Input of Protein to Lake Water Microcosms Affects Expression of Proteolytic Enzymes and the Dynamics of Pseudomonas spp.

Abstract

The objective of this study was to determine how an input of protein to lake water affects expression of a proteolytic potential and influences the abundance and composition of a specific group of bacteria. Pseudomonas spp. were chosen as a target group that can be recovered on selective growth media and contain both proteolytic and nonproteolytic strains. Amendment with 2 mg of casein per liter increased total proteinase activity (hydrolysis of [³H]casein) by 74%, leucine-aminopeptidase activity (hydrolysis of leucine-methyl-coumarinylamide) by 133%, bacterial abundance by 44%, and phytoplankton biomass (chlorophyll a) by 39%. The casein amendment also increased the abundance of culturable Pseudomonas spp. by fivefold relative to control microcosms but did not select for proteolytic isolates. Soluble proteins immunochemically related to the Pseudomonas fluorescens alkaline proteinase, AprX, were detected in amended microcosms but not in the controls. The expression of this class of proteinase was confirmed exclusively for proteolytic Pseudomonas isolates from the microcosms. The population structure of Pseudomonas isolates was determined from genomic fingerprints generated by universally primed PCR, and the analysis indicated that casein amendment led to only minor shifts in population structure. The appearance of AprX-like proteinases in the lake water might thus reflect a general induction of enzyme expression rather than pronounced shifts in the Pseudomonas population structure. The limited effect of casein amendment on Pseudomonas population structure might be due to the availability of casein hydrolysates to bacteria independent of their proteinase expression. In the lake water, 44% of the total proteinase activity was recovered in 0.22-μm-pore-size filtrates and thus without a direct association with the bacteria providing the extracellular enzyme activity. Since all Pseudomonas isolates expressed leucine-aminopeptidase in pure culture, proteolytic as well as nonproteolytic pseudomonads were likely members of the bacterial consortium that metabolized protein in the lake water.

Bacterial growth in pelagic ecosystems is supported by a complex mixture of organic compounds (43), among which proteins appear to be important (7, 27, 40). Bacteria cannot assimilate proteins directly (37) but depend on extracellular and/or cell-associated enzyme systems to liberate protein-bound amino acids for assimilation and metabolic processes (5). During a so-called proteolytic cascade, proteins are broken into smaller fragments by proteinase enzymes (endopeptidases), and these peptides serve as substrates for exopeptidases (e.g., aminopeptidase) with affinities to release terminal amino acids (28). Extracellular proteinase activity is thus important for the initial cleavage of proteins.

Addition of protein to sea or lake water stimulates bacterial growth and leucine-aminopeptidase (LAP) activity (13, 38, 50). Following an input of protein, Pinhassi et al. (38) found that five populations of bacteria proliferated, while the abundance of another 10 populations was more stable, as evident from a whole-genome hybridization between environmental DNA and DNA from pure cultures. The authors proposed a link between the shift in the structure and function of the community. However, they also recognized that proteolytic activity should be traced directly to the enzyme-producing populations to prove causal relationships. As yet, no studies have done that for proteolytic enzymes in aquatic environments.

Pseudomonads are found in many aquatic ecosystems by both culture-dependent (11, 12, 17) and culture-independent techniques (9, 10, 16, 25, 39). Pseudomonads are known as early colonizers of “new” habitats, such as developing root systems and food products, indicating an opportunistic growth strategy in response to available nutrient resources. In general, they are also easy to culture on nutrient-rich agar media. The genus Pseudomonas comprises both proteolytic and nonproteolytic strains (53). Several proteolytic strains are well characterized due, e.g., to their deterioration of milk (8) and meat products (29). Several proteinase enzymes have been characterized (14), and antibodies have been raised to some of them (2, 33, 47). The above properties make Pseudomonas an attractive target group for studies that address how protein amendment can affect the expression of a proteolytic potential and influence the dynamics and composition of specific bacterial populations.

In this study we aim to determine how protein amendment affects expression of a proteolytic potential and influences the abundance and population structure of Pseudomonas spp. in lake water. Microbial dynamics (direct and culture-dependent estimates of microbial abundance) and enzyme activities (proteinase and LAP) were followed in lake water microcosms. The abundance of Pseudomonas spp. was followed specifically (Gould S1 agar), and expression of the AprX-like Pseudomonas proteinase was detected immunochemically in the microcosms. The population structure of Pseudomonas was examined using genomic fingerprints generated by universally primed PCR (UP-PCR).

MATERIALS AND METHODS

Sampling.

On 3 April 2000, water was collected from mesotrophic Lake Esrum, Denmark (23). Within 2 h, microbiological analyses (see below) were initiated in the laboratory, and subsequently microcosms of 2.5 liters of lake water were established. Duplicate microcosms were amended with 2 mg of casein sodium salt (Sigma, St. Louis, Mo.) per liter to increase the pool of biodegradable dissolved organic matter by approximately 2.5-fold relative to the natural level (44), assuming a C:N ratio of 5 for casein. The casein stock solution had been dialyzed against water from the Milli-Q purification system (Millipore Corporation, Bedford, Mass.) to remove eventual low-molecular-weight compounds. Another two unamended microcosms served as controls. The microcosms were incubated at 15°C in a 16-h light–8-h dark cycle on a shaker at 100 rpm and sampled daily during the following 4 days.

Abundance of phytoplankton and bacteria.

Phytoplankton were collected on Whatman GF/C filters and frozen. Chlorophyll a was extracted with 96% ethanol and measured spectrophotometrically to indicate the phytoplankton biomass (20).

Samples for direct counts of bacteria were stained for 15 min with the DNA-binding fluorophore SYBR green (Molecular Probes, Leiden, The Netherlands), fixed in 2% buffered formaldehyde (final concentration), and stored at 5°C for less than 2 weeks. Bacteria were counted with a FacsCalibur flow cytometer (Becton Dickinson, Brøndby, Denmark) using a fixed concentration of 2-μm-diameter fluorescent beads (Molecular Probes) as internal standards for the volumes analyzed.

Proteinase and aminopeptidase activity.

Proteinase activity (PRTase) was measured as the turnover of ³H-methylated casein (% h⁻¹) according to Keil and Kirchman (24). ³H-methylated casein (0.5 μCi μg⁻¹) was generated by reductive methylation (48) of dialyzed (0.2 M borate buffer [pH 8.9]) casein sodium salt (Sigma) with 32 Ci mmol of B³H₄⁻¹ (Amersham Pharmacia Biotech., Little Chalfont, England) and formaldehyde (Sigma). Subsequently, protein-bound ³H was rinsed by several washes in 10-kDa-cutoff Microcon microconcentrators (Amicon, Beverly, Calif.). Triplicate samples of lake water were amended with ³H-methylated casein (ca. 10 μg of C liter⁻¹) and incubated for 3 h at 20°C. Incubations were stopped by additions of nonlabeled casein (2 g liter⁻¹) and trichloroacetic acid (TCA) (5% final concentration) to precipitate unhydrolyzed protein on ice for >20 min. TCA-soluble radioactivity (protein fragments and other degradation products) was separated from precipitated protein by centrifugation (10,000 × g, 10 min, 5°C) and quantified with a Beckman LS 1801 liquid scintillation counter (Beckman Instruments, Inc., Fullerton, Calif.) by use of Lumasafe (Lumac LSC B.V., Groningen, The Netherlands). Parallel blank samples were incubated for either <1 min with lake water or 3 h with autoclaved Milli-Q water. These two approaches gave similar background levels of radioactivity in the assay.

Coefficients of variation averaged 3.4% (standard deviation [SD] = 2.2%) for triplicate samples or blanks. Corrected for blank values, radioactivity in supernatants was divided by total radioactivity added and normalized to incubation time. No corrections were done for bacterial assimilation of ³H in accordance with Keil and Kirchman (24). The turnover rate (% h⁻¹) measured under these conditions was constant during an incubation of at least 4 h (data not shown).

LAP activity was measured with l-leucine-4-methyl-coumarinylamide hydrochloride (ICN, Costa Mesa, Calif.) (Leu-MCA). Cleavage of the peptide bond between Leu and MCA releases fluorescent 7-amino-4-methylcoumarin (AMC) with proportional increases in fluorescence, as described by Hoppe (19). Samples were amended with 0.25 mM Leu-MCA and incubated in the dark at 20°C. Fluorescence was read with a Kontron SFM 25 fluorometer (Kontron AG, Zürich, Switzerland) and calibrated against standard solutions of free AMC (ICN). The formation of hydrolysates was normalized to the time of incubation to express hydrolysis rate as micromolar per hour. No increase in fluorescence was observed in parallel blank samples of autoclaved Milli-Q water.

Immunochemical detection of AprX-like Pseudomonas proteinase.

To test whether Pseudomonas contributed to the proteolytic activity in the microcosms, proteins in the water were concentrated, and the presence of the alkaline Pseudomonas proteinase (AprX) was estimated immunochemically. Polyclonal rabbit immunoglobulin G (IgG) (antibody) raised toward a purified 46-kDa proteinase of P. fluorescens UP206 was provided by A. C. Magee and D. A. McDowell, University of Ulster, United Kingdom (33).

The specificity of the antibody was tested by Western blot analysis of proteolytic culture supernatants of the following strains: P. fluorescens strain ON2 (54), P. aeruginosa DSM 50071, Erwinia chrysanthemi DSM 4610, Serratia marcescens ATCC 990, and Proteus mirabilis S31F2 (Department of Veterinary Microbiology, The Royal Veterinary and Agricultural University, Frederiksberg, Denmark). These bacteria produce proteinases homologous to the AprX proteinase of P. fluorescens CY091 (31, 54). Western blot analysis was also performed with supernatants from selected proteolytic and nonproteolytic supernatants from Pseudomonas isolates from Lake Esrum. Nonproteolytic control strains included P. fluorescens DF57 (45), P. fluorescens ON2-pd5 (54), and Escherichia coli DSM 498. The strains were grown overnight in ZoBell medium, Luria broth, or minimal medium with glucose, glutamine, and casein to induce expression of proteolytic activity (54). Culture supernatants were harvested following centrifugation (6,000 × g, 5 min), and protein concentration (bicinchoninic acid protein assay; Pierce, Rockford, Ill.) and proteinase activity (41) were measured before Western blot analysis.

Sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) was performed in 12% acrylamide gels (Novex, San Diego, Calif.). Details for SDS-PAGE and Western blotting are described by Kragelund et al. (26). Indigenous phosphatase activity was inactivated by a 30-min incubation in phosphate-buffered saline (pH 2.6). Following washing and blocking steps, nitrocellulose membranes with transferred proteins were incubated overnight with the rabbit antibody at dilutions of 1:5,000 or 1:10,000. Rabbit antibody was detected following 3-h incubations with a 1:1,000 dilution of anti-rabbit immunoglobulins conjugated with alkaline phosphatase (Dako, Glostrup, Denmark) and a color reaction with 5-bromo-4-chloro-3-indolylphosphate (Sigma) and nitroblue tetrazolium (Sigma).

For the detection of the AprX-like Pseudomonas proteinase in the microcosms, water samples were passed through 0.22-μm-pore-size Stericup-GS filter system (Millipore) and concentrated approximately 400-fold by use of Ultrafree-CL and Centriplus centrifugal filter units (Millipore) with nominal molecular size limits of 30 and 10 kDa, respectively. Subsequently, SDS-PAGE, blotting, and immunodetection were carried out as described above.

Cultivation of bacteria.

Bacteria were cultured on two solid media, Gould S1 (15) and 1/10 strength ZoBell 2216E agar (0.5 g of peptone, 0.1 g of yeast extract, 15 g of agar per liter of lake water) following 10-fold dilutions in 0.9% NaCl. Gould S1 is highly selective for Pseudomonas spp. due to the antibiotic trimethoprim and the detergent sodium lauryl sarcosine (15, 21, 22, 26), while the ZoBell medium is a general medium supporting the growth of a wide variety of bacteria.

Phenotypic and genotypic characterization of Pseudomonas.

A collection of Pseudomonas isolates was established for subsequent phenotypic and genotypic characterization. At the start of the experiment, 96 colonies from the Pseudomonas-specific Gould S1 agar were streaked to purity. On day 5, this procedure was repeated for the four microcosms to provide a collection of 480 Pseudomonas isolates. The strain collection was stored in equal volumes of ZoBell medium and glycerol at −80°C.

Pseudomonas isolates were screened for extracellular proteinase activity on casein in skim milk agar (41). To screen for LAP activity, cultures grown overnight in liquid ZoBell medium were incubated with 0.25 mM Leu-MCA for 3 h in the dark at 20°C. Fluorescence was quantified before and after the incubation by use of an LS 50B luminescence spectrometer (Perkin Elmer Ltd.); see details for LAP activity above. Additionally, some isolates were characterized by their ability to utilize 95 distinct carbon sources in Biolog GN2 assays (Biolog Inc., Hayward, Calif.). Strains were streaked on tryptic soy broth agar (Difco Laboratories, Detroit, Mich.) prior to the inoculation in duplicate Biolog GN2 plates at a final cell density of optical density at 600 nm (OD₆₀₀) = 0.1. Color reactions following the utilization of specific carbon sources were quantified after 24 h of incubation (OD₅₉₀ read by EL312 automated microplate reader; Bio-Tek Instruments Inc., Winooski, Vt.). Wells with OD₅₉₀ values 40% above the blank wells were scored as positive if a purple color response was also visible. Strains were identified by matching the patterns of sole-carbon-source utilization to the GN database release 4.01C running under the Biolog MicroLog 1 system release 4.01B (Biolog Inc.).

Genomic fingerprints of the isolates were obtained by UP-PCR (4). A universal primer consists of a universal sequence (5′ end) and a variable sequence (3′ end) without homology to any known gene position. This design stabilizes the primer annealing at relatively high temperatures and the generation of reproducible PCR products without prior knowledge of the target organism (4). DNA was extracted from cultures grown overnight in liquid ZoBell medium by 10 min of incubation at 94°C and freezing at −20°C. Then 1 μl of cell extract was mixed with 40 ng of L15/AS19 primer (32), 0.6 U of F-501 DyNAzyme II DNA polymerase (Finnzymes, Espoo, Finland), and 18 μl of reaction buffer. The reaction buffer contained 1× F-511 Dynazyme buffer (Finnzymes), 2 mM MgCl₂, and 0.4 mM deoxynucleoside triphosphate mix (New England Biolabs Inc.,). The PCR was carried out in a GeneAmp PCR System 9700 thermal cycler (Perkin Elmer, Norwalk, Conn.) with the following settings: 94°C for 3 min, followed by 33 cycles of 92°C for 50 s, 53°C for 70 s, and 70°C for 60 s, and a final extension prolonged by 2 min. PCR products were separated by electrophoresis in 2% (wt/vol) agarose gels. Gels were loaded with 100-bp ladders (Amersham Pharmacia Biotech Europe GmbH, Denmark) as molecular size standards. PCR products were grouped manually by use of digitized photographs overlaid by a size marker.

Data analysis.

Abundances and biochemical measures in duplicate microcosms were compared by t test (two-tailed at P < 0.05) following logarithmic transformations to equalize the variances (42). Distributions of Pseudomonas isolates among UP-PCR groups were analyzed by the following indices (3): diversity H = −Σ p_i ln p_i, where p_i is the frequency of isolate in the ith UP-PCR group. H was normalized by the richness in groups (S) to obtain an index of evenness E = H/ln S. Diversity indices were compared by t tests (42) where continuity of the variables was assumed from the relatively high number of isolates being distributed among a relatively high number of UP-PCR groups. The distribution of UP-PCR groups was also compared by χ² tests. We focused specifically on the 10 most abundant UP-PCR groups, but for χ² statistics, relevant joint groups had to be defined to meet the required minimum of five observations per class in each sample (42).

RESULTS

Microbial dynamics in microcosms.

Initially, the lake water contained 14 μg of chlorophyll per liter. In the control microcosms (dashed lines), chlorophyll increased and peaked at 39 μg liter⁻¹ on days 3 to 4 and decreased to 31 μg liter⁻¹ at the end of the experiment (Fig. 1A). Direct counts of bacteria increased from 1.8 × 10⁶ to 5.3 × 10⁶ ml⁻¹ from days 1 to 3 and declined subsequently to 3.3 × 10⁶ ml⁻¹ (Fig. 1B). In parallel, colony counts on ZoBell agar increased from 5.2 × 10³ to 20 × 10³ CFU ml⁻¹, whereas colony counts of Pseudomonas (Gould S1) decreased from 13 to 5.4 CFU ml⁻¹ (Fig. 1C). Finally, enzyme activities were stimulated: PRTase increased gradually from 5.8 to 11.5% h⁻¹, and LAP activity increased from 0.5 to 2.2 μM h⁻¹ (Fig. 1D).

FIG. 1.

Microbial dynamics in control (○) and casein-amended (▪) microcosms. Shown are chlorophyll (A), direct counts of bacteria (B), CFU on ZoBell agar and the Pseudomonas-specific Gould S1 agar (C), enzyme activity of PRTase and LAP (D). Values are averages of duplicate microcosms, with standard deviations shown by vertical lines (except day 1). Significant differences between casein-amended and control microcosms are indicated by asterisks (two-tailed t test, P < 0.05).

Amendment with 2 mg of casein per liter increased PRTase activity from 6 to 12% h⁻¹ by day 2, and a significantly higher level than in the control microcosms persisted until day 5. LAP also increased significantly but peaked 1 day later at 3.4 μM h⁻¹. Subsequently, LAP activity was similar to the control (ca. 2 μM h⁻¹) (Fig. 1D). By day 3, direct and culturable counts of bacteria responded to the casein enrichment with peaks at 7.6 × 10⁶ cells ml⁻¹ and 5.3 × 10⁴ CFU ml⁻¹, respectively, but this significant difference relative to the control microcosms disappeared during the subsequent days (Fig. 1B and C). In contrast, counts of Pseudomonas remained fairly constant from days 3 to 5 at 30 CFU ml⁻¹, which is a fivefold higher abundance than in the control microcosms (Fig. 1C). Finally, chlorophyll peaked at a significantly higher level of 52 μg liter⁻¹ on day 4 (Fig. 1A). The casein effect relative to control microcosms was replicated in an independent experiment (data not shown).

PRTase and LAP activity was measured in unfiltered and 0.22-μm-pore-size-filtered samples to address the significance of cell-associated enzyme activity. Both at day 1 and at day 5, PRTase and LAP in the filtrates equaled 44% ± 17% and 3.8% ± 4.7% of the total activities, respectively. The different distributions of enzyme activities were confirmed on more sampling occasions (data not shown).

Immunochemical detection of AprX-like Pseudomonas proteinase.

PRTase activity in the microcosms was derived from a diverse microbial community. To address whether pseudomonads express a proteolytic potential in lake water, we set out to detect the AprX-like Pseudomonas proteinase immunochemically. An antibody to the AprX proteinase of P. fluorescens CY091 (33) reacted with 50- to 60-kDa proteins in proteolytic culture supernatants from Pseudomonas isolates obtained from Lake Esrum (Fig. 2). These isolates represented the proteolytic UP-PCR groups A+, D+, E+, F+, G+, H+, and I+ (see below). A band of approximately 51 kDa was found for the proteolytic control strain P. fluorescens ON2 (54), and a relatively broad band of approximately 55 kDa was found for a mixed culture supernatant derived from the 96 Pseudomonas isolates obtained from Lake Esrum at day 1 (Fig. 2).

FIG. 2.

Western blot to detect AprX-like Pseudomonas proteinase in supernatants from overnight-grown Pseudomonas cultures. Proteolytic activity was detected in supernatants from all proteolytic strains. Lanes 1 to 10, representative isolates of the 10 most abundant UP-PCR groups, named A through J, with suffixes to indicate proteolytic activity on skim milk agar (+ or −). Lane 11, mixture of supernatants from 96 proteolytic and nonproteolytic strains isolated at day 1. Lane 12, proteolytic supernatant from P. fluorescens ON2. At the left is shown the migration of protein size standards. Bitmap pictures from separate digital gel scans were combined graphically.

Nonproteolytic supernatants from Pseudomonas isolates belonging to UP-PCR groups B−, C−, J−, and W− did not contain material that reacted with the antibody. Furthermore, no antibody reactivity was observed for the nonproteolytic control strains P. fluorescens ON2-pd5, P. fluorescens DF57, and E. coli DSM 498 or for the proteolytic supernatants of P. aeruginosa, E. chrysanthemi DSM 4610, S. marcescens ATCC 990, and P. mirabilis, all of which produce proteinases with some homology to the P. fluorescens AprX proteinase (see Materials and Methods).

In the lake water sampled at day 1, the proteinase antibody reacted with material migrating at 73 kDa in concentrated samples of 0.22-μm-pore-size-filtered lake water. At day 5, this band was less intense, but a double band of approximately 50 to 55 kDa appeared in water from the casein-amended microcosms (Fig. 3). This doublet did not appear in the control microcosms (data not shown). The 50- to 55-kDa doublet was similar to that observed by Western blot analysis of proteinases from Pseudomonas isolates from Lake Esrum (Fig. 2). Hence, these data indicate that the AprX-like Pseudomonas proteinase was expressed in the casein-amended microcosms.

FIG. 3.

Western blot to detect AprX-like Pseudomonas proteinase in approximately 400-fold-concentrated 0.22-μm-pore-size-filtered water from the microcosms. Lane 1, Lake Esrum at day 1. Lanes 2 and 3, duplicate microcosms amended with casein at day 5.

Pseudomonas population structure.

To examine the population structure of Pseudomonas in the microcosm experiment, 469 isolates from Gould S1 agar were screened for proteinase (skim milk assay) and LAP activity and grouped by UP-PCR genomic fingerprints. The population structure for Pseudomonas was highly diverse, as 138 distinct UP-PCR groups were defined for these isolates. The 10 most abundant UP-PCR groups consisted of between 80 and 7 isolates and were named A+, B−, and C− through J− (Fig. 4). The suffix specifies the presence or absence of proteolytic activity on skim milk agar (+ or −). UP-PCR groups were specifically proteolytic or nonproteolytic, as 97% of the isolates matched the prevailing phenotype of a given group, whereas LAP was expressed by all isolates grown on liquid ZoBell medium and could not be used to differentiate the isolates (Table 1). The remaining 212 isolates were distributed in 128 groups. For statistical analyses, these 128 groups were joined as proteolytic and nonproteolytic residual groups, i.e., Res+ and Res−, respectively. A robust identification by the Biolog GN2 assay (similarity index > 0.5) was obtained for isolates representing some of the major UP-PCR groups: P. fluorescens biotype F (A+), P. fluorescens biotype G (G+), P. currogata (C−), and P. synxantha (D+, E+, and F+).

FIG. 4.

UP-PCR genomic fingerprint of representative isolates of the most abundant UP-PCR groups. The groups are named A through J with a suffix for their ability to produce proteinase activity on skim milk agar (+ or −). The migration of size standards is indicated by horizontal dotted lines. Bitmap pictures from separate digital gel scans were combined graphically.

TABLE 1.

Population structure of Pseudomonas isolates from Lake Esrum microcosms^a

Parameter	Day 1	Control, day 5	Casein, day 5
Peptidase expression (% of isolates)
PRTase	84b	60 (1.5)c	57 (0.7)c
LAP	100b	100 (0)b	100 (0)b
UP-PCR groups
Diversity	3.74b	3.15 (0.11)c	2.81 (0.13)c
Evenness	0.94b	0.86 (0.01)c	0.81 (0.03)c
Richness	53b	38.5 (6.4)bc	32.0 (1.4)c

^aSampling at day 1 provided 90 isolates. Duplicate casein-amended or control lake water microcosms were established. At day 5, 94 to 96 isolates were recovered from casein-amended and control microcosms. Isolates were screened for PRTase and LAP expression and grouped according to UP-PCR genomic fingerprints. Data for enzyme expression are given as the average percentages (standard deviation) of isolates positive for the listed activities. The distribution of strains in UP-PCR groups is summarized by diversity indices. Similar superscript letters indicate that values are statistically similar (one-tailed t tests, P < 0.05). 

At day 1, 84% of the Pseudomonas isolates were proteolytic. The most abundant UP-PCR groups (D+, F+, and H+) comprised between 6 and 9% of the isolates. Groups A+, B−, C−, E+, G+, I+, and J− each accounted for 0 to 3% of the isolates. The remaining isolates were assigned to the groups Res+ (56%) and Res− (11%) (Fig. 5). At day 5, proportions of proteolytic pseudomonads were reduced to 57% ± 1% and 60% ± 2% of the isolates in the casein-amended and control microcosms, respectively. The decrease from the initial 84% was significant (P < 0.001) but independent of the casein amendment (P = 0.47). Diversity, evenness, and richness decreased between days 1 and 5, almost consistently at P < 0.05. The lowest diversity indices were found in casein-amended microcosms, but they were not significantly different from values from control microcosms (Table 1).

FIG. 5.

Relative distribution of Pseudomonas isolates grouped by similar UP-PCR genomic fingerprints at days 1 and 5 (control and casein amendment). Group names are indicated with a suffix for their ability to express proteinase activity on skim milk agar (+ or −). Groups represented by small numbers of isolates were joined to clarify the significant changes in the population structure; see text for explanation and statistics. Day 1 and 5 samples comprise 90 and 189 of 190 isolates, respectively.

The distribution of UP-PCR groups changed between days 1 and 5 (Fig. 5). Groups A+, B−, and C− became more abundant, whereas Res+ became less dominant (P < 0.001). Consistent shifts were not evident for Res− or groups D+ through J−. Casein amendment only affected the abundance of group B−, as the higher frequency of B− following the casein amendment was significant (P < 0.02) relative to the control microcosms. A statistical analysis of the distribution of UP-PCR groups in casein-amended and control microcosms is weakened by the high incidence of groups containing less than five observations. Consequently, joint groups were defined to test for broader similarities between replicates and treatments. Two relevant contingency tables, [Res+; Res−; (ADEFGHI)+; (BCJ)−] and [Res±; (ABC); (DEFGHIJ)], both showed similarity between replicate microcosms (P > 0.76), but significant differences between casein-amended and control microcosms (P < 0.03). Hence, the population structure of Pseudomonas shifted during the microcosm experiment, but only minor compositional differences were related significantly to the casein amendment.

DISCUSSION

Dynamics in microcosms.

The addition of protein (2 mg of casein per liter) appeared to induce a cascade of events in the microbial community. PRTase activity was rapidly mobilized, as increased turnover rates of [³H]casein were recorded in spite of the amendment with unlabeled casein (Fig. 1D). The action of PRTase provided protein hydrolysates for exopeptidases, e.g., LAP, and the subsequent metabolism of protein-bound amino acids was indicated by a transient increase in LAP and bacterial abundance. Finally, the biomass of phytoplankton increased, which indicated that casein-bound nitrogen was made available for phytoplankton growth (Fig. 1).

The culturable Pseudomonas population never exceeded 10² CFU ml⁻¹ on Gould S1 agar. For comparison, direct probing techniques have revealed cell abundances between 10¹ and 10⁴ ml⁻¹ for species/serotypes of Pseudomonas (10, 39, 51) or the RNA group I genus (25). Gould S1 agar is specific for Pseudomonas, but plate counts may be lower than on media without selective agents (15, 26, 35). Therefore, a culturing bias is likely to have affected the Pseudomonas recovery (1).

The stimulation of bacterial abundance and LAP activity following an amendment with protein was also found by Pinhassi et al. in coastal mesocosms (38). To our knowledge, however, no studies have previously demonstrated the sequential mobilization of PRTase and LAP activity. Enzymes with affinity to cleave either interior (PRTase) or terminal peptide bonds (LAP and other exopeptidases) are clearly distinguished in the nomenclature of enzymes (52). Hence, the cleavage of [³H]casein and the peptide-like Leu-MCA model substrates monitor enzymes responsible for distinct steps in the mobilization of protein for bacterial growth.

Numerous studies have shown negligible levels of LAP activity in 0.2-μm-pore-size filtrates (6). PRTase activity associated with cells (i.e., ectoenzymes) and larger particles was also prevalent in a study of oligotrophic seawater (18). This indicates a strategy where bacteria tend to concentrate the formation of hydrolysates near the cell surface (5, 18, 49). We also found negligible activity of LAP (4%) in 0.22-μm-pore-size filtrates, but nearly half of the total activity of PRTase (44%) was detected in the same size fraction. Hence, in our more eutrophic system, PRTase was not strongly cell associated, so that hydrolysates of dissolved proteins were available to virtually all bacteria independent of their expression of proteinase.

Immunochemical detection of AprX-like Pseudomonas proteinase.

Representative isolates of the proteolytic UP-PCR groups A+ and D+ through I+ all produced a single extracellular protein in pure cultures with size and immunoreactivity (Fig. 2) comparable to the AprX-like proteinase of P. fluorescens UP206 (33). The 50- to 60-kDa size range of these putative AprX proteinases is consistent with those reported for purified proteinases from P. fluorescens (14). Proteolytic supernatants from nonpseudomonads with related proteinase systems (31) did not react with the antibody, and lack of reactivity was also observed for proteolytic supernatants from P. aeruginosa DSM 50071. Hence, the antibody was specific for a subgroup of the alkaline proteinases in Pseudomonas, as shown for comparable proteinase antibodies (2, 47). Nevertheless, the grouping by UP-PCR profiles and subsequent Western blot analysis of representative isolates strongly indicates that AprX-like proteinases were widespread among proteolytic pseudomonads in Lake Esrum, because the 10 most abundant UP-PCR groups accounted for 55% of the isolated strains.

Proteins with size and immunoreactivity comparable to the AprX-like proteinase of P. fluorescens UP206 (33) were found in concentrated 0.22-μm-pore-size filtrates from the casein-amended (Fig. 3) but not from the control microcosms (data not shown). The antibody also reacted with 73-kDa proteins in the samples. This larger size class of proteins was not recovered from pure cultures, even in a mixture of supernatants from 96 strains isolated at day 1 (Fig. 2). Therefore, we consider the 50- to 55-kDa bands in the Western blot prime evidence that the AprX-like proteinase of Pseudomonas was present in the casein-amended microcosms. Accordingly, it appears that Pseudomonas contributed to the increase in bulk proteinase activity following an input of protein (Fig. 1). To the best of our knowledge, no studies have previously provided comparable direct evidence for expression of proteinase by a specific group of bacteria in a complex community.

The appearance of the AprX-like Pseudomonas proteinase might reflect that the expression of proteolytic activity from resident pseudomonads was stimulated by the casein amendment (5). It has also been suggested that changes in the expression of enzyme activity might reflect changes in the composition of the population (34), i.e., that the AprX-like Pseudomonas proteinase was detected because strains producing this class of enzymes proliferated in response to the casein amendment. These two fundamental mechanisms controlling the expression of enzyme activity were addressed by an analysis of the abundance and population structure of Pseudomonas isolates.

Abundance and composition of Pseudomonas spp.

UP-PCR genomic fingerprints provided a specific and reproducible system to discriminate isolates at the subspecies level without prior knowledge of their identity (4). The grouping of isolates by UP-PCR fingerprints demonstrated a highly diverse population structure of Pseudomonas in Lake Esrum (Table 1; Fig. 5). Complexity of pseudomonad populations has so far not been analyzed in aquatic systems, whereas studies in soil, using fingerprints generated by repetitive sequence primed PCR, have revealed complex population structures (21, 30). A high diversity of Pseudomonas has been explained by the ability of this group of bacteria to differentiate in niches and by a flexibility regarding exchange of mobile genetic elements and genomic recombination (46).

Casein amendment led to a fivefold-higher abundance of Pseudomonas than found in control microcosms (Fig. 1), but the impact on the Pseudomonas population structure was limited (Table 1 and Fig. 5). Between days 1 and 5, proportions of proteolytic phenotypes decreased together with the diversity indices, which was explained mainly by the proliferation of UP-PCR groups A+, B−, and C− (Fig. 5). Hence, a pronounced shift in the population structure of Pseudomonas cannot explain the detection of the AprX-like proteinase in casein-amended microcosms. We suggest that the casein amendment stimulated the expression of the AprX-like Pseudomonas proteinase, eventually for strains with high growth potential in the microcosms, e.g., A+. A more detailed evaluation of the origin of the AprX-like proteinase is not feasible at present due to the broad specificity of the antibody, narrow size range of AprX-like proteins, and high diversity of Pseudomonas isolates.

The proliferation of nonproteolytic phenotypes in the casein-amended microcosms is in accordance with the hypothesis that casein hydrolysates were available to bacteria independent of their proteinase expression. All Pseudomonas isolates from our study tested positive for LAP activity (Table 1), and alanine-aminopeptidase is also a general function for gram-negative bacteria (36). Hence, proteolytic as well as nonproteolytic pseudomonads were likely members of the bacterial consortium that metabolized protein in the lake water. Future studies will be important to address long-term effects of protein amendment. Also, it is important to compare responses of bacteria in different habitats, e.g., associated with organic particles versus freely suspended in the water.