Poly(3,4-ethylenedioxythiophene):polystyrene sulfonate (PEDOT:PSS), Nafion solution, dimethyl sulfoxide (DMSO), phosphate-buffered saline solution (PBS), glutaraldehyde, and 3-glycidoxypropy-trimethoxysilane were purchased from Sigma-Aldrich. Yeast extract and NaCl were procured from Fisher Scientific. Tryptone was purchased from G-Biosciences. Activated carbon (AC) was purchased from Cabot Corporation. A nickel conductive spray for the cathode was purchased from MG Chemicals. Graphite ink was obtained from Ercon, Inc. Filter paper was purchased from the General Electric Company (Whatman 3MM CHR). Poly(methyl methacrylate) (PMMA) sheets were procured from McMaster-Carr.

Very recently, the Whiteside group created a novel method that could co-fabricate electronic and microfluidic structures (called “electrofluidic” structures) on paper (Hamedi et al., 2016a,b). They used the water-dispersed conducing polymer mixture PEDOT:PSS as a conductive ink. Upon drying of the ink patterned on paper, the conductive reservoir was formed without blocking the pores of paper, which was still hydrophilic. This technique finds the best-fit solution for our paper-based microbial electrogenic applications by providing a fluidic component that allows mass transfer of ions and an electrical wire that allows conduction of electrons generated from bacterial metabolism. A 20 µl mixture of 1 wt% PEDOT:PSS and 5 wt% DMSO is pipetted into each anode reservoir and air-dried for 8 h (Figure 3A). To further increase anode reservoir hydrophilicity, 20 µl of 2 wt% 3-glycidoxypropy-trimethoxysilane is added to the reservoir and air-dried. This technique creates a conductive, biocompatible and porous scaffold for the exoelectrogens placed in each paper reservoir to efficiently transfer electrons to the anode. Furthermore, this task is the first example of using the electrofluidic engineering technique and revolutionizes the power generation of the MFC.

Air-cathode MFCs offer the best promise for a single-sheet paper-based bacterial screening array, since the cathodic chambers for holding the chemical catholyte can be removed from the system, which reduces device complexity, simplifies fabrication steps, and increases power generation (Lee and Choi, 2015; Fraiwan et al., 2016). Moreover, oxygen is readily accessible, sustainable, and environmentally friendly. In this work, an AC-based air-cathode on paper substrate is constructed with carbon paint/nickel spray that provides structural support and functions as a current collector (Figure 3A). The carbon paint with AC catalysts (15 mg/cm²) in a binder solution is screen-printed on a Nickel-sprayed paper zone. The binder solution is prepared by mixing 150 µl of deionized water, 600 µl of isopropanol, and 1200 µl of 5 wt% Nafion solution.

The 8-well MFC sensor array is designed on a single sheet of the filter paper (Figure 2). The hydrophilic reservoir is first defined with hydrophobic wax boundaries on the paper simply by using a commercially available solid-wax Xerox Phaser printer and letting the wax penetrate through the paper with heat treatment (Gao and Choi, 2017). The hydrophobic wax melted into the paper substrate (130°C for 2 min) defines individual MFC units for batch-fabrication and strengthens the paper fiber matrix while retaining its flexibility (Figure 3A). The wax is used as the PEM, as it provides the hydrophobic property of the paper, separating the anodic compartment form the cathode and allowing protons to pass through efficiently. The vertical or horizontal penetration depth of the melted wax is carefully controlled by adjusting the temperature and heating time (130°C for 45 s). The air-cathode is formed on the wax-based PEM. A mixture of PEDOT:PSS and DMSO is pipetted into the defined anodic regions to make it function as an electrofluidic, and 3-glycidoxypropy-trimethoxysilane is added to improve the hydrophilicity. The fabricated single-sheet MFC array is sandwiched between PMMA frames with electrical contacts and is attached with a spray adhesive (3M Super 77, multipurpose adhesive) (Figures 3B,C). Two PMMA frames function as the electrical contacts for anode and cathode and as a supporting structure (Figure 3B). The PMMA frames are micromachined by laser cutting (Universal Laser System VLS 3.5). Then, graphite ink is screen-printed on both sides of PMMA frame for the anode and nickel is sprayed for the cathode. Carbon-based anode material is chosen for the biocompatibility, while nickel is selected for the cathode for high conductivity. After screen printing and spray deposition, the PMMA frames are baked in ventilated oven at 65°C for 2 and 1 h, respectively. A window is made on the cathode PMMA frame to allow the use of freely available oxygen in the air as the electron acceptor.

To demonstrate the MFC array for studies of bacterial electrogenicity, Pseudomonas aeruginosa was selected as our model organism. P. aeruginosa has been thoroughly studied with its complete genome sequenced, and there are well-established techniques for their genetic manipulation. For the main microbial EET mechanisms, namely, (a) direct contact, (b) shuttling compound, and (c) conductive type IV pilus nanowires P. aeruginosa utilize direct contact electron transfer (mediatorless) and shuttling compounds (mediators) as pyocyanin and pyorubrin (Hernandez and Newman, 2001; Choi, 2015). Our working hypothesis is that the genetic modifications of microbial metabolic/signaling pathways and surface structures greatly affects the electrochemical activity of P. aeruginosa strains, and the MFC array provides fast, reliable, accurate information for a quantitative understanding of the resulting electrogenicity. In this work, wild-type PAO1 and five of its strategically engineered mutants pilT, fliC/pilA, bdlA, lasR/rhlR, and pmpR were tested using the proposed screening platform while deionized water and Luria-Broth (LB) medium are used as the negative controls. The mutants are engineered using classical allelic replacement techniques with sucrose counter-selection (Hoang et al., 2000). Microorganisms are first cultivated in LB medium for 24 h at 37°C, and then centrifuged and re-suspended in fresh LB medium with the cell titers controlled by monitoring the optical density at 600 nm. The detailed information of five genetically engineered P. aeruginosa and the rationale behind the genetic modifications for this work is described below in detail.

Using an 8-channel pipette, 22 µl of bacterial suspension was simultaneously introduced into each MFC unit in the array. The strong capillary adsorption of the paper promotes bacterial attachment to the anode while the conductive paper reservoirs harvest electrons generated by bacterial respiration. A Scotch Magic™ tape covers the anode reservoir after injecting the sample to prevent evaporation. A data acquisition system (NI, USB-6212) with a customized LabVIEW interface was used to measure the potential between anode and cathode every 1.44 s. To determine the current generated by the MFC units, external resistors (470 kΩ) are individually connected between the anodes and cathodes, and current flows through the resistors are calculated by Ohm’s law.

After testing, the MFC array was disassembled and the single-sheet paper layer was gently rinsed with PBS. Immediately after rinsing, the paper layer was immersed in 2% glutaraldehyde solution at 4°C overnight. Dehydration was achieved by submerging the layer in increasing concentrations of ethanol (50, 70, 80, 90, 95, and 100%) for 5 min each. The paper layer was then examined using a FESEM (Supra 55 VP, Zeiss).

Surface morphology, micro-pores, and the presence of attached bacterial cells of the bare paper and the PEDOT:PSS-treated paper reservoir were assessed by SEM analysis. As shown in Figure 4, the PEDOE:PSS-coated paper shows the same morphology of the bare papers without blocking the paper pores and this macro-porous structure ensures a large surface area and efficient mass transfer to and from the bacteria. The PEDOT:PSS polymer is also very useful as an anode material because of its facile processibility, hydrophilicity, biocompatibility, and stability, features of which lead to fast and dense bacterial adhesion (Jiang et al., 2015). Furthermore, the polymer modification creates more activation centers on anodic surfaces to improve bacterial adhesion and their electrocatalytic activity, increasing electron transfer efficiencies (Guo et al., 2015). Figure 4D shows a higher density of bacteria in the PEDOT:PSS-coated paper reservoir than that on the bare paper alone (Figure 4B), which is well reflected in the polarization curve and output power measured as a function of current (Figure 5). The polarization curve and power outputs are drawn based on the maximum values measured at a given external load. The maximum power density of the PEDOT:PSS-coated paper MFC is about 0.4 µW/cm², which is twice that of the bare paper MFC. Using the polarization curve, we can also estimate internal resistances from the external resistor values where the maximum power densities are obtained. The bare paper anodic reservoir shows the substantially higher internal resistance (800 kΩ) compared to the PEDOT:PSS MFC (200 kΩ). The PEDOT:PSS-coated paper reservoir has a sheet resistance of about 386 Ω/square determined by a four-point probe measurement. The conductivity is enhanced by treatment with the polar solvent, DMSO. Because the conductivity of the paper reservoir is strongly related to the resistance to the microbial electron transfer, the conductive paper modification is helpful for reducing internal resistance, thus ultimately increasing the MFC power performance and providing sensitive power assessment of electrogenic bacteria.

Before we characterized bacterial electrogenicity of the different bacteria strains, we first tested the reliability of the MFC array by measuring the OCV variation from individual device units. The OCVs of the 8-well MFCs generates 1.7% variation with LB medium, which is far less than that of other MFC arrays (25%) and even our previous MFC arrays (2.5%) (Choi et al., 2015). This low percent deviation is mainly because of (i) consistent cathodic reactions from device to device by accessing freely available oxygen in the air and (ii) simplified device configuration, fabrication, and operation by integrating all components in a single sheet of paper. After we confirmed that our MFC array has such a low percentage deviation, we began our experiments with different anodic samples. Eight samples were prepared: deionized water, LB medium, wild-type PAO1, pilT, fliC/pilA, bdlA, lasR/rhlR, and pmpR. After the anodic samples were introduced into the MFC units and operated under open circuit mode for 15 min, the OCVs from individual MFC units in the array were first compared and the experiments repeated three times (Figure 6A). The OCVs range from 92 to 118 mV with the SD of less than 2%. while the MFC units with water and LB medium have very low OCVs. The highest OCV value is obtained from pilT while noticeably second highest is produced from wild-type PAO1.

To evaluate bacterial electrogenicity, an external resistor (470 kΩ) is used to connect to each MFC unit and the current output through the resistor is calculated 20 min after the load connection (Figure 6B). The negative controls generate a certain amount of current because of ions present in the samples but significantly less current outputs than the other bacterial samples. The current generations revealed significant differences between the respective MFC units. The highest current was observed from the pilT mutant, followed by the wild-type PAO1 while the other mutants, fliC/pilA, bdlA, and lasR/rhlR show the similar electrogenic capabilities but much lower than the pilT mutant. The pmpR mutant yielded the lowest current.

In the paper reservoir of our previous 48-well MFC for holding the bacterial inoculum, only a few cells adjacent to the anode contributed to the current because the non-conducting paper matrix cannot be an electron acceptor (Figure 1) (Choi et al., 2015). The bacterial EET occurred within a short distance of the anode. Thus, we could not harvest all the electricity that would seem to be available from all the bacteria dispersed throughout a paper reservoir, impairing the device performance, which is inevitable even given the most current technological advances. To overcome this issue, a highly conductive reservoir directly connected to the anode is required. Although electrically conducting papers are available, it is quite challenging to pattern non-conducting papers with conducting materials while enabling them to have open pores and hydrophilic features for subsequent liquid sample introduction. Co-fabricating conducting paths and microfluidic channels structures, namely electrofluidic structures (Hamedi et al., 2016a,b) on paper, creates the best-fit reservoir to efficiently transfer electrons from all the bacteria on the paper of the anode. The application of the electrofluidic technique with PEDOT:PSS over the patterned small paper reservoir of the MFC results in a faster and higher power generation only with microliter sample volume compared to conventional non-conducting paper-based devices. This innovation will be essential in creating a large-scale MFC array for high-throughput and rapid bacterial electrogenic screening.

As an OCV of a device reflects its thermodynamic balance between anode and cathode and all the cathode reactions of the MFC are the same in the array, the OCVs obtained from individual MFCs show the thermodynamic differences of bacterial strains in oxidizing organic substrates. The measured voltages vary significantly between the MFC units, which clearly indicate different thermodynamic reactions of the bacterial strains (Figure 6A). This result indicates that the overproduction of the type IV pili by the pilT mutant would enhance bacterial interactions with the anode and reduce the anode overpotential compared to other strains. It should be noted that our previous work with the different bacterial strains had little influence on the OCVs because of their low sensitivity in the non-conductive paper reservoirs (Choi et al., 2015). Although our proposed MFC array provides a reliable screening platform for exoelectrogens with low percentage variation, the OCV values generated from our MFC array are much lower than those of typical MFCs (300–600 mV), which is highly related to overall MFC sensitivity and power assessment. This is mainly due to two fundamental challenges: (i) oxygen diffusion into the paper reservoir is more severe than the conventional MFC device and oxygen abiotically reacts with the anode (Mukherjee et al., 2013) and (ii) an air-cathode configuration suffers from high cathodic overpotential (Xie et al., 2013). Applying a novel oxygen-impermeable membrane for the device and an innovative cathode with high-performance [e.g., solid-state cathodic electron acceptors (Xie et al., 2013)] will be a potential solution to realize a more practical and powerful tool for accurate and parallel analysis of electrogenic bacteria.

For the hyperpiliated pilT mutant, our results demonstrate that hyperpiliation of the type IV pili increases the cell-to-cell, cell-to-surface adhesion and promotes early stage biofilm formation (Giltner et al., 2006), consequently enhancing the bacterial electricity generation (Figure 6B). This is in good agreement with our previous report and other studies (Mukherjee et al., 2013; Choi et al., 2015; Shreeram et al., 2016). The direct EET efficiency of the pilT mutant is further boosted by the 3D conductive anode reservoir which provides high surface area for enhanced bacterial attachment and interaction with the anode. The fliC/pilA mutant that lacks flagella and type IV pili expectedly generated less current than the pilT mutant and wild-type PAO1, which is also supported by previous work (Mukherjee et al., 2013). Despite the hypothesis that it may increase current generation by forming a “less-dispersive” biofilm, the dispersion-deficient bdlA strain produces lower current than the wild-type strain. This finding can be explained by the work showing that the bdlA mutant shows less enzymatic activity and lacks various proteins such as proteolysis and lipid hydrolysis in contrast to the wild-type PAO1 (Li et al., 2014). Accordingly, the metabolic efficiency of bdlA mutant could be reduced in the amino acid/peptide-rich LB medium. In a similar manner, the lasR/rhlR mutant also lacks certain proteolytic enzymes, such as LasA and LasB, that are responsible for staphylolysis and elastase activity. As the LasA and LasB proteases are regulated by LasR and partially under RhlR control, the lasR/rhlR double mutant fails to degrade milk proteins in an agar medium containing skim milk (Dekimpe and Deziel, 2009). Another factor that could negatively affect MFC performance is the bacteria migration toward nourishment-rich environment. The bdlA and lasR/rhlR mutants all exhibit limited responses toward nutrient-induced stimulus. Further investigations are required to fully understand the effects of these genetic engineering techniques on MFC current generation.

Table 1 summarizes specifications of prior MFC screening arrays and compares our MFC array to them. Although the electrochromic-based approach provides higher throughput screening capability than the conventional MFC-based method, the technique is limited to specific bacterial species whose EET enables electrochromic color changes. Our recent paper-based MFC screening tools offer several advantages over the conventional technique, including high-throughput and rapid power assessment without using external fluidic pumps or tubes. Also, a low percentage deviation between MFCs can be obtained from simplified and reliable batch-fabrication. This work further improves the potential of the paper-based MFC tools by using air-cathode and electrofluidic technique. However, this platform will not provide general analysis for all type of microorganisms having different electron transfer mechanisms. We need to further study what bacterial electron transfer mechanisms through the paper matrix can be preferably investigated in a more effective way.