Drahos DJ, Barry GF, Hemming BC, Brundt EJ, Kline EL, Skipper HD, Kluepfel DA (1992) Spread and survival of genetically marked bacteria in soil. In: Day MJ, Fry JC (eds) Environmental release of genetically engineered and other microorganisms. Cambridge Univ Press, Cambridge, pp 147-160.

Gile JD, Collins JC, Gillett JW (1982) Fate and impact of selected wood preservations in a terrestrial model ecosystem. J Agric Food Chem 30: 295-301.

Gillett JW, Gile JD (1976) Pesticide fate in terrestrial laboratory ecosystems. Inter J Environ Stud 10: 15-22.

Heath RT (1979) Holistic studies of an aquatic microcosm. Int J Env 1: 87-93.

Henschke RB, Henschke EJ, Schmidt FRV (1991) Monitoring survival and gene transfer in soil microcosms of recombinant Escherichia coli to represent an industrial production strain. Appl Microbiol Biotechnol 35: 247-252.

Iwasaki K, Uchiyama H, Yagi O (1994) Survival and impact of genetically engineered Pseudomonas putida harboring mercury resistance genes in soil microorganisms. Biosci Biotech Biochem 58: 156-159.

Jaffee B, Phillips R, Muldoon A, Mangel M (1992) Density-dependent host-pathogen dynamics in soil microcosms. Ecology 73: 495-506.

Kluepful DA, Tonkyn, DW (1992) The ecology of genetically altered bacteria. In: Tjamos ES, Papavizas GC, Cook RJ (eds) Biological control of plant diseases. Plenum Press, New York, pp 407-413.

Naik GA, Bhat LN, Chopade BA, Lynch JM (1994) Transfer of broad-host-range antibiotic resistance plasmids in soil microcosms. Current Microbiol 28: 209-215.

Parmelee RW, Wentsel RS, Phillips CT, Simini M, Checkai RT (1993) Soil microcosm for testing the effects of chemical pollutants on soil fauna communities and trophic structure. Environ Toxicol Chem 12: 1477-1486.

Richaume A, Smit E, Faurie G, van Elsas J (1992) Influence of soil type on the transfer of plasmid RP4p from Pseudomonas fluorescens to introduced recipient and to indigenous bacteria. FEMS Microbiol Ecol 168: 123-129.

Sanders WM (1976) EPA needs in microcosm research. Intern J Environ Studies 10: 3-5.

Schindler JE, Waide JB, Waldron MC, Hains JJ, Schreiner SP, Freedman FL, Benz SL, Pettigrew DR, Schissel LA, Clark PJ (1980) A microcosm approach to the study of biogeochemical systems. In: Geisy JP (eds) Microcosms in ecological research. USDOE Symposium Ser4g 52 NTIS, Springfield, VA, pp 192-203.

Wolf DC, Dao TH, Scott HD, Lavy TL (1989) Influence of sterilization methods on selected soil microbiological physical and chemical properties. J Environ Qual 18: 39-44.

Wollum AG II (1982) Cultural methods for soil microorganisms. In: Page AL, Miller RH, Keeney DR (eds) Methods of soil analysis 2: chemical and microbiological properties, 2nd edn. American Society of Agronomy, pp 781-801.

Angle Table 1

Angle Table 2

Angle fig 1

Angle fig 2

Angle fig 3

Angle fig 4