INTRODUCTION

Top Abstract Introduction Materials and Methods Results and Discussion References

Understanding the transport and behavior of bacteria in porous media and other environments has broad implications in diverse areas, ranging from agriculture to groundwater quality, risk assessment, and bioremediation. Investigations in this area have been hampered by limitations of microbial tracking methods that have hindered other studies in microbial ecology, such as reliably distinguishing organisms of interest from the background microbial community and specifically detecting and quantifying small numbers of specific bacteria in environmental samples. Nonetheless, numerous experiments have been performed to monitor the behavior and movement of bacterial cells through porous media under conditions of saturated flow. Data from these studies indicate that a variety of physical, chemical, and biological factors affect bacterial transport characteristics (for reviews, see references l, 8, and 11). Most assessments of microbial transport have employed stained bacterial cells (9), relied on selective-plating approaches based on natural or engineered bacterial traits (6, 10), utilized immunological detection methods (22), or applied DNA-based detection strategies employing molecular techniques to detect sequences specific to the organism of interest (21). While providing valuable data, such approaches may not be applicable or appropriate in all cases. For example, DNA-binding dyes, such as acridine orange and DAPI (4',6'-diamidino-2-phenylindole), may influence characteristics of bacteria, such as viability and surface properties, and thus affect transport behavior (17). There is also ample evidence that many species of bacteria can enter a viable but nonculturable state under environmental conditions rendering simple culture-based detection of those cells inadequate (16, 19). Further, it is not always feasible to employ organisms that are genetically engineered to facilitate monitoring, particularly for in situ studies.

We report a novel bacterial tracking approach, based on altering the stable carbon isotope value (¹³C) of bacterial cells, which provides specific and sensitive detection and quantification of those cells in environmental samples. The rationale is that bacteria cultured on growth substrate enriched in ¹³C will differ only in their stable isotopic signature and will otherwise be physically and genetically unaltered. The presence and transport of introduced organisms in groundwater samples collected downfield of an injection site are indicated by mass spectrometric analysis of groundwater samples, where the presence of ¹³C-enriched bacteria produces a distinct increase in the ¹³C value of the sample. A regression model relating known numbers of ¹³C-enriched bacteria to the ¹³C values of groundwater samples allows quantification of the number of labeled organisms present. While ¹³C-enriched growth substrates have previously been introduced into soils and sediments to help identify organisms actively metabolizing that substrate (2, 7, 18), we report the first application of isotopic enrichment of whole cells to facilitate monitoring their transport following introduction into the environment.

This method was developed in support of in situ bacterial transport experiments in a coastal plain aquifer that were bounded by the following parameters. (i) The test organism had to be indigenous to the aquifer. (ii) Use of radioactive labels was not allowed. (iii) The test organism could not be resistant to clinically important antibiotics. (iv) The tracking methodology had to facilitate good survival and detection of the test organism during the course of the experiment. (v) The test organism had to be detectable against the indigenous bacterial community. An important advantage of the technique we describe is the ability to specifically detect an introduced, unmodified bacterial strain in the presence of the indigenous bacterial community (including that same organism in its unenriched form, i.e., background of "itself"). This approach has the additional advantages of low sensitivity levels (~2 labeled bacteria/ml) in groundwater samples and allowing precise quantification of the introduced organisms in each sample. Finally, exhaustive sampling can be performed in the field followed by spot analysis of samples to identify regions of interest in space and time (in this case, the times and locations of bacterial breakthrough), assuring that important data points are captured and analytical efforts are minimized.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results and Discussion References

Study site description. The bacterial transport site is located on the southern end of the Eastern Shore of Virginia (i.e., the tip of the Delmarva Peninsula) and has been described in detail elsewhere (3). Briefly, the sediments are comprised of unconsolidated to weakly cemented sand that is well sorted and ranges from fine- to medium-grained and pebbly sand. The sediments were deposited by wind-, wave-, and tide-driven currents.

Bacterial strains and culture conditions. The three strains utilized in these experiments were isolated from groundwater samples obtained from the bacterial transport site. The strains used in these experiments were initially isolated from site groundwater (SGW) as previously described (3). Strains F3T3, PL2W31, and DA001 are aerobic heterotrophs and have been determined to be most closely related to Stenotrophomonas maltophilia (D. Balkwill, personal communication), Arthrobacter globiformis (W. E. Holben and W. P. Kovacik, unpublished data), and Comamonas testosteroni (Holben and Kovacik, unpublished data), respectively, based on partial 16S rDNA sequence analysis. For these experiments, each culture was grown to late log phase in R2A broth at 25°C and then brought to 15% (vol/vol) glycerol and stored at 70°C. To initiate each experiment, the appropriate glycerol stock culture was streaked onto R2A agar and incubated at either 25°C for strains F3T3 and DA001 or 37°C for strain PL2W31, which was found to grow robustly at this temperature. Direct enumeration of cells in the cultures was based on fluorescence microscopy of DAPI-stained cells, performed as described by Schallenberg et al. (20). Enumeration of PL2W31 CFU on selective medium was achieved by spread plating of appropriate dilutions of samples onto R2A agar supplemented with 50-µg/ml nalidixic acid (R2A plus nal agar), to which this strain is naturally resistant.

Isotopic enrichment. To produce ¹³C-enriched F3T3 or PL2W31 bacterial cells, colonies growing on agar were scraped off with a sterile loop and thoroughly suspended to an optical density at 550 nm (OD₅₅₀) of 5.0 in Oyster Artificial Groundwater (OAGW), which is based on the groundwater chemistry of the site and contains per liter 60 mg of MgSO₄ · 7H₂O, 20 mg of KNO₃, 36 mg of NaHCO₃, 48 mg of CaCl₂ · 2H₂O, 50 mg of Ca(NO₃)₂ · 4H₂O, 25 mg of CaSO₄ · 4H₂O, and 28 mg of NaH₂PO₄ · H₂O. The cell suspension was then diluted 1:100 in OAGW supplemented with 0.1% of uniformly (98 to 99%) labeled [¹³C]glucose (Cambridge Isotope Laboratories, Andover, Mass.), resulting in an initial OD₅₅₀ of 0.05. The cultures were incubated with shaking at 250 rpm on a rotary shaking platform at 25°C for 72 h. Culture conditions for strain DA001 were essentially the same as those used for F3T3 and PL2W31, except that M9 medium (14) amended with 0.1% uniformly labeled [¹³C]acetate (Cambridge Isotope Laboratories) was used in lieu of OAGW amended with glucose. In this way, all biomass that accumulates during culturing is based on [¹³C]glucose or [¹³C]acetate as the sole source of carbon and energy. Following incubation, the cells were harvested by centrifugation at 16,000 × g at 10°C for 10 min and then resuspended in the same volume of unsupplemented OAGW or M9, as appropriate, to remove unincorporated glucose or acetate. This wash step was repeated, and the cells were resuspended in one-fourth volume of unsupplemented OAGW or M9, as appropriate, and starved at room temperature in the dark for 48 to 72 h prior to use in experiments. The degree of ¹³C enrichment in the cultures was determined by isotopic analysis of an aliquot (1.0 ml) of the washed bacterial suspension as described below.

Production of lysed cell material. The experiment comparing intact cells to lysed cell material employed ¹⁴C-labeled intact cells and ¹⁴C-labeled lysed cells. To ensure direct comparability of results, the intact and lysed cells were produced from a single [¹⁴C]acetate-labeled culture as follows. DA001 cells were cultured on [¹⁴C]acetate, washed, and starved as described above. Following the starvation period, the cell suspension was washed twice to remove any label that may have been released to the medium and then split into two 50-ml aliquots, each containing 10¹⁰ cells/ml. One aliquot of the suspension served as the intact cells for injection, while the other part was lysed via sonication. Lysis was achieved using a Sonifier Cell Disruptor Model W-350 (Branson Sonic Power Co., Danbury, Conn.). The sonicator was fitted with a microtip, and the cell suspension was lysed at full intensity on a 65% duty cycle. Sonication was continued until direct microscopic enumeration indicated that cell numbers were reduced from 10¹⁰ cells/ml to fewer than 10⁵ cells/ml (i.e., >99.99% lysis).

Stable isotopic analysis. Aqueous samples from sediment cores and groundwater from the in situ injection experiment were collected as described in the appropriate sections below. It should be noted that these groundwater samples also contain essentially constant sources of background carbon (e.g., indigenous microbes and other sources of particulate organic carbon), which contribute to the total carbon measured. Total bacteria and other particulates in the aqueous samples (core eluent or groundwater samples) were collected by centrifugation at 100,000 × g for 30 min at 20°C, and the supernatant was carefully decanted by aspiration. This regimen ensured that all bacteria in the samples (including indigenous microbes in groundwater samples) were recovered. The pellets were resuspended in 1.0 ml of OAGW and then transferred to 10-mm quartz tubes, lyophilized to dryness, and retained frozen (20°C) until isotopic analysis was performed. Analysis of samples containing ¹³C-enriched bacteria was performed using a modified Dumas combustion (13). Purified gases were obtained by cryogenic gas separation, and isotopic determinations were performed using a PRISM stable-isotope-ratio mass spectrometer (VG Isogas Ltd.). As most samples from the aquifer were within the range of natural-abundance samples encountered in a variety of other studies, stable-carbon-isotope values are presented as
where R is ¹³C/¹²C and the standard is Vienna Peedee Belemnite.

Quantification of cell numbers based on ¹³C signature. To quantify bacteria and determine the sensitivity of detection, the relationship between ¹³C values and cell numbers was established. Decadal serial dilutions of known numbers (based on direct microscopic enumeration) of ¹³C-enriched cells were made such that 10⁷, 10⁶, 10⁵, 10⁴, 10³, 10², 10¹, and 0 cells/ml were present in a total volume of 50 ml of SGW from the Oyster, Virginia, site. This protocol provided a dilution series against the natural bacterial population existing in the groundwater, which was used in intact-core and in situ experiments. These samples were prepared and analyzed for stable isotope analysis as described above.

Radioactive isotopic analysis. When radioactively labeled cells were used, 4 ml of groundwater sample was either added directly to 16 ml of EcoLite scintillation fluid (ICN Biomedicals, Inc., Costa Mesa, Calif.) or filtered through a 0.45-µm-pore-size type HA filter (Millipore Corporation, Bedford, Mass.). In the latter case, the filter (with trapped bacteria) and the filtrate were added to separate scintillation vials containing 16 ml of EcoLite scintillation fluid. Each sample was mixed vigorously and then subjected to LSC for replicate 10-min counts using a Beckman LS6500 (Beckman Instruments, Inc., Fullerton, Calif.).

Quantification of cell numbers based on ¹⁴C label. To quantify the numbers of radioactively labeled cells in samples, the relationship between ¹⁴C counts per minute and cell numbers was established. This was accomplished essentially as for the ¹³C-based samples, except that the dilution series of known cell numbers was subjected to LSC analysis as described in the previous section. Assuming that significant detection of ¹⁴C was twice the normal background signal for ¹⁴C, the detection limit for ¹⁴C-labeled cells or cell material was approximately 10⁵ cells/ml (data not shown).

Intact-core experiments. Native-matrix intact-core experiments were performed to assess the feasibility of and validate our approach for specific detection of introduced bacteria in porous media. Sediment cores and groundwater used in these experiments were taken from the transport site. Intact cores (7.5 by 70 cm) were taken from an exposed outcrop (the "borrow pit") which comprised the same lithologies as did the flow field that was used for the in situ experiments (15). Each exposed core end was trimmed to provide a final length of 50 cm and capped with a ported polyvinyl chloride endcap milled and screened to provide uniform access to the entire core diameter for influent and effluent water. Prior to initiation of experiments, cores were perfused extensively with 5 to 10 pore volumes (PV) of SGW in an upflow configuration. Isotopically enriched ¹³C or ¹⁴C cells (or an equivalent number of lysed cells) were introduced at a density of approximately 10⁹/ml (exact concentrations were precisely determined for each experiment) in 0.5 PV of SGW at a flow rate of 5.0 ml/min in the upflow configuration. Samples (0.1 PV each) were collected from the core outlet during this process by using a Bio-Rad Model 2128 fraction collector (Bio-Rad Laboratories, Hercules, Calif.). Individual samples were processed for enumeration of added cells by selective plating, stable isotopic analysis, or radioactive isotopic analysis as indicated.

In situ experiment. A large-scale in situ injection experiment was conducted at Oyster, Virginia, with ¹³C-enriched bacteria. The Oyster site has been extensively characterized physically and chemically, and descriptions of the site and the injection experiment are available elsewhere (3). Briefly, ¹³C-enriched PL2W31 bacteria were introduced into the flow field via an injection well over a 3.5-m interval spanning several depositional layers in the aquifer (5.5 to 9.0 m below the ground surface). Approximately 100 liters of bacteria suspended in SGW at a density of 10⁷/ml was introduced at a constant rate over a 30-h interval. Groundwater samples were taken from a series of multilevel samplers by using peristaltic pumps as described previously (3). Samples were collected over the course of 19 days postinjection from the array of 10 multilevel recovery wells, each with 11 sampling points at 0.5-m intervals between 5.5 and 10 m below the ground surface. Total bacteria in the groundwater samples were harvested by centrifugation and processed for isotopic analysis as described above.

	RESULTS AND DISCUSSION

Top Abstract Introduction Materials and Methods Results and Discussion References

We achieved high degrees of ¹³C enrichment for three different bacterial strains (Table 1) by using either labeled ¹³C glucose or ¹³C acetate as the sole source of carbon and energy. These data demonstrate that culturing on highly enriched (98 to 99%) ¹³C substrates results in bacterial cultures with isotopic signatures up to 5 orders of magnitude higher (76,794) than the background signature for microbes from the bacterial transport site (consistently 25 ± 1 based on analysis of multiple samples). Individual cultures varied somewhat in the degree of ¹³C enrichment despite similarity in culturing conditions, presumably reflecting natural variation in the degree of substrate uptake and incorporation into biomass during microbial growth. Variation in the degree of ¹³C enrichment between cultures, however, is not problematic, since a regression model relating isotope values to numbers of bacteria is developed for each experiment.

Our regression models define the relationship between the log of the number of cells per milliliter and log (¹³C + 25) (Fig. 1). Subtraction of the ¹³C value of the background community (25) from the isotope ratio of our samples was necessary because a logarithmic function can be performed only on a positive real number. The correlation coefficient (r²) for the regression model presented was 0.95 (Fig. 1). To estimate the limit of detection for our technique, we considered an increase of 3 above the ¹³C value of the background community in the sample to reflect the lower detection limit for enriched bacteria. In this case, the regression models predict our limit of detection to be ~100 bacterial cells per 50-ml sample. This corresponds to ~2 bacterial cells per ml of groundwater, which, to our knowledge, exceeds the sensitivity levels achieved by any other published bacterial detection strategy for use in environmental samples. For example, lower limits of detection of bacteria in aquifer material or sediments previously reported are 10⁴ to 10⁵ cells/ml for direct microscopic enumeration (8), 10³ cells/ml for selective plating approaches (6; unpublished observations), 10³ to 10⁴ cells/ml for immunological detection (21), and 10³ cells/ml for PCR-based detection (20).

View larger version (12K): [in this window] [in a new window]   FIG. 1.   Representative regression analysis of 13C-enriched bacteria. A separate regression analysis was performed for each experiment and the equation used to determine the number of 13C-enriched cells in a sample based on the 13C value for that sample. The regression shown is for an enrichment of strain PL2W31. Error bars indicate the standard errors of the mean for three replicate samples. The R2 value for the regression analysis is given.

To evaluate our approach, ¹³C-enriched strain PL2W31 was employed in an intact-core experiment, and data obtained from stable-isotope analysis were compared to results obtained by selective plating (on R2A plus nal agar) of the same groundwater eluent samples. The rationale was that, although selective-plating data may be less accurate than those from some other assays due to concerns with culturability under selective conditions, colonies observed on R2A plus nal agar would result from viable PL2W31 cells that had passed through the core. Thus, comparing breakthrough behavior (breakthrough curves) for ¹³C-labeled cells to the selective-plating data allowed us to assess whether ¹³C-labeled cells behave as viable PL2W31 cells do in this experiment. Both tracking methodologies produced similar breakthrough curves, except that the selective-plating data underestimated the number of cells in the sample (Fig. 2). This is consistent with our prior findings that strain PL2W31 had a very low plating efficiency in that less than 1% of the cells detected by direct microscopic enumeration were able to form colonies (data not shown). Thus, the similarity in breakthrough kinetics as determined by the two methods suggests that stable-isotope values are a good predictor of bacterial transport behavior. This is supported by data from a second experiment to validate this approach.

View larger version (20K): [in this window] [in a new window]   FIG. 2.   Comparison of breakthrough curves in an intact-core experiment as determined by selective plating on R2A plus nal medium (nalR) and by the stable isotope detection approach (13C). Sample numbers correspond to 0.1-PV fractions eluted from the intact core.

The next experiment employed strain DA001, a less adhesive and therefore better field transport candidate strain than PL2W31. Strain DA001 was recovered from the study site using a selective screening process (3). The objective of this experiment was to demonstrate that the isotope values of samples obtained from bacterial injection experiments reflect transport of intact bacterial cells and not cell fragments or soluble cell components possibly released by the demise of added bacteria. To accomplish this, an intact-core experiment using SGW was performed to compare the breakthrough behavior of isotopically labeled intact cells and isotopically labeled cell lysate. For this experiment, cells of strain DA001 were enriched in ¹⁴C by growth on [¹⁴C]acetate. Since there is no a priori reason to expect differences in the behavior of cellular macromolecules synthesized using carbon in the form of ¹²C, ¹³C, or ¹⁴C, provided that it is incorporated into bacterial cells in the same chemical form (in this case acetate) and under the same conditions, the ¹⁴C isotope should function as a faithful analog of ¹³C enrichment.

For this experiment, a single culture of DA001 was grown on [¹⁴C]acetate and then split into two equal aliquots, of which one was thoroughly lysed by sonication. The cell lysate and the intact cells were injected into separate intact cores representing the same lithology, and the breakthrough of label was monitored. To determine whether labeled carbon was being released from intact bacterial cells during the course of the experiment, and whether label from lysed cells ended up in the aqueous fraction of the eluent, samples were subjected to filtration using 0.45-µm-pore-size filters and the filter and filtrate were assayed separately. Interestingly, the data indicated that while approximately half of the intact DA001 cells (C/C0 = 0.46) passed through the core, essentially all of the lysed cell material was retained by the sediments (C/C0 = 0.045) (Fig. 3). These data are consistent with prior findings which indicate that soils and sediments bind nucleic acids (5, 23), proteins (12), viruses (4), and presumably other cell-free biological macromolecules quite strongly.

View larger version (40K): [in this window] [in a new window]   FIG. 3.   Results of injection of intact cells versus lysed cell material in an intact-core experiment. (A) Breakthrough of intact cells. Labeled material in the eluent samples captured by the 0.45-µm-pore-size filter (filter counts), labeled material in the eluent samples that passed through the 0.45-µm-pore-size filter (filtrate counts), and total counts in the sample were enumerated. (B) Breakthrough of lysed cell material in an intact-core experiment. Labeled material in the eluent samples captured by the 0.45-µm-pore-size filter (filter counts), labeled material in the eluent samples that passed through the 0.45-µm-pore-size filter (filtrate counts), and total counts in the sample were enumerated. (Each sample was analyzed four times by LSC for 10 min, and values for the standard errors where label was detected were always less than 5%. Note the different y axis scales in panels A and B.

The data for the intact cells also demonstrate that essentially all of the carbon incorporated into cellular biomass is retained by the cells and not released or otherwise lost to the aqueous medium during the course of the experiment, since 99% of the ¹⁴C in those samples was retained by the filter and not released to the filtrate (Fig. 3A). Where the breakthrough curves for intact cells and lysed cell material have common features (i.e., the filter data), the lysed-cell-material values were only 8.7% of the intact-cell values (Fig. 3). Since >99.99% cell lysis was achieved for this experiment, this presumably represents a cellular fraction (e.g., cell wall or membrane fragments) that behaves like intact cells. For the nonparticulate cell material (i.e., recovered in the filtrate), the material that did break through was somewhat delayed relative to the intact cells and represents 57% of the total lysed cell material recovered (Fig. 3B). Since (i) cells isotopically enriched by growth on acetate as the sole source of carbon and energy do not release significant amounts of label, (ii) intact cells are readily transported while lysed cell material is largely retained by the sediment, and (iii) the small proportion of lysed cell material that did break through appeared predominantly in the filtrate, this stable-isotope-monitoring approach appears to reliably and sensitively indicate transport of intact bacterial cells.

This approach was employed in an in situ injection experiment using ¹³C-enriched PL2W31 cells. Representative data showing variation in breakthrough curves downfield of the injection well are shown in Fig. 4. Data obtained from a depth of 7.5 m at samplers 1 and 2 (0.5 and 1.0 m downfield from the injection well, respectively) are presented in Fig. 4A, while data obtained from the 8.5-m depth at these same samplers are presented in Fig. 4B. Comparison of Fig. 4A and B shows that at a single sampler, the breakthrough dynamics at 7.5- and 8.5-m depths were considerably different. By contrast, a comparison of the two curves within a single panel shows that the breakthrough curves at a single depth interval are very similar for both samplers. Although transport dynamics differ between depths, at a single depth interval they are propagated, as evidenced by the shapes of the breakthrough curves downfield. The observed differences in breakthrough kinetics between depths are presumably related to lithological variations in the aquifer sediments (15). The observation that >99.9% of the injected bacteria (in terms of peak concentration at sampler 1) appeared to be retained within the first 0.5 m of sediment in situ is consistent with the findings from the intact-core experiment using this same bacterial strain (Fig. 2).

View larger version (25K): [in this window] [in a new window]   FIG. 4.   Representative data from an in situ bacterial transport experiment on the Eastern Shore of Virginia. (A) Data from a depth of 7.5 m. , data from sampler 1 (0.5 m downfield from the injection well); , data from sampler 2 (1.0 m downfield from the injection well). (B) Data from a depth of 8.5 m at the same two samplers. Same symbols apply.

The bacterial tracking approach described here, which employs stable isotope enrichment of bacteria, offers a sensitive and specific means to monitor introduced bacteria in environmental settings without requiring radioactivity, stains or dyes, genetic alteration, or selective culture methods to facilitate tracking. The ability to culture and subsequently detect stable isotope-enriched bacteria offers a mechanism whereby an unmodified natural population can be identified when introduced to a mixed-species assemblage in the field. This approach has the added advantage of allowing detection of an unmodified indigenous strain that has been introduced to an environment, even if the numbers of added ¹³C-enriched cells in the sample are below the background levels for that same organism in its unenriched form.

The data presented provide (i) a demonstration of the ability to enrich microbes with ¹³C to isotopic signatures 4 to 5 orders of magnitude above the background signature, (ii) regression models relating ¹³C values to numbers of bacteria and low-end sensitivity of detection, (iii) a comparison of results obtained with this approach and by selective plating which detects only culturable intact cells, (iv) a demonstration that the isotopic signature detected accurately represents intact added bacteria that have been transported through the porous medium, and (v) representative field data on in situ bacterial transport derived from detection of ¹³C-enriched bacteria. Isotopically enriched microbes should also prove valuable in other experiments aimed at determining the fate, growth rates, predation, and turnover rates of bacteria in situ.