The isolation and microbial community analysis of hydrogen producing bacteria from activated sludge


B. Jin, Center for Water Science and Systems, University of South Australia, SA 5095, Australia. E-mail:


Aims:  To profile the fractions of bacteria in heat-treated activated sludge capable of producing hydrogen and subsequently to isolate those organisms and confirm their ability to produce hydrogen.

Methods and Results:  Profiling the community composition of the microflora in activated sludge using 16S rRNA gene-directed polymerase chain reaction–denaturing gradient gel electrophoresis suggested that a majority of bacteria were various Clostridium species. This was confirmed by clone library analysis, where 80% of the cloned inserts were Clostridium sp. A total of five isolates were established on solid media. Three of them, designated as W1, W4 and W5, harboured the hydrogenase gene as determined by PCR and DNA sequence analysis (99% similarity). These isolates were similar to Clostridium butyricum and Clostridium diolis as determined by 16S rRNA gene sequence. A maximum hydrogen production yield of 220 ml H2 g−1 glucose was achieved by W5, which was grown on improved mineral medium by batch fermentation without pH adjustment and nitrogen sparging during fermentation. Accumulation of malic acid and fumaric acid during hydrogen fermentation might lead to higher hydrogen yields for W4 and W5. W1 is the first reported Clostridium species that can tolerate microaerobic conditions for producing hydrogen.

Conclusion: Clostridium species in heat-treated activated sludge were the most commonly identified bacteria responsible for hydrogen production. Specific genetic markers for strains W1, W4 and W5 would be of great utility in investigating hydrogen production at the molecular level. Two previously described primer sets targeting hydrogenase genes were shown not to be specific, amplifying other genes from nonhydrogen producers.

Significance and Impact of the Study: Clostridium species isolated from heat-treated activated sludge were confirmed as hydrogen producers during dark hydrogen fermentation. The isolates will be useful for studying hydrogen production from wastewater, including the process of gene regulation and hydrogenase activity.


The increasing global shortage of fossil fuels, climate change, environmental pollution and associated health problems are all driving the need for alternative forms of energy (Das and Veziroğlu 2001). Due to its environmental friendly nature and high energy yield (143 kJ g−1), which is about 2·75 times greater than hydrocarbon fuels (Boyles 1984), hydrogen is a promising fuel for the future. Hydrogen can be produced by either biological, photo-electrochemical or thermochemical processes. In comparison with other biological hydrogen production processes (e.g. using algae), dark fermentation appears to be the most attractive method because: (i) it can continually produce H2 without the need for light; (ii) it can use a variety of cheap carbon sources, such as starch contained in wastewater; (iii) by-products of the fermentation include butyric, lactic and acetic acid, which have alternative commercial value; and (iv) the bacteria used are anaerobic and, therefore, do not require costly sparging with O2 (Nath and Das 2004). The most common bacteria used in dark fermentation to produce H2 are Clostridium sp. (Yokoi et al. 2001; Liu and Shen 2004), Enterobacteriaceae (Podestăet al. 1997), Thermoanaerobacterium sp. (Zhang et al. 2003; Shin et al. 2004), Citrobacter sp. (Oh et al. 2002) and Rhodopseudomonas palustris (Oh et al. 2003). These organisms have the ability to produce H2 anaerobically in pathways coupled either to fermentation or CO oxidation (Posewitz et al. 2005).

Commercialization of dark fermentation has been hampered in the past due to low yields of H2 and relatively high production costs. Much work has attempted to elucidate the most suitable conditions for hydrogen production. However, yields are still low (Lee et al. 1999). To date, much of research has concentrated on optimization of reactor and fermentation conditions (Nath and Das 2004), and little work appears to have focused on the microbial community structure in activated sludge and how this impacts on H2 production. Recently, using polymerase chain reaction–denaturing gradient gel electrophoresis (PCR-DGGE), Chang et al. (2006) reported that clostridia represented the largest component of the microbial community in activated sludge that was generating H2. In addition, they also reported a positive relationship between mRNA levels of the hydrogenase gene and the level of hydrogen production. However, the identity of the main hydrogen producers is yet to be determined, and it is not known whether the production yield can be increased by manipulation of the community composition.

An important first step towards understanding the roles of various bacteria in the hydrogen production process is to identify and quantify the relevant bacteria, in particular those responsible for the bulk of the activity. Methods employed for microbial community investigation can generally be divided into culture-dependent or culture-independent methods. Traditional culture-dependent methods result in the characterization and isolation of bacteria that can be cultured on selective media. However, about 99% of environmental bacteria remain ‘unculturable’ and are still undescribed (Amann et al. 1995). Culture-independent approaches such as PCR–DGGE and 16S rRNA gene clone libraries compensate the shortcomings of traditional culture methods by providing the structure of microbial communities irrespective of how conducive the organisms are to the culture (You et al. 2000; van der Gucht et al. 2005).

In this study, the bacterial community profile of a H2 producing process was assessed by 16S rRNA gene-directed PCR–DGGE, and subsequent identification of the major bands was performed after cloning and DNA sequence analysis. A 16S rRNA gene clone library also confirmed the dominance of Clostridium species in this process. Organisms conducive to culture were isolated onto solid media and their potential to produce hydrogen was investigated by PCR amplification of the hydrogenase gene. The bacterial isolates containing the hydrogenase gene were confirmed to produce hydrogen during fermentation.

Materials and methods

Seed activated sludge

Seed activated sludge was collected from anaerobically digested sludge at the Bolivar Wastewater Treatment Plant, South Australia. To eliminate methane producing bacteria, the sludge was heat treated by three cycles of boiling at 100°C for 15 min as described by Wang et al. (2004). The pretreated sludge was seeded into sterilized potato wastewater (provided by Smiths Chips Ltd, Australia) in an anaerobic reactor with an 800 ml working volume. Total Kjeldahl nitrogen (TKN) of potato wastewater was 3·2 g kg−1. The total solid content was 0·01% and the starch content of potato wastewater was adjusted to 1%. The initial pH of fermentation was 6·2.

DNA extraction

Genomic DNA from pure cultures was extracted by rapid freeze-thawing (Kawai et al. 2002). Cell suspensions underwent three cycles of boiling at 100°C for 5 min followed by freezing in liquid nitrogen for 2 min. After being resuspended in 100 μl InstaGene™ Matrix (Bio-Rad Laboratories Inc., Hercules, CA), a cycle was carried as follows: 56°C for 10 min, 99°C for 10 min and then 56°C for 4 min in a GeneAmp PCR system 2400 (Applied Biosystems, Norwalk, CT, USA). Activated sludge from the exponential phase of hydrogen fermentation was taken in sterile 1·5 ml microfuge tubes, immediately closed and stored at −20°C until further processing. For DNA extraction, the sludge was centrifuged at 18 000 g (4°C for 10 min), and the resulting pellet was resuspended with 100 μl 10 mmol l−1 Tris–HCl, pH 8·5. DNA was subsequently extracted by using the Ultra Clean™ Soil DNA Kit (MoBio Laboratories Inc., Solana Beach, CA, USA), which has been described previously for the extraction of DNA from activated sludge (Kreuzinger et al. 2003). The extraction was confirmed by gel electrophoresis. DNA was stored at −20°C until further processing.

PCR–DGGE profiling of microflora

Nested-PCR amplification was performed as described by Hoefel (Hoefel et al. 2005). Primer set 27 F/1492R was used for the first round of amplification, followed by primer set 357 F-GC/518R for the second round (Table 1). Amplified DNA was examined by horizontal electrophoresis in 100 ml 1% (W/V) agarose gel with 5 μl aliquots of PCR product stained with 10 μl SYBR® Safe DNA gel stain (Invitrogen, Carlsbad, CA, USA). Before the addition of the DNA polymerase and template, the PCR mix was exposed to UV for 10 min to reduce the level of contaminating bacterial DNA that might be present in any of the reagents (Goldenberger and Altwegg 1995).

Table 1.   Primers used for PCR amplification in this study
PrimerSequence (5′–3′)TargetReference
E1fGCTGATATGACAATAATGGAAGAAHydrogenase geneChang et al. (2006)
E1rGCAGCTTCCATAACTCCACCGGTTGCACCHydrogenase geneChang et al. (2006)
L1fAAATCACCACAACAAATATTTGGTGCHydrogenase geneChang et al. (2006)
L1rACATCCACCAGGGCAAGCCATTACTTCHydrogenase geneChang et al. (2006)
27FAGAGTTTGATYMTGGCTCAGBacteria 16S rDNASuzuki and Giovannoni (1996)
357FCCTACGGGAGGCAGCAGBacteria 16S rDNAMuyzer et al. (1993)
357F-GCGC clamp-CCTACGGGAGGCAGCAGBacteria 16S rDNAMuyzer et al. (1993)
786FGATTAAATACCCTGGTAGBacteria 16S rDNANeimark and Kocan (1997)
518RATTACCGCGGCTGCTGGBacteria 16S rDNAMuyzer et al. (1993)
1492RTACGGYTACCTTGTTACGACTBacteria 16S rDNASuzuki and Giovannoni (1996)

DGGE was performed using the Bio-Rad D-GENE™ Denaturing Gel Electrophoresis System (Bio-Rad) as previously described (Hoefel et al. 2005). Nested-PCR products were applied directly to 8% (wv−1) polyacrylamide gels in 1x TAE [40 mmol l−1, Tris–base, 0·12% (vv−1) acetic acid (glacial), 1·0 mmol l−1 EDTA (pH 8·0)] with a denaturing gradient ranging from 40 to 70% (where 100% denaturant contained 7 mol l−1 urea and 40% formamide). Amplified DNA from Escherichia coli, Aeromonas hydrophila and Staphylococcus epidermidis was used as a DNA reference marker. After 16 h of electrophoresis at a constant voltage of 60 V at 56°C, gels were stained with a 1x SYBR Green I solution (FMC BioProducts; Rockland, ME, USA) in 1·TAE for 2 h. A gel documentation and analysis system (EDAS 290 with Kodak version 3·5·3 1D Image Analysis Software, Eastman Kodak Company, Rochester, NY, USA) was used to capture digital images of the stained gels.

Bands excised from the DGGE gel were cleaned using an UltraClean™ GelSpin DNA Purification Kit (MoBio Laboratories Inc.) and cloned into E. coli using the TOPO TA Cloning Kit (Invitrogen), following the manufacturer’s instructions. Plasmid inserts were re-amplified with the primer set 357 F-GC/518R and the products analysed by DGGE against the original nested PCR–DGGE profile. Inserts were sequenced by the Australian Genome Research Facility Ltd (Brisbane, QLD, Australia) using a BigDye Terminator Sequencing Reaction Kit (Applied Biosystems, Norwalk, CT, USA). Sequence chromatograms were analysed using SeqMan II version 4·05 (Dnastar, Madison, WI, USA). Sequence similarity searches were performed using the Basic Local Alignment Search Tool (BLAST) program to search the National Center for Biotechnology Information (NCBI) sequence database (

Phylogenetic analysis

Full-length 16S rDNA sequences from Clostridium roseum (Acc. No. Y18171), Clostridium diolis (Acc. No. AJ458417), Clostridium acetobutylicum (Acc. No. U16165), Clostridium beijerinckii (Acc. No. X68179), Clostridium saccharoperbutylacetonium (Acc. No. U16122), Clostridium butyricum (Acc. No. X68178), Clostridium punicem (Acc. No. X71857) and Clostridium tyrobutyricum (Acc. No. M59133) were obtained from GenBank. Additional sequence data for W1, W4 and W5 were generated by sequencing the 27 F, 357 F, 786 F and 1492R amplicon. Sequences were aligned using ClustalX version 1·8 (Thompson et al. 1997). Phylogenetic analyses were conducted on an approximate 1401 nucleotide fragment of the 16S rRNA gene using MEGA 3·1 (Kumar et al. 2004). In the case of W5, the 5′ end was truncated by 116nt, and C. roseum was truncated at the 3′ end by 34nt. Distance-based analyses were conducted using Tamura-Nei distance estimates and the Neighbor Joining algorithm with the pair-wise deletion option selected. Maximum parsimony analysis was conducted using all sites and close neighbour interchange for tree construction, with a search level of 1 and random addition of trees with 10 replications. All analyses were conducted using 1000 bootstrap replicates to determine the robustness of the resulting tree topologies.

Direct cloning and characterization of amplicons

Nested-PCR amplicons were cloned into E. coli directly using the TOPO TA Cloning® Kit (Invitrogen, CA, USA) following the manufacturer’s instructions. Plasmid DNA was extracted from clones by QIAprep® Spin Miniprep Kit (Qiagen GmbH, Germany) and amplified by 16S rRNA gene specific primer set 357 F and 518R. After analysis by DGGE, amplicons with distinct migration positions were sequenced as described above.

Isolation of culturable bacteria

Two types of media were used to isolate and culture bacteria from hydrogen-producing activated sludge: nutrient agar and blood agar (PP2001, Columbia, HBA, Oxoid). The plates were cultured in a MK3 Anaerobic Work Station (Don Whitley Scientific Limited, West Yorkshire, England) with 10% hydrogen and 90% nitrogen gas atmosphere at 35°C for 48 h. DNA was extracted from colonies picked randomly and amplified using primer set 357 F-GC and 518R, and then analysed by DGGE. As described above, amplicons with distinct migrations were sequenced.

Hydrogenase gene amplification by PCR

The two primer sets of Chang et al. (Chang et al. 2006) were used to amplify segments of the hydrogenase gene (Table 1). These primer sets, which target the same hydrogenase gene, were designed to be partially overlapping to verify if any resulting amplicons came from the same domain. The reaction conditions were modified to reduce nonspecific amplification as follows: 95°C for 10 min to activate the Taq polymerase (Amplitaq Gold, Applied Biosystems, Foster, CA, USA), followed by 40 cycles of denaturation (30 s at 94°C), annealing (1 min at 53°C), and extension (1 min at 72°C), with a final extension at 72°C for 10 min. Amplicons were examined by horizontal electrophoresis in a 1% agarose gel. Amplicons of the expected sizes were cleaned using an UltraCleanTM PCR Clean-up Purification Kit (MoBio, Laboratories Inc.) and sequenced as described above.

Measurement of hydrogen production, gas and organic acid analyses

Mineral medium [100 mmol l−1 phosphate, 3 g yeast extract per litre and 10 g glucose per litre (Jung et al. 1999)] was employed to grow bacterial isolates and to test their hydrogen production at 35°C (using a water bath) under anaerobic conditions, according to work described previously (Oh et al. 2002, 2003). Improved mineral medium was designed and employed as higher hydrogen yield and shorter fermentation period were observed when isolates grew in the medium that contained higher nitrogen concentration in this research. It consisted of 100 mmol l−1 phosphate, 17 g pancreatic digest of casein per litre, 3 g papaic digest of soybean meal per litre and 10 g glucose per litre. The working volume was 400 ml. The initial pH was 7·0. The inoculum was grown in the same medium in the anaerobic work station and transferred anaerobically at late exponential phase with a sterile hypodermic disposable syringe. Nitrogen purging of the medium was used before fermentation to help maintain anaerobic environment (Nath and Das 2004). Biogas produced was collected by the water release method.

The biogas was sampled using a glass syringe and analysed on a CP-3800 gas chromatograph (Varian Inc. CA, USA) equipped with a thermal conductivity detector. Two stainless steel columns were used together to detect H2, CO2 and CH4. The gas sample with carrier gas (argon at a flow rate of 30 ml min−1) was run through the first column (1·0 m × 1/8′′ × 2·0 mm SS packed with Molecular Sieve 5A, 80 = 100 mesh, Varian Inc) for 7 min and then the flow direction was changed to a second column and run for 5·5 min (5·0 m × 1/8′′ × 2·0 mm SS packed with Hayesepn, 80 = 100 mesh, Varian Inc.). This combination of columns allowed the testing of H2, O2, CO, CO2 and CH4. During the measuring period, the working temperatures of injector, detector, and column were kept at 50°C, 140°C and 40°C, respectively. Organic acids were analysed by HPLC using a ROA Organic Acid Column (Phenomenex, 300 × 7·8) and a refractive index detector (Varian, Model 350). The mobile phase was 4 mmol l−1 H2SO4 at a flow rate of 0·5 ml min−1 and the column temperature was 50°C. The results were the means of duplicate or triplicate determinations.


PCR–DGGE profiling of hydrogen-producing bacteria in activated sludge

Total genomic DNA was extracted from hydrogen-producing activated sludge at late exponential phase when maximum hydrogen production was observed. This DNA was used as a template for profiling the bacterial community composition by 16S rRNA gene-directed PCR–DGGE. As shown in Fig. 1, the community profile consisted of four major bands and possibly several faint, less well-defined bands, indicating a relatively low bacterial diversity within the community. This result is consistent with other investigations on hydrogen production by activated sludge, which showed that only a few bacterial species survive heat or acid pretreatment (Wang et al. 2004; Chang et al. 2006). The major DGGE bands were excised, cloned into E. coli and verified against the original DGGE profile following re-amplification using the original PCR–DGGE protocol. DNA sequence analysis revealed that band A was 96% similar to a species of Bacillus; band B was 99% similar to C. butyricum (Acc. No. X68178); band C was 99% similar to Clostridium sp. (Acc. No. AY082483); and band D was 100% similar to C. butyricum (Acc. No. X68178).

Figure 1.

 PCR-DGGE profile of bacteria in hydrogen-producing activated sludge. DNA extraction was performed in quadruplicate and amplified by nested PCR. The major bands are shown by arrows.

16S rRNA gene clone library

DGGE identified four dominant bacteria in the hydrogen-producing activated sludge. Further attempts to identify the presence of other less-abundant bacteria were performed with the construction of a 16S rRNA gene clone library using the products of the same PCR that had been analysed by DGGE. Twenty-seven clones were isolated and characterized by PCR–DGGE. Cloned inserts with unique DGGE migrations were sequenced and identified as C. butyricum (Acc. No. AY442812, 100% similarity), Clostridium sp. (Acc. No. AY082483, 100% similarity), Aneurinibacillus aneurinilyticus (Acc. No. AB211018, 100% similarity), Clostridium sp. (Acc. No. DQ168199, 100% similarity), and Clostridium sp. (Acc. No. AY082483, 96% similarity). The bulk of clones (80%) in the library belonged to the genus Clostridium. The bacterial community structure determined by the 16S rDNA clone library method was slightly different to the PCR–DGGE result, possibly because the library has the potential to detect minor species within the population that fail to amplify sufficiently to be detectable by PCR–DGGE.

Isolation of culturable bacteria from hydrogen-producing activated sludge

Following culture onto solid medium, colonies were randomly selected and characterized using the primer set 357 F-GC/518R and PCR–DGGE. Results suggested that five different bacterial species were isolated and designated W1, W2, W3, W4 and W5. The isolates differed in colony shape and haemolytic characteristics on blood agar (Table 2) and were identified using 16S rRNA gene sequence analysis. One isolate, designated W1, was similar (99%) to C. diolis (Acc. No. AJ458417) and grew only on nutrient agar. The other four isolates, designated W2 through W5, were similar to Bacillus sp. (Acc. No. AF169535, 96%), Swine manure bacterium (Acc. No. AY167932, 100%), C. butyricum (Acc. No. X68178, 99%) and C. butyricum (Acc. No. X68178, 99%) grew only on blood agar (Table 2). The 16S rRNA gene sequences of isolates W1, W4 and W5 have been deposited onto GenBank with accession numbers DQ831125, DQ831126 and DQ831124, respectively.

Table 2.   Summary of the isolates
Isolate16S rDNA sequence resultsColony characters*Haemolytic typeHydrogenase gene targeted primers PCR resultsHydrogen production yield (ml H2 g−1 glucose)†
Closest similar sequenceSimilarityClosest similar sequenceSimilarity
  1. *As W1 cannot grow on blood agar, the colony characteristics were determined by nutrient agar culture. Others were all determined by blood agar culture.

  2. †Data are shown for hydrogen production yield with both mineral medium and improved mineral medium.

W1Clostridium diolis, Acc. No. AJ45841799% (1375/1383)Round and clear edge, 1 mm, cream-Hydrogenase, Clostridium diolis isolate Z2, Acc. No. AY65273296% (110/115)63/150
W2Bacillus sp., Acc. No. AF16953596% (1329/1379)Round and clear edge, 2 mm, gray-L-serine dehydratase (alpha chain), Bacillus subtilis subsp. subtilis str. 168, Acc. No. CAB1345985% (101/118)0
W3Swine manure bacterium, Acc. No. AY167932100% (1296/1296)Irregular shape and edge, 4 mm, white+Ferredoxin: 4Fe-4S ferredoxin hydrogenase, Alkaliphilus metalliredigenes QYMF, Acc. No. ZP_0080042567% (95/141)0
W4Clostridium butyricum, Acc. No. X6817899% (1434/1435)Irregular shape and edge, 8 mm, gray-Hydrogenase, uncultured Clostridium sp., Acc. No. AAT7684799% (129/130)70/167
W5Clostridium butyricum, Acc. No. X6817899% (1078/1079)Irregular shape and edge, 2 mm, gray-Hydrogenase, uncultured Clostridium sp., Acc. No. AAT7684799% (146/147)86/220

Hydrogenase gene-targeted PCR analyses

In order to test their potential to produce hydrogen, isolates were screened by PCR for the presence of a hydrogenase gene using primer sets E1f/E1r and L1f/L1r. All of the isolates were PCR positive using primer set E1f/E1r, while only W3, W4 and W5 were PCR positive using primer set L1f/L1r. The amplicons were sequenced and identified by searching the NCBI sequence database using BLAST (GenBank accession numbers AAT76847, CAB13459, ZP_00800425, AAT76847 and AAT76847 for W1 to W5 respectively). The results (Table 2) suggest that primer sets E1f/E1r and L1f/L1r are not specific for hydrogenase genes. For primer set E1f/E1r, the amplicon of W2 was similar to a dehydratase gene, and the amplicon of W3 had a relatively low similarity (67%) to a 4Fe-4S ferredoxin hydrogenase gene, which had not been reported before to catalyze the hydrogen synthesis reaction. For primer set L1f/L1r, the amplicon of W3 gave the same result as primer set E1f/E1r and it failed to amplify the hydrogenase gene of W1. The amplicons from isolates W1, W4 and W5 had high similarity with hydrogenase genes from Clostridium species (Table 2).

Hydrogen fermentation

The five isolates were seeded into mineral medium to test their ability to produce hydrogen. Only W1, W4 and W5 could produce hydrogen after adjustment of the initial pH (from 4·0 to 7·5) and fermentation temperature (from 25°C to 40°C) (data not shown). Hydrogen production yields for W1, W4 and W5 were 63 ml H2 g−1 glucose, 70 ml H2 g−1 glucose and 86 ml H2 g−1 glucose, respectively (Table 2). The production yields were improved to 150 ml H2 g−1 glucose, 167 ml H2 g−1 glucose and 220 ml H2 g−1 glucose, respectively, using an improved mineral medium. Using the improved mineral medium, it was possible to decrease the fermentation time from 48 h to 24 h. As assessed by HPLC, the main organic acid by-products of W1 were acetic and butyric acids, while W4 and W5 produced acetic, butyric, malic and fumaric acids. The production of malic and fumaric acids may correlate to the higher hydrogen production yields because more protons would be available (Hallenbeck and Benemann 2002). For W4 and W5, strict anaerobic conditions were required, which included nitrogen purging of the medium and fermentation device before fermentation. However, unlike W4 and W5, isolate W1 did not appear to be as fastidious in its requirements and did not require strict anaerobic conditions for growth and hydrogen production.

Phylogenetic analysis of isolated strains

The three isolates capable of hydrogen production were further characterized by phylogenetic analysis using 16S rRNA gene sequences from a variety of Clostridium species associated with hydrogen production (Ogino et al. 2005; Liu et al. 2006; Zhang et al. 2006). A larger fragment of the 16S rRNA gene was amplified and sequenced from isolates W1, W4 and W5 using primers 27 F, 357 F, 786 F and 1492R. Phylogenetic analysis (Fig. 2) suggested that isolates W4 and W5 are closely related to (or the same species as) C.butyricum. Whilst most similar to C. diolis, isolate W1 was placed in an unresolved cluster of species including C. roseum, C. diolis, C. acetobutylicum and C. beijerinckii, and would possibly represent a distinct species.

Figure 2.

 Phylogenetic tree resulting from Neighbor Joining analysis of 16S rDNA sequences of the three isolates and published sequences of Clostridium species known to produce hydrogen. Percent bootstrap support (≥50%) from 1000 replicates is indicated at supported nodes for distance-based and parsimony analyses, respectively.


In this study, culture independent methods (PCR–DGGE and 16S rDNA clone library construction) and culture dependent isolation techniques were used to profile a hydrogen-producing microbial community in heat-treated activated sludge and to isolate and identify potential hydrogen producers. The ability of isolates to produce hydrogen was subsequently confirmed. The results of 16S rDNA-targeted profiling suggested that spore formers (in particular Clostridium sp.) were dominant in the hydrogen producing sludge. The bacteria identified in the 16S rDNA clone library were not 100% in accordance with the PCR–DGGE results. This difference may have been caused by different sensitivities and biases of the two methods (Rainey et al. 1996; Vallaeys et al. 1997; Wintzingerode et al. 1997).

The measurement of hydrogenase gene expression using primer sets E1f/E1r and L1f/L1 could provide an activity index for hydrogen-producing systems (Chang et al. 2006). Sequencing results for DNA amplified using these primer sets suggested that they are not strictly specific for hydrogenase genes because a gene encoding a dehydratase was amplified. This lack of specificity makes these primers poor candidates for measuring hydrogenase gene expression and would lead to an overestimation of system activity.

The hydrogen production yields for isolates W1, W4 and W5 were much higher than those for Clostridium species reported elsewhere under the same reaction conditions (Kumar and Das 2000; Oh et al. 2002, 2003; Shin et al. 2004). The shorter fermentation periods required for these isolates make them more attractive for use in industrial fermentation processes. These three isolates will allow more detailed investigation of the relationship between hydrogen yield and by-products composition. One hypothesis is that the higher hydrogen yield produced by these isolates is associated with the production of both malic and fumaric acids, which have not been reported previously as by-products accumulated during fermentation. This may result in the release of more protons (Hallenbeck and Benemann 2002). Further investigation on three hydrogen-producing isolates could be done to compare the metabolic pathways for hydrogen production in these three isolates and to determine whether the pathways can be manipulated to produce useful by-products while at the same time improving the hydrogen yield, which will make bio-hydrogen production more profitable and practicable.

Another interesting result was the isolation of W1, which appeared to grow and produce hydrogen in nonstrict anaerobic conditions. Isolates W4 and W5, as well as other species of Clostridium reported in the literature to produce hydrogen, require strict anaerobic conditions. The requirement for strict anaerobic conditions greatly increases the production costs because of the relatively high price of reducing agents, such as argon, nitrogen, hydrogen gas and L-cysteine·HCL, required to remove trace amounts of oxygen from the medium, and the additional constraints placed on reactor design and operation (Fabiano and Perego 2002; Kapdan and Kargi 2006). A mixed fermentation of Enterobacter aerogenes with C. butyricum has been trialed to solve this problem, as E. aerogenes could remove oxygen from the reactor, but E. aerogenes became dominant, which led to lower hydrogen yields (Yokoi et al. 2001). A Clostridium species with the ability to tolerate dissolved oxygen will be highly beneficial and make the dark fermentation more economical and practicable. To our knowledge, there are two possible reasons to explain this oxygen tolerance. Firstly, the hydrogenase generated by W1 may be more tolerant to oxygen compared with other hydrogenases, Secondly; there may be a protective system against oxygen presence during hydrogen fermentation. Although the hydrogen production yield by W1 is lower compared with that of W4 and W5, more work focusing on the differences between their respective hydrogenase genes and gene regulation will be helpful to determine whether there is any relationship between oxygen tolerance and hydrogen yield or whether the hydrogen metabolic pathway can be modified to improve yields in oxygen-tolerant hydrogen producers.


We are very grateful for research assistance given by Zhanying Zhang, Zhihui Bai and Richard Haas from the Water Environment Biotechnology Laboratory at the University of South Australia and technical support from staff at the Australian Water Quality Centre, SA Water Corporation. This work was financially supported by an International Linkage Fellowship (LX0560210) from the Australian Research Council. Thanks are due to Bolivar Waste Water Treatment Plant, SA Water Corporation, for providing activated sludge and Smiths Chips Ltd, Australia, for providing potato wastewater.