In situ detection of protein-hydrolysing microorganisms in activated sludge


  • Yun Xia,

    1. Department of Biotechnology, Chemistry and Environmental Engineering, Section of Environmental Engineering, Aalborg University, Aalborg, Denmark
    Search for more papers by this author
  • Yunhong Kong,

    1. Department of Biotechnology, Chemistry and Environmental Engineering, Section of Environmental Engineering, Aalborg University, Aalborg, Denmark
    Search for more papers by this author
  • Per Halkjær Nielsen

    1. Department of Biotechnology, Chemistry and Environmental Engineering, Section of Environmental Engineering, Aalborg University, Aalborg, Denmark
    Search for more papers by this author

  • Editor: Michael Wagner

Correspondence: Per Halkjær Nielsen, Department of Biotechnology, Chemistry and Environmental Engineering, Section of Environmental Engineering, Aalborg University, Sohngaardsholmsvej 57, DK-9000 Aalborg, Denmark. Tel.: +45 96358503; fax: +45 96350558; e-mail:


Protein hydrolysis plays an important role in the transformation of organic matter in activated sludge wastewater treatment plants, but no information is currently available regarding the identity and ecophysiology of protein-hydrolysing organisms (PHOs). In this study, fluorescence in situ enzyme staining with casein and bovine serum albumin conjugated with BODIPY dye was applied and optimized to label PHOs in activated sludge plants. A strong fluorescent labeling of the surface of microorganisms expressing protease activity was achieved. Metabolic inhibitors were applied to inhibit the metabolic activity to prevent uptake of the fluorescent hydrolysates by oligopeptide-consuming bacteria. In five full-scale, nutrient-removing activated sludge plants examined, the dominant PHOs were always different morphotypes of filamentous bacteria and the epiflora attached to many of these. The PHOs were identified by FISH using a range of available oligonucleotide probes. The filamentous PHOs belonged to the candidate phylum TM7, the phylum Chloroflexi and the class Betaproteobacteria. In total they comprised 1–5% of the bacterial biovolume. Most of the epiflora-PHOs hybridized with probe SAP-309 targeting Saprospiraceae in the phylum Bacteroidetes and accounted for 8–12% of the total bacterial biovolume in most plants and were thus an important and dominant part of the microbial communities.


Microbial hydrolysis is a process through which macromolecules are hydrolysed to oligomers and monomers by microbial activity. It plays an important role not only in natural biogeochemical cycles, but also in engineered systems such as activated sludge wastewater treatment plants (Henze & Mladenovski, 1991). Organic substances found in domestic wastewater typically consist of 40–60% proteins, 25–50% polysaccharides and 10–30% lipids (Nielsen et al., 1992) and all macromolecules are hydrolysed into their monomers or oligomers such as amino acids, peptides, monosaccharides and long-chain fatty acids before being further degraded. Microbial hydrolysis, therefore, is the essential step in the degradation of organic matter in wastewater treatment plants, and it is often the rate-limiting step for processes related to nitrogen and phosphorus removal (Dueholm et al., 2001; Morgenroth et al., 2002).

Microbial hydrolysis is carried out by exoenzymes excreted by hydrolysing microorganisms. Several studies have focused on the activity of different exoenzymes and the location of the enzymes in activated sludge (Frolund et al., 1995; Gessesse et al., 2003). It is now clear that hydrolytic enzymes are primarily found associated with the cell surfaces where hydrolysis and release of partly degraded macromolecules are repeated until hydrolytic fragments are small enough to be assimilated by the microorganisms (Confer & Logan, 1998; Goel et al., 1998).

Activities of different exoenzymes have been reported in activated sludge. These include activities of proteases, galactosidases, glucosidases, lipases, chitinases and phosphatases (Teuber & Brodisch, 1977; Boczar et al., 1992; Nybroe et al., 1992). The surface-associated activity of most of these enzymes can be visualized using the ELF® 97 approach (Enzyme-labeled Fluorescence, Molecular Probes). The principle is that once a nonfluorescing conjugated substrate is hydrolysed it forms a fluorescent precipitate on the surface of bacteria excreting the exoenzyme. The ELF approach has been used in combination with FISH using 16S rRNA gene-targeting oligonucleotide probes to identify various exoenzyme-producing microorganisms in activated sludge. Some members of the phylum Bacteroidetes expressed phosphatases (Kloeke & Geesey, 1999), while some Thiothrix sp. filaments and Bacteroidetes filaments showed both phosphatase and esterase activities (Kragelund et al., 2005). The filamentous bacterium Microthrix parvicella expresses lipases and is believed to be involved in lipid hydrolysis (Nielsen et al., 2002). However, no method is available for the determination of protease activity on the surfaces of protein-hydrolysing organisms (PHOs) even though protease is the most dominant exoenzyme in activated sludge (Frolund et al., 1995). It is important to extend our knowledge of protein degradation and the microorganisms involved because the amino acids resulting from protein hydrolysis are important carbon and energy sources for polyphosphate-accumulating organisms, which are not only responsible for P-removal, but are also involved in N-removal via denitrification (Kong et al., 2004, 2005).

In this study we have tested a fluorescence in situ enzyme staining approach using BODIPY fluorescein-labeled (FL) substrates to stain PHOs in activated sludge followed by their identification by FISH probing. BODIPY dyes are substituted 4-bora-3a, 4a-diaza-s-indacene derivatives and if covalently bound to proteins they form stable conjugates in which the BODIPY-dye fluorescence can be quenched almost completely. The quenched fluorescence is released in the presence of proteases (Jones et al., 1997). Based on this, BODIPY dye-labeled substrates [casein and bovine serum albumin (BSA)] have been used to measure protease activity of pure enzymes (Welder et al., 2002) and recently also directly in pure cultures of the Gram-negative Porphyromonas gingivalis (Yoshioka et al., 2003). However, besides a precipitation of fluorescent BODIPY dye on the surface of P. gingivalis some intracellular uptake of fluorescent hydrolysates has also been reported, causing the bacterial cells to fluoresce (Yoshioka et al., 2003). This means that the surface-associated protease activity can only be visualized on microorganisms excreting protease (PHOs) in complex ecosystems if the fluorescent signals resulting from hydrolysis of BODIPY FL casein and BSA can be distinguished from those caused by uptake of fluorescent hydrolysates. In this paper, we describe how dominant casein- and BSA-hydrolysing microorganisms in activated sludge are identified by a new approach based on fluorescence in situ enzyme staining with BODIPY FL substrates combined with FISH.

Materials and methods

Activated sludge sources

Activated sludge samples used in this study were obtained from five Danish wastewater treatment plants, Aalborg West (AAV), Aalborg East (AAE), Egaa, Aabybro and Skagen, from February 2006 to July 2006. Aabybro is a nitrogen removal plant consisting of nitrifying and denitrifying tanks, while the others have also enhanced biological phosphorus removal with a Biodenipho® configuration (Seviour et al., 1999). The Skagen plant mainly treats industrial wastewater from the fish industry to a volume equivalent to that from a human population of 280 000. The Aalborg West, Egaa and Aalborg East plants mainly treat domestic wastewater serving communities with populations of 220 000, 100 000 and 100 000, respectively. Aabybro mainly treats domestic wastewater from smaller towns with a total population of 9800. Fresh sludge samples were taken from the nitrifying tanks and transferred to the laboratory within 0.2–1 h.

BODIPY FL proteins and incubation conditions

BODIPY FL casein in EnzChek® Protease Assay Kit (E6638) and BODIPY FL BSA in DQ™ Green BSA special packaging (D-12050) were purchased from Molecular Probe (VWR International, Roedovre, Denmark). The working solutions were all prepared as recommended by the provider. Incubations of activated sludge with BODIPY FL substrates were carried out in 10-mL serum bottles and the final reaction volume used was 600 μL. Then, 400 μL fresh mixed liquor of activated sludge [4 g L−1 mixed liquor suspended solids (MLSS)] was transferred into 1.5-mL Eppendorf tubes and spun down (4500 g for 10 min). The supernatant was discarded. Depending on the final substrate concentration, different amounts of freshly prepared 1 × Tris-HCl buffer (10 mM, pH 7.8) were added and the mixed liquor was transferred into a 10-mL serum bottle where the mixture was supplemented by a working solution of BODIPY FL casein or BSA to a final volume of 600 μL. When necessary, the inhibitors sodium azide, sodium fluoroacetate (from Sigma-Aldrich Logistik Gmbh, Germany) and sodium iodoacetate (from Merck KGaA, Darmstadt, Germany) were added (from a 50 × concentrated stock solution) at three different inhibitor combinations. Set one included iodoacetate+fluoroacetate+azide at final concentrations of 2, 1 and 2 mM, respectively, set two 3, 2 and 4 mM, and set three 4, 3 and 8 mM (final concentrations, respectively). Once all the reagents had been added, the serum bottles were immediately wrapped in aluminum paper to exclude light and mixed on a rotating disk (220 r.p.m. at 20±1°C). If inhibitors were used, sludge samples were incubated with inhibitors for 20 min before BODIPY FL casein or BSA was added.

In some experiments, preincubation of activated sludge samples with unlabeled casein or BSA was carried out in 100-mL flasks under aerobic conditions. Fifty millilitres of fresh sludge (4 g L−1 MLSS) was incubated with 200–500 mg L−1 (final concentration) calcium caseinate (Arla Foods Company, Denmark) or BSA (Sigma-Aldrich Logistik Gmbh) on a rotating disk shaking at 220 r.p.m. at 20±1°C for 1–4 or 12 h and washed three times with supernatant from the original activated sludge sample before being used to incubate with a labeled substrate.

Microscopical examination

Activated sludge samples incubated with BODIPY FL substrates were spread on a gelatine-coated three-well (10–15 μL in each well) Teflon printed slide (Electron Microscopy Sciences, Hatfield, PA) in a darkroom and allowed to air-dry at room temperature. The slides were mounted with CITI fluor (Citifluor Ltd, London, UK) and examined microscopically. The fluorescent signals of sludge samples stained with BODIPY FL casein and BSA (excitation/emission maxima of c. 505/513 nm) were visualized and captured through the FLUOS filter (excitation 450–490 nm, emission 515 nm) of an epifluorescence microscope (Axioskop 2 Plus, Zeiss) equipped with a charge-couple device (CCD) camera (CoolSNAP HQ, Photometrics, Oberkochen, Germany).

Enzyme staining combined with FISH

Enzyme staining was combined with FISH probing to identify PHOs. The fluorescent signals of stained sludge samples were examined as described above. The bacterial cells stained for proteases were detected microscopically and their positions on three-well gelatine-coated slides were recorded. CITI fluor on the slides was washed away by gentle rinsing (for 1 min) with 70% ethanol before fixation in either 4% paraformadehyde (for Gram-negative bacteria) or in 50% ethanol (for Gram-positive bacteria) for 2 h. When necessary, slides for FISH probing of Gram-negative bacteria were also fixed in 50% ethanol for 2 h instead of in paraformaldehyde. FISH was carried out according to Amann (1995). Cy3-labeled oligonucleotide probes were used. FISH signals of the bacteria of interest were examined after relocation. To identify the PHOs attached to filamentous bacteria (epiflora-PHOs) detected by BODIPY FL casein and BSA at least 50 filaments with epiflora in different microscopic fields were examined by FISH probing in each sludge sample. It was not possible to perform FISH before the enzyme assay as the fixation procedure inhibited the activity of exoenzymes.

The following oligonucleotide probes were applied: EUB338 mix [a mixture of EUB338 (Amann et al., 1990), EUB338II and EUB338III (Daims et al., 1999)] targeting most bacteria, CFX1223 (Bjornsson et al., 2002) and GNSB941 (Gich et al., 2001) targeting most members of the phylum Chloroflexi, CFX784 and CFX109 (Bjornsson et al., 2002) targeting some members of the genus Anaerolinea and some members of the family Chloroflexaceae, respectively, all within the phylum Chloroflexi, CHL1851 targeting the filamentous bacterium Eikelboom Type 1851 of the genus Roseiflexus, phylum Chloroflexi, Aqs997 (Thomsen et al., 2004) targeting Aquaspirillum-related bacteria in the class Betaproteobacteria, phylum Proteobacteria, TM7-905 (Hugenholtz et al., 2001) targeting candidate phylum TM7, MPA-645 targeting Candidatus ‘Microthrix parvicella’ in the phylum Actinobacteria (Erhart et al., 1997), CF319a plus CF319b targeting the CytophagaFlavobacteria group in the phylum Bacteroidetes (Manz et al., 1996), SAP-309 targeting most members of the Saprospiraceae in the phylum Bacteroidetes (Schauer & Hahn, 2005), PAOmix (a mixture of PAO651, PAO846 and PAO432) targeting betaproteobacterial Rhodocyclus-related polyphosphate-accumulating organisms (PAOs) (Crocetti et al., 2000), and GB targeting the GB group of glycogen-accumulating organisms (GAOs) in the class Gammaproteobacteria (Kong et al., 2002). Detailed information of all these probes is given in probeBase (Loy et al., 2003). All probes used were Cy3 labeled except for EUB338mix, which was FLUOS labeled.

Determination of the percentage of filamentous PHOs in a probe-defined group or phylum

The percentage of filamentous PHOs in a probe-defined group or phylum of a sludge sample was determined by counting the number of filamentous PHOs among at least 50 filaments hybridizing with a specific probe. This was done by first recording the positions of filamentous PHOs in different microscopic fields followed by FISH probing with a specific probe. Each counting was repeated three times using different sludge samples and the percentages obtained for each sludge sample were expressed as a percentage range (Table 1).

Table 1.   Identification of filamentous PHOs stained with BODIPY FL casein and BSA by using FISH probing in wastewater treatment plants
Probe nameSpecificityPHOs and their presence in different plants
  • *

    ‘+’, ‘++’ and ‘+++’ indicate that <10%, 10–50% and >50%, respectively, of 50 filaments hybridizing with a specific probes were PHOs. Each counting was repeated three times using different sludge samples.

  • A dash indicates that no PHO was found in 50 filaments hybridizing with a specific probe in three independent experiments.

TM7-905Candidate phylum TM7+*+++*++*
GNSB941&CFX1223Most members in the phylum Chloroflexi++++++++
CFX784Some members of the genus Anaerolinea, phylum Chloroflexi
CHL1851Type 1851 of the genus Roseiflexus, phylum Chloroflexi
CFX109Family Chloroflexales, phylum Chloroflexi+++
CF319a+bCytophaga–Flavobacteria group, phylum Bacteroidetes
MPA-645Microthrix parvicella, phylum Actinobacteria
Aqs997Aquaspirillum-related bacteria, phylum Proteobacteria++++++++++

Biovolume determination

The biovolume of epiflora-PHOs was determined by FISH on a very thin layer of biomass by measuring the percentages of area fluorescing with probe SAP-309 vs. the percentages of area fluorescing with the EUBmix probe in the same microscopic field. The biovolumes of the filamentous and the coccoid PHOs were preliminarily estimated by measuring the percentages of fluorescent area caused by the filamentous PHOs and the coccoid PHOs, respectively, after enzyme staining (by carefully choosing those fields with mainly filamentous or coccoid PHOs) vs. the percentages of area fluorescing with EUBmix after relocation. At least 10 microscopic fields (× 1000) were analysed for each enumeration against the original fields.


Application and optimization of BODIPY FL substrates in activated sludge samples

Two BODIPY FL substrates, BODIPY FL casein and BSA, were applied in this study to detect protease activity in activated sludge. BODIPY FL casein was first applied on sludge samples from the AAV plant on a 1 : 1 ratio [biomass in 300 μL of sludge sample (4 g L−1 MLSS) was resuspended in 1 × digestion buffer to the original volume+300 μL working solution of BODIPY FL casein] based on the protocol recommended by the provider. No protocol was provided for BODIPY FL BSA, so this was also used according to the protocol proposed for casein. After 1 h of incubation with BODIPY FL casein or BSA at a final substrate concentration of 2.50 mg g−1 MLSS some fluorescently labeled bacterial cells were visible microscopically. They were mainly large coccobacilli, some filamentous bacteria and some epiphytic bacteria attached to filamentous bacteria, but a strong fluorescence background was also observed. No fluorescent cells and no background were observed in control samples treated at 100°C for 15 min, indicating that casein and BSA in untreated samples were hydrolysed releasing BODIPY dye-labeled products. The same test was repeated for sludge samples from the AAE, Abybro, Egaa and Skagen plants, and similar results were obtained. These initial experiments indicated that BODIPY FL substrates could be used to label or stain PHOs and/or consumers of fluorescent hydrolysates in activated sludge.

To find the optimal incubation time such that the fluorescently stained cells could be observed with a low background, microscopical examination of the stained samples was carried out every 15 min for 180 min. Dominance of fluorescing cells with different morphotypes was observed during the observation period. After 10–15 min many attached bacteria on filaments (epiflora) (Fig. 1a and b) and filaments of different morphotypes were strongly stained in all five plants investigated. After 30–60 min some large coccobacilli were also stained, while some epiflora began to show weaker labeling. After 120 min (up to 180 min) many large coccobacilli were strongly stained. The background fluorescence level increased with incubation time but fluorescent cells could still be observed after c. 3 h. Similar results were obtained using BODIPY FL casein or BSA.

Figure 1.

 BODIPY fluorescence enzyme staining and FISH images of activated sludge. The epiflora shown in panel (a) (bright field image) are stained with BODIPY FL BSA in panel (b) (fluorescent image). The epiflora stained with BODIPY FL BSA in panel (c) (fluorescent image) hybridize with gene probe SAP-309 (targeting most members of Saprospiraceae in phylum Bacteroidetes) in panel (d) (FISH image). Scale bar in panel (a) represents 10 μm in panels (a) and (b). Scale bar in panel (c) represents 10 μm in panels (c) and (d). in panels (c) and (d) same bacterial cells are indicated by the arrows labeled with the same number.

To distinguish the PHOs from the hydrolysate-consumers, which may also fluoresce due to their uptake of fluorescent hydrolysates, different metabolic inhibitors were added in the incubations to prevent uptake and ensure that all BODIPY dye-stained bacteria were PHOs (casein- or BSA-hydrolysing microorganisms). Iodoacetate, fluoroacetate and azide were added to inhibit glycolysis (Bickis & Quastel, 1965), the tricarboxylic acid (TCA) cycle (Peters et al., 1953) and the electron transport chain (Myers & Nealson, 1988). Three sets of inhibitor conditions with iodoacetate+fluoroacetate+azide were tested (see concentrations in Materials and methods). The three sets all gave the same results in BODIPY FL casein and BSA incubations: only the filamentous bacteria and the epiflora and a few coccobacilli were stained during 3 h of incubation in activated sludge from all five treatment plants. This indicated that the epiflora and some filamentous bacteria were the main PHOs able to hydrolyse casein and BSA, while the large coccobacilli stained in incubations without adding inhibitors were probably consumers of hydrolysates (amino acids or oligopeptides). The fluorescence intensity observed for the potential PHOs, however, decreased as the inhibitor concentrations increased and only weak signals could be observed at the highest concentrations. No significant difference in the background fluorescence level at different inhibitor concentrations was observed. Therefore, in all experiments medium concentrations of metabolic inhibitors were added to visualize only PHOs.

To rule out the possibility that the protease activity of some casein- or BSA-hydrolysing organisms might require longer induction time or that their in situ protease activity be too weak to be detected within 3 h of enzyme staining incubation, preincubation with unlabeled casein or BSA was carried out before incubation with a BODIPY FL substrate to induce or enhance the protease activity of all potential PHOs. Preincubation periods of 1–4 or 12 h under aerobic conditions were tested. The results showed that the same morphotypes were still stained in sludge samples from all plants investigated, indicating that the main casein- and BSA-hydrolysing organisms were stained after 10–180 min of incubation, making preincubation unnecessary. This also implied that other microorganisms in activated sludge could not be induced to produce protease within these time periods.

In order to have a strong fluorescent signal of the PHOs and a low background fluorescence level, the substrate concentration of BODIPY FL casein and BSA was optimized by examining microscopically the signal-to-noise ratio at various substrate concentrations (0.25, 0.50, 0.63, 0.75, 1.00, 1.25, 1.50, 1.88 and 2.50 mg g−1 MLSS) using sludge samples from AAV and Egaa. Fluorescently labeled cells from both casein and BSA hydrolysis could be observed at concentrations as low as 0.5 mg g−1 MLSS after only 10 min of incubation. The fluorescent signals from cells and from the background increased with substrate concentration and the optimum concentrations were found to be between 0.50 and 1.25 mg g−1 MLSS after 15–30 min incubation. The change of incubation conditions did not change the PHO morphotypes stained during 3 h of incubation.

Identification of PHOs in different plants using enzyme staining combined with FISH

FISH probing using different oligonucleotide probes was used to identify the PHOs and hydrolysate-consumers. The filamentous PHOs stained with BODIPY FL casein and BSA were phylogenetically diverse, and comprised bacteria from different phyla (Table 1). They included filamentous bacteria belonging to the candidate phylum TM7 hybridizing with probe TM7-905, phylum Chloroflexi hybridizing with a mixed probe of GNSB941 and CFX1223, and class Betaproteobacteria hybridizing with probe Aqs997 targeting Aquaspirillum-related bacteria, but they did not include filamentous bacteria belonging to the CytophagaFlavobacteria group within the phylum Bacteroidetes hybridizing with a mixture of CF319a and CF319b. Some of the filamentous PHOs also hybridized with a more specific Chloroflexi probe CFX109 targeting the family Chloroflexaceae, but none hybridized with the probe CHL1851 specifically targeting Eikelbooms Type 1851, and the probe CFX784 targeting some members of the genus Anaerolinea. There were many actinobacterial filamentous Candidatus ‘Microthrix parvicella’ present at most plants (targeted by probe MPA-645), but they did not stain with BODIPY FL casein or BSA. The fractions of filamentous PHOs in a probe-defined group or phylum were also determined using BODIPY FL BSA and the results are also presented in Table 1.

In all the plants investigated 10–50% of the filamentous bacteria belonging to the phylum Chloroflexi hybridizing with a mixture of probes GNSB941 and CFX1223 were PHOs. In Aabybro more than 50% of the Chloroflexaceae filaments hybridizing with CFX109 (only present in Aabybro) were PHOs. Several Aquaspirillum-related filaments hybridizing with Aqs997 were PHOs in plants AAV and AAE (10–50%) and more than 50% were involved in protein hydrolysis in Egaa and Aabybro. The PHO fraction of phylum TM7 filaments hybridizing with probe TM7-905 varied significantly in the different plants: 10–50% in Aabybro, higher in AAE (>50%), lower in AAV (<10%) and absent in Egaa.

All the filamentous PHOs except for the TM7 PHOs, which were not found at Egaa, were present in all plants investigated. The biovolumes of different filamentous PHOs were investigated (Table 2) and they comprised 1–5% of the total bacterial biovolume.

Table 2.   Abundance of different PHO morphotypes in four wastewater treatment plants
PlantBiovolume (%) of different PHO morphotypes
AAV, Egaa and Aabybro8–121–2<1

Almost all epiflora on the various types of filaments hybridized (Fig. 1c and d) with probe SAP-309 targeting the family Saprospiraceae, phylum Bacteroidetes. This was observed by examining epiflora-PHOs on at least 50 epiphytic filaments in each sludge sample (Fig. 1b and c). Most of these were stained with BODIPY FL casein and BSA in the presence of inhibitors, indicating that they were PHOs and showing that almost all epiflora-PHOs belonged to Saprospiraceae. They did not, however, hybridize with the commonly used broad probes (CF319a+b) for the CytophagaFlavobacteria group in the Bacteroidetes. The epiflora-PHOs were present and abundant in all plants investigated except in AAE and accounted for 8–12% of the total bacteria biovolume (Table 2). The epiflora-PHOs were in all treatment plants attached to filamentous microorganisms belonging to the Chloroflexi (probe GNSB941 plus CFX1223), TM7 (probe TM7-905), Aquaspirillum-like bacteria (probe Aqs997) and some unidentified filaments.

We also attempted to identify the large coccobacilli stained positively with BODIPY FL casein and BSA in incubations without inhibitors indicating uptake of amino acids and/or oligopeptides. Some of the large coccobacilli were Rhodocyclus-related PAOs as they hybridized with the probe PAOmix, and some were GAOs belonging to the Gammaproteobacteria hybridizing with probe GB. They were present in 3–7% of the total bacterial biovolume in most treatment plants (Kong et al., 2006), indicating that they could be important amino acid and/or peptide consumers.

In FISH probing of Gram-negative bacteria using samples fixed with paraformaldehyde we found that FISH signals were generally weaker after enzyme staining than before. This could be due to a coating of cell walls with BODIPY dye-labeled hydrolysates thereby limiting the ability of the probe to penetrate or preventing the probes from penetrating the bacteria after paraformaldehyde treatment, and resulting in a weak signal or no signal. To confirm this, 50% ethanol was used to fix the sludge samples after enzyme staining. Fixation was carried out 1–7 h before FISH probing to determine the optimal fixation time. Strong FISH signals were observed after 1–2 h of ethanol fixation for all probes.


Use of BODIPY FL substrates to detect PHOs in activated sludge

BODIPY dye-labeled proteins have previously mainly been used to study the protease activity of purified enzymes (e.g. Welder et al., 2002), and only recently have they been applied also to study protease activity directly in a bacterial culture (Yoshioka et al., 2003). Our study has shown that BODIPY FL proteins can also be used successfully to label and identify PHOs in combination with FISH in activated sludge and thus gain more knowledge regarding the organisms involved in hydrolysis and consumption of proteins in complex microbial communities.

Both casein and BSA are used widely in studies of protein hydrolysis in relation to wastewater systems and are assumed to represent the proteins present in wastewater fairly well (Morgenroth et al., 2002). According to the provider (Molecular Probe, VWR International, Roedovre, Denmark) BODIPY FL casein can be used to detect a wide range of proteases including serine, acid, metallo and sulfhydryl proteases, thus ensuring reliable labeling of most bacterial PHOs. Information regarding the enzymes that are detected by BODIPY FL BSA is not available, but they may be similar to BODYPI FL casein as we found the same organisms in the samples investigated to be active for both proteins.

Several studies have previously shown that many exoenzymes in microbial aggregates are mainly associated with the cell surfaces (Confer & Logan, 1998; Goel et al., 1998; Kloeke & Geesey, 1999; Nielsen et al., 2002) and this was also observed for the proteases. Once quenched BODIPY dye-labeled casein and BSA were hydrolysed, fluorescent precipitates attached onto the surfaces of bacteria excreting the proteases thus enabling labeling of PHOs. However, some hydrolysates were released to bulk water, as also observed by Yoshioka et al. (2003), so bacteria able to take up these oligopeptides were also labeled. Furthermore, this release caused an increase in background fluorescence. Although protease activity is mainly associated with cell surfaces, the possibility that some bacteria primarily excrete protease into bulk liquid (and thus provide some background activity) cannot be ruled out. These bacteria would be missed from our PHO screening.

Inhibitors were successfully used to block the metabolic activity of all bacteria in the activated sludge, so potential consumers of the hydrolysates were inactivated and production of new exoproteases inhibited, allowing only detection of active PHOs at the time the incubation was initiated. Iodoacetate, fluoroacetate and azide were employed to inhibit glycolysis, the citric acid cycle (TCA cycle) and the electron transport chain, respectively, and activated sludge samples were incubated with inhibitors for at least 20 min before a BODIPY FL substrate was added. Rhodocyclus-related PAOs and some GAOs belonging to the Gammaproteobacteria (GB bacteria) were not stained with BODIPY FL casein or BSA in the presence of these inhibitors, but were fluorescent in the absence of inhibitors, indicating that they were not PHOs, but amino acid or oligopeptide consumers. This result agrees with the findings from previous studies (Kong et al., 2004, 2006) where microautoradiography combined with FISH showed that these two groups are capable of taking up certain amino acids. However, it was also noticed that some amino acid-utilizing actinobacterial PAOs (Kong et al., 2005), which were also present in the sludge samples examined, were not positively stained (results not shown). It is not clear whether it was because the amino acids they took up were not labeled by BODIPY dye or that they could not take up amino acids conjugated with BODIPY dye. By contrast, the epiflora and different filaments were still stained in the presence of the inhibitors, indicating strongly that extracellular proteases of these bacteria were present before the incubations started as proteases could not be produced in the presence of the inhibitors. Thus, the protease activity detected in this study for activated sludge was the in situ activity and reflected the real activity in wastewater treatment plants.

Some of the inhibitors used in this study not only inhibited the uptake of oligopeptides or amino acids, but also to some extent the activity of exoproteases. It was observed that the fluorescence intensity of the stained PHOs decreased with increased inhibitor concentrations. It is known that iodoacetate is an effective inhibitor of cysteine proteases (Vincents et al., 2004) whereas the effect of fluoroacetate and azide on protease activity is less clear. A decrease in the fluorescence intensity of the stained PHOs was often seen after 45–60 min of incubation. This may be due to dissolution of precipitated hydrolysates and inhibition of proteases caused by accumulation of hydrolysates. Therefore, the results show that it is important to test the inhibitor concentrations before they are applied to a new microbial community.

The background fluorescence levels increased with higher substrate concentrations due to the release of more hydrolysates with BODIPY dye into bulk liquid, thereby masking the fluorescence signals from PHOs. Therefore, for a new type of microbial community it is necessary to optimize substrate concentration and staining time. Another problem related to the quality of the FISH signals of Gram-negative PHOs and consumers of amino acids and/or oligopeptides was noticed. The staining generally caused the FISH signal to weaken and sometimes no FISH signal could be seen. This was probably due to problems associated with the inability of the probes to penetrate into the paraformaldehyde-fixed cells, and ethanol fixation was required to improve the penetration of the gene probes.

Given the requirements and optimizations discussed above, the new approach of combining FISH with BODIPY FL proteins is a very powerful method to identify the main functioning PHOs in activated sludge plants and other ecosystems. Furthermore, we also tried to quantify the protease activity per probe-defined population and compare this with the overall protease activity of the total community, as measured by the increase in fluorescence over time produced from BODIPY FL proteins or from methylumbelliferyl (MUF)-labeled substrates (Frolund et al., 1995). However, this proved to be very difficult owing to the increase in background fluorescence level and large variations in the exoenzymatic activity of the individual cells within a certain population, resulting in large variations in cellular fluorescence signal.

Identity and ecophysiology of the PHOs

A range of oligonucleotide probes were selected based on data obtained from our long-term study of these treatment plants, and these were then used in FISH probing to identify the PHOs. The dominant PHOs in three of the four plants investigated were rods attached as epiflora on different types of filamentous microorganisms. Most of them hybridized with probe SAP-309 targeting the family Saprospiraceae, phylum Bacteroidetes. They typically consisted of 8–12% of the bacterial biovolume. Epiphytic growth on filamentous microorganisms is commonly observed in activated sludge treatment plants, but their identity has never been revealed (Eikelboom, 2002; Thomsen et al., 2002). The SAP-309-targeted epiphytic bacteria have most likely not been detected in previous studies because they did not hybridize with the commonly applied probe (CF319a+b) targeting the CytophagaFlavobacterium group of the Bacteroidetes. However, coverage of the probe SAP-309 is relatively broad (Schauer & Hahn, 2005) and in addition to the epiflora, the probe also hybridized with some filamentous bacteria in the wastewater treatment plants investigated in this study. The detailed phylogeny of the epiphytic Saprospiraceae, therefore, awaits more detailed studies. However, they seem to form a unique ecological niche where by attaching themselves to filaments they form a three-dimensional ‘screen’ suited to trapping particulate organic matter, facilitating their role as PHOs (Fig. 1a–d).

The filaments attached with epiflora-PHOs were diverse, belonging to at least the phyla Chloroflexi and TM7 (candidate phylum) and class Proteobacteria. They could be identified as morphotypes commonly present in activated sludge from numerous studies worldwide (e.g. Eikelboom, 2002) and described as Eikelboom's Type 0041 (primarily from candidate phylum TM7), Type 1851 (primarily from Chloroflexi) and Type 1701 (mainly Aquaspirillum-related bacteria; Thomsen et al., 2006). Hence, the results strongly indicate that most filamentous bacteria in activated sludge with epiphytic growth are involved in protein degradation. However, the extent to which they were involved is not yet clear, and requires more data regarding the ecophysiology of the epiflora and their host, in particular the ecological relationship between them.

FISH probing also showed that several different filamentous bacteria were PHOs. Being less abundant but more diverse than the epiflora-PHOs in most of the sludge samples examined, the filamentous PHOs belonged again to the phylum Chloroflexi, candidate phylum TM7 and Aquaspirillum-related bacteria (Table 1). Most of these probes are very broad so the phylogeny and detailed ecophysiology of the filamentous PHOs require further investigation. For example, the filamentous PHOs hybridizing with CFX109 were only present in Aabybro, an N-removal plant consisting of nitrification and denitrification tanks, but not in the three other biological P removal plants containing not only nitrification and denitrification tanks but also anaerobic tanks (Biodenipho configuration). This suggests that a configuration-related selection may exist for the PHOs hybridizing with CFX109. The ecophysiology of Aquaspirillum-related bacteria has recently been investigated (Thomsen et al., 2004, 2006) and these filamentous organisms seem to be specialized in the consumption of amino acids, fitting well with the results obtained here showing them to be PHOs. Common to all these filamentous PHOs was that in most plants only a few of them had epiphytic bacteria. Whether this is due to the presence of different closely related filamentous species with different preferences for epiphytic growth (most probes applied were broad) or other factors, such as growth conditions, is not known.

PHOs with a morphotype of small cocci were also observed occasionally, but in much lower numbers than the epiflora and filamentous PHOs. They were not important PHOs in the sludge samples tested and no further effort was attempted to identify them. However, we realize that the amount of epiflora, filamentous and coccoid PHOs in each plant only represent the situation over a limited period (4–5 months), so the relative importance of these PHOs in a specific plant may change with time. However, the results obtained in this study indicate that only a few bacterial groups or populations were specialized PHOs and thus seem to be the key players in the turnover of protein in many activated sludge plants.


This study was supported by the Danish Technical Research Council in the framework program ‘Identification and characterization of uncultured bacteria involved in hydrolysis and fermentation in nutrient removal plants’ (grant number 26-04-0115). Caroline Kragelund and Jeppe L. Nielsen are acknowledged for helpful suggestions and discussions. We also thank Jane Ildal for her technical support.