Introduction

It is well established that environmental and pathogenic bacteria are capable of growing as complex, surface-attached communities enmeshed in extracellular polymeric substances (EPS) known as biofilms [1]. In microbial fuel cells (MFCs), electrogenic (power-generating) bacteria that utilize multiple carbon sources can generate direct current electricity in planktonic (free-swimming or floating) and especially biofilm (surface-attached) cultures [2]. An ideal MFC biofilm occurs when bacteria bind tenaciously to an MFC electrode (either anode or cathode) at high densities in an open, porous structure that allows for nutrient/waste exchange by biofilm bacteria. The microbes should also be highly electroactive (capable of rapidly transporting electrons between cells and the electrode) either directly via direct surface attachment (e.g., mediatorless) or via soluble redox-active mediators [3, 4]. The anodic biofilm bacteria should ideally be able to oxidize a myriad of organic molecules. Thus, the Gram-negative rod, Pseudomonas aeruginosa (PA), is an ideal organism that fulfills these requirements, being highly competitive with other organisms in both soil and aquatic environments. PA forms robust biofilms on multiple surfaces including the anode of MFCs [5,6,7] and is highly electrogenic using both pilus-mediated (mediatorless) and phenazine-based (mediator) conductance [8, 9].

In PA, the surface appendages (flagella and type IV pili) are required for optimal biofilm formation [10]. However, there are hundreds of genes that affect various aspects of biofilm formation, including attachment, cell division, micro/macro-colony formation, matrix biosynthesis, and dispersion (active release of biofilm bacteria back to the free-swimming planktonic phase) [11]. One gene, pilT, encodes an ATPase, which, when absent, increases the number of type IV pili (polar, retractable, conductive appendages that enhance electron transfer to the anode [6]). In addition, a pilT mutant is incapable of a surface mode of movement known as twitching motility. Chiang and Burrows previously reported that suppression of twitching motility enhances biofilm formation by promoting surface attachment and cell-to-cell adhesion (for model, see Fig. 1, [12]). We recently demonstrated that a PA pilT mutant generates nearly three times the direct current electricity of wild-type bacteria using urine as a carbon source [5]. The pilT mutant also has been shown to be impaired in virulence when compared with wild-type bacteria [13].

Fig. 1
figure 1

a Predicted bacterial biofilm structure upon pilT, bdlA or pilT bdlA mutations. b Predicted biofilm structure of each strain. The green color represents the bacteria, while the light blue color represents the biofilm matrix. c Predicted dispersion patterns in response to chemoattractants or chemorepellents in a flow from left to right. The conditions depicted are meant to demonstrate the response of biofilm bacteria, not necessarily the conditions in the MFCs used in these studies

Full size image

Another gene discovered in 2006, coined bdlA (for biofilm dispersion locus), is critical for bacterial dispersion from mature biofilms [14]. The pilT bdlA double mutant has all the properties of a pilT mutant mentioned above, but the bdlA chemotaxis gene is inactivated. Based on previous studies using PA [14, 15], the process of chemotaxis generally is vital for the bacteria to disperse from a biofilm when the carbon sources are scarce or even in abrupt abundance. The bdlA mutation can impede the dispersion process in bacteria enmeshed in biofilms even upon an immediate tenfold increase in the concentration of glutamate in the medium [14]. As the suppression of dispersion could lead to thicker biofilms, we postulated that the pilT bdlA double mutant would enhance the electrogenic potential when compared to wild-type and the pilT mutant organisms. The pilT mutant alone has been shown to generate substantially higher power than wild-type bacteria [5]. Figure 1 depicts the bacterial structure (A), predicted biofilm structure (B) and how the strains could potentially react to the sudden addition of a chemoattractant or chemorepellent (C). This work explores the electrogenic potential of three mutants of PA, namely a hyperpiliated pilT mutant, a chemotaxis-suppressed bdlA mutant and a pilT bdlA double mutant.

Methods

Experimental setup

Cylindrical MFC reactors (110 ml anode/cathode chambers) were constructed using (poly (methyl methacrylate), PMMA) as previously described [5]. Graphite felt (GF, 4 cm × 0.8 cm × 0.6 cm, Bay Composites, MI) served as the anode, while Pt/carbon cloth (3 cm × 0.8 cm, Fuel Cell Store, Texas) served as the cathode. A Nafion 117 (Fuel Cell Store) proton exchange membrane divided the anode and cathode chambers. All materials were sterilized and pre-treated as discussed in our previous work [5]. The catholyte was 50 mM potassium ferricyanide (pH 7.0), while the anolyte was Luria (LB) broth (10 g tryptone, 5 g yeast extract, 5 g NaCl per liter). Rich media such as LB broth or nutrient broth are used frequently in MFC research to foster not only robust bacterial growth, but also biofilm formation and, ultimately, enhanced power generation [5, 16].

Inoculation of MFCs

PA wild-type strain PAO1, a wound isolate and annotated sequenced strain (www.pseudomonas.com) [17], was maintained indefinitely as a frozen stock at − 80 °C in 1 ml of stationary phase bacteria grown in LB medium and 1 ml of 45% sterile glycerol until use. Isogenic pilT, bdlA and pilT bdlA mutants were constructed by insertional mutagenesis using a gentamicin and/or tetracycline resistance cassettes followed by sucrose counter-selection to resolve co-integrates as previously described [18]. All strains from frozen stocks were first grown on LB agar plates under aerobic conditions for 24 h at 37 °C. An isolated colony was inoculated into 5 ml of LB broth and incubated with continuous shaking at 37 °C for 12–16 h. One ml of the stationary phase suspension was then used to inoculate 110 ml of fresh LB broth in the MFC anode. MFCs were maintained at room temperature (~ 21–23 °C) after inoculation and all the electrochemical studies were performed between days 1–5 post-inoculation.

Data procurement

Electrochemical data (Fig. 2) were collected using a Gamry potentiostat (PCI4300-32034) as previously described [5]. Voltage and the current measurements were used to determine the MFC power density curves (Fig. 2a) and electrochemical impedance spectroscopy (EIS) was also performed (Fig. 3). Under ideal circumstances, EIS is used to elucidate the observations made through polarization studies and calculate the total resistance of the full cell. EIS was performed at the open circuit voltage with a frequency range varying from 0.01 to 100 kHz. The amplitude of the used AC voltage was 10 mV. All EIS and polarization were performed at room temperature (21–23 °C).

Fig. 2
figure 2

Electrochemical data for pilT (red), pilT bdlA (orange), wild type PAO1 (blue), and bdlA (green): a full-cell power curves using graphite felt (12 cm2) anode and a Pt/carbon cathode. b Polarization data. The poor performance of the bdlA MFC is due to high internal resistance (large initial slope)

Full size image
Fig. 3
figure 3

Bode plot comparing the electrogenic performance of wild-type PAO1and mutant strains. The Bode plot implies that the intermediate performance of the pilT bdlA mutant is not the result of averaging pilT and bdlA electrochemical responses. Rather, the bdlA mutant displays a qualitatively different electrochemical equivalent circuit reflected by the absence of an RC circuit element with a characteristic frequency (\( = 1/{\text{RC}}) \) of about 30 Hz

Full size image

Twitching motility assays

Twitching motility is a form of surface-associated movement by bacteria such as PA that is based upon extension and retraction of polar type IV pili [19]. This process is directly linked to the ability of PA to form robust biofilms, as strains lacking type IV pili form little or no in vitro-cultured biofilms [10, 20]. Overnight bacterial plate-grown cultures of PA strains PAO1, pilT, bdlA or pilT bdlA were stab-inoculated through 1% LB agar (3 mm thickness) to the bottom of the Petri dishes. The plates were then incubated for 2 days at 37°C, followed by incubation at room temperature until the agar was less than 1 mm thick. The agar was removed with tweezers and the twitching motility zone on the surface of the Petri dish was visualized by staining with 0.1% Coomassie Brilliant Blue R-250.

Crystal violet staining assay and colony forming unit (CFU) assessment of bacterial biofilm formation

Overnight bacterial cultures were diluted evenly in 200 μl of LB broth in 96-well polystyrene microtiter dishes and incubated at 37 °C for 24 h. Each plate containing all four bacteria biofilm cultures (PA, bdlA, pilT and pilT bdlA) was removed from the incubator daily from days 1–5 to investigate the amount of attached biofilm bacteria using the crystal violet (CV) staining assay developed by O’Toole and Kolter [10]. Briefly, the bacterial biofilms were washed twice with phosphate-buffered saline (PBS) after the unattached bacteria were decanted. For crystal violet staining, 200 μl of 0.1% crystal violet was added to each well and incubated at room temperature for 20 min. The crystal violet was removed and the adherent biofilms washed twice with water. Finally, 200 μl of 95% ethanol was added to each well to dissolve the crystal violet bound to the peptidoglycan layer of biofilm bacteria. After incubation for 30 min at room temperature, the level of purple color in each plate was determined by spectrophotometric analysis at O.D.595 using a microplate reader. For CFU counts, PBS-washed and serially diluted samples were diluted tenfold and 10 μl of each dilution was pipetted on LB agar plates and CFU was recorded after 18 h incubation at 37 °C.

Results and discussion

Electrogenic performance evaluation of LB-fed MFCs

To investigate the electrogenic performance of the four selected bacterial strains, full-cell polarization analyses were performed. The cell voltage was decreased from open circuit voltage (OCV) to near zero. Figure 2a shows full-cell power curves comparing wild-type PAO1 to the three mutant strains. The power curves are calculated from the measured current–voltage curves in Fig. 2b. Table 1 captures the performance parameters: peak power density, short-circuit current density and fill factor. The fill factor is the ratio of maximum power output of the cell to the theoretical maximum power output of the cell (product of OCV and short circuit current). The fill factor can also be directly correlated to the performance of the mutant. The error bars on the bdlA and pilT bdlA mutants are based on multiple measurements. For strain PAO1 and the pilT mutant, we used the standard deviation observed in our previous work [5].

Table 1 MCF performance parameters
Full size table

The pilT mutant outperformed wild-type strain PAO1, while the bdlA mutant unexpectedly underperformed based on our initial hypothesis that the double pilT bdlA mutant would be the most electrogenic of the four strains. The electrogenic output of the pilT bdlA fell between the bdlA and pilT mutants, but still outperformed strain PAO1. Figure 2b indicates that the poor performance of the bdlA MFC is due to high internal resistance (refer to the initial slope). Although the bdlA cell produced the highest voltage at open circuit, the voltage dropped precipitously as the bacteria began to produce current.

Based on the aforementioned results, understanding the differences between the electrogenic performance of the pilT versus the bdlA and pilT bdlA mutants emerged as a critical feature of this study. The pilT mutant overexpresses type IV pili on the poles of the bacterium, which enhances electron transfer to the anode [5]. In addition, the pilT and bdlA mutants are incapable of twitching motility [12, 14], a form of surface-associated motility known to enhance biofilm formation by promoting surface attachment and cell-to-cell adhesion [12]. It was initially hypothesized that construction of another mutation within the PA bdlA gene in the pilT mutant background would further improve the power output, the logic being that suppressed chemotactic movement would impede dispersion from the already hyper-electrogenic pilT biofilm, thereby enhancing the electrogenic properties of the double pilT bdlA mutant. However, we observed that the performance of the bdlA mutant was below the performance of the wild-type strain PAO1. The pilT bdlA mutant performance fell between pilT and bdlA, but was still greater than that of PAO1.

EIS measurements

This section examines the electrochemical impedance spectroscopy (EIS) data from each of the four strains. The measured values of the imaginary impedance (Zimaginary(\( f \))) and real impedance (Zreal(\( f \))) were used to calculate the magnitude of the impedance (\( \left| Z \right| \)) at different frequencies, \( f \). The Bode plot in Fig. 3 displays the frequency dependence of \( \left| Z \right| \). Several observations emerged from the Bode plot. We attribute the impedance at highest frequency (105 Hz) to solution resistance (RS) and the impedance at zero frequency (0 Hz) to total resistance (RT). RT is considered to be the sum of all types of resistances in a full-cell (i.e., anodic, cathodic) solution and the biofilm resistance. [5] Since the data only extended to 0.01 Hz, we can only state that RT > \( \left| Z \right| \) (0.01 Hz). \( \left| Z \right| \) at 0.01 Hz and the solution resistance (Rs) are listed in Table 2.

Table 2 Solution, anodic and total resistance 48 h after inoculation as calculated from Fig. 3
Full size table

Several observations ensued from the EIS data. The larger \( \left| Z \right| \) of the bdlA mutant confirms the observation made from Fig. 2b. High internal resistance has been observed to deleteriously affect the performance of MFCs [5]. The Bode plot also implies that the intermediate performance of the pilT bdlA mutant is not the result of averaging pilT and bdlA electrochemical responses. Rather, the bdlA mutant displayed a qualitatively different electrochemical equivalent circuit reflected by the absence of an RC circuit element with a characteristic frequency (= 1/RC) of about 30 Hz. Although the origin of this circuit element is not understood, the resulting EIS profile may be a diagnostic tool to probe the impact of genetic modification on microbial electrochemistry. In the double pilT bdlA mutant, for example, the resistance associated with the 30-Hz element was enhanced compared to all other strains. This resistance contributes to RT and thus reduces the performance. Finally, after 48 h, the peak power of the pilT bdlA mutant dropped rapidly: at 55 h the peak power is comparable to wild-type strain PAO1 (not shown). As the peak power dropped, the resistances of all the EIS circuit elements increased (compare 48 and 55 h data in Fig. 3). This change could also indicate increased levels of dead or metabolically inactive bacteria.

Twitching motility and biofilm formation

To better comprehend the differences in the power output of each strain, twitching motility patterns as well as overall biofilm formation were examined. The primary reason behind the lower power output of the bdlA mutant was hypothesized to be a higher total MFC internal resistance. The higher internal resistance could be due to thicker biofilms with metabolically inactive/dormant bacteria or defective twitching motility. To test this hypothesis, we first measured twitching motility in each of the four strains to confirm that the three mutant strains are defective in twitching motility (Fig. 4). The pilT, bdlA and pilT bdlA mutants were all defective. The absence of twitching motility therefore does not differentiate the strains, since the pilT mutant generated a higher power density than wild-type bacteria [5] whereas the bdlA generates the lowest.

Fig. 4
figure 4

Twitching motility by various PA strains. The four PA strains were stabbed to the base of 1% LB agar medium and the twitching motility zone allowed to develop for about 5 days. The twitching zone was amplified by staining of the plates with 0.1% Coomassie Brilliant Blue R-250

Full size image

Thicker biofilms with inactive bacteria could potentially impede the transfer of electrons through the biofilms leading to a drop in the power output. To examine this possibility, bacterial viability was determined by measuring colony forming units (CFU) counts in biofilms cultured by each organism. Our results showed no difference in viable bacteria (Fig. 5a). Finally, to measure biofilm density, the crystal violet biofilm assay was again utilized [10]. This assay is commonly used to measure biofilm formation because of the direct relationship between O.D.595 and surface-attached bacterial titers. Figure 5B indicates that the bdlA biofilm is almost twice as thick as that of strain PAO1. Since bacterial viability from our CFU count was similar in the two biofilms, it is likely that the bdlA biofilm harbors dormant or dead bacteria at the electrode surface, which explains the lower performance of this strain when compared with the other strains. Film thickness, however, is not sufficient to explain all the observations. On the basis of thickness alone the pilT and pilT bdlA mutants should perform similarly, whereas Fig. 2 shows that the pilT mutant is a substantially better electrogen. Further work is required to understand other factors contributing to the deteriorating performance of the bdlA mutant.

Fig. 5
figure 5

CFU count and crystal violet assay of bacteria biofilms in MFC. a CFU count of 3-day-old bacteria biofilm in 96-well plates grown at room temperature (mean ± SE of three replicates). b Crystal violet (0.1%) was used to stain 3-day-old bacterial biofilm in 96-well plates grown at room temperature and the O.D. at 595 nm recorded using a microplate reader (mean ± SE of three replicates)

Full size image

Conclusions

We observed that a pilT mutant of PA yielded a power output of 5.0 times that of the wild-type strain PAO1 and 3.2 times the pilT bdlA mutant (pilT > pilT bdlA > PAO1 > bdlA). The enhanced output of the pilT mutant was due to reduced twitching motility and overexpression of polar conductive pili relative to wild-type bacteria. The bdlA mutation shows increased biofilm thickness in addition to reduced twitching motility. Since all four strains showed similar CFU counts, the increased thickness of the bdlA mutant was correlated with an increased population of dead or dormant bacteria. Such bacteria increase biofilm resistance, especially if concentrated near the electrode. The reduced power output and the increased total resistance confirmed the presence of a “dead cell layer”.

Industrial relevance

Downstream industrial studies, although highly desirable, are currently not feasible due to a variety of issues affecting the overall performance of MFCs. These include but are not limited to (1) non-optimized electrogenic bacteria, (2) low power output, (3) non-optimal fuel cell design, (4) internal resistance, (5) biofouling and (6) scalability issues. Here, we offer a very promising solution, at least from the improvement of bacterial electrogen perspective. Using synthetic biology (SynBio) techniques, which marry electrical Boolean logic gates with bacterial genetic logic gates, we argue that MFC researchers can dramatically improve the overall power density of future MFCs. Our research team is currently using SynBio technology to construct highly controlled gene circuits designed to (1) optimize carbon skeleton utilization, (2) enhance biofilm formation while limiting nutrient-impermeable exopolysaccharide production, and (3) optimize electron transfer from conductive pilus “nanowires” and from overproduction of redox-active mediators, pyocyanin and pyorubrin. Another issue for the use of PA in applied MFCs is the potential to cause human, animal or plant disease. In this case, we would create strains that are avirulent (unable to cause disease). Many genes in PA are required for full virulence in various animal models. Constructing mutations in genes such as the global regulator gacA [21] and the suhB gene [22] would eliminate any concerns for spread of disease from applied MFCs.