Ecophysiology of Uncultured Filamentous Anaerobes Belonging to the Phylum KSB3 That Cause Bulking in Methanogenic Granular Sludge

Abstract

A filamentous bulking of a methanogenic granular sludge caused by uncultured filamentous bacteria of the candidate phylum KSB3 in an upflow anaerobic sludge blanket (UASB) system has been reported. To characterize the physiological traits of the filaments, a polyphasic approach consisting of rRNA-based activity monitoring of the KSB3 filaments using the RNase H method and substrate uptake profiling using microautoradiography combined with fluorescence in situ hybridization (MAR-FISH) was conducted. On the basis of rRNA-based activity, the monitoring of a full-scale UASB reactor operated continuously revealed that KSB3 cells became active and predominant (up to 54% of the total 16S rRNA) in the sludge when the carbohydrate loading to the system increased. Batch experiments with a short incubation of the sludge with maltose, glucose, fructose, and maltotriose at relatively low concentrations (approximately 0.1 mM) in the presence of yeast extract also showed an increase in KSB3 rRNA levels under anaerobic conditions. MAR-FISH confirmed that the KSB3 cells took up radioisotopic carbons from [¹⁴C]maltose and [¹⁴C]glucose under the same incubation conditions in the batch experiments. These results suggest that one of the important ecophysiological characteristics of KSB3 cells in the sludge is carbohydrate degradation in wastewater and that high carbohydrate loadings may trigger an outbreak of KSB3 bacteria, causing sludge bulking in UASB systems.

Activated sludge wastewater treatment plants often suffer from operational disorders, particularly bulking, which is caused by the excessive growth of certain filamentous bacteria (21). A number of morphologically and phylogenetically distinct types of filamentous microorganisms that are associated with activated sludge bulking have been characterized (13, 14, 21). Similarly, the flotation and subsequent washout of sludge caused by filamentous bulking have also been reported for anaerobic digestion systems (anaerobic sludge bulking) (1, 16, 25, 29, 36, 39, 40). Because the high-rate treatment of organic wastewater using modern anaerobic bioreactors such as upflow anaerobic sludge blanket (UASB) systems strongly depends on the retention of high concentrations of biomass in the reactor vessel (as sludge granules), the bulking of the sludge severely affects treatment performance, often leading to a complete loss of performance.

The phenomena of anaerobic sludge bulking reported so far were caused by the overgrowth of certain filamentous microbes. To date, Anaerolinea thermophila has been identified as a causative agent of anaerobic sludge bulking in a thermophilic (55°C), methanogenic UASB system (29, 30, 37). In addition, we have recently identified uncultured filamentous bacteria affiliated with the candidate phylum KSB3 (also known as the candidate phylum GN6 [15]) to be a causative agent of anaerobic sludge bulking (39). KSB3 filaments were abundant in the bulking sludge in a mesophilic UASB reactor that had been treating high-strength organic wastewater discharged from an isomeric sugar-manufacturing factory (39). We predicted that the KSB3 filaments are heterotrophs that ferment sugars since (i) KSB3 phylotypes were found at high frequencies in the sludges fed with sugar-processing wastewaters (isomerized sugar processing) (20, 39) and (ii) KSB3 cells were found only in the outermost layer of the healthy sludge granules, possibly degrading primary substrates in wastewater, such as carbohydrates (39). However, because the organisms are very fastidious and uncultivable in the laboratory, their detailed ecophysiology and metabolisms remain unclear.

In this study, a polyphasic approach consisting of various rRNA-based tests was performed to gain further insight into the physiological traits of the KSB3 filamentous bacteria in methanogenic sludge.

MATERIALS AND METHODS

Reactor operation.

Samples of granular sludge were collected from a mesophilic (32°C to 42°C), full-scale UASB system (reactor volume, approximately 750 m³) that was fed with actual organic wastewater (pH range, 12.0 to 13.0) discharged from an isometric sugar-manufacturing factory (39). Prior to feeding to the UASB system, the pH of the wastewater was adjusted to pH 8.5 to 9.0 by the addition of NaOH. The system consisted of two reactor vessels (reactors I and II, approximately 375 m³ each), in which the same wastewater was treated in parallel. The performances of both reactors were examined for 75 days of continuous operation. During this period, the reactor temperature, methane formation, and influent and effluent characteristics (i.e., chemical oxygen demand [COD], volatile fatty acid [VFA] level, carbohydrate concentration, and protein concentration) were monitored. Overall, reactors I and II exhibited good treatment performance, with 90 to 95% COD removal efficiencies at COD loadings of 8 to 17 kg COD m⁻³ day⁻¹, and most of the COD removed was converted to methane in the experimental period (methane conversion rate of >80%). Sludge granules (1 to 2 mm in diameter) that settled well were retained in reactors I and II, but a fluffy bulking sludge similar to white cotton (>2 cm), composed mainly of KSB3 filaments, was occasionally and suddenly formed in both reactors during the experimental period (19, 39).

Microorganisms, media, and cultivation.

Sphaerobacter thermophilus strain S6022^T was obtained from the Deutsche Sammlung von Mikroorganismen (DSM) und Zellkulturen GmbH (Braunschweig, Germany). Cells of S. thermophilus were aerobically cultivated in the medium (medium 467) in accordance with the instructions of the DSM and used for RNA extraction (see below). A basal medium for the cultivation of KSB3 bacteria was prepared according to methods described in previous reports (28, 38).

Sequence-specific small-subunit (SSU) rRNA cleavage with oligonucleotides and RNase H.

The rRNA-based quantification of selected microbial groups, the domains Bacteria and Archaea and a phylum, KSB3, was performed by the RNase H method. The extraction and quantification of RNA from sludge samples were performed as described previously (20, 34). Extracted RNA samples were quantified with a RiboGreen RNA quantification kit (Invitrogen, Carlsbad, CA) and a spectrofluorophotometer apparatus (model RF-1500; Shimadzu, Kyoto, Japan). For the RNase H reaction, the hybridization mixture contained 1 μl of RNA template (ca. 1 μg μl⁻¹), 1 μl of each scissor probe solution (10 pmol μl⁻¹), 2.5 μl of 10× hybridization buffer (250 mM Tris-HCl, 10 mM EDTA, 250 mM NaCl), and a given amount of formamide (pH 7.5) (20, 34). The mixture was heated at 95°C for 1 min to denature RNA molecules and then incubated at 50°C for 1 min. The cleavage reaction was initiated by adding 5 μl of 10× enzyme mixture (200 mM Tris-HCl [pH 7.5], 100 mM MgCl₂, 125 mM NaCl, 10 mM dithiothreitol [DTT], 300 μg μl⁻¹ bovine serum albumin [BSA], 5 U μl⁻¹ Escherichia coli RNase H [Takara, Shiga, Japan]) and immediately incubating the mixture at 50°C for 15 min. To terminate the digestion reaction, 25 μl of 3× stop solution (30 mM EDTA, 0.9 M sodium acetate [pH 7.0]) was added to the mixture. The RNA mixture was then deproteinized by washing with acid phenol-chloroform-isoamyl alcohol (25:24:1, vol/vol/vol). The supernatant was collected by centrifugation, and the RNA was precipitated by ethanol and dissolved in 4 μl of nuclease-free water.

The quantity and integrity of intact and digested 16S rRNA fragments were evaluated by using an Agilent (Palo Alto, CA) Bioanalyzer 2100 with an RNA 6000 Nano kit (Agilent) in accordance with the manufacturer's instructions. The percentage of digested 16S rRNA was calculated as described previously (20, 34). Each cleavage experiment was performed in triplicate.

The rRNA-targeted oligonucleotide probes used in this study are listed in Table 1. All the oligonucleotide probes were purchased from Tsukuba Oligo Service Co. (Tsukuba, Japan). The specificity and optimum conditions for the 3KSB1238 probe, specific for the phylum KSB3 (39), were evaluated by using 16S rRNAs transcribed from a KSB3-related 16S rRNA gene clone (DDBJ accession no. AB218870) (39) and the 16S rRNA gene of S. thermophilus strain S6022^T, which has 3-base mismatches against the complementary sequence of the 3KSB1238 probe as positive and negative controls, respectively. The 16S rRNAs were transcribed in vitro from clonal rRNA genes using a T7 RiboMAX Express large-scale RNA production system (Promega Corp., Madison, WI) as described previously (34).

TABLE 1.

Concentrations of carbohydrates in influent and effluent wastewaters

Substance	Concn of carbohydrate (mg liter−1) in:
	Influent wastewater	Effluent wastewater from reactor I	Effluent wastewater from reactor II
Saccharides
Glucose	16	<0.5	<0.5
Galactose	4.6	<0.5	<0.5
Mannose	<0.5	<0.5	<0.5
Fructose/arabinose	23	<0.5	<0.5
Rhamnose	0.6	<0.5	<0.5
Xylose	15	<0.5	<0.5
Ribose	15	<0.5	<0.5
Galactosamine	4.4	<0.5	<0.5
Sucrose/glucosamine	2.3	<0.5	<0.5
Lactose	14	<0.5	<0.5
Cellobiose	<0.5	<0.5	<0.5
Maltose/mannosamine	550	1.2	1.9
Isomaltose	1.4	<0.5	<0.5
Organic acids
Lactate	358	<0.1	4.9
Formate	103	<0.1	<0.1
Acetate	332	17.8	25.7
Propionate	103	<0.1	16.3
n-Butyrate	44.4	<0.1	<0.1

Examinations of KSB3 rRNA responses with spike-in substrate in batch cultivation tests.

For batch experiments with spike-in substrates, intact sludge granules were collected from reactor II, gently washed with phosphate-buffered saline (120 mM NaCl, 4 mM NaH₂PO₄, and 10 mM Na₂HPO₄ [pH 7.2]), and anaerobically cultivated with a basal medium (28, 38) supplemented with spike-in substrates at different concentrations. Anaerobic cultivation was performed at 37°C in 50-ml serum vials containing 20 ml of the medium (pH 7.2 at 25°C) under an 80% (vol/vol) N₂-20% (vol/vol) CO₂ atmosphere with gentle shaking (60 rpm). RNA was extracted from (i) sludge granules before batch cultivation and (ii) sludge granules after batch cultivation. For both the samples, the wet weight of the sludge samples (mg wet) was determined after the complete removal of moisture with an aspirator for 5 min and was used for calculating the amount of RNA extracted from a unit of sludge granules in sludge (i.e., μg RNA g wet weight of sludge⁻¹). Cultures were maintained at 37°C until approximately half of the spike-in substrate was degraded and were then immediately subjected to RNA extraction (Table 2). Changes in the amounts of KSB3-related RNA during the cultivation were estimated by using the following equation: changes in the amount of KSB3 16S rRNA in sludge (%) = [(d/100 × f) − (e/100 × g)]/(d/100 × f) × 100, where d and e are the percentages of KSB3-related 16S rRNA in total 16S rRNA (percent) before and after cultivation, respectively, and f and g are the amounts of total RNA in sludge (μg RNA g wet weight of sludge⁻¹) before and after cultivation, respectively. Each batch experiment was performed in duplicate.

TABLE 2.

Changes in KSB3 RNA levels in batch cultivation tests with spike-in substrates

Organic matter	Concn (mg liter−1)	Incubation time (h)	Response of KSB3-related RNAa	Mean change in amt of KSB3-related RNA after spike-in substrate tests (%) ± SD
Short incubation in absence of yeast extract
Lactate	18	16	−	−45 ± 9
Acetate	540	16	−	−58 ± 6
Propionate	360	48	−	−80 ± 9
Butyrate	260	48	−	−33 ± 2
Glucose	18	12	−	−14 ± 4
Fructose	18	12	−	−71 ± 2
Maltose	34	12	−	−81 ± 5
Yeast extract	100	12	−	−73 ± 9
Yeast extract	250	32	−	−28 ± 5
Peptone	250	32	−	−10 ± 5
Short incubation in presence of yeast extract (0.01%)
Lactate	18	16	−	−42 ± 5
Acetate	540	16	−	−33 ± 7
Propionate	360	48	−	−85 ± 4
Butyrate	260	48	−	−37 ± 2
Glucose	18	12	+	67 ± 11
Fructose	18	12	+	60 ± 10
Caramel	18	12	−	−52 ± 2
Maltose	3.4	12	−	−74 ± 6
Maltose	34	12	+	156 ± 38
Maltose	340	12	+	25 ± 1
Maltose	3,400	12	−	−58 ± 6
Maltose	34,000	12	−	−68 ± 3
Maltotriose	50	12	+	23 ± 2
Maltotetraose	67	12	−	−10 ± 10
Dextran	1,000	12	−	−90 ± 3
Corn steep liquor	1,000	12	−	−42 ± 16
Cornstarch	1,000	12	−	−20 ± 16

^a+, positive response; −, negative response.

Microautoradiography and fluorescence in situ hybridization.

Substrate uptake tests using radiolabeled [U-¹⁴C]maltose, [U-¹⁴C]glucose, [U-¹⁴C]acetate, and [U-¹⁴C]pyruvate were performed with KSB3 cells in the sludge granules harvested from reactor II. These substrates were spiked into the medium to obtain a final radioactivity of 10 μCi mmol⁻¹. After washing with phosphate-buffered saline, approximately 40 mg of UASB granular sludge was added to the anaerobic medium (3 ml) supplemented with the unlabeled substrates to obtain a final concentration of 0.1 mM, with a radioactivity of 370 kBq, and anaerobically cultivated at 37°C with gentle shaking at 60 rpm for 6 h. A washing step with UASB granules and subsequent liquid scintillation counting for radioactivity of the biomass were conducted as described previously (12, 22). Cell fixation of the sludge samples and whole-cell in situ hybridization of dispersed sludges were performed in accordance with a previously described method (27). A 16S rRNA-targeted probe, 3KSB1238, labeled with Cy3 fluorescent dye was used for in situ hybridization experiments (39). The sludge samples were simultaneously stained with Syto 60 (Invitrogen) for 10 min in the dark to visualize total cells. Microautoradiography (MAR) was carried out directly on a coverslip as described previously (12, 22). After fluorescence in situ hybridization (FISH) experiments, an autoradiographic liquid film emulsion (LM-1; GE Healthcare) was used. The optimal exposure time was determined to be 2 days. A Carl Zeiss (Vertrieb, Germany) model LSM510 confocal laser scanning microscope (CLSM) was used for the observation of MAR-FISH.

Enrichment and cultivation of KSB3 cells.

For the enrichment and cultivation of KSB3 bacteria, the fluffy part of the sludge was serially diluted in solid or liquid medium (described above) and was then cultivated under anaerobic conditions at different pHs (pH 5.0, 6.0, and 7.0) and different temperatures (25°C, 37°C, and 45°C). Agar and gellan gum were used as gelling regents for the preparation of the solid medium. The following carbohydrates (0.1 mM each unless specified otherwise) were tested for the growth of KSB3 cells in medium supplemented with yeast extract (0.01%): glucose, mannose, galactose, fructose, rhamnose, N-acetyl-glucosamine, glucosamine, sorbitol, trehalose, lactose, lactitol, maltose (0.1, 1, and 10 mM), sucrose, and raffinose. The following electron acceptors, which are generally used for cultivating denitrifiers and S reducers, were also tested with the addition of the substrates described above: 20 mM nitrate, 20 mM sulfate, 1 mM sulfite, and 20 mM thiosulfate. The cultivation of KSB3 cells under anaerobic conditions was tested in the absence or the presence of the external electron acceptors mentioned above. To check cells grown in all cultures, cell fixation and FISH experiments using the 3KSB1238 probe, specific for KSB3 organisms, were performed according to a method described in a previous report (39).

Analytical methods.

Short-chain fatty acids and other organic acids were determined by using high-pressure liquid chromatography (HPLC) with a Shimadzu SCR-102-H column and a Shimadzu CDD-6A conductivity detector (10). The total carbohydrate concentration was determined colorimetrically by the anthron method (7). Carbohydrates in wastewater were identified by HPLC using a Shimadzu Shim-Pack ISA-07 column (eluent, 0.1 M potassium borate [pH 8.0] and 0.4 M potassium borate [pH 9.0]; column temperature, 65°C) and a Shimadzu model RF-10A spectrofluorometric detector. Proteins were determined by a method described previously by Lowry et al. (17). The COD was determined with a standard dichromate method (5). The measurement of mixed-liquor suspended solids (MLSS) was performed in accordance with a method described previously (5). Methane and other gaseous fractions in biogas were determined by gas chromatography (with a GC-8A instrument [Shimadzu], a GL Science model 370 gas chromatograph with thermal conductivity detection, packing material [Unibeads C], and a column temperature of 145°C) with a thermal conductivity detector.

RESULTS

Optimization of a scissor probe for rRNA-based quantification by the RNase H method.

For the quantification of KSB3-type populations in sludge samples on the basis of rRNA, we first optimized the reaction conditions for the RNase H method using a KSB3-targeted probe (3KSB1238 probe). The 3KSB1238 probe successfully cleaved KSB3 16S rRNA at a formamide concentration of 25% (cleavage coefficient, 0.93), while the 16S rRNA of Sphaerobacter thermophilus (a nontargeted reference strain with 3-base mismatches against the probe sequence) was not cleaved at formamide concentrations higher than 15% (see Fig. S1 in the supplemental material). Thus, an optimum formamide concentration of 25% was chosen for the 3KSB1238 probe (Fig. S1 and Table S1).

Real-time quantification of the KSB3 population in sludge of UASB reactors operated continuously.

First, we monitored the activity of KSB3 cells based on rRNA in a full-scale UASB system for which severe filamentous bulking events were frequently observed (39). By continuously operating the system fed with actual organic wastewater, the response of KSB3 rRNA was examined in relation to operational conditions and wastewater compositions. The changes in the population sizes of members of the Bacteria, the Archaea, and KSB3 in the sludges of full-scale UASB reactors (reactors I and II) were evaluated during an operation time of 75 days using the RNase H method. To identify the factors affecting the population size of KSB3 filaments, the daily changes in reactor temperature, loading, methane formation, MLSS of the retained sludges, and characteristics of the influent and effluent streams were simultaneously monitored for both the reactors (Fig. 1 and see Fig. S2 in the supplemental material). The influent stream contains mainly saccharides, and these compounds are degraded to form methane in reactors I and II. During the experimental period, the pHs and temperatures in reactors I and II were maintained at approximately 6.7 and 37°C, respectively (Fig. S2A and S2B). The COD loading rate for the reactors was also maintained at approximately 10 kg COD m⁻³ day⁻¹ (Fig. 1A), and the reactors exhibited good COD (>80%) (Fig. S2D) and VFA (Fig. S1E to S1G) removals. The carbohydrate concentration in the wastewater significantly fluctuated, ranging from approximately 170 to 1,500 mg liter⁻¹, but the carbohydrates in the wastewater were almost completely removed in the reactors (Fig. 1B).

FIG. 1.

Monitoring of reactor parameters (COD loading rate and carbohydrate concentration) in influent and effluent wastewaters together with the bacterial, archaeal, and KSB3 population abundances (as revealed by the RNase H method) in full-scale UASB reactors (reactors I and II) operated continuously. (A) Daily changes in COD loading rates in reactors I (closed squares) and II (open triangles). CODcr loading rate indicates a daily loading rate of wastewater on the basis of COD measured by the standard dichromate method (5). (B) Daily changes in carbohydrate concentrations in influent wastewater (open circles) discharged from an isometric sugar-manufacturing factory and those in effluent wastewaters that were discharged from reactors I (open triangles) and II (closed squares). The line (carbohydrate concentration above 500 mg liter⁻¹ in the wastewater) in the figure indicates the point above which an increase in the KSB3 population was observed. (C) Changes in bacterial and archaeal population abundances in reactor I (open circles, Bacteria; open triangles, Archaea) and reactor II (closed squares, Bacteria; closed diamonds, Archaea) operated continuously. (D) Change in relative KSB3 population abundances in the total prokaryotic populations in the same reactors I (open triangles) and II (closed squares) operated continuously. All experiments for C and D were run in triplicate, and error bars show standard deviations.

Throughout the experiment, the relative abundances of members of the domains Bacteria and Archaea ranged from 73 to 86% of the total 16S rRNA for the Bacteria and from 22 to 30% for the Archaea; the ranges remained constant throughout the experimental period (Fig. 1C). During the experimental period, it was found that fluffy, white, cotton-like sludges were formed on days 18 to 40 and at around day 70, similarly to the case of the filamentous bulking which we previously observed (39). However, the fluffy sludge formation caused no serious washout of the sludges retained in the reactors, and the MLSS of the sludges did not significantly change throughout the experiment (see Fig. S2C in the supplemental material). In the sludges, the KSB3 population was estimated to be in the range of 15 to 55% of the total 16S rRNA level (Fig. 1C). The fluffy sludge formation seemed to be linked with the increase in the KSB3 population; i.e., the fluffy sludge was formed when the KSB3 population increased to 30 to 55% (Fig. 1). It was further indicated that the increase in the KSB3 population (>30%) was linked with the increase in the carbohydrate concentration in the wastewater (the carbohydrate loading to the reactors); carbohydrate concentrations above 500 mg liter⁻¹ in the wastewater seemed to cause an increase in the KSB3 population (Fig. 1). It was also noted that the decrease in the carbohydrate concentration (<500 to 600 mg liter⁻¹) might lead to a decrease in the KSB3 population (Fig. 1). Other parameters, such as pH, temperature, COD loading rate, VFA level, and protein concentration, did not seem to be linked with the change in the KSB3 population (Fig. 1 and Fig. S2). These observations suggest that carbohydrate loading is an important factor affecting the abundance of KSB3 cells in the sludges. In the wastewater with a carbohydrate concentration of approximately 650 mg liter⁻¹, the major components of the carbohydrates were identified to be maltose and mannosamine (550 mg liter⁻¹), with relatively low concentrations (5 to 20 mg liter⁻¹) of glucose, fructose/arabinose, ribose, xylose, and lactose (Table 1).

Ecophysiological traits of KSB3 cells.

To determine the types and concentrations of substrates preferred by KSB3 cells, KSB3 RNA responses were analyzed by using carbohydrates that were found to be present in the wastewater (Table 1) and VFAs as the spike-in substrates in batch cultivation tests (Table 2). Regarding the substrates tested, KSB3 bacteria showed positive RNA responses (i.e., an increase in their RNA level during the batch cultivation) when the sludges were anaerobically cultivated with maltose, glucose, fructose, and maltotriose in the presence of yeast extract for 12 h. The addition of yeast extract to the medium was always required for the positive KSB3 RNA response (Table 2). The spike-in test with maltose (concentration, 0.1 mM [34 mg liter⁻¹]) gave the best result for increasing the KSB3 RNA level. Maltose at concentrations ranging from 34 to 340 mg liter⁻¹ (0.1 to 1 mM) showed increases in the KSB3 RNA level (25 to 156% increases in their RNA levels), while the concentrations outside this range resulted in decreases in the KSB3 RNA level. These results suggest that KSB3 filaments utilize relatively low concentrations of disaccharides (e.g., maltose) and monosaccharides (e.g., glucose and fructose) in the granular sludge.

To determine whether the carbohydrates that were identified to activate KSB3 RNA were actually taken up by the KSB3 cells, MAR-FISH experiments with KSB3 cells were performed. MAR-FISH experiments with [¹⁴C]glucose or [¹⁴C]maltose (each 0.1 mM) showed that the KSB3 cells took up radiocarbons under the same incubation conditions as those for the spike-in substrate tests (Fig. 2). The MAR-positive KSB3 cells account for almost all of the 3KSB1238-reactive cells when [¹⁴C]maltose was used but only half of those when [¹⁴C]glucose was used (data not shown). To identify the cross-feeding that is apprehended for the MAR-FISH experiments, fermentation products from glucose (0.1 mM) and maltose (0.1 mM) by the UASB sludge biomass were identified in the presence of yeast extract (0.01%) and bromoethanesulfonate (5 mM, for inhibiting methanogenesis), resulting in the formation of acetate and pyruvate as major products from these substrates (data not shown). MAR-FISH experiments with [¹⁴C]acetate or [¹⁴C]pyruvate showed that the KSB3 cells did not take up these by-products.

FIG. 2.

MAR-FISH experiments on microbial cells in dispersed granular sludges of reactor II with [¹⁴C]maltose (A to C) and [¹⁴C]glucose (D to F). (A and D) Photomicrographs of whole cells stained with Syto 60 (bar, 20 μm). (B and E) FISH images of the same field as those in A and D showing filamentous cells hybridized with the KSB3-specific 3KSB1238 probe (bar, 20 μm). (C and F) MAR images of the same fields as those in A and D indicating the uptake of [¹⁴C]maltose (C) and [¹⁴C]glucose (F) (bar, 20 μm).

Attempt at cultivation of KSB3 cells.

We attempted to cultivate KSB3 bacteria in the solid and liquid media supplemented with yeast extract (0.01%) and with relatively low concentrations of carbohydrates, such as monosaccharides and disaccharides, especially maltose at the concentration (0.1 to 1 mM) suggested to be preferable for KSB3 cells in the analyses shown above, under anaerobic conditions. Attempts were also made at cultivation under various pH and temperature conditions. FISH analyses with the 3KSB1238 probe were performed to check whether cells grown in the cultures are KSB3 bacteria. However, it was not successful for detecting KSB3 cells in any grown cultures (data not shown). In the almost all media, Streptococcus-type bacteria grew immediately for 1 to 2 days. In some media, filamentous bacteria affiliated with the class Anaerolineae of the phylum Chloroflexi along with other bacteria showing rod or coccus morphotypes were also found (data not shown).

DISCUSSION

Our study provided clear evidence that KSB3 filaments in UASB sludge are the primary fermenters degrading carbohydrates, as previously speculated, and that the increase in the carbohydrate loading triggers an outbreak of the KSB3 filaments, causing anaerobic sludge bulking (39). Furthermore, maltose was identified to be the most preferable substrate for KSB3 cells in this study. Low concentrations (34 to 340 mg liter⁻¹) of maltose were found to be taken up by KSB3 filaments, as determined by MAR-FISH, and to increase KSB3 rRNA levels in batch cultivation experiments, suggesting that maltose supports the metabolism (and possibly the growth) of KSB3 cells. Maltose was actually found to be the main component of the carbohydrates in the wastewater. The concentration of carbohydrates in the wastewater examined was relatively high (>300 mg liter⁻¹); however, considering the fact that the carbohydrates in the wastewater are rapidly diluted when they are fed into the bottom of the UASB reactor vessels, the carbohydrate concentrations in the bottom of the reactor vessels may be the same as those (0.1 to 1 mM [34 to 340 mg liter⁻¹]) that are preferable for KSB3 cells. On the basis of the results obtained in this study, the following physiological features of the KSB3 filaments are suggested: (i) they are anaerobic, (ii) they are primary fermenters degrading maltose and glucose in the presence of yeast extract, and (iii) they have preferable concentrations of substrates (0.1 to 1 mM) for metabolism (and growth). We attempted to cultivate KSB3 cells using anaerobic media supplemented with maltose at the preferable concentrations several times, but all the attempts have not been successful to date. It is still not clear why the cultivation of KSB3 cells is so difficult despite the fact that the growth can be stimulated (based on RNA) using the same medium. However, one may speculate that KSB3 filaments require constant carbohydrate loading for growth, as observed for the reactor; i.e., a continuous supply of carbohydrates (maltose) at low concentrations may be an important factor for the selective enrichment of KSB3 filaments. Considering such traits of KSB3 cells, one possible way to cultivate and enrich them is to use a constant-flow chamber instead of batch cultures in the laboratory, which are generally used for anaerobic cultivation.

In a previous study, it was demonstrated that anaerobic sludge bulking was caused by an outbreak of KSB3 filaments (39). This study showed the link between the carbohydrate (mainly maltose) loading to the UASB system and the outbreak of KSB3 cells. It has long been noted that anaerobic bulking phenomena caused by the outbreak of filamentous cells are likely related to changes in carbohydrate loading (1, 6, 26, 36, 40), but there has been no direct evidence indicating a link between microbial community changes causing bulking and changes in the reactor operation parameters. Under the conditions of the present UASB system, carbohydrate concentrations higher than 400 to 500 mg liter⁻¹ may trigger an increase in the population of KSB3 filaments, and a KSB3 population higher than 30 to 40% may cause the formation of fluffy sludges, which can initiate bulking. Such indicators may be limited only to the present system and may not be common in other anaerobic wastewater treatment systems. However, our findings further noted that the acidification step prior to UASB treatment (methanogenic process) might be important for the control of the bulking of sludge granules when high-strength organic wastewater containing high concentrations of carbohydrates is treated anaerobically, as was suggested previously (1, 6, 26, 36, 40). Other than the two-step UASB systems consisting of acidification and methanogenic processes, the anaerobic baffled reactor system may be a good alternative for the anaerobic treatment of high-strength organic wastewater containing carbohydrates at high concentrations (>400 mg liter⁻¹) (3).

Thus far, more than 30 phylotypes belonging to the candidate phylum KSB3 with approximately 15% sequence divergence have been identified, with no cultured representatives. They have been retrieved from anaerobic sludge (19, 39), soil (11), marine sediments (4, 9, 15, 32), and lakes (24). Among them, four phylotypes have been found in the same UASB system (for example, DDBJ accession number AB218870); all these phylotypes shared more than 99% similarity in their 16S rRNA gene sequences (39). Thus, the KSB3 cells in the UASB system were tentatively considered a single “species,” and their ecophysiology was elucidated in this study. It remains unclear whether the physiological properties of KSB3 cells in this study are common to all the phylotypes of the KSB3 phylum. However, similar phylotypes found in other UASB systems (>98% 16S rRNA gene sequence similarity with the sequence reported under accession number AB218870) (19) may have similar physiological traits, because those systems also treated wastewater containing high concentrations of carbohydrates.

Lastly, it should be noted that the rRNA-based batch cultivation test with spike-in substrates used in this study is a useful means of exploring the substrate utilization profile of uncultured microbes in situ. To analyze the physiological traits of uncultured cells, MAR-FISH (23), FISH combined with stable-isotope Raman microscopy (8, 35), and FISH combined with stable-isotope secondary-ion mass spectrometry (35) are powerful tools, and they have contributed to unveiling the physiologies of many uncultured microbes (e.g., see references 8, 18, and 33). In contrast, because rRNA-based batch cultivation tests allow us to work with bulk samples to estimate substrate ranges and preferable concentrations in very simple, rapid, cost-effective, and quantitative ways, such techniques should be very powerful to gain insight into the ecophysiology of uncultured microorganisms and thus provide hints for cultivating uncultured microorganisms.