Biofilm microbial community of a thermophilic trickling biofilter used for continuous biohydrogen production


  • Yeonghee Ahn,

    Corresponding author
    1. Department of Chemical and Biomolecular Engineering, Korea Advanced Institute of Science and Technology, Guseong-dong, Yuseong-gu, Daejeon 305–701, Republic of Korea
    Search for more papers by this author
  • Eun-Jung Park,

    1. Department of Chemical and Biomolecular Engineering, Korea Advanced Institute of Science and Technology, Guseong-dong, Yuseong-gu, Daejeon 305–701, Republic of Korea
    Search for more papers by this author
  • You-Kwan Oh,

    1. Department of Chemical and Biochemical Engineering, Pusan National University, San 30 Jangjeon-dong, Kumjung-ku Pusan 609–735, Republic of Korea
    2. Institute for Environmental Technology and Industry, Pusan National University, San 30 Jangjeon-dong, Kumjung-ku Pusan 609–735, Republic of Korea
    Search for more papers by this author
  • Sunghoon Park,

    1. Department of Chemical and Biochemical Engineering, Pusan National University, San 30 Jangjeon-dong, Kumjung-ku Pusan 609–735, Republic of Korea
    2. Institute for Environmental Technology and Industry, Pusan National University, San 30 Jangjeon-dong, Kumjung-ku Pusan 609–735, Republic of Korea
    Search for more papers by this author
  • Gordon Webster,

    1. Cardiff School of Biosciences, Cardiff University, Main Building, Park Place, Cardiff CF10 3TL, Wales, United Kingdom
    2. Cardiff School of Earth, Ocean and Planetary Sciences, Cardiff University, Main Building, Park Place, Cardiff, CF10 3YE, Wales, United Kingdom
    Search for more papers by this author
  • Andrew J. Weightman

    1. Cardiff School of Biosciences, Cardiff University, Main Building, Park Place, Cardiff CF10 3TL, Wales, United Kingdom
    Search for more papers by this author

  • Edited by E. Baggs

*Corresponding author. Tel.: +82 42 869 3941; fax: +82 42 869 3910., E-mail address:


Molecular methods were employed to investigate the microbial community of a biofilm obtained from a thermophilic trickling biofilter reactor (TBR) that was operated long-term to produce H2. Biomass concentration in the TBR gradually decreased as reactor bed height increased. Despite this difference in biomass concentration, samples from the bottom and middle of the TBR bed revealed similar microbial populations as determined by PCR-DGGE analysis of 16S rRNA genes. Nucleotide sequences of most DGGE bands were affiliated with the classes Clostridia and Bacilli in the phylum Firmicutes, and the most dominant bands showed a high sequence similarity to Thermoanaerobacterium thermosaccharolyticum.


Microbial H2 production can be either photosynthetic or non-photosynthetic. Facultative or obligate anaerobes perform non-photosynthetic (or fermentative) H2 production. Fermentative H2 production is light independent and generally faster. To date, most studies on continuous processes for fermentative H2 production have employed suspended culture systems such as a conventional stirred tank reactor (CSTR) [1–8] since these systems are relatively simple and easy to operate. However, washout of cells results in low biomass in these systems.

Considering the low biomass in suspended systems, employing biofilms would appear to be a better approach for fermentative H2 production since biofilms accommodate higher biomass and fermentation rate [9–11]. Besides, a reactor employing a biofilm generally shows more stable performance as biofilm cells are more resistant to changes in environmental conditions (e.g., pH, temperature, organic load, etc.) [9–13].

Trickling biofilter reactors (TBR) employ biofilms formed on supporting matrices packed inside the reactors [14]. Biofilm in TBR can degrade organic compounds in wastewaters that are trickled over the biofilm. Thermophilic (45–65°C) TBR can take advantage of characteristics of biofilm and thermophilic bacteria to achieve high and stable production of H2. Thermophilic bacteria show a higher degradation rate of organic substances than mesophilic (30–40°C) bacteria under anaerobic conditions [15]. Since H2 is less soluble at high temperature, thermophilic TBR can reduce partial pressure of H2 and alleviate inhibition of H2 production [16,17]. Thermophilic TBR can also take advantage of high temperatures of wastewaters discharged from industries such as canneries, distilleries, and food process plants.

Information on the microbial community in fermentative H2-producing reactors is necessary to better understand and improve the process. However, only a few studies on the microbial community in such reactors have been reported [5,18,19], although many studies have reported physicochemical aspects of the process [4,6–11,20,21]. Despite many advantages of a TBR system for H2 production, microbial community in this system has not been reported. Therefore, the aim of this study was to investigate the microbial community in a thermophilic TBR that showed superior H2 production rate and long-term stability of performance. This is the first report on microorganisms in a thermophilic TBR used for continuous H2 production.

2Materials and methods

2.1Operation of TBR and sampling of biofilm

Biofilm samples were obtained from a bench-scale (inner diameter 8 cm; length 75 cm; working volume, 2 l) thermophilic TBR [22] packed with fibrous polymeric matrices (size, 2 cm × 1.5 cm × 1.5 cm; Model DSRN, Dockil Felt Industry, Ltd., Seoul, Korea). Inoculum of the TBR was obtained from a bench-scale thermophilic CSTR used for H2 production. Biomass concentration and H2 production activity of the inoculum were 0.72 g VSS l−1 and 49 mmol H2 l−1 d−1, respectively. The CSTR was initially inoculated with heat-treated anaerobically digested sludge obtained from a wastewater treatment plant. The feed for the CSTR and the TBR was a glucose-based synthetic medium [22]. Feed solution was purged with N2 gas and filter-sterilized before introduction into the reactor.

TBR was operated stably at 55–64°C for 234 d. The maximal volumetric H2 production rate and H2 yield were 1050 ± 63 mmol H2 l−1 d−1 and 1.11 ± 0.12 mol H2 mol−1 glucose, respectively [22]. Typical composition of biogas from the reactor was as follows (v/v): H2, 53 ± 4%; CO2, 47 ± 4%. Methane in the biogas was not detected throughout the normal operation of the TBR, although methane was detected during the initial start-up period. Major organic acid products were lactate, n-butyrate, and acetate. Except the initial start-up period, lactate concentration was the highest among the organic acids produced by the TBR during the experiment [22].

The H2 production activity of the TBR was extremely sensitive to oxygen and therefore opening the reactor to take biomass samples was not possible during the operation of the TBR. Supporting matrix samples were taken from different heights of the TBR bed at the end (day 234) of the reactor operation. Biofilm cells released from the supporting matrices were used for microbial analyses. Volatile suspended solid (VSS) was measured according to the Standard Methods [23] as described previously [22].

2.2DNA extraction

Two pieces of supporting matrices were placed into 30 ml of phosphate-buffered saline (PBS; 0.13 M NaCl, 10 mM sodium phosphate buffer, pH 7.2) and vortexed vigorously for 5 min to release cells from the matrices. Cells were harvested and resuspended in 10 ml of PBS. Total genomic DNA was extracted from the cells harvested from 1 ml of the biofilm resuspension. In addition, cells harvested from 8 ml of TBR inoculum were also rinsed with PBS and used for DNA extraction. A soil DNA isolation kit (Mo Bio Labs. Inc., Solana Beach, CA) was used to extract DNA according to the manufacturer's instructions.

2.3Polymerase chain reaction

The extracted DNA was used as template in polymerase chain reaction (PCR) for DGGE analysis and methanogen detection. For DGGE analysis, 16 S rRNA gene fragments were amplified with PCR primers 357f-GC and 518r [24], producing ?194-bp PCR products (corresponding to positions 341–534, Escherichia coli numbering, [25]). PCR products of the correct size were confirmed by agarose gel electrophoresis prior to DGGE.

Two sets of mcrA gene primers were employed to detect methanogens; ME1 and ME2 [26], and the primers developed by Luton et al. [27]. Gene mcrA encodes α-subunit of methyl coenzyme M reductase, the key enzyme of methanogenesis. PCR conditions were as previously described [26,27]. Positive controls in the mcrA PCR employed total DNA extracted from the reference strains; Methanobacterium bryantii DSM 863T, Methanosarcina barkeri DSM 10131, Methanosaeta concilli DSM 3671T, Methanospillium hungatei DSM 864T, and Methanococcus jannashii DSM 2661T, belonging to the orders Methanobacteriales, Methanosarcinales, Methanomicrobiales, and Methanococcales, respectively. Negative controls were Pseudomonas putida DNA and no added template DNA.

PCR mixtures contained 20 pmol of primers (synthesized by MWG, Germany), 1 × reaction buffer, 0.5 mM each dNTP, 3 mM MgCl2, 1.25 U Taq DNA polymerase (Bioline Ltd., London, UK) and 1.0 μl of template DNA made up to 50 μl with molecular grade water. The PCR employed 1 ng of template DNA from each reference strain. Five microliters of PCR products were analyzed on 1% (w/v) agarose gel using a standard electrophoresis protocol.

2.4DGGE analysis

DGGE was performed using the PCR-amplified 16S rRNA gene fragments to characterize the microbial community in the reactor. DGGE condition was optimized by changing electrophoresis time to separate the amplified DNA fragments. PCR products were separated using a DCode System (Bio-Rad, Hercules, CA) at 200 V, 60°C for 4 h. Samples (?200 ng) were loaded on a 10% (w/v) polyacrylamide gel (acrylamide: N, N′-methylenebisacrylamide, 37.5:1, Bio-Rad) in 1× TAE buffer (2 M Tris base, 1 M glacial acetic acid, 50 mM EDTA, pH 8.0). The denaturing gradient in the gel was generated by mixing two stock solutions of 10% polyacrylamide containing 40% and 60% denaturant: 100% denaturant was 7 M urea and 40% (w/v) formamide. Denaturing gradient increased in the direction of electrophoresis. After electrophoresis, the gel was stained with GreenStar™ (Bioneer Co., Daejeon, Korea) and DNA was visualized on a UV transillumination table. Digitized DGGE images were obtained using Scion Image software (Scion Co., Frederick, MD). DNA band intensities of DGGE images were determined using SigmaGel (Jandal Scientific, San Rafael, CA). Major DNA bands were excised from DGGE gels and re-amplified by PCR for nucleotide sequencing as described previously [28]. Nucleotide sequences were analyzed using the CHIMERA CHECK program of the Ribosomal Database Project II to screen for and eliminate chimeric sequences [29]. Nucleotide sequences were also screened against GenBank database using BLASTN (version 2.2.10) [30] to identify the most similar sequences in the database.

2.5Whole cell hybridization and DAPI (4′,6-diamidino-2-phenylindole) staining

Biofilm formed on the supporting matrices was used for whole cell hybridization and DAPI staining. Biofilm suspension was obtained as described above. An appropriate volume of biofilm suspension and TBR inoculum was taken and cells were rinsed with PBS. Rinsed cells were fixed and used for whole cell hybridization and DAPI staining as described previously [31]. Fluorescently labeled probes (synthesized by ThermoHybaid GmbH, Ulm, Germany) used in this study were ARC915 for domain Archaea[32] and LGC354A-C for part of Firmicutes[33]. Probes were 5′-end labeled with tetramethyl rhodamine. Hybridized or stained cells were then visualized using a Zeiss Axiolab (Jena, Germany) with a 50 W mercury lamp. Images were obtained with a digital camera (Model AxioCam; Zeiss) mounted on the microscope. Cell counting was done with at least 10 random microscopic fields with more than 100 cells per field and MS Excel was used for statistical analysis.

3Results and discussion

3.1Abundance of biomass

Biomass amount (VSS) in the TBR gradually decreased as bed height increased from the bottom of the bed (Fig. 1). VSS (g l−1) found at different heights (from the bottom of the bed) was in the following ranges: 0 cm, 24.1 ± 3.0; 14 cm, 19.6 ± 1.2; 26 cm, 17.7 ± 2.8; 40 cm, 1.6 ± 0.5. Biomass observed at the bottom was 26.6% higher than that found at the height of 26 cm. Biomass measured at the top of the bed was 93.4% less than that found at the bottom of the bed. The decreasing trend seems to be related with the flow rate of the biogas. As the height increases, the gas flow becomes faster and a higher shear results. The highest linear gas velocity which will be observed at the top of the reactor is estimated to be 1.4 cm min−1 when the TBR is operated at a hydraulic retention time (HRT) of 2 h and 20.6 g glucose l−1[22]. A low biomass developed at the top section of the reactor could be attributed to the high shear rate caused by recirculation liquid flow and gas flow [14].

Figure 1.

Abundance of microorganisms in the reactor bed. Bed heights 0 and 40 cm indicate bottom and top of the reactor bed, respectively. Error bars represent SDs of the mean; n= 5.

The biomass concentration (17.7–24.1 g VSS l−1) observed in the TBR demonstrated that the system was able to accommodate a higher biomass than previously observed in the other systems used for fermentative H2 production [6–11]. However, the TBR showed no clogging throughout the operational period. Since the TBR was a Pyrex glass column, any clogging would have been observed. Shear stress caused by fast biogas flow (1.4 cm min−1) and recirculation liquid flow (210 l d−1) through the TBR seems to remove excess biomass [14] as evidenced by biomass observed in the effluent from the TBR. Effluents from the TBR showed biomass concentration in the range of 0.32–0.77 g VSS l−1 throughout the experiment. No apparent clogging in the TBR could have been due to slower growth of thermophilic bacteria comprising the microbial community in the TBR. In contrast, a mesophilic TBR used for H2 production exhibited clogging due to fast microbial growth (data not shown).

3.2DGGE profiles of microbial community

DGGE was performed to compare 16S rRNA gene fragment profiles of biofilm cells obtained from two different heights of the TBR bed. Despite the differences in biomass concentrations with bed height, samples from 0 and 26 cm heights of the TBR bed revealed similar DGGE profiles (Fig. 2), suggesting similar microbial populations at these heights.

Figure 2.

DGGE profile of 16S rRNA gene fragments. The fragments were PCR-amplified from the total DNA extracted from biofilm cells in the TBR used for H2 production. Lanes: 1, biofilm sample taken from the middle (26-cm high from the bottom) of the reactor; 2, biofilm sample taken from the bottom of the reactor; 3, TBR inoculum. Arrows indicate DNA bands that were excised and analyzed for nucleotide sequences.

The DGGE profiles of TBR samples were also compared with that of the inoculum and clear differences in DGGE profiles were observed. TBR samples showed two strongly stained distinctive bands along with some additional weakly stained bands whereas the TBR inoculum showed at least eight distinctive bands with some other minor bands. The difference in DGGE banding pattern suggests that long-term operation of the TBR caused a change in the microbial community composition of the inoculated culture. The two strongly stained bands observed in DGGE profiles of TBR samples suggest the occurrence of strong enrichment culture in the TBR.

Major bands of the TBR and inoculum samples were phylogenetically related to Firmicutes (Table 1). Majority of the DGGE bands were affiliated with the classes Clostridia and Bacilli in the phylum Firmicutes. In the class Clostridia, DGGE bands closely related to the genera Thermoanaerobacterium and Mitsuokella were frequently observed. DNA bands affiliated with T. thermosaccharolyticum appeared in biofilm cells of TBR while DNA bands affiliated with M. jalaludinii appeared in the inoculum. H2 production activity of these two bacterial species has previously been reported [34,35] and 16S rRNA gene fragments or isolates affiliated with these two bacteria have been found in other anaerobic reactors used for thermophilic H2 production [5–7].

Table 1.  Characteristics of 16S rDNA fragments obtained from DGGE gel
DGGE bandaGenBank search result
Closest match (Accession No.)Isolated environment of closest matchSequence similarity (%)Taxonomic description (class) 
  1. aName of the DGGE bands in Fig. 2.

0–1Uncultured bacterial clone w.4 (AY212556)Manure97Bacteroides
0–6Uncultured bacterial clone 3C3d–18 (AB034087)Rumen92Clostridia
0–8Mitsuokella jalaludinii strain M9 (AF479674)Cattle rumen96Clostridia
0–11M. jalaludinii strain M9 (AF479674)Cattle rumen98Clostridia
0–12M. jalaludinii strain M9 (AF479674)Cattle rumen94Clostridia
0–13Uncultured bacterial clone pPD6 (AF252319)Compost93Bacilli
1–2Thermoanaerobacterium thermosaccharolyticum strain D120–70 (AF247003)Extraction juice from sugar beet factory98Clostridia
1–3T. thermosaccharolyticum strain D120–70 (AF247003)Extraction juice from sugar beet factory98Clostridia
1–7T. thermosaccharolyticum strain D120-70 (AF247003)Extraction juice from sugar beet factory97Clostridia
4–1Uncultured bacterial clone pPD6 (AF252319)Compost97Bacilli

Nucleotide sequence of the most strongly stained band (band 1–3) observed in the two TBR samples was most similar (98% similarity) to the 16S rRNA gene of T. thermosaccharolyticum strain D120–70 (Accession No. AF247003) [36] in the GenBank database. Another dominant band 1–2 was also most closely related (98% sequence similarity) with T. thermosaccharolyticum strain D120-70. However, bands 1–2 and 1–3 differed by 1 base in their nucleotide sequences which may represent different strains or different copies of the 16S rRNA gene in the same strain. Comparison of DGGE band intensities in each lane showed reproducible data; constantly, the intensities of bands 1–2 and 1–3 in the TBR samples were 18.1 ± 0.3% and 23.9 ± 0.8%, respectively (n= 3). Intensities of the bands 0–1 and 0–8 observed in the TBR inoculum were 13.0 ± 0.1% and 11.9 ± 0.1%, respectively (n= 3).

Observing the two strong major bands affiliated with T. thermosaccharolyticum suggests that T. thermosaccharolyticum related organisms are highly enriched under the TBR conditions and that they play an important role in H2 production. Stable and high H2-production by the TBR was shown in the previous report [22], despite various parameters (pH, temperature, HRT, and COD load) applied to the reactor. Maintaining high biomass of acidogenic bacteria such as T. thermosaccharolyticum in the reactor could be a basis for the performance.

3.3Whole cell hybridization

Whole cell hybridization and DAPI staining were employed to analyze microorganisms present at different height of the reactor bed and in the inoculum. DAPI staining was employed to comparatively analyze morphology of microbial community and enumerate total cells while whole cell hybridization was performed to analyze specific phylogenetic groups of microorganisms.

Epifluorescence microscopic observation of DAPI-stained cells revealed that microorganisms in the TBR inoculum had different morphology from those found in TBR samples (Fig. 3). Long rod-shaped microorganisms were frequently found in the samples containing cells released from TBR biofilm. On the other hand, the inoculum consisted of microorganisms with various morphology, (long or short) rods and cocci. The result of epifluorescence microscopy supported DGGE results that samples obtained from TBR biofilm and TBR inoculum showed different microbial populations resulting from long-term operation of the TBR.

Figure 3.

DAPI staining (A and D) and whole cell hybridization (B and E) of microorganisms with the probes LGC354A-C. A–C, inoculum of TBR; D–F, cells released from biofilms formed on supporting matrix samples taken at the bed height of 26 cm; C and F, overlaid images of DAPI staining and whole cell hybridization. All images were viewed by epifluorescence microscopy (1000×).

Although methane was not detected throughout the normal operation of the reactor, we cannot eliminate the possibility that the TBR might contain methanogens as potential H2-consumers. Whole cell hybridization was performed using probe ARC915 specific for the domain Archaea where methanogens belong [32]. Cells hybridized with the ARC915 probe were not detected in the samples obtained from the inoculum and TBR, suggesting few or no methanogens in the samples. This result was consistent with the gas analysis results that methane was not detected throughout the normal operation of the TBR.

The H2 production rate (1050 ± 63 mmol H2 l−1 d−1) of the TBR is superior to the other fermentative H2-production systems reported to date. The H2 yield (1.11 ± 0.12 mol H2 mol−1 glucose) of the TBR is reasonable but not superior [2,3]. In fermentative H2 production, reduced end products such as alcohols and lactic acid represent hydrogen that has not been liberated as hydrogen gas [16]. Thus, the production of reduced end products is associated with low H2 yield [16,22,37]. Phylogenetically, lactic acid (producing) bacteria (LAB) can be divided into two groups [38,39]; most LAB belong to Firmicutes, while some other LAB belong to Actinobacteria. In the phylum Firmicutes, the order Lactobacillales contains the most important genera of LAB; Streptococcus, Lactobacillus, Lactococcus, Enterococcus, Pediococcus, and Leuconostoc.

Among the major organic acids produced by the TBR, lactic acid concentration was the highest [22]. TBR samples contained low number (less than 5% of total cells) of cells detected by the probes LGC354A-C, as determined by whole cell hybridization (Fig. 3). The hybridized cells are likely to belong to members of the Bacillales, considering the probes are specific for the Bacillales and Lactobacillales orders of the Firmicutes[33] and that DGGE analysis performed in this study revealed 16S rRNA genes related to members of the Bacillales but not Lactobacillales. Therefore, the results of whole cell hybridization and DGGE suggest that other microorganisms and not members of Lactobacillales play a role in producing lactate in the TBR.

3.4PCR detection of methanogens

This study employed two different sets of mcrA-specific PCR primers to detect methanogens in the TBR and inoculum samples. Repeatedly, no amplification was obtained with the two sets of primers used (data not shown), suggesting very few or no methanogens in the TBR and inoculum samples, confirming the results of the whole cell hybridization and gas analysis. However, all reference methanogens used in this study showed PCR products of the correct sizes, except when ME1/ME2 primers and M. concilli DSM 3671T total DNA were employed, supporting a previous report that these two sets of primers have different phylogenetic coverage of methanogens [40].

However, the potential biases associated with DNA extraction and PCR amplification [41,42] may limit the quantitative conclusions that can be drawn from the PCR amplification of mcrA. Nonetheless, the results of gas analyses and whole cell hybridization suggested that H2 consumption by methanogens was negligible in the TBR. However, possibility of H2 loss via CO2-reductive acetogenesis [43] cannot be excluded considering the fact that TBR showed 27.8% of conversion efficiency of glucose to H2 (based on theoretical yield of 4 mol H2 mol−1 glucose when acetate is the sole byproduct [2,3]).

Now that we have our first insights into the microbial community of a thermophilic TBR used for continuous H2 production, it is clear from this study that bacteria related to T. thermosaccharolyticum play a major role in H2 production under the TBR conditions. Previous studies [5–7,44] that employed different carbon sources and reactor types also suggested the importance of these bacteria in thermophilic H2 production. High concentration of T. thermosaccharolyticum affiliated bacteria was successfully maintained within the TBR to show superior H2 production rate. Considering that T. thermosaccharolyticum can produce lactate by fermenting glucose [34,45] and that lactate concentration was the highest among the organic acids produced by the TBR, strategies should be developed to avoid lactate formation to increase H2 production yield since the pathway leading to lactate is not related to H2 production [16].


This work was sponsored by Korea Research Foundation Grant (KRF-2003–003-D00230).