Enterococcus faecalis Gene Transfer under Natural Conditions in Municipal Sewage Water Treatment Plants

Abstract

The ability of Enterococcus faecalis to transfer various genetic elements under natural conditions was tested in two municipal sewage water treatment plants. Experiments in activated sludge basins of the plants were performed in a microcosm which allowed us to work under sterile conditions; experiments in anoxic sludge digestors were performed in dialysis bags. We used the following naturally occurring genetic elements: pAD1 and pIP1017 (two so-called sex pheromone plasmids with restricted host ranges, which are transferred at high rates under laboratory conditions); pIP501 (a resistance plasmid possessing a broad host range for gram-positive bacteria, which is transferred at low rates under laboratory conditions); and Tn916 (a conjugative transposon which is transferred under laboratory conditions at low rates to gram-positive bacteria and at very low rates to gram-negative bacteria). The transfer rate between different strains of E. faecalis under natural conditions was, compared to that under laboratory conditions, at least 10⁵-fold lower for the sex pheromone plasmids, at least 100-fold lower for pIP501, and at least 10-fold lower for Tn916. In no case was transfer from E. faecalis to another bacterial species detected. By determining the dependence of transfer rates for pIP1017 on bacterial concentration and extrapolating to actual concentrations in the sewage water treatment plant, we calculated that the maximum number of transfer events for the sex pheromone plasmids between different strains of E. faecalis in the municipal sewage water treatment plant of the city of Regensburg ranged from 10⁵ to 10⁸ events per 4 h, indicating that gene transfer should take place under natural conditions.

The discovery that a bacterial phenotype can be “transformed” into a new phenotype dates back to the time when the nature of genetic material was not yet known (17). It was assumed for a long time that gene transfer between different species of microorganisms is a very rare event at best; later, this view changed. From gene-protein comparisons it was concluded that, “the available evidence suggests that interspecific transfer of genes has occurred between the three major groups of organisms: archaebacteria, eubacteria and eucaryotes” (33). There is very strong evidence that gene transfer occurs even between distantly related bacteria (25, 26), which supports the idea that effective mechanisms prevent uncontrolled gene transfer, which otherwise might even interfere with the species concept.

The most spectacular example of gene transfer in nature involves the ability of Agrobacterium tumefaciens strains to genetically engineer plants to produce the nutritive compounds nopalines and opines by transferring and integrating the so-called T-part of their endogenous Ti plasmids into plant genomes (14). Interestingly, the principal mechanisms used for Ti-mediated T-DNA transfer to plant cells and for bacterial conjugation seem to involve similar systems (23). Nevertheless, T-DNA transfer from A. tumefaciens to plants is one of the few examples of gene transfer in nature. In almost all reports dealing with gene transfer the workers performed experiments under laboratory conditions; e.g., the proof that T-DNA can be transferred from A. tumefaciens to the yeast Saccharomyces cerevisiae was obtained in this way (4).

For bacterium-bacterium gene transfer it has been postulated that strains of the gram-positive eubacterium Enterococcus faecalis “(the ‘Escherichia coli’ of the gram-positive bacteria?) may serve as reservoirs of genetic information available for passage to other streptococci and even other genera by conjugal processes” (5). This hypothesis stems from the fact that various genetic elements have been found to be present in great numbers in this species; these elements include transposons, so-called conjugative transposons, cryptic plasmids, resistance plasmids, and so-called sex pheromone plasmids (for reviews see references 5, 8, and 36). Some of these elements, the sex pheromone plasmids, have a very restricted host range; except for pIP964, which reportedly also replicates in Enterococcus faecium and Listeria monocytogenes (30), the sex pheromone plasmids are restricted to E. faecalis. Resistance plasmids apparently have a broad host range for gram-positive bacteria; conjugative transposons are found primarily in gram-positive bacteria, but they also can occur in gram-negative bacteria. Indeed, the data for conjugal transfer of genes and genetic elements between different strains of E. faecalis and between this bacterium and other bacterial species (in both directions [i.e., to and from E. faecalis]) under laboratory conditions are too numerous to be cited here (for an overview see reference 6).

On the other hand, reports of transfer of genetic elements from or to E. faecalis in nature are rare. Resistance plasmid pAMβ1 has been shown to transfer among E. faecalis, E. faecium, and Lactobacillus reuteri in the digestive tracts of mice (27). (This plasmid also transfers between Lactobacillus curvatus strains in fermenting sausages [35].) Inducible transfer of conjugative transposon Tn1545 from E. faecalis to Listeria monocytogenes in the digestive tracts of gnotobiotic mice has also been reported (12).

Here, we define natural conditions as conditions under which bacteria are kept and grown under conditions prevalent in the natural biotope (e.g., the digestive tract of an animal or a municipal sewage water treatment plant with rather low water temperatures, partially toxic components, etc.). Conditions under which conjugation between two strains occurs on a filter which is placed on a (rich medium) agar plate incubated at the optimal growth temperature are referred to as laboratory conditions.

Since no data on E. faecalis gene transfer in sewage water treatment plants are available, we performed such studies in the sewage water treatment plants of two Bavarian cities, Munich (Klärwerk Marienhof; Dietersheim) and Regensburg (Klärwerk Barbing), Germany. Our data indicate that under the natural conditions of the Regensburg plant ca. 10⁶ to 10⁹ gene transfer events between different E. faecalis strains should take place per day.

MATERIALS AND METHODS

Bacterial strains.

E. faecalis OG1X (Sm^r, containing no mobile genetic element) (20) was used as the source of donor strains. Various derivatives of OG1X, which contained sex pheromone plasmids pAD1 (34) and pIP1017 (19), broad-host-range plasmid pIP501 (18), and conjugative transposon Tn916 (15) alone or in all possible combinations, were constructed by standard conjugation techniques (13, 15). E. faecalis FA2-2 (Rif^r Fus^r, containing no mobile genetic element) (7) was used as the recipient. The following bacteria (species were identified by routine tests performed at University Hospital at Regensburg) were used as potential recipients for the genetic elements listed above: Escherichia coli, Bacillus subtilis, Enterococcus durans, Staphylococcus aureus, Streptococcus mutans, and Streptococcus sanguis. In some cases spontaneous antibiotic-resistant derivatives of these organisms had to be selected on Todd-Hewitt broth (THB) plates so that we would be able to test for conjugative transfer of the genetic elements used. Since transconjugants were obtained in none of these experiments, the details of the experiments are not described here.

Experiments in sewage water treatment plants.

Experiments in activated sludge basins of municipal sewage water treatment plants had to be performed in a microcosm (shown schematically in Fig. 1) so that there would be a sterile compartment (composed of dialysis tubing) which could be placed at various positions in the plants and protected from leakage caused by debris. Each Nadir dialysis bag (Roth, Karlsruhe, Germany) had a diameter of 50 mm and a nominal pore size of 2.5 nm. It was protected from damage by a stainless steel cage having a pore size of ca. 0.75 mm. Experiments performed with phages present in the sewage water at the Regensburg plant showed that the dialysis bags were phage proof. Linear DNA, introduced into a microcosm (which was preequilibrated in the activated sludge basin for 2 h) via a T connector and a sterile syringe, was degraded to nondetectable levels after 20 h; this did not occur if the microcosm was preequilibrated in sterile 10 mM Tris-Cl (pH 7.5)–0.1 mM EDTA buffer. We concluded, therefore, that our setup allowed an equilibrium between the dialysis bag contents and the soluble components of sewage water to occur (within 90 min), but that phages were excluded by the dialysis bag.

FIG. 1.

Schematic diagram of the microcosm used for experiments in the activated sludge digestors. The whole assembly could be varied in length, by using nuts and bolts, from 15 to 45 cm and was sterilized with the dialysis tubing connected to the polypropylene end caps via autoclavable O rings. The arrows indicate connections via silicone tubing to the recirculation system.

All parts of the microcosm were constructed of noncorrosive materials (the nuts, bolts, and protective cage were stainless steel; the dialysis tubing, O rings, and end caps were autoclavable plastic); the whole assembly, filled with 10 mM Tris-Cl (pH 7.5)–0.1 mM EDTA, was autoclaved prior to each experiment. Silicone tubes attached to the inlet and outlet were closed immediately after sterilization; they were connected in the plant to a recirculating pump (Isamatec type MV-Z). The recirculating system contained pH, temperature, and O₂ electrodes (WTW, Weilheim, Germany) and was rinsed for 1 h prior to use with 70% ethanol. Then the recirculating system and the microcosm were connected, and the liquid was replaced via two T connectors by sterile 1% NaCl. This assembly was circulated (at a rate of 600 ml/min) in the activated sludge basin for at least 2 h, after which the system was inoculated via one T connector and a sterile syringe. The total volume of the recirculating system plus microcosm was ca. 600 ml, and the microcosm contained ca. 300 ml. An equilibration time of ca. 90 min for the microcosm contents and sewage water was determined by performing experiments in which Tris-Cl buffer (pH 9.5) was used and the pH was measured.

Donor strains of E. faecalis (OG1X with various combinations of genetic elements) and recipient strain FA2-2 were grown at 37°C separately in THB (Oxoid, Wesel, Germany) to an optical density at 600 nm of ca. 0.6, collected by centrifugation, washed with 1% NaCl containing 10 mM Tris-Cl (pH 7.5), and resuspended in the same buffer. The cells were kept on ice for a maximum of 45 min and were mixed immediately before inoculation into the microcosm to obtain a donor/recipient ratio of 0.1. A maximum volume of 20 ml was used for inoculation; the syringe and T connector were rinsed with another 20 ml of buffer to mix all of the cells into the recirculating system. After 3 min of recirculation, the first sample (the zero-time sample) was removed with a syringe from the second T connector by first withdrawing 20 ml of waste and then collecting an appropriate amount. Samples were collected 1, 4, and 20 h after inoculation.

For experiments in anoxic sludge digestors cells were prepared as described above. Only very slight turbulence occurred in the digestors, and thus there was no potential danger of damage to the dialysis tubing. Therefore, donor and recipient cells were preincubated in separate parts of a sterile dialysis bag for 90 min; the separating clamp was removed after 90 min, and the contents were mixed and incubated for various times. The contents of the dialysis bag were mixed every hour by 3- to 5-fold inversion.

Determinations of gene transfer efficiency.

The numbers of donor cells and recipient cells were determined for each sample by counting the number of colonies resistant to streptomycin (500 μg/ml) and the number of colonies resistant to rifampin (25 μg/ml) and fusidic acid (25 μg/ml), respectively, for various sample dilutions. Transconjugants were identified on THB agar containing fusidic acid onto which various sample dilutions were plated. These plates also contained 5% defibrinated horse blood (Oxoid, Augsburg, Germany) for detection of pAD1-encoded cytolysin (3), 1,000 μg of kanamycin per ml for detection of pIP1017, 10 μg of erythromycin per ml for detection of pIP501, and 10 μg of tetracycline per ml for detection of Tn916.

Gene transfer efficiency was calculated by determining the number of transconjugants per donor cell by using the conventional definition. It should be noted, however, that this calculation was to a certain extent problematic for pAD1, since the pAD1-encoded cytolysin kills recipients over extended periods of time and therefore the donor/recipient ratio is not constant over time. Cell titers were determined for donor strains, recipient strains, and transconjugants for each experiment in parallel, and the duplicate values varied by a maximum factor of 3. The numbers shown in the tables below are mean values from these duplicate determinations of transfer efficiencies. Also, under natural conditions variables like sewage water temperature, oxygen saturation, and chemical composition of the wastewater, etc., influence the results; in our experiments even greater variances occurred. Therefore, no extensive statistical analyses were done; we note that the experiments performed in the Regensburg plant (see Fig. 3) were performed in triplicate for the activated sludge basin and in duplicate for the anoxic sludge digestor. The values (determined in duplicate, as described above) which we obtained in these experiments varied by a maximum factor of 15. In the case of the Munich plant all of the experiments were performed only once; the identities of transconjugants containing pIP501 or Tn916, both of which appeared at unexpectedly high rates, were determined by Southern hybridization (see below).

FIG. 3.

Comparative data for gene transfer efficiency in the Regensburg sewage water treatment plant. Efficiencies of conjugation for sex pheromone plasmid pIP1017, broad-host-range plasmid pIP501, and conjugative transposon Tn916 are shown. For the columns labelled pIP501 (+ pIP1017) and Tn916 (+ pIP1017) the values are values for transfer of the first genetic element from a donor strain also harboring pIP1017. Values obtained under laboratory conditions, obtained in the anoxic sludge digestor (35°C), obtained during the summer in the activated sludge digestor (20 to 25°C), and obtained during the winter in the activated sludge digestor (14 to 16°C) are indicated. Open bars indicate that no transconjugant was obtained and indicate the detection limit.

A small inhibitory effect of the biotope sewage water treatment plant on E. faecalis was observed; reduction of the donor and recipient cell titers by a maximum factor of 2.7 was detected in the activated sludge basin, while the maximum factor for the anoxic sludge digestor was 16.3. This reduction occurred after incubation for between 4 and 20 h, but was negligible between zero time and 4 h (data not shown). Since under laboratory and natural conditions gene transfer events increased with time up to 5 h (data not shown), most experiments were performed for 4 h. This also allowed us to definitely conclude that experiments in the sewage water treatment plant were performed under sterile conditions. In the few cases in which sterility problems occurred (due to dialysis membrane damage), the problem very clearly was detected as a reduction in donor and recipient titers by at least 5 orders of magnitude, due to the massive occurrence of phages lytic for E. faecalis in the sewage water (data not shown).

In experiments in which less than 10 transconjugants were observed, these were examined not only by determining the antibiotic resistance profile, but also by isolating total genomic DNA as described by Muscholl et al. (28) and probing by Southern hybridization by using the enhanced chemiluminescence protocol (Amersham, Braunschweig, Germany) and the following probes: two 1,600- and 700-bp HindIII fragments of pWM401 (37) specific for pIP501 and one 1,700-bp HindIII-Asp718 fragment of plasmid pAM120 (16) specific for Tn916. In the case of pIP1017, transconjugants could be detected by pulsed-field gel electrophoresis (PFGE) of BamHI-restricted total genomic DNA (see Fig. 2) and the sex pheromone-induced clumping reaction. In the case of pAD1-containing transconjugants, we used not only the hemolytic phenotype, but also the sex pheromone-induced clumping reaction for identification.

FIG. 2.

Identification of FA2-2 transconjugants containing pIP1017 by PFGE. Conventionally purified and BamHI-restricted total genomic DNA was analyzed by PFGE. Lanes 1 to 3, three independent clones, identified as FA2-2::pIP1017 transconjugants; lane 4, FA2-2; lane 5, OG1X; lane 6, OG1X::pIP1017. Lane 7 contained λ DNA restricted with Asp718 (30, 17, and 1.5 kb). The two arrows on the left indicate the positions of bands specific for FA2-2 genomic DNA; the arrow on the right indicates the position of a pIP1017-specific fragment.

RESULTS AND DISCUSSION

Gene transfer efficiencies under laboratory conditions.

Gene transfer efficiencies under laboratory conditions were measured for the following two reasons: (i) reference data were needed and (ii) potential deviations of gene transfer efficiency values due to different settings (laboratory or natural conditions) had to be determined.

Transfer of single genetic elements under laboratory conditions was within the expected size range (Table 1) (13). Cotransfer of two genetic elements was clearly observed, with the sex pheromone plasmids showing a synergistic effect at least for Tn916 (the conjugative transposon was transferred at least 10- to 100-fold better from a strain containing not only Tn916 but also pAD1 or pIP1017). Since no cotransfer of three genetic elements was observed, such data are not listed in Table 1.

TABLE 1.

Efficiencies of gene transfer between different E. faecalis strains under laboratory conditions^a

Genetic element(s)	Transfer efficiency in rich liquid medium	Transfer efficiency on rich medium agar plates
pAD1	1.9 × 10−1	2.1 × 10−1
pIP1017	1.4 × 10−1	1.0 × 10−1
pIP501	5.4 × 10−8	3.8 × 10−6
Tn916	2.0 × 10−9	7.9 × 10−9
pAD1/pIP501	4.4 × 10−2/4.7 × 10−7 (1.2 × 10−7)	2.4 × 10−2/1.2 × 10−6 (9.3 × 10−7)
pIP1017/pIP501	2.3 × 10−3/9.6 × 10−7 (5.1 × 10−7)	8.0 × 10−2/1.8 × 10−4 (6.6 × 10−5)
Tn916/pIP501	7.5 × 10−8/2.9 × 10−8 (3.1 × 10−8)	1.2 × 10−7/9.0 × 10−7 (1.3 × 10−7)
pAD1/Tn916	5.0 × 10−2/2.4 × 10−7 (7.7 × 10−7)	5.4 × 10−3/1.0 × 10−6 (9.5 × 10−7)
pIP1017/Tn916	2.7 × 10−3/2.4 × 10−7 (2.6 × 10−8)	7.1 × 10−2/1.7 × 10−6 (6.0 × 10−7)
pIP501/Tn916	2.9 × 10−8/7.5 × 10−8 (3.1 × 10−8)	9.0 × 10−7/1.2 × 10−7 (1.3 × 10−7)

^aIn cases in which two genetic elements are listed (e.g., pAD1/pIP501) the values are the transfer efficiencies of the elements measured separately. The efficiencies of transfer of both genetic elements (i.e., concurrent transfer of the two elements) are given in parentheses. 

We also determined if broad-host-range plasmid pIP501 and the conjugative transposon Tn916 were transferred under laboratory conditions into the gram-negative bacterium Escherichia coli or into the gram-positive bacteria B. subtilis, E. durans, Staphylococcus aureus, Streptococcus mutans, and Streptococcus sanguis. These experiments were deliberately performed in liquid culture, because data for transfers on the surfaces of filters were available, at least for some species (2). In contrast to previously reported transfer efficiencies ranging from 10⁻⁶ to 10⁻⁸ for Tn916 on solid surfaces, we did not observe any transfer in liquid medium (our detection limit was ca. 10⁻⁸); very probable mating pair formation is not disturbed on solid surfaces, in contrast to preparations incubated in liquid. Since gene transfer into the potential recipients was also not observed in a few orienting studies when we used the microcosm in the activated sludge basin of the Regensburg plant, no further studies of gene transfer into other species in sewage water treatment plants were done.

Gene transfer efficiency in the sewage water treatment plants.

As Table 2 shows, the results obtained under laboratory conditions were comparable for Munich and Regensburg; therefore, the differences obtained in experiments performed in the activated sludge basins of the Munich and Regensburg municipal sewage water treatment plants, especially the differences observed with pIP501 and Tn916, were interpreted to be significant. Potential factors which might have been responsible for the lower gene transfer efficiencies in the Regensburg plant than in the Munich plant are discussed below.

TABLE 2.

Transfer efficiencies of various genetic elements from E. faecalis OG1X to strain FA2-2 in liquid

Conditions	Transfer efficiency
	pAD1	pIP1017	pIP501	Tn916
Laboratory, Munich	5.5 × 10−1	2.1 × 10−1	5.4 × 10−7	5.2 × 10−9
Laboratory, Regensburg	3.4 × 10−1	1.1 × 10−1	1.9 × 10−7	9.3 × 10−9
Activated sludge basin, Munich	NAa	5 × 10−5	8.8 × 10−4	5.2 × 10−5
Activated sludge basin, Regensburg	NA	1.4 × 10−7	<8.1 × 10−9	<5 × 10−9

^aNA, not applicable, but clearly less than 10⁻⁵ (see text). 

For sex pheromone plasmid pAD1 no definite transfer efficiencies were measured under natural conditions; they clearly were less than 10⁻⁵. This stems from the fact that in the case of pAD1 no direct selection for transconjugants was possible; rather, the phenotype hemolysis of erythrocytes had to be determined. Not a single hemolytic colony was detected in these experiments, even if samples were plated on agar plates (22.5 by 22.5 cm). Our data clearly indicated that the activity of the sex pheromone system was reduced in the activated sludge basins of both plants by 4 to 6 orders of magnitude or that this system was not active at all. This could have been due to various reasons. (i) Active sex pheromone was rapidly diluted from the microcosm during the experiment. (Experiments in which synthetic sex pheromone cAD1 was added to the microcosm at a titer of at least 100,000 together with E. faecalis OG1X::pAD1 did not result in any clumping reaction. From experiments in which we used the microcosm in sterile buffer we concluded that active sex pheromone diffuses too fast into the surrounding liquid [sewage water in the activated sludge basin] to induce a clumping reaction [data not shown]). (ii) In addition, degradation of the inducing peptide in activated sludge seems to play a role: 25 ml of filter-sterilized activated sludge inactivated synthetic sex pheromone at a titer of 10,000 within 30 min at 37°C, but not if the preparation was pretreated for 5 min at 100°C. Such inactivation did not occur if sex pheromone and filter-sterilized activated sludge were separated by a dialysis bag (data not shown). (iii) Furthermore, induction of an aggregation substance (leading to the bacterial clumping reaction, resulting in very effective gene transfer specific for sex pheromone plasmids under laboratory conditions) by the sex pheromone is not observed at temperatures below ca. 12°C. To our knowledge, the latter effect has not been observed before; it represents a true temperature regulatory phenomenon. Under laboratory conditions transfer efficiencies for pAD1 and pIP1017 in liquid culture dropped from ca. 10⁻¹ at 37°C to 10⁻⁷ at temperatures below ca. 12 to 15°C. Since donor strains in which an aggregation substance was preinduced at 37°C showed normal transfer efficiencies of sex pheromone plasmids at 10°C, the regulatory phenomenon seems not to involve the actual conjugation process.

In the case of broad-host-range plasmid pIP501 transfer efficiencies under laboratory conditions were in the expected range; they dropped below the detection limit in the Regensburg plant, but were extraordinarily high in the Munich plant. The same phenomenon was observed for Tn916, which had measured transfer efficiencies ca. 10-fold lower than those for pIP501. Because these data were rather surprising, we checked transconjugants obtained in the Munich plant for the presence of pIP501 or Tn916 by Southern hybridization. The identity of the transconjugants as FA2-2 transconjugants was tested by performing PFGE with conventionally purified genomic DNA. The fact that OG1X and FA2-2 could be differentiated clearly in such experiments is shown in Fig. 2. In this case not only was differentiation between OG1X and FA2-2 possible, but differentiation between OG1X::pIP1017 and FA2-2::pIP1017 was also possible.

There were several potential reasons for the extraordinarily high transfer efficiencies observed for pIP501 and Tn916 in the activated sludge basin of the Munich plant compared to the Regensburg plant. (i) Most of the experiments in the Munich plant were performed during a very unusual weather situation in the fall of 1993 (heavy rain showers and even snowfall in October), which resulted in very low temperatures in the activated sludge basin and also unusually dilute sewage water. The Tn916 experiments had to be performed at ca. 8°C, and all other experiments at the Munich plant were performed at temperatures between 10 and 13°C. In the Regensburg plant, experiments in the activated sludge basins were performed at water temperatures between 20 and 25°C (comparative data for different temperatures at the Regensburg plant are shown in Fig. 3). (ii) The efficiencies of aeration at the Munich and Regensburg plants differed extensively. An O₂ saturation level of 1.4 to 2.2 mg/ml was measured at the Regensburg plant, while the values at the Munich plant were consistently more than 10 mg/ml. (iii) Although it is not possible to directly compare the compositions of the two sewage waters, it should be noted that the Munich plant receives relatively more industrial wastewater than household wastewater compared to the Regensburg plant (indicated by a ca. fivefold-higher load of heavy metals). Taken together, our data seem to indicate that under severe stress situations very unusual (high) gene transfer efficiencies can occur.

Comparative data for the Regensburg sewage water treatment plant.

Figure 3 summarizes data for gene transfer efficiencies at different temperatures in the activated sludge basin and the anoxic sludge digestor of the Regensburg sewage water treatment plant (for comparison results obtained in the Regensburg laboratory are shown). Since no actual numbers could be determined for pAD1 under natural conditions (the values were <10⁻⁵ [see above]), data for pAD1 under natural conditions are not included. While the experiments performed in the activated sludge basin were performed with the microcosm system (to simulate the extensive mixing in the basin), incubations in the sludge digestor were in dialysis bags, the contents of which were mixed every hour (to mimic the minor turbulence in the digestor). Experiments under laboratory conditions were performed at 37°C; data for the activated sludge basin were collected during the summer at temperatures between 20 and 25°C and during the winter at temperatures between 14 and 16°C. For the anoxic sludge digestor only one data set was collected, since the temperature was constant (35°C) throughout the year.

It is evident that (i) in all cases gene transfer efficiencies were highest under laboratory conditions; (ii) gene transfer in the activated sludge basin was more efficient at higher temperatures (i.e., during the summer) than at lower temperatures; (iii) the gene transfer efficiencies in the anoxic sludge digestor were higher than those in the activated sludge basin (this effect can be attributed to two facts, the higher temperatures in the anoxic sludge digestor and the far less intense mixing in the digestor, which did not disrupt potential mating pairs); and (iv) the presence of sex pheromone plasmids enhances transfer of other genetic elements, especially under laboratory conditions (under natural conditions this effect is less pronounced).

Our data showing that transfer efficiencies were highest under laboratory conditions were mainly due to the fact that conjugation efficiencies in THB are temperature dependent. The actual values for pIP1017 were 1 × 10⁰, 2 × 10⁻³, and 1.5 × 10⁻⁷ for experiments performed at 37, 25, and 10°C, respectively. Nevertheless, other factors, such as growth medium, also play a role, since for experiments that were performed at 37°C under laboratory conditions but in which filter-sterilized activated sludge was used instead of THB the efficiency of pIP1017 conjugation was 1.2 × 10⁻⁴.

Calculation of endogenous transfer efficiencies for E. faecalis in the Regensburg plant.

In all of the experiments described above we used comparable bacterial concentrations under laboratory and natural conditions, which resulted in comparable data. Therefore, the concentrations of E. faecalis introduced into the microcosm (experiments performed in the activated sludge basin) or the dialysis bags (experiments performed in the anoxic sludge digestor), up to 10⁷ donor cells/ml and 10⁸ recipient cells/ml, were 3 to 4 orders of magnitude higher than the actual numbers in the sewage water treatment plants. For different locations in the Regensburg plant we determined the following E. faecalis titers: 2.4 × 10³ CFU/ml at the point of entry into the plant; 7.5 × 10² CFU/ml in the activated sludge basin; 4.8 × 10⁴ CFU/ml at the entrance of the anoxic sludge digestor; 4.7 × 10² CFU/ml at the exit of the anoxic sludge digestor; and 2.8 × 10⁰ CFU/ml at the plant exit.

Since transfer efficiencies under natural conditions were highest for pIP1017, experiments were performed with this genetic element in the activated sludge basin and the anoxic sludge digestor by using reduced titers of donor and recipient cells. We still had to use titers of donor cells (OG1X::pIP1017) of 10⁴ to 10⁵ to detect single transconjugants (our assumption that 10% of all E. faecalis cells could represent potential donors is supported by data from other workers [9]). The results of these experiments are shown in Fig. 4. Since the limit of detection for gene transfer was ca. 100-fold higher than the actual concentrations of potential E. faecalis donor cells, extrapolations had to be done to calculate potential gene transfer rates. The extrapolated, maximal gene transfer efficiencies were ca. 10⁻⁸ for the conditions in the activated sludge basin (the actual E. faecalis titer was ca. 10³ for donor plus recipient cells) and the anoxic sludge digestor (the actual E. faecalis titer was ca. 10⁴ for donor plus recipient cells). By using these values the maximal numbers of potential transconjugants originating in 4 h under natural conditions could also be calculated (Table 3).

FIG. 4.

Calculation of maximal transfer frequencies for pIP1017 at E. faecalis titers present in the Regensburg plant. Transconjugation rates (numbers of transconjugants per donor cell) were determined independent of the donor titer (note that the recipient cell titer was in each case 10-fold higher). Since the limit of detection of transconjugants was higher than the actual E. faecalis cell number in either the anoxic sludge digestor or the activated sludge basin, extrapolations had to be made to obtain calculated transconjugation rates for total E. faecalis titers of 10³ (activated sludge basin) and 10⁴ (anoxic sludge digestor).

TABLE 3.

Calculated maximal transfer efficiencies for pIP1017 in the Regensburg sewage water treatment plant

Biotope	Extrapolated maximal gene transfer efficiency	Vol of biotope (ml)	Total no. of E. faecalis cells	Maximal calculated no. of transconjugants
Activated sludge basin	10−8	3 × 1013	3 × 1016	3 × 105
Anoxic sludge digestor	10−8	6 × 1012	6 × 1016	3 × 108

The maximal numbers of calculated transconjugation events under natural conditions, which range from 10⁵ to 10⁸, might appear to be high at a first glance; it should be noted, however, that under these conditions only ca. 10⁻⁸ and 10⁻¹¹ of all E. faecalis cells in the anoxic sludge digestor and the activated sludge basin would be transconjugants.

We want to emphasize that the above-mentioned calculations rely on several numbers which could not be measured directly and therefore represent only rough estimates. In addition, conventional statistical analyses of data obtained under natural conditions seem not to make sense to us, since too many variables, which cannot be controlled, influence the actual numbers (see above).

A final comment relates the results of our study to other results, including the results described below. (i) Our finding that under very special conditions (e.g., 30°C below the optimal growth temperature in the Munich plant), unexpectedly high efficiencies of gene transfer were observed is at least to some extent supported by other findings. Heat treatment resulted in enhanced gene transfer from Escherichia coli to various coryneform bacteria (31), as did other stress situations, such as exposure to organic solvents or detergents and pH shifts (32). (ii) Our finding that gene transfer efficiencies were highest under laboratory conditions is corroborated by the results of comparisons of efficiencies of gene transfer from Alcaligenes eutrophus to Variovorax paradoxus under laboratory conditions and in soil (29) and by the fact that gene transfer rates between marine bacteria were higher if the strains were encapsulated in microbeads than if the genes were transferred in marine water (1). (iii) Gene transfer occurs under natural conditions not only in soil, marine water, and sewage water treatment plants, but also in other biotopes. For example, transduction via phages occurs between Pseudomonas aeruginosa strains on the surfaces of leaves (21); gene transfer occurs between the fish-pathogenic bacterium Aeromonas salmonicida and a human Escherichia coli isolate in raw salmon on a cutting board (see reference 22 for various other natural settings); and gene transfer between physically isolated bacteria is enhanced by the presence of burrowing earthworms as a biological factor which facilitates cell-to-cell contact (10).

Finally, workers also have to be aware of the fact that not all genes are transferred with the same efficiency (11) and the fact that gene transfer mechanisms other than conjugation also can be extremely efficient in various biotopes (24).

Conclusions.

Gene transfer efficiencies between different strains of E. faecalis were highest under laboratory conditions; under natural conditions in municipal sewage water treatment plants the efficiencies dropped up to 6 orders of magnitude, but still were measurable. Gene transfer from E. faecalis to other bacterial species could not be detected in liquid media. Calculations to determine potential gene transfer rates between different strains of E. faecalis in the municipal sewage water treatment plant of the city of Regensburg resulted in maximal values of ca. 10⁸ events per day. Even with a 100-fold reduction (killing under natural conditions) 10⁶ transconjugants per day would still be released into the River Danube. These transconjugation events cannot be avoided since every person excretes E. faecalis.

The argument that the possibility of gene transfer has to be totally excluded in genetically engineered bacterial strains does not make sense if gene transfer occurs in nature. A case-by-case discussion of safety demands still seems to be justified; this discussion should take into account the enormous wealth of evidence that gene transfer occurs under natural conditions and the few cases in which actual numbers have been measured.