|Home | About | Journals | Submit | Contact Us | Français|
Variation in the hydrogen production rate was consistent with the succession of dominant bacteria during the batch fermentation process. Denaturing gradient gel electrophoresis (DGGE) of 16S rRNA genes and quantitative analysis of the hydA genes at both the DNA and mRNA levels confirmed that Clostridium perfringens was the most dominant hydrogen producer in the bioreactor.
Biological hydrogen production is expected to be an important strategy in the development of clean and sustainable alternative energy sources (6). Anaerobic mixed cultures are frequently used as inocula in hydrogen production (3, 4, 10). Compared with pure cultures, mixed cultures have the advantage of more technical feasibility and the potential for using complex carbohydrates as substrates (14), possibly because different members of the bacterial community play complementary or mutually beneficial roles in utilizing substrates, providing growth factors, eliminating feedback inhibition, etc. (12, 17). Many sources of natural microflora, including compost (20), different sludges (4, 10, 24), and soils (21), have been used as inocula for hydrogen production after inactivation of hydrogenotrophic methanogens. However, whether pretreated natural microflora is the most efficient mixed culture for hydrogen production is still unknown. Understanding the relationships between variation in microbial composition and its hydrogen production efficiency is the first step in reconstructing more-efficient hydrogen-producing consortia.
The 16S rRNA gene has been widely used as a universal molecular marker (7, 15). The Fe-hydrogenase gene (hydA), which is usually involved in proton reduction (H2 production) to dispose of excess reducing equivalents (1, 2, 22) in Clostridium spp. and sulfate reducers, has recently been used as a molecular marker to distinguish potential hydrogen-producing bacteria in mixed cultures (3, 8, 23). Therefore, in this study, the V3 regions of the 16S rRNA gene and the hydA gene were used as biomarkers to investigate the succession of the bacterial community during hydrogen production in a batch culture.
Hydrogen production was conducted in a 10-liter continuously stirred tank reactor (CSTR) with a 7-liter working volume and no specific requirement to remove oxygen from the culture and headspace. Cattle dung compost was pretreated by boiling for 5 min and then used as an inoculum. The fermentation reaction mixture was kept at pH 5.8, with stirring at 100 rpm. Other fermentation conditions for batch hydrogen production from 18 g liter−1 sucrose were the same as in a previous study (27). The evolved biogas was collected by the water displacement method. The concentrations of hydrogen in the biogas were quantified as previously described (18). A total of 25.6 liters of hydrogen was evolved in the batch fermentation, corresponding to a hydrogen yield of 3.11 mol H2 mol sucrose−1 or 200 ml H2 g sucrose−1. The reported high hydrogen yields from sucrose in dark fermentation were 3.8 mol H2 mol sucrose−1 by a Clostridium pasteurianum strain and 2.73 mol H2 mol sucrose−1 by microflora (5, 25).
Over 72 h, the hydrogen-producing process was divided into five phases according to the hydrogen production rate: the lag phase (0 to 12 h after inoculation), exponential phase (12 to 16 h), stationary phase (16 to 24 h), early decline phase (24 to 36 h), and late decline phase (36 to 72 h) (Fig. (Fig.1).1). The oxidation reduction potential (ORP) in the medium was traced with a built-in ORP probe (Mettler-Toledo, Columbus, OH). The ORP value was above zero during the hydrogen-producing lag phase, dropped dramatically from 148 mV to −246 mV when hydrogen production was started, and then remained below −156 mV in the hydrogen-producing exponential and stationary phases (Fig. (Fig.11).
Genomic DNA was extracted, using the sodium dodecyl sulfate (SDS)-based DNA extraction procedure (26). The V3 regions of the bacterial 16S rRNA genes were PCR amplified and analyzed by denaturing gradient gel electrophoresis (DGGE) as previously described (16). During the fermentation process, the bacterial compositions of all DNA samples clustered into three groups, based on similarities among the DGGE profiles as assessed with Quantity One software, version 4.4 (Bio-Rad) (Fig. (Fig.2B).2B). The three groups were consistent with the variations in hydrogen production rates and ORP values during the hydrogen-producing process.
All of the sequenced fragments recovered from clear DGGE bands were most closely affiliated with the 16S rRNA genes of species from the genera Bacillus and Clostridium (Fig. (Fig.2A2A and Table Table1).1). It appeared that boiling the cattle dung compost for 5 min had killed almost all non-spore-forming microorganisms. As the bioreactor was not sparged with N2 to remove oxygen in the headspace and in the culture, the ORP was above 100 mV at the initial stage of fermentation, which is why Bacillus spp., facultative anaerobes, dominated in the lag phase. With the growth of Bacillus spp., oxygen was exhausted, the ORP began to decrease rapidly, and an anaerobic environment suitable for the growth of anaerobic and hydrogen-producing bacteria began to appear (9). Clostridium beijerinckii and Clostridium perfringens were found to be the dominant species in the exponential and stationary phases. Clostridium lundense and other Clostridium spp. began to appear in the decline phases, when hydrogen production had stopped and hydrogen uptake was observed (Table (Table11 and Fig. Fig.11).
At least five Clostridium species were detected by DGGE analysis during the hydrogen-producing process in our bioreactor, suggesting that the Clostridium species were the major hydrogen producers in this study, as in other anaerobic bioreactors (13, 19). Based on the conserved regions of the H domains in the hydA genes from nine Clostridium species (see Table S1 in the supplemental material), a primer set for fragments with an average length of 200 bp (see Table S2 and Fig. S1A in the supplemental material) was designed and used in quantitative PCR to monitor the quantity of potential hydrogen producers and the expression levels of their hydA mRNAs. Total RNA was extracted, using an RNeasy kit (Qiagen, Hilden, Germany) (11). Reverse transcription of isolated RNA was carried out with random primers, using an iScript cDNA synthesis kit (Toyobo, Osaka, Japan). The calibration curves against the cycle threshold (CT) for the hydA gene showed a good linear relationship over six orders of concentration, from 1.0 × 101 to 1.0 × 106 copies μl−1, with a R2 value of 0.9999.
The copy number of the hydA genes at DNA level, which represents the abundance of the potential hydrogen producers, was very low in the lag phase and then increased to approximately 1.0 × 106 copies μl−1 in the exponential phase. Several hours after the bioreactor stopped evolving hydrogen, the copy number of the hydA genes decreased dramatically, to less than 1.0 × 105 copies μl−1 (Fig. (Fig.3A3A).
The copy number of the hydA genes per reaction volume at the cDNA level, which reflects the total expression level of hydA mRNA per reaction volume (hydA mRNA units μl−1), peaked in the exponential and stationary phases (Fig. (Fig.3B),3B), consistent with the highest instant hydrogen production rate and the dramatic decrease in the ORP (Fig. (Fig.1).1). By contrast, the relative hydA mRNA expression level per cell was highest in the early exponential phase (12 h), while no hydrogen production was observed (Fig. (Fig.3B).3B). The rapid increase in the abundance of hydrogen producers in the late exponential phase might be related to the highest total expression level of hydA mRNA.
Another primer set for fragments with an average length of 695 bp (see Table S2 and Fig. S1B in the supplemental material) was designed based on the H domains in the hydA genes and used to investigate the diversity and composition of the hydA genes in the bioreactor by sequence analysis. In total, 13 different hydA gene types were identified from the libraries (Table (Table2).2). No potential hydrogen producers and no hydA mRNA expression were detected in the early lag phase. Hydrogen production was still not detected 12 h after inoculation; however, the hydA genes of C. perfringens, C. beijerinckii, and two Paenibacillus D14-like species were detected in the hydA DNA library from this phase, but only the hydA mRNAs from C. perfringens and C. beijerinckii were detected in the corresponding hydA cDNA library. In the exponential and stationary phases, C. perfringens became the dominant hydrogen producer and showed active expression of the hydA mRNA. In the early decline phase, three hydA gene types closely matched to that of a Clostridium 7_2_43 FAA-like species appeared, and the relative abundance of C. perfringens decreased significantly, but its hydA mRNAs were still dominant in the hydA cDNA library. In the late decline phase, six new Clostridium-like species were detected at either the DNA or cDNA level. The hydA mRNAs of C. perfringens decreased to the lowest level, and the hydA cDNA library became dominated by the hydA mRNAs from these newly emerging species.
Clostridium peptidivorans, C. lundense, and Clostridium vincentii were identified in the decline phase by sequencing the obvious bands on 16S rRNA gene DGGE profiles. However, sequencing of the hydA DNA and cDNA libraries from the early and late decline phases indicated nine newly emerging types of hydA, namely, types 5 to 13. As there are no reported sequences for the hydA genes from C. peptidivorans, C. lundense, and C. vincentii, it is difficult to determine which of the three possible Clostridium species are represented by the above newly emerging hydA genes. In the decline phases, nearly no hydrogen was produced, hydrogen uptake was observed when the ORP rose to about 100 mV, and a low level of expression of hydA mRNA was detected. This phenomenon might be because the Fe-hydrogenases are bidirectional hydrogenases (1) that catalyze hydrogen uptake when the metabolic system is more likely to accept electrons.
Both C. perfringens and C. beijerinckii contributed to hydrogen production in the bioreactor, but C. perfringens was the most dominant hydrogen producer there. The hydrogen production rate and ORP values were consistent with the succession of dominant bacterial species during the different phases of the hydrogen-producing process. It appears that rapid and complex bacterial succession occurs even within a batch culture during hydrogen production.
Nucleotide sequences of the hydA and 16S rRNA genes from clone libraries were deposited in GenBank under accession numbers GQ915138 to GQ915262 and GU735571 to GU735638, respectively.
This work was funded by the National Natural Science Foundation of China (30670042), the Knowledge Innovation Key Program of the Chinese Academy of Sciences Project (grant no. KSCX2-YW-G-022), and the High-Tech Research and Development Program of China (863: 2007AA021302).
Published ahead of print on 19 March 2010.
†Supplemental material for this article may be found at http://aem.asm.org/.