Comparison of Extracellular Enzyme Activities and Community Composition of Attached and Free-Living Bacteria in Porous Medium Columns

Abstract

Free-living and surface-associated microbial communities in sand-packed columns perfused with groundwater were compared by examination of compositional and functional characteristics. The composition of the microbial communities was assessed by bulk DNA extraction, PCR amplification of 16S ribosomal DNA fragments, separation of these fragments by denaturing gradient gel electrophoresis, and sequence analysis. Community function was assessed by measurement of β-glucosidase and aminopeptidase extracellular enzyme activities. Free-living populations in the aqueous phase exhibited a greater diversity of phylotypes than populations associated with the solid phase. The attached bacterial community displayed significantly greater β-glucosidase and aminopeptidase enzyme activities per volume of porous medium than those of the free-living community. On a per-cell basis, the attached community had a significantly higher cell-specific aminopeptidase enzyme activity (1.07 × 10⁻⁷ nmol cell⁻¹ h⁻¹) than the free-living community (5.02 × 10⁻⁸ nmol cell⁻¹ h⁻¹). Conversely, the free-living community had a significantly higher cell-specific β-glucosidase activity (1.92 × 10⁻⁶ nmol cell⁻¹ h⁻¹) than the surface-associated community (6.08 × 10⁻⁷ nmol cell⁻¹ h⁻¹). The compositional and functional differences observed between these two communities may reflect different roles for these distinct but interacting communities in the decomposition of natural organic matter or biodegradation of xenobiotics in aquifers.

Functional differences between free-living and particle- or macroaggregate-associated bacteria have been documented in the water columns of aquatic (14, 31, 37) and marine (20, 32, 35, 39, 40) environments. Because cell attachment is known to strongly but unpredictably affect cell physiology (29, 41), there has been uncertainty whether these functional differences are due to changes in functional expression or differences in community composition (12, 22, 32, 35). Independent reports of compositional differences in free-living and attached bacteria in marine water columns (1, 5, 12) and two studies in which aspects of both structure and function of these two communities were measured (10, 35) indicate that compositional differences may be responsible (in part) for the observed functional differences.

The current working model in aquatic and marine environments is that attached and unattached bacteria comprise two distinct but interacting communities, shaped by differential access to nutrients and susceptibility to predation (14, 20, 23, 31, 32, 35, 39, 40). For example, bacteria attached to particulate matter suspended in water columns display greater extracellular enzyme activity than their free-living counterparts (20, 22, 31, 32, 35, 39). However, uptake by the attached populations of the products of this extracellular activity appears to be uncoupled from the enzyme activity, creating a supply of usable organic carbon to free-living bacteria in the surrounding water (22, 31, 39).

Differences in rates of polymer hydrolysis, substrate uptake and assimilation, biomass production, and predation for attached and free-living bacteria may result in considerably different roles for these communities in material and energy transfer in aquatic and marine environments. The quantity and quality of organic carbon and the mean size of suspended particles may influence observed partitioning of functions between attached and free-living bacteria (1, 10, 14, 32, 35, 40). Saturated sediments and freshwater aquifers are two-phase environments that differ fundamentally from water columns in the nature of their organic carbon and particle attachment sites. The conceptual model that defines roles for attached and free-living microorganisms in water column biogeochemical cycles has not been extended to aquatic and marine sediments (20, 30) or aquifers.

In aquifer environments, reports suggest compositional and potential functional differences between microorganisms suspended in the groundwater and those attached to the geologic medium (4, 16, 26, 27). Surfaces in aquifers contribute a much larger area per unit volume in aquifers than that associated with particles and macroaggregates suspended in water columns. Groundwater tends to have lower concentrations of natural organic matter, primarily composed of larger, polymeric molecules, than surface waters. The relative importance of attached and free-living aquifer communities to organic matter decomposition has not been investigated. Polymer hydrolysis by extracellular enzymes is generally recognized to be the rate-limiting step in organic material decomposition (9) and should be of particular importance in aquifers, where there is little directly usable organic matter.

We examined β-glucosidase and aminopeptidase activities of attached and free-living communities in sand-packed columns perfused with groundwater. We used laboratory columns to access paired samples of groundwater and porous medium, which are difficult to obtain in a defensible manner from the field, and commence immediate analysis. The objective of the study was to compare β-glucosidase and aminopeptidase activities between attached and free-living aquifer populations and to relate activities to community composition based on 16S ribosomal DNA (rDNA) profiles and sequence analysis. The effect of trichloroethylene (TCE) treatment on these measures was assessed. This study initiates the extension of the conceptual model for partitioning of energy and material cycling between attached and free-living bacteria in water columns to aquifer environments.

MATERIALS AND METHODS

General experimental design and statistical analyses.

The structure and function of attached and free-living microbial communities were examined using sand-packed columns perfused with groundwater. Two treatment groups, each consisting of triplicate columns, were established: (i) those that received addition of ca. 10 mg of TCE liter⁻¹ in the groundwater (treated) and (ii) those that received no addition of TCE (control). Following a period of equilibration and TCE exposure (see following section), sand and groundwater samples were analyzed for extracellular enzyme activities and banding patterns and sequence analysis of 16S rDNA amplification fragments.

Effects due to TCE treatment were tested independently on attached and free-living communities using a one-way analysis of variance with significant differences defined at the P = 0.05 level. Comparisons of enzyme activity and number of 16S rDNA bands between attached and free-living communities were performed with a two-tailed, paired Student's t test, with significant differences defined at the P = 0.05 level. Values of measured variables are reported in the text as means ± 1 standard deviation.

Column construction and operation.

Glass columns (5.03-cm internal diameter by 28.58 cm long; Ace Glass, Inc., Vineland, N.J.) fitted with nylon endplates, Teflon O-rings, and screened outlet ports were packed with 2-cm lifts of graded, clean quartz sand (particle diameter, 0.60 to 0.85 mm; 99.8% silicon dioxide; ASTM 20/30 sand [unground silica]; U.S. Silica, Berkeley Springs, W.Va.) mixed with filter-sterilized (pore size, 0.2 μm) groundwater from well USGS-103 (see description below). Sand was previously rinsed in distilled water to remove fine particles and sterilized by autoclaving three times at 121°C for 60 min. Polyethylene tubing (1/8-in. [ca. 0.3-cm] diameter), nylon, and stainless steel fittings were used to connect the columns to a peristaltic pump, source reservoir, and waste container (Fig. 1). Column components were sterilized by either autoclaving or surface disinfection with 70% ethanol or dilute bleach solution, followed by thorough rinsing with sterile, deionized water.

FIG. 1.

Schematic relating the experimental set-up. Eleven days after onset of treatment, groundwater samples from column effluents were collected, and sand samples were collected by destruction of the columns. PEG, polyethylene glycol.

During packing, appropriate weight and dimensional measurements were made to obtain porosity and bulk density values. After the initial packing and saturation, the column feed was changed to unfiltered groundwater collected from the eastern Snake River Plain aquifer. Groundwater in the eastern Snake River Plain aquifer is described as a calcium-sodium-bicarbonate type that is slightly alkaline (ca. pH 8), oligotrophic, and saturated with dissolved oxygen (42). The groundwater was collected (following standard three-well-volume purging) from a 213-m depth in an uncased well (USGS-103) located in a pristine location where total organic carbon is about 0.3 mg liter⁻¹ (R. Bartholomay, U.S. Geological Survey [USGS], personal communication). Columns were operated in the up-flow mode at a nominal flow velocity of 1.5 m day⁻¹, resulting in an approximate discharge rate of 0.75 ml min⁻¹ per column.

The groundwater source reservoir was maintained at 4°C to minimize enrichment effects. Backpressure (ca. 22 lb/in²) was applied to the outlet side of the column system using backpressure valves (Swagelok Inc., Solon, Ohio) to reduce degassing of the groundwater as it warmed from 4°C in the reservoir to the 22°C operating temperature of the columns. Groundwater was pumped through the columns for 15 days before TCE was added to the treatment group. Concentrated (1,000 mg liter⁻¹) TCE solution in filtered groundwater was pumped into the columns from composite bags (2-L Cali-5-Bond bags; Calibrated Instruments Inc., Hawthorne, N.Y.) to achieve a dissolved TCE concentration of about 10 mg liter⁻¹. The control columns were similarly operated with additional inflow containing only filtered groundwater and no TCE.

Following a subsequent time period of 11 days, groundwater samples for microbiological analyses of unattached organisms were collected from the effluent tubing using sterile polypropylene centrifuge tubes placed on ice. Residual water was removed from the columns by a gentle stream of filtered nitrogen gas, and composite samples for attached organisms were collected by combining subsamples taken from the top, middle, and lower parts of the column. Enzyme analyses and DNA recovery and storage commenced immediately upon column disassembly.

The stability of the experiment in progress was assessed by daily monitoring of column flow rates and outlet pressures and periodic enumeration of total cells in column effluents. Stabilization of cell counts in the column effluents during the experiment was used to indicate the development of a steady state between cell attachment and detachment (excluding consideration of significant cell growth in this oligotrophic system). The two periods of equilibration (following initial column saturation and following treatment) were comparable to equilibration times (based on cell densities in column effluents) observed in two preliminary studies and to that found by Lehman et al. (26), who also examined community carbon source utilization profiles of column effluents.

Percent moisture determination of sand.

Sand moisture content was measured by standard gravimetric methods on a composite sample from each column.

TCE analysis.

TCE analyses were performed on extracts (solid-phase microextraction technique; Supelco Inc., Bellefonte, Pa.) of groundwater and sand composite samples collected in volatile organic carbon sample vials. Analyses were performed using an HP5890 Series II gas chromatograph (Hewlett Packard, Inc., Boise, Idaho) equipped with a flame ionization detector and an Rtx-624 column (Restek Corp., Bellefonte, Pa.).

Total cell counts.

Samples of groundwater and sand were fixed in 2% formalin solution (final concentration) and refrigerated (4°C) prior to analysis (24). Aliquots of the fixed samples were filtered under vacuum onto 0.2-μm-pore-size, black polycarbonate membrane filters with cellulose-acetate support filters. The total number of cells was enumerated by direct counts of acridine orange-stained (0.01%, 2 min) cells (19) using epifluorescent illumination on a Nikon E-600 light microscope equipped with a xenon lamp and a Nikon EF-4 B-2E/C filter cube (Nikon Inc., Melville, N.Y.). A minimum of 10 fields each containing 200 cells were counted on each filter. Cell counts were expressed on a milliliter basis for the groundwater samples and on a gram dry weight basis for the sand samples. The free-living and sand-associated cell densities were converted to cubic centimeters of porous medium using porosity and bulk density measurements performed during the construction of the columns.

Extracellular enzyme activities.

β-Glucosidase and aminopeptidase activities were estimated by assay with the fluorogenic substrates 4-methylumbelliferyl-β-d-glucoside (MUF-glu; Sigma M3633; Sigma-Aldrich, St. Louis, Mo.) and l-leucine-4-methylcoumarinyl-7-amide hydrochloride (MCA-leu; Sigma L2145), respectively. The assays were performed as described by Hoppe (21) at saturating levels of added substrate (250 μM) to estimate maximum velocities of substrate hydrolysis. For groundwater samples, substrates were added directly to aliquots of groundwater, and for sand samples, substrates were added to a 1:3 dilution (wt/vol) of sand in filtered, deionized water. Incubations were performed in 50-ml polypropylene tubes with shaking (150 rpm) at 22°C in the dark. The enzyme assay solution volume was adjusted to permit 3-ml volumes (excluding particles for the sand samples) to be aseptically removed for periodic fluorimetric (Hitachi F-2000 fluorimeter, Hitachi LTD., Tokyo, Japan) analysis over 48 h (β-glucosidase) or 72 h (aminopeptidase) in quartz cuvettes.

The pH of each subsample was adjusted to 10.3 by adding 0.6 ml of Tris-HCl (pH 10.3) prior to fluorescence measurement. The following controls were performed for each substrate: labeled substrate was added to filtered, deionized water (abiotic control), and labeled substrate was added to autoclave-sterilized samples of the groundwater and sand (sterile control). No extraction of the fluorochrome or centrifugation of the particles was necessary due to the lack of fluorochrome sorption by the sand and rapid settling of the sand particles by gravity. Standard curves for each fluorochrome were constructed using an appropriate range of known concentrations added to the respective sample matrices, with modification of pH to 10.3 before reading. Plots of fluorescence intensity over time were constructed, and the initial rates of hydrolysis (fluorescence units per hour) were calculated from the linear portion of these curves. Fluorescence units were converted to nanomoles of substrate based on a linear regression of the standard curve. Hydrolysis rates were expressed per milliliter or dry gram of sample, per cubic centimeter of porous medium, and per cell (cell-specific activity).

DNA extraction.

For nucleic acid analysis of free-living organisms, approximately 950 ml of effluent groundwater from each column was collected on ice in a 1-liter glass bottle and then immediately filtered onto a 47-mm hydrophobic (previously made hydrophilic by washing with ethanol) Durapore filter (Millipore Corp., Bedford, Mass.), which was frozen at −70°C. Each filter was placed in a bead-beating tube with 1.5 g of 0.1-mm zircon beads (BioSpec Products, Inc., Bartlesville, Okla.), 600 μl of 24:1 chloroform-isoamyl alcohol, and 800 μl of lysis solution (0.1 M NaCl, 10 mM EDTA, 10 mM Tris [pH 8.0], 2% sodium dodecyl sulfate), mixed using a BioSpec bead beater (2,500 rpm, 5 min), and iced, and the process was repeated.

Following deposition of the beads by pulse centrifugation, the filter was aseptically removed from the tube while the filter surface was washed with 400 μl of ice-cold Tris-EDTA (TE; 20 mM Tris, 2 mM EDTA, pH 8.0). The tube was centrifuged (15,000 × g, 2 min), and the supernatant was transferred to a new tube. The beads were then reextracted with 800 μl of cold TE, and the supernatants from the two extractions were combined. The homogenized supernatants were split into two tubes, and the aqueous phases were extracted with equal volumes of cold phenol-chloroform-isoamyl alcohol (25:24:1), hand-mixed by inversion for 2 min, and centrifuged (15,000 × g, 5 min). The aqueous phases were transferred to new tubes and extracted with 24:1 chloroform-isoamyl alcohol, inversion mixed, and centrifuged. The aqueous phases were collected, extracted with 2-butanol (37), and combined, and their volume was reduced with butanol to 500 μl. Ice-cold 100% ethanol (1 ml) was added to the extracts, which were held at −20°C overnight.

DNA was pelleted by centrifugation (15,000 × g, 20 min) at 4°C, the supernatant was removed, and the pellet (usually invisible) was dried for 4 h in a laminar flow hood. The dried pellet was resuspended in 50 μl of 10% TE (2 mM Tris, 0.2 mM EDTA, pH 8.0), and 10 μl was run on a 1% agarose gel to examine DNA quality and quantity. Sephadex G50 spin columns (Amersham Pharmacia Biotech, Piscataway, N.J.) were used to remove an unidentified PCR inhibitor that was coextracted with the DNA in pilot studies.

For nucleic acid analysis of sand-associated organisms, composite sand samples (ca. 75 g each) were extracted by a physical-chemical extraction method designed for large masses (43). The method was employed with a few modifications, including use of a stainless steel blender head and blender instead of the bead beater used by the authors, final resuspension of the DNA pellet in 1% TE (0.2 mM Tris, 0.02 mM EDTA, pH 8.0), and reduction of the aqueous volume to 50 μl using butanol extractions (36). The 50 μl of DNA solution was passed through a Mo Bio spin filter (Mo Bio Laboratories, Inc., Solana Beach, Calif.) and processed using steps 12 to 20 of the protocol. This last step was employed to remove trace concentrations of PCR inhibitors (e.g., EDTA, phenol, and alcohols) from the DNA solution.

PCR-DGGE.

The primers used in the PCR amplification were 16S rDNA universal bacterial primers 341F with a GC clamp and 907R (7). A touchdown approach modified from that of Muyzer et al. (33) was employed, consisting of 5 min at 94°C following the hot start, then 29 cycles with denaturation at 94°C for 1 min, annealing for 1 min at various temperatures, and extension for 3 min at 72°C. The annealing temperature in the first round was 65°C, followed by two rounds each at 1°C lower than the previous round, and finally ten rounds at 55°C. A final extension for 7 min at 72°C was employed. The total reaction mixture consisted of 100 μl with the following ingredients: 0.05% IgePal (Sigma, St. Louis, Mo.), 63.5 or 72.5 μl of nuclease-free water (the former for sand samples, the latter for water samples), 1× PCR buffer, 1.5 mM MgCl₂, 5 U of Taq DNA polymerase (Promega Corporation, Madison, Wis.), 0.25 μM each primer (Operon Technologies, Alameda, Calif.), 0.25 mM deoxynucleoside triphosphates (dNTP; Boehringer-Mannheim/Roche Molecular Biochemicals, Indianapolis, Ind.), and 1 and 10 μl of DNA template for water and sand samples, respectively.

Denaturing gradient gel electrophoresis (DGGE) analyses were conducted using 25 μl of sand and water bacterial PCR products loaded into a 30 to 70% and a 40 to 55% urea-formamide-polyacrylamide gel. A Bio-Rad DCode universal mutation detection system (Bio-Rad Laboratories, Hercules, Calif.) was run at 65 V for 15 h at 60°C to separate the fragments. Gels were stained with ethidium bromide for 30 min, destained for 10 min, and photographed under UV illumination using an Alpha Innotech MultiImage II system (San Leandro, Calif.).

Sequencing and taxonomic affiliations.

Discrete bands were excised from the DGGE gels with sterile razor blades and placed into 100 μl of 10% TE in a 2-ml bead beater tube. The bands were spun into the tubes and incubated overnight at 37°C while slowly shaking at 70 rpm. A 1-μl aliquot was removed from each tube and used in PCR as above with no cell lysis or hot start step (i.e., all PCR reagents mixed and thermal cycler program commenced at the initial 94°C denaturation). PCR was performed using the same methods as above for the original samples, and PCR products were purified with Wizard kits (Promega Corporation, Madison, Wis.). Sequencing was performed by the DNA Sequencing and Synthesizing Facility at Washington State University. Sequence analyses were conducted for products from both forward and reverse primers using the BLAST (3) and RDP II (28) Internet tools.

RESULTS

The bulk density of the packed columns was calculated to be 1.69 ± 0.02 dry g cm⁻³ (n = 4), with a porosity of 0.36% ± 0.01% (n = 4). Dissolved TCE concentrations in the control and TCE-treated columns were <0.02 mg liter⁻¹ and 11.4 ± 3.41 mg liter⁻¹, respectively. There was no significant TCE treatment effect on cell densities of the groundwater effluents (P = 0.255, n = 3) or of the sands (P = 0.308, n = 3). When data from both the control and TCE-treated groups were combined, the cell densities were 3.97 × 10⁵ ± 1.15 × 10⁵ cells ml⁻¹ (n = 6) for the column effluent groundwater and 4.47 × 10⁶ ± 3.77 × 10⁵ cells (dry g)⁻¹ (n = 6) for the sand. The total cell densities were significantly greater for the sand, on a dry gram basis, than the water (P < 0.001, n = 6), and the difference increased when the cell densities were expressed per cubic centimeter of porous medium (groundwater, 1.43 × 10⁵ cells cm⁻³; sand, 7.51 × 10⁶ cells cm⁻³).

Negligible β-glucosidase and aminopeptidase activities were detected in the abiotic and sterilized controls. Plots of fluorescence intensity for the β-glucosidase and aminopeptidase activity assays in groundwater and sand had a sigmoidal shape, with a period of low or undetectable activity, a period of linear increase, and a period of diminishing or no increase (Fig. 2A and B). β-Glucosidase extracellular enzyme activity averaged 0.656 nmol ml⁻¹ h⁻¹ for the groundwater and 2.72 nmol (dry g)⁻¹ h⁻¹ for the sand (n = 6; columns from both treatments were combined for paired comparisons after no significant TCE treatment effects on β-glucosidase activity were found for either groundwater [P = 0.972, n = 3] or sand [P = 0.661, n = 3]) (Fig. 3).

FIG. 2.

Development of fluorescence related to β-glucosidase (A) and aminopeptidase (B) activity over 72 h (β-glucosidase) or 138 h (aminopeptidase) of incubation of sand and groundwater samples from independent replicate columns. Error bars represent 1 standard deviation (n = 6).

FIG. 3.

β-Glucosidase hydrolysis rates by attached and unattached communities expressed in three different ways. Error bars represent 1 standard deviation (n = 6).

Aminopeptidase activity averaged 0.0196 nmol ml⁻¹ h⁻¹ for the groundwater and 0.473 nmol (dry g)⁻¹ h⁻¹ for the sand (n = 6; columns from both treatments were combined for paired comparisons after no significant TCE treatment effect on aminopeptidase activity was found for either groundwater [P = 0.273, n = 3] or sand [P = 0.307, n = 3]) (Fig. 4).

FIG. 4.

Aminopeptidase hydrolysis rates by attached and unattached communities expressed in three different ways. Error bars represent 1 standard deviation (n = 6).

Using values for bulk density and porosity, activities were converted to express activity per cubic centimeter of porous medium. Sand had significantly higher (P < 0.001, n = 6) values for both β-glucosidase and aminopeptidase activity than groundwater when these activities were expressed either as mass or volume of porous medium. On a cell-specific basis, aminopeptidase activity was also significantly higher (P = 0.01, n = 6) for the sand-associated organisms (1.07 × 10⁻⁷ nmol cell⁻¹ h⁻¹) than for those suspended in the groundwater (5.02 × 10⁻⁸ nmol cell⁻¹ h⁻¹). In contrast, cell-specific β-glucosidase activity was significantly higher (P < 0.001, n = 6) for the unattached bacteria suspended in the groundwater (1.92 × 10⁻⁶ nmol cell⁻¹ h⁻¹) than for the bacteria associated with sand (6.08 × 10⁻⁷ nmol cell⁻¹ h⁻¹). Nanomoles of substrate can be converted to nanograms of carbon using the factor 72 ng of C nmol⁻¹, which relates the carbon content of the organic moiety to the concentration of the fluorochrome (21).

The ratio of aminopeptidase to β-glucosidase activities averaged 0.033 ± 0.012 for the groundwaters and 0.191 ± 0.07 for the sands. The aminopeptidase/β-glucosidase ratio was significantly higher (P = 0.002, n = 6) for the sand samples than for the groundwater samples (the ratio was invariant with respect to the activity units). The consistently higher aminopeptidase activities and lower β-glucosidase activities for the attached organisms than in the unattached organisms could also be seen by examination of water/sand ratios for the two activities. The water/sand ratios were approximately 5.5 times higher for β-glucosidase than for aminopeptidase activities, a difference that was significant regardless of the units used (P < 0.005, n = 6) (data not shown).

An average of 13.3 ± 0.8 distinct bands and 5.6 ± 1.5 bands was obtained from groundwater and sand samples, respectively (Fig. 5). The difference in the number of bands was significant (P < 0.001, n = 6). Most of the bands derived from populations attached to sand migrated similarly to those derived from the groundwater samples. Forty-four of the 47 bands identified based on differential migration patterns were successfully excised from the gels. The 16S rDNA bacterial sequences from the sand and groundwater samples fell within the α-, β- and γ-subclasses of the Proteobacteria and the Flexibacter-Cytophaga-Bacteroides groups (Table 1).

FIG. 5.

Digital image of DGGE gel showing 16S rDNA fragment migration patterns for sand samples with no TCE treatment (lanes 1 to 3), sand samples receiving TCE treatment (lanes 4 to 6), groundwater samples with no TCE treatment (lanes 7 to 9), and groundwater samples receiving TCE treatment (lanes 10 to 12). Bands that were successfully excised from the gel and sequenced are numbered; these numbers correspond to those used in Table 1, where the taxonomic affiliations of the sequences are reported.

TABLE 1.

Sequence affiliation of numbered bands shown in Fig. 5

Band	Samplea	Phylogenetic group	Highest sequence matchb
1	Water	β-Proteobacteria	Thiobacillus aquaesulis (0.399)
2	Both	Flexibacter-Cytophaga-Bacteroides	Cytophaga sp. strain 50.n (AB013834) (0.600)
3	Water	γ-Proteobacteria	Coxiella burnetti subgroup (AB015255) (0.426)
4	Water	β-Proteobacteria	Hydrogenophaga palleronii (0.848)
5	Both	Cyanobacteria and chloroplasts	Olisthodicus luteus (chloroplast) (0.563)
6	Water	γ-Proteobacteria	Legionella beliardensis strain MontbeliardA1 (0.834)
7	Water	α-Proteobacteria	Afipia broomeae strain G5958 (0.533)
8	Water	β-Proteobacteria	Azoarcus sp. strain BH72 (AF011344) (0.753)
9	Both	β-Proteobacteria	Hydrogenophaga palleronii (1.000)
10	Both	α-Proteobacteria	Azospirillum sp. strain ASP-1 (0.422)
11	Water	β-Proteobacteria	Hydrogenophaga palleronii (0.918)

^aSample type from which sequence was retrieved, water or both water and sand.

^bSequence in the RDP database with the highest match to retrieved sequences, with the similarity index indicated in parentheses after the accession number, where available.

The 11 numbered bands with sequence affiliations include duplicate bands selected from the gels but exclude DNA retrieved from some bands that was unsuitable for sequence analysis and sequences which appeared unreliable (>60 wobbles per 600 bp). Considering only the named bands, seven were unique to groundwater samples and none were unique to the sand samples. With the exception of band 1, which represented a free-living population from one of the three columns receiving TCE, there was no difference in the occurrence of bands due to TCE.

DISCUSSION

In analysis of laboratory columns packed with sand and perfused with groundwater, it was found that (i) attached and free-living communities both actively produced aminopeptidase and β-glucosidase extracellular enzymes, (ii) community aminopeptidase and β-glucosidase activities were higher in attached communities than in free-living communities, (iii) cell-specific β-glucosidase activity was higher for free-living communities while cell-specific aminopeptidase activity was higher for attached communities, (iv) the aminopeptidase/β-glucosidase activity ratio was higher for attached communities than for free-living communities, (v) greater numbers of and unique 16S rDNA sequences were recovered from free-living versus attached communities, and (vi) there was no significant effect of TCE on enzyme activity or population structure.

Free-living bacteria in the columns had generally higher β-glucosidase activity and lower aminopeptidase activity than water column-suspended bacteria (8, 31, 32, 35, 38). The enzyme activity levels displayed by free-living bacteria in the columns indicate that planktonic aquifer organisms may not be dead or compromised cells, as suggested by some authors (2, 15, 17, 18, 34). Enzyme activities for attached bacteria in the columns were in the low range of values for marine sediments (13), equivalent to activities in lake sediments (6) and generally higher than values reported for particle-associated bacteria (not macroaggregates) suspended in the water columns of freshwater and marine samples (31, 32, 35).

As in other environments (22, 31, 32, 35, 39), we found significantly higher extracellular enzyme activities on a per mass or per volume basis for the attached aquifer communities than for the unattached communities. The greater volumetric activity of attached aquifer communities incorporates both the relative number of cells per unit of mass (attached and unattached) and the percentage of volume that is occupied by solids versus water and has general implications for biotransformations in the subsurface.

Both unattached and attached aquifer organisms expressed overall higher β-glucosidase activities than aminopeptidase activities; the ratios of aminopeptidase to β-glucosidase activity averaged 0.033 for the groundwaters and 0.191 for the sands in our study. This observation contrasts with the ratio of these two enzyme activities (>1.0) that can be calculated from studies that reported both enzyme activities (8, 31, 32, 35, 38). Fabiano and Danovaro 13) reported an aminopeptidase/β-glucosidase activity ratio of approximately 10 in marine sediments, which they describe as being very low.

The low ratios of aminopeptidase to β-glucosidase activity observed for both aquifer communities may indicate differences in the quantity and quality of organic carbon substrates in the subsurface. Suspended aquifer bacteria had consistently lower aminopeptidase/β-glucosidase activity ratios and significantly higher cell-specific β-glucosidase activities than the attached bacteria. The higher cell-specific β-glucosidase activity for suspended bacteria reinforces their potential significance in biogeochemical transformations in aquifers.

The apparent propensity for unattached bacteria to have higher hydrolysis rates for polysaccharides and lower hydrolysis rates for polypeptides than their attached counterparts suggests specific roles for these two communities in the decomposition of classes of carbon compounds. We have observed a similar affinity for carbohydrates by the unattached communities and for amino acids by the attached communities when measuring catabolic potentials for these communities in laboratory microcosms (26) and with colonization devices incubated in an aquifer (F. S. Colwell and R. M. Lehman, unpublished observations). There are few published data for any environment to interpret this apparent substrate preference, and its significance for biogeochemical cycling in the subsurface remains undetermined.

It should be noted that the extracellular activities were measured with saturating substrate concentrations incubated at 22°C compared to incubations with lower substrate concentrations conducted at lower temperatures. Our incubation times were relatively long, with the period between 24 and 72 h generally reflecting linear development of fluorescence. This comparatively long incubation time reflects activity potentials but also allows the activity of the unattached and attached bacteria to be measured, rather than that of previously produced enzymes that may have partitioned between dissolved and particulate phases.

The 16S rDNA banding patterns and sequence identity suggest compositional differences between the two communities that may be responsible for differences in extracellular enzyme activities. A greater number of Bacteria sequences were retrieved from the groundwater than the sand, and there were several sequences unique to the free-living community (Fig. 5, Table 1). Similarly, a greater number of Archaea sequences (ca. 5 per column) were recovered from the groundwater than from the sand (<1 per column) (data not shown). These findings corroborate compositional differences between attached and unattached communities observed in studies of aquifers (4, 16, 27) and in freshwater (10, 11) and marine water (1, 5, 10, 12, 35) columns.

Relationships between community function and community structure have been observed in a series of ponds (25) and in seawater mesocosms (35). The unique sequences found in the groundwater reinforce the notion that suspended populations may have unique functions in the subsurface and that they are not inactive or injured subsets of the attached populations. An equitable approach was attempted to facilitate comparison of attached and free-living communities; however, it was necessary to use different DNA extraction approaches on the sand and the groundwater. The total numbers of cells that were subjected to DNA extraction were similar for groundwater and sand (ca. 10⁸ cells), but the different matrices represent different extraction difficulties.

Larger quantities of DNA template were used for PCR on the sand samples to overcome the lower quantity of relatively sheared DNA that was extracted from sand than from groundwater. Despite the known presence of gram-positive organisms in groundwater from the Snake River Plain aquifer (26) and the use of bead-beating methods, no gram-positive sequences were obtained in this study. Therefore, the efficiency of the DNA extraction and subsequent bias in the PCR step may have contributed to the compositional differences observed between the attached and unattached communities. Thus, it remains possible that the differences in extracellular enzyme activity may be related to differences in functional expression.

These activity measurements were made on attached and free-living communities in packed sand columns perfused with groundwater and not on authentic depth-paired samples taken from an aquifer. However, this investigation represents the first attempt at comparing attached and free-living aquifer communities with respect to their role in organic matter decomposition. It appears that attached and free-living aquifer communities are two distinct communities that possess complex interactions with respect to population exchange and functional role. Obviously one community cannot be considered without the other, and the partitioning of organisms and functions must be described as tendencies that respect the source of all subsurface organisms (original sediment deposition or subsequent transport from elsewhere).

The concept of two interacting communities with different roles (i.e., polymer hydrolysis), regardless of the source of origin of the cells, is the current working model in aquatic and marine environments (14, 20, 23, 31, 32, 35, 39, 40). In addition to study of appropriately matched field samples of groundwater and geologic medium, measurements of carbon uptake, assimilation and mineralization are required to outline a conceptual model for the roles of attached and unattached aquifer communities in carbon cycling and other biotransformations.