Shaping of bacterial community structure in microbial fuel cells by different inocula

Abstract

Understanding how the community structure of anode biofilms is shaped is important for enhancing the performance of microbial fuel cells (MFCs). Activated sludge (AS), garden soil (GS), wastewater (WW) and river sediment (RS) were inoculated into single-chamber MFCs to assess the effects of inocula on the power outputs and microbial communities of MFCs. MFCs with different initial inocula showed differences in acclimation time and power density. MFC-RS (river sediment inoculum) obtained the maximum power density (744.8 mW m⁻²), followed by MFC-AS, MFC-GS and MFC-WW. Illumina Miseq sequencing of the 16S rRNA gene and comparative analyses indicated that the microbial community structure of established anode biofilms was clearly differentiated from that in the initial inocula. Principal component analysis (PCA) proved that MFC-AS and MFC-GS were closely clustered and were separated from MFC-WW and MFC-RS. The majority of the dominant populations of MFC-RS were affiliated with Azoarcus (45.20%). The most dominant genus belonged to Flavobacterium (14.18%) in MFC-AS, Geobacter (14.40%) in MFC-GS and Azovibrio (11.11%) in MFC-WW, respectively. This study implies that different inocula substantially influenced the community structure of the anode biofilms of MFCs.

---

1. Introduction

The microbial fuel cell (MFC) is a promising technology due to its simultaneous pollutant removal and energy production from wastewater or solid waste.¹ Exoelectrogenic microorganisms oxidize organic matter and extracellularly transfer electrons to the electrode, this environmentally-friendly process produces electricity without the combustion of fossil fuels.^2,3 Understanding electron transfer of exoelectrogenic microorganisms and optimizing MFC configurations are important for enhancing the power generation of MFCs. In previous studies, researchers made efforts to improve the power output from many aspects. A simplified MFC without a proton exchange membrane had higher power generation due to a lower internal resistance.⁴ Electrode materials (carbon felt, graphite fiber, carbon cloth etc.) and operational conditions including substrate, temperature, pH and external resistance etc. have been extensively studied.^5–11 In order to use the electrical current, different types of microbial electrochemical systems (MESs) derived from MFCs such as microbial electrolysis cells (MECs), microbial desalinization cells (MDCs) and microbial reverse-electrodialysis cells (MRCs) were developed.^12–15 However, the capacity of the microbial biofilm for extracellular electron transfer is key for improving the performance of MESs.¹⁶

Extracellular electron transfer of fixed-configuration microbial electrochemical systems (MESs) depends upon the community composition of the biofilm. Niche-based deterministic factors such as temperature, pH and light conditions play a significant role in shaping the microbial community structure of MESs.^17–19 A recent study demonstrated that stochastic assembly plays a dominant role in determining community structure in MECs.²⁰ Both deterministic and stochastic factors play important roles in shaping the microbial community structure of biofilms in MESs. Understanding the relationship between ecosystem function and community structure is important for improving MES configurations and enhancing electron transfer.

Although some exoelectrogenic bacteria have been isolated from MFCs, our understanding of microorganisms capable of extracellular electron transfer in natural environments is still deficient.² Community analyses prove that more diverse populations exist in electrode biofilms. Exploring unknown exoelectrogenic microbes in natural environmental or engineered systems will facilitate a useful insight into electron transfer. Some recent studies showed interspecific interactions between exoelectrogens and non-exoelectrogens in MFCs for soil bioremediation.^21–23 Wastewater or activated sludge is used frequently as an inoculum while mixed culture MFCs have been developed as a novel wastewater treatment technology.¹ Almost all previous studies on MFC inoculum were performed using two-chamber MFCs. Inocula (wastewater, waste sludge, defined mixed- or pure culture) obviously influenced the power production and internal resistance of two-chamber MFCs.^10,24,25 In order to enhance phenanthrene degradation in MFCs, the community composition of electrode biofilm was shaped by supplementing Pseudomonas aeruginosa into mixed cultures.²⁶

A recent study analyzed the microbial community of MFCs with two kinds of wastewater inoculum using denaturing gradient gel electrophoresis (DGGE) of the 16S rRNA gene.⁹ Community analysis by conventional molecular tools such as DGGE and clone libraries of the 16S rRNA gene provides incomplete information due to limited throughput and resolution. Although previous studies showed that inoculum influenced the power generation of MFCs, how the initial inoculum influenced the community structure is still unknown based on high throughput sequencing. The community structure of microbial biofilms may be shaped by different initial inocula, which will result in different electricity generation by MFCs.

In this study, the effect of different inocula from natural consortia and wastewater on the performance of microbial fuel cells was investigated. The bacterial communities of the anode biofilms of MFCs and initial inocula were assessed by sequencing 16S ribosomal RNA (rRNA) gene amplicons with the Illumina MiSeq technology.

2. Materials and methods

2.1 MFC configuration and operation

A cubic single-chamber MFC was made from polymethylmethacrylate (PMMA) (cylindrical chamber volume of 25 mL) as previously described.⁶ A graphite fiber brush was used as the anode.²⁷ The air-cathode was made from carbon cloth (7 cm² projected area) with a layer of platinum catalyst and three polytetrafluoroethylene (PDFE) diffusion layers.⁵

Four types of initial inocula were used in MFCs. Natural consortia of river sediment (RS) were obtained from the Songhua River in Harbin. Garden soil (GS) was obtained from shrubs in Harbin. Wastewater (WW) and activated sludge (AS) were obtained from the primary clarifier and secondary sedimentation tank of Wenchang Wastewater Treatment Plant of Harbin, respectively. The MFC reactors were fed a nutrient medium containing 1 g L⁻¹ sodium acetate as substrate in a 50 mM phosphate buffer solution (PBS) (11.55 g Na₂HPO₄·12H₂O, 2.77 g NaH₂PO₄·2H₂O, 0.31 g NH₄Cl and 0.13 g KCl)²⁸ amended with 1.25 mL L⁻¹ mineral solution and 0.5 mL L⁻¹ vitamin solution. All MFC reactors were operated in fed-batch mode, and every batch cycle was regarded as a period. The solution was replaced when the voltage decreased to lower than 0.05 V. All MFC reactors were operated in a constant temperature room set at 30 °C.

2.2 Calculations and analyses

The cell voltage across an external resistor of 1000 Ω in the MFCs was collected automatically every 30 min by a multichannel data acquisition system (Model 2700 with 7702 module, Keithley Instruments Inc., USA) and then connected to a personal computer via PCI interface. The polarization curve was measured by changing the external resistance from 3000 to 50 Ω. Power density and coulombic efficiency (CE) were calculated as previously described.⁶

2.3 Illumina sequencing analysis of 16S rRNA gene amplicons

After the MFCs steadily operated for more than 2 months, brush anodes were cut and fragmented by sterile scissors.¹⁷ Genomic DNA of the anode biofilms and initial inocula was extracted using the PowerSoil DNA Isolation Kit (Mo Bio Laboratories, Inc., Carlsbad, CA) according to the manufacturer’s instructions. Bacterial 16S rRNA gene targeting the hypervariable regions of V4–V5 were amplified using a pair of universal primers as follows: 515F (5′-GTGCCAGCMGCCGCGG-3′) and 907R (5′-CCGTCAATTCMTTTGAGTTT-3′). Individual samples were barcoded for pooling multiple samples in one run of an Illumina Miseq (Illumina, San Diego, CA).

2.4 Analysis of 16S rRNA sequencing data

The resultant sequencing reads were analyzed using Quantitative Insights into Microbial Ecology (QIIME) software (http://qiime.org).²⁹ Operational taxonomic units (OTUs) were determined at 97% similarity levels using UPARSE software (http://drive5.com/uparse/). A representative sequence from each OTU was selected for taxonomic identification using the Silva database (http://www.arb-silva.de) and the Ribosomal Database Project (RDP) classifier (http://rdp.cme.msu.edu) with a 0.80 confidence threshold. Diversity indices, the species richness of the Chao1 estimator and the ACE estimator of each sample were generated in MOTHUR (http://www.mothur.org). The heat map of genus level was generated using a hierarchical clustering algorithm. Principal component analysis (PCA) and hierarchical cluster analysis were performed to visualize and interpret differences in microbial community structure between data sets.

3. Results

3.1 Power production of MFCs with different inocula

All MFCs showed voltage production after acclimation and there was an obvious difference in lag phase after MFCs were inoculated with different inocula (Fig. 1). The maximum peak voltage of the MFC with active sludge inoculum (MFC-AS) reached more than ∼540 mV after 4 days, faster than that obtained by other MFCs, suggesting that mixed culture from sludge inoculum may have a high vitality. After 6 days the MFC with river sediment inoculum (MFC-RS) and MFC with wastewater inoculum (MFC-WW) reached maximum peak voltage but the MFC with garden soil inoculum (MFC-GS) did not. A smaller proportion of exoelectrogenic microorganisms in the soil ecosystem may result in longer lag growth. When all MFC reactors produced reproducible current, the maximum power density of 744.8 ± 11.7 mW m⁻² was obtained by MFC-RS, with 666.3 ± 19.1 mW m⁻² for MFC-AS, 654.7 ± 30.1 mW m⁻² for MFC-GS and 536.1 ± 6.3 mW m⁻² for MFC-WW (Fig. 2).

Fig. 1 Voltage generation (external resistance of 1000 Ω) of MFCs with different inocula. MFC-AS, MFC-RS, MFC-WW and MFC-GS were inoculated using active sludge, river sediment, wastewater and garden soil, respectively.

Fig. 2 Polarization curves of MFCs with different inocula. Error bars represent standard deviation based on measurements from duplicate reactors in three cycles.

3.2 COD removal and coulombic efficiency

All MFCs with different inocula had similar COD removal rates, and the maximum COD removal rate of 95.7% was obtained by MFC-AS, followed by MFC-WW (95.4%), MFC-GS (94.3%) and MFC-RS (93.6%) after operating for 20 days. Although the MFC-AS achieved the highest COD removal rate, it did not show the maximum power density, suggesting that some non-exoelectrogens in the anode biofilm resulted in lost energy production. The coulombic efficiency (CE) of MFC-AS, MFC-RS, MFC-WW and MFC-GS was 27.4%, 28.6%, 29.8% and 31.2%, respectively.

3.3 Bacterial diversity of the anode biofilms and initial inocula

All high-quality reads of 98 263 (average length of 402 bp) were obtained from the Illumina Miseq sequencing of the 16S rRNA gene to identify the operational taxonomic units (OTUs) of eight individual samples (Table 1). A total 362 (MFC-AS), 169 (MFC-RS), 164 (MFC-GS) and 124 OTUs (MFC-WW) were determined at a level of 97% similarity. The initial inocula showed a higher relative abundance of OTUs, with 594 (GS), 478 (AS), and 403 (RS) OTUs, but WW showed 156 OTUs. The species richness of MFC biofilms was lower than that of the original inocula except for the MFC-WW biofilm. GS inoculum had a greater species richness than WW, RS and AS inocula. MFC-AS had the highest diversity (Shannon index, 4.04 and Simpson, 0.0429), followed by MFC-GS, MFC-WW, and MFC-RS (Table 1).

Table 1 Number of reads, operational taxonomic units (OTUs), Abundance-based Coverage Estimator (ACE), estimator Chao1, Shannon and Simpson indices obtained from different samples at 97% nucleotide identity

Samples	Reads	OTUs	ACE	Chao1	Shannon	Simpson
WW	11 608	156	182	189	3.21	0.0894
GS	14 896	594	605	607	5.58	0.0068
RS	9291	403	474	492	4.48	0.0313
AS	11 876	478	512	510	4.75	0.0278
MFC-WW	10 860	124	185	200	3.63	0.043
MFC-GS	11 230	164	275	217	3.63	0.0453
MFC-RS	15 927	169	293	238	2.5	0.2276
MFC-AS	12 575	362	443	476	4.04	0.0429

---

3.4 Comparative analysis of microbial community structures

Hierarchical clustering and heatmap analysis were used to identify the differences of eight bacterial community structures (Fig. 3). The heatmap based on genus level showed clear distinctions of community structure between each anode biofilm and inoculum. Hierarchical cluster analysis of OTUs (on the top of the heatmap) indicated that the anode biofilms of MFCs differed from the initial inocula, suggesting obvious shaping of the community structure after enrichment of exoelectrogens in MFCs. The principal component analysis (PCA) showed that the anode biofilms and inocula were well separated, with 21.91% and 19.28% variation explained by PC1 and PC2, respectively (Fig. 4). MFC-AS and MFC-GS were closely clustered and were separated from MFC-WW and MFC-RS.

Fig. 3 Hierarchical clustering and heatmap analysis of eight bacterial community structures based on Illumina sequencing of the 16S rRNA gene. Hierarchical cluster analysis is based on the OTUs abundance on the top of the heatmap. The heatmap is plotted at genus level. The bar on the bottom represents scale of the relative abundance.

Fig. 4 Principal component analysis (PCA) based on operational taxonomic units of anode biofilms of MFCs and initial inocula. Inocula and MFCs are clustered in the parallelogram and ellipse, respectively.

3.5 Community compositions of the anode biofilms and initial inocula

In terms of the assignment at the phylum level, Proteobacteria and Bacteroidetes were predominant in all communities, followed by Firmicutes except in GS and AS inocula. Proteobacteria apparently composed most of the sequences at 72% (MFC-RS), 69.33% (MFC-WW), 51.79% (MFC-GS), 34.42% (MFC-AS), 54.39% (WW), 22.05% (GS), 41.41% (RS) and 38.47% (AS), respectively (Fig. 5(a)). Among all phyla, Bacteroidetes accounted for 20.12% (MFC-RS), 11.29% (MFC-WW), 16.81% (MFC-GS), 35.99% (MFC-AS), 28.7% (WW), 15.05% (GS), 30.49% (RS) and 21% (AS), respectively.

Fig. 5 Microbial community structures of the MFC anodes and inocula at the phylum (a) and class (b) level. Items with relative abundance lower than 1% of total composition were classified into the “others” group.

The majority of classes or subclasses belonged to Betaproteobacteria in MFC-RS (55.80%), MFC-GS (20.77%), MFC-WW (33.66%), AS (27.53%) and RS (19.45%) (Fig. 5(b)). The predominant classes were affiliated with Flavobacteria (19.22%) in MFC-AS, Acidobacteria in GS (11.63%), Epsilonproteobacteria (29.26%) and Bacteroidia (27.59%) in WW, respectively.

The difference in dominant populations was more distinct at genus level (Fig. 6). The taxonomic assignments of genera were significantly shifted after the anode biofilms were enriched from different initial inocula. The populations accounting for less than 4% of relative abundance were designated as “others”. The majority of dominant populations in MFC-RS were affiliated with Azoarcus (45.20%). The dominant populations in MFC-AS belonged to Flavobacterium (14.18%), Stenotrophomonas (11.96%) and Geobacter (6.73%), compared to Geobacter (14.40%) and Victivallis (8.62%) in MFC-GS, and Azovibrio (11.11%), Alishewanella (9.43%) and Pseudomonas (7.97%) in MFC-WW.

Fig. 6 Relative abundance of dominant genera in microbial communities of the MFC anode biofilms and inocula. Genera with relative abundance lower than 4% of the total composition were classified into group “Others”.

4. Discussion

The effect of inocula (natural consortia and mixed culture) on the performances of single-chamber air-cathode MFCs was investigated. Our results were similar with previous report that showed the effect of inoculum types on the power density of MFCs.³⁰ At the beginning of start-up, MFCs inoculated with activated sludge and wastewater inocula showed higher peak voltages, compared to MFCs with RS and GS inocula (Fig. 1). After several cycles, the peak voltage of MFC-RS gradually increased. Compared to other MFCs, MFC-RS obtained a maximum power density. Our results implied that exoelectrogens are widely found in natural habitats and wastewater treatment bioreactors resulting in easily obtainable inoculum for MFCs.

Community analyses based on Illumina Miseq sequencing of the 16S rRNA gene indicated that different inocula resulted in a difference in community composition in the anode biofilms.³¹ PCA analysis proved that MFC-AS and MFC-GS were closely clustered and were separated from MFC-WW and MFC-RS, which also obtained similar maximum power densities (Fig. 4). These results implies that the difference in power densities may arise from the different community structures of anode biofilms. Microbial community structures of biofilms in MFCs were shaped by inocula, however stochastic factors may also play a role in biofilm establishment as previously described.²⁰

Although MFC-RS obtained the maximum power density, the population diversity of the anode biofilm was low (Table 1), presumably exoelectrogens were predominant in the community. The relative abundance of Geobacter (a well-known exoelectrogen) in MFC-RS only accounted for 1.46%, compared with 6.73% (MFC-AS) and 14.40% (MFC-GS) (Fig. 6). A nitrogen fixing bacterium Azoarcus (45.20%) was the predominant genus in MFC-RS, which was found as the dominant population in the anode biofilm in a previous report.³² The capacity for extracellular electron transfer of Azoarcus should be further tested in the future. The functions of uncultured and low abundance bacteria are unclear in anode biofilms, suggesting large quantities of unknown exoelectrogens may be present in the natural environment.

5. Conclusions

The effect of different inocula including river sediment, activated sludge, garden soil and wastewater on microbial communities and the performances of MFCs was investigated. The MFC with river sediment inoculum (MFC-RS) achieved the maximum power density, followed by MFC-AS, MFC-GS and MFC-WW. Community analyses indicated that different inocula resulted in a difference in community composition in the anode biofilms. The dominant populations of anode biofilms differed obviously from those in the initial inocula. The principal component analysis (PCA) of OTUs showed that MFC-AS and MFC-GS were closely clustered and were separated from MFC-WW and MFC-RS. The results confirmed that the different initial inocula influenced community structures and power generation of MFCs.

Acknowledgements

This research was supported by National Natural Science Foundation of China (No. 31270004, 51422805), the State Key Laboratory of Urban Water Resource and Environment (Harbin Institute of Technology) (No. HC201110), the Fundamental Research Funds for the Central Universities (No. HIT.BRETIII. 201232).

References

1. H. Liu, R. Ramnarayanan and B. E. Logan, Environ. Sci. Technol., 2004, 38, 2281–2285.
2. B. E. Logan and J. M. Regan, Trends Microbiol., 2006, 14, 512–518.
3. D. R. Bond and D. R. Lovley, Appl. Environ. Microbiol., 2003, 69, 1548–1555.
4. H. Liu and B. E. Logan, Environ. Sci. Technol., 2004, 38, 4040–4046.
5. S. Cheng, H. Liu and B. E. Logan, Electrochem. Commun., 2006, 8, 489–494.
6. J. Jia, Y. Tang, B. Liu, D. Wu, N. Ren and D. Xing, Bioresour. Technol., 2013, 144, 94–99.
7. Y. Ahn, M. C. Hatzell, F. Zhang and B. E. Logan, J. Power Sources, 2014, 249, 440–445.
8. I. Ieropoulos, J. Winfield and J. Greenman, Bioresour. Technol., 2010, 101, 3520–3525.
9. J. Yu, Y. Park and T. Lee, Bioprocess Biosyst. Eng., 2014, 37, 667–675.
10. A. L. Vázquez-Larios, O. Solorza-Feria, G. Vázquez-Huerta, F. Esparza-García, N. Rinderknecht-Seijas and H. M. Poggi-Varaldo, Int. J. Hydrogen Energy, 2011, 36, 6199–6209.
11. F. Zhang, T. Saito, S. A. Cheng, M. A. Hickner and B. E. Logan, Environ. Sci. Technol., 2010, 44, 1490–1495.
12. H. M. Wang and Z. J. Ren, Biotechnol. Adv., 2013, 31, 1796–2180.
13. R. D. Cusick, M. Hatzell, F. Zhang and B. E. Logan, Environ. Sci. Technol., 2013, 47, 14518–14524.
14. F. Harnisch and U. Schroder, Chem. Soc. Rev., 2010, 39, 4433–4448.
15. H. M. Wang, J. D. Park and Z. J. Ren, Environ. Sci. Technol., 2015, 49, 3267–3277.
16. Y. Kim and B. E. Logan, Environ. Sci. Technol., 2011, 45, 5834–5839.
17. D. Xing, S. Cheng, J. M. Regan and B. E. Logan, Biosens. Bioelectron., 2009, 25, 105–111.
18. L. Lu, N. Ren, X. Zhao, H. Wang, D. Wu and D. Xing, Energy Environ. Sci., 2011, 4, 1329–1336.
19. Z. He, Y. L. Huang, A. K. Manohar and F. Mansfeld, Bioelectrochemistry, 2008, 74, 78–82.
20. J. Zhou, W. Liu, Y. Deng, Y. Jiang, K. Xue, Z. He, J. van Nostrand, L. Wu, Y. Yang and A. Wang, mBio, 2013, 4, e00584-12.
21. L. Lu, T. Huggins, S. Jin, Y. Zuo and Z. J. Ren, Environ. Sci. Technol., 2014, 48, 4021–4029.
22. L. Lu, D. Xing and Z. J. Ren, Bioresour. Technol., 2015, 195, 115–121.
23. X. Li, X. Wang, Z. J. Ren, X. Zhang, N. Li and Q. Zhou, Chemosphere, 2015, 141, 62–70.
24. A. Yadav, P. Panda and B. Bag, Energy Sources, Part A, 2013, 35, 1828–1835.
25. A. S. Mathuriya, Environ. Technol., 2013, 34, 1957–1964.
26. O. Adelaja, T. Keshavarz and G. Kyazze, Eng. Life Sci., 2014, 14, 218–228.
27. B. E. Logan, S. Cheng, V. Watson and G. Estadt, Environ. Sci. Technol., 2007, 41, 3341–3346.
28. D. Wu, D. Xing, X. Mei, B. Liu, C. Guo and N. Ren, Int. J. Hydrogen Energy, 2013, 38, 15568–15573.
29. J. G. Caporaso, J. Kuczynski, J. Stombaugh, K. Bittinger, F. D. Bushman, E. K. Costello, N. Fierer, A. G. Pena, J. K. Goodrich, J. I. Gordon, G. A. Huttley, S. T. Kelley, D. Knights, J. E. Koenig, R. E. Ley, C. A. Lozupone, D. McDonald, B. D. Muegge, M. Pirrung, J. Reeder, J. R. Sevinsky, P. J. Tumbaugh, W. A. Walters, J. Widmann, T. Yatsunenko, J. Zaneveld and R. Knight, Nat. Methods, 2010, 7, 335–336.
30. H. Lin, X. Wu, C. Miller and J. Zhu, Biomass Bioenergy, 2013, 54, 170–180.
31. C. Gao, A. Wang, W. Wu, Y. Yin and Y. Zhao, Bioresour. Technol., 2014, 167, 124–132.
32. J. R. Kim, S. H. Jung, J. M. Regan and B. E. Logan, Bioresour. Technol., 2007, 98, 2568–2577.

---