A Bioinformatic Approach to Select Species for Testing in Biofuel Cells

Abstract:

Biological fuel cells (BFCs) use organic catalysts and fuels to derive electricity from nearly any biological process.  Applications include a suite of self-powered implantable medical devices, wastewater treatment plants, and biosensors.  Power density and longevity limitations of BFCs have prevented their widespread adoption.  A novel BFC design incorporating bioregenerative microorganism(s) is under design.  A plethora of candidate organsims are available for testing, but electrochemical resources limit the initial testing to 3 species.  A program using perl and python languages was constructed to identify organisms which can secrete high REDOX potential laccases for electrochemical testing.  

Introduction:

Biological fuel cells (BFCs) use organic catalysts and fuels to derive electricity from nearly any biological process. These cutting edge power generation devices hold promise for the next generation of self-powered wastewater treatment plants and implantable medical devices.  They use highly conductive materials such as carbon nanotubes to steal the electrons from the oxidation or reduction of organic matter by REDOX enzymes to generate current.  Depending on the enzymes used, useful side products can also be collected such as drinking water or even high-value chemicals.  Current limitations in the form of high costs and low power production have confined biofuel cells to the laboratory.  

In 2014 Sane et al. reported on the use of crude laccase from Trametes versicolor in a BFC cathode with a higher open circuit potential (OCP), and longer run time than comparable platinum cathodes.  "Crude" refers to the fermentative product achieved without further purification steps, the best known of which is wine.  The fermentative broth produced by T. versicolor and other fungi allows for the low cost construction of high performance fuel cells.  The factors influencing the success of an organism for power generation in a BFC like the one reported by Sane et al. are the REDOX potential of the laccase and the concentration of laccase in the culture filtrate. 

The REDOX potential of laccase determines the voltage of the fuel cell.  Voltage is analagous to electrical pressure, an increase in REDOX potential translates to an increase in power produced.  Pardo et al. found that the REDOX potential of laccases are directly determined by the axial amino acid (AA) at the T1 copper binding site.  A leucine imparts a REDOX potential of around 400mV vs SHE, a methionine imparts a REDOX potential of around 600mV vs SHE, and a phenylalanine imparts a REDOX potential of 800+mV vs SHE.   Using the single letter amino acid code: L = low, M= med., F = high.  ThÖny-Meyer et al. found that the T1 copper motif is highly conserved region of 12 amino acid residues: HCH...H....L/M/F where '.' is any amino acid residue.  This allows regular expressions to be used to identify laccases of differing REDOX potentials.  

The other important factor influencing the success of an organism in power generation is the concentration of laccase produced.  Laccase needs to be comparatively and quantitatively abundant in the culture filtrate.  Fungi are specifically chosen for biofuel cells because of their natural ability to secrete proteins.  Fungal secretion is a complex process which is not fully understood, but posttranslational modifications are known to be crucial in the targeting of eukaryotic proteins.  Despite the limited theoretical knowledge about fungal secretion, empirically derived data has shown N-Glycosylation to be an accurate predictor of fungal protein secretion, most notably laccase.  The open source program Secretome.P identifies N-Glycosylation sites via the regular expression: N[^P](S|T)[^P].  Where [^P] indictes any amino acid other than proline and (S|T) denotes either a serine or threonine residue in that position.  The abundance of this regular expression was used as a rough indicator of secretion.  

Previous results have indicated that Fusarium oxysporum and Trametes versicolor secrete the most laccase, with Pleurotus ostreatus at nearly 1/10 the production capacity of the other species tested.  To further refine the selection of organisms for incorporation into the BFC cathode chamber a screening method was devised and carried out in silico.  The program, entitled fasta_seq_regex is available here and identifies the axial amino acid at the T1 site, reported by ThÖny-Meyer to influence the REDOX potential of the laccase isoform.  Additionally the program counts the number of N-Glycosylation sites on each laccase sequence which is indicative of the secretion potential of the laccase.  

Fig. 1 flowchart for the fasta_seq_regex project.   

High REDOX potential laccases were identified using fasta_seq_regex and the regular expression HCH...H....F (fig.3).  162 sequences were identified, the majority of which were white rot fungi. The program was tested using the commercial regular expression tester regex101.  Regex101 also reported 162 matches.  The output file matched.fasta was renamed to high_E0_matched.fasta and rerun through the fasta_seq_regex using the regular expression N[^P](S|T)[^P].  List of high REDOX potential laccases by number of N-Glycosylation sites: matched.fasta.  The candidate species are given in table 1 below.  Interestingly no high REDOX potential laccases contained N-Glycosylation sites utilizing the serine motif: S-matched.fasta

Table 1 List of species identified as secreting high potential laccases.  Click table headers to sort by column.  

Methods:

I used the SPYDER python IDE and the quick install for biopython, DWIM Perl for Windows, and theopen-source NCBI tool FASTA match RegEx.  Fusarium oxysporum, Trametes versicolor, and Pleurotus ostreatus laccase sequences were used as the sample data set to test the accuracy of the program.  Fragment-free sequences (fig. 3) were retrieved from UniProt protein database using the following search term on the UniProKB database: laccase taxonomy:"Fungi [4751]" length:[100 TO 1000].  

I decided to use perl to do the sequence parsing as I was able to modify FASTA match RegEx for my purposes. In the deconstruction of the code I found that the FASTA match regex relied upon the JLS:fasta.pm dependency to parse the fasta file was into individual array entries using FASTA.pm.  The original program only searched the defline of the FASTA file, so it was modified to search the entirety of each entry.  The original code I modified is commented out in lines 242-262, and my modification is in 262-274.  To run the program simply download the new FASTA.pm from the github repository, and replace the original Fasta.pm in the modules folder.  

The 1331 sequences from UniProt were trimmed to an average length of 453 AA (fig. 3).  These sequences were renamed to laccase.fasta and were run through the program to remove medium and low REDOX potential laccases using the following parameters:  

cd "C:\install\FASTA_Match_Regular_Expression_1.0"

perl fasta_matchregex.pl -i C:\FastaIO\laccase.Fasta -o C:\FastaIO\Matched.Fasta -r HCH...H....F

Matching other regular expressions is easily achieved simply by modifying the text after the -r modifier.  Searching other fasta files is also possible, this time by modifying the file path after the -i modifier.  If a different output filename is desired the path following the -o modifier my be modified.  

Using this versatility, the output file 'matched.fasta' was renamed and used as the input file for the secretion prediction.  The high REDOX potential laccases were were run through the program to remove laccases without secretion tags using the following parameters:

cd "C:\install\FASTA_Match_Regular_Expression_1.0"

perl fasta_matchregex.pl -i C:\FastaIO\high_E0.Fasta -o C:\FastaIO\Matched.Fasta -r N[^P](S|T)[^P]

Fragment free laccase sequences:

laccase.fasta

High REDOX potential laccases:

High_E0_matched.fasta

Secreted high REDOX laccases:

T-matched.fasta

Discussion:

As illustrated in Fig. 4 all three test species were identified as high REDOX potential laccases.  T. versicolor is a medicinal white rot fungus, P. ostreatus is a choice edible mushroom, and F. oxysporum is a fast-growing ascomycete with a promising assortment of biotechnological uses and a robust genetic background.  Electrochemical data is unavailable for Fusarium oxysporum, but it is the only high REDOX potetial Ascomycete identified.  This is encouraging because the diploid nature of Basidiomycetes can make them difficult to genetically modify.    Kano et al. state that it is unlikely that an Ascomycete laccase has as high of a REDOX potential as a Basidiomycete laccase, but F. oxysporum seems to be the most promising candidate yet.  Future work will directly compare the electrochemical performance of the species used in the sample data set, F. oxysporum, T. versicolor, and P. ostreatus.   

Consistent with the enzyme assay results, (Evans unpublished data) the program identified the Pleurotus ostreatus as the least likely fungus to secrete laccase (fig. 5).   Though oddly enough only one of the test species, Fusarium oxysporum, was identified as having high secretion potential (table 1).  This is inconsistent with the enzymatic activity data obtained previously which showed that F. oxysporum and T. versicolor secrete laccase at approximately the same levels.  This indicates the simple methodology of secretion prediction was breaking down, and that a more robust statistical methodology is needed to model a complex biological processes such as secretion.  In the future the regular expression identification of secretion potential will be replaced by secretome.P's more advanced secretion prediction algorithms.  

Finally, the program is robust enough to search any fasta file for any regular expression.  The program can use any standard format fasta file and any regular expression can be input into the command line interface.  This enables motif finding using regular expressions in DNA, RNA, and protein sequences from a single platform.  

References:

1. Reiss, R., Ihssen, J., Richter, M., Eichhorn, E., Schilling, B., & Thöny-Meyer, L. (2013). Laccase versus laccase-like multi-copper oxidase: a comparative study of similar enzymes with diverse substrate spectra. PloS one, 8(6), e65633.
2. Pardo, I., & Camarero, S. (2015). Laccase engineering by rational and evolutionary design.Cellular and Molecular Life Sciences, 1-14.
3. Arechederra RL, Minteer SD. Organelle-based biofuel cells: immobilized mitochondria on carbon paper electrodes. Electrochim Acta 2008;53:6698--6703.
4. Calabrese Barton S, Kim H, Binyamin G, Zhang Y, Heller A. Electroreduction of O2 to water on the “wired” laccase cathode. J Phys Chem B 2001;105:11917--11921.
5. Atanassov P, Apblett C, Banta S, Brozik S, Calabrese Barton S, Cooney MJ, Liaw BY, Mukerjee S,Minteer SD. Enzymatic biofuel cells. ElectrochemSoc Interface 2007;16:28--31.
6. Kano K, Ikeda T. Bioelectrocatalysis, powerful means of connecting electrochemistry to biochemistry and biotechnology. Electrochemistry 2003;71:86--99.
7. Vaddiraju S, Tomazos I, Burgess DJ, Jain FC, Papadimitrakopoulos F. Emerging synergy between nanotechnology and implantable biosensors: a review. Biosens Bioelectron 2010;25:1553--1565.
8. Heller A. Miniature biofuel cells. Phys Chem Chem Phys 2004;6:209--216.
9. Heller A. Potentially implantable miniature batteries. Anal Bioanal Chem 2006;385: 469--473.
10. D'Souza, T. M., Boominathan, K., & Reddy, C. A. (1996). Isolation of laccase gene-specific sequences from white rot and brown rot fungi by PCR. Applied and Environmental Microbiology, 62(10), 3739-3744.