The interest in redox-regulated gene expression stems from the notions and the expertise we have gained in all the previous chapters, where we have learnt how to control a working electrode potential, to drive the redox state of immobilized and freely diffusing molecules, etc.

Now we are in a position to try to design a possible electrical approach to the control of gene expression, but, to achieve such an ambitious goal, we first have to select a suitable biological system. A plausible choice appears to be to focus on photosynthetic bacteria of the genus Rhodobacter. The reasons for this choice are connected with a number of interesting and suitable features that one can find in these bacteria, apart from the prerequisite of featuring redox-dependent gene expression molecular machinery. Indeed, as we have seen in section 6.2, variation in oxygen tension is responsible for modulating the oxidative state of the inner bacterial compartment that activates the change of metabolism in the microorganism and makes it a facultative photosynthetic bacterium.

The ability that these bacteria have to switch from photosynthesis to respiration is at the core of a straightforward way to monitor the products of their genes. Indeed, the photosynthetic state (low oxygen tension) tends to enhance expression of those genes that encode for proteins involved in light harvesting and in light-induced charge separation (i.e., LHI, LHII, reaction centers), along with the connected pigments, whereas high oxygen tension tends to depress their expression. Therefore, both R. sphaeroides and R. capsulatus (see section 6.2) are convenient objects to monitor in their metabolic variation, simply by following the variation of the optical absorption bands related to their pigments (bacteriochlorophyll monomers and dimers, bacteriopheophytins) brought about by the proteins of photosynthesis.

Other positive aspects of the genus at issue are its robustness and relative ease of growth, which make them good candidates for repeated experimentation in non-conventional environments.

Affecting the inner redox state of a bacterium by direct electrochemistry implies a “hot-wiring” of the bacterial cytoplasm with an external voltage source, e.g., a potentiostat. A direct physical connection is, however, difficult to implement and doable only one cell at a time; as such it appears of limited technological interest.

We seek, indeed, an approach suitable to affect, contemporarily, the inner redox state of a large bacterial population. A plausible possibility to achieve this goal is to use redox molecular relays or mediators whose oxidation state can be conditioned by the interaction with an electrode in the culture medium, whose potential is driven by a potentiostat. These mediators have to be able to freely diffuse across the bacterial membrane, penetrating the cytoplasm and affecting its redox state. By doing so, they play the role of a metal wire piercing the bacterial membrane and directly affecting the redox state inside the cell.

The role of these mediators has to be that of mimicking the effect of oxygen tension variations in modulating gene expression. Therefore, as we have seen in section 6.2, they have to be able to affect the redox state of thioredoxins that in turn differentially bind gyrases, resulting in a different supercoiling of the bacterial chromosome, hence in a modulation of the transcription of the photosynthesis genes.

The overall project appears quite ambitious and its various steps are best faced if we try to dissect the process into its basic constituents. For each step or phase, a specific approach will be sought and solutions will be suggested often using a crucial interplay between top-down and bottom-up approaches. In the next sections we will analyze in detail the principal main phases of a process whose completion can lead to implementing the sought electrical control.

Low-molecular-weight redox species can assist the shuttling of electrons between the intracellular bacterial space and an electrode. However, there are many important requirements that such a mediator should satisfy in order to provide efficient electron transport from/to the bacterial thioredoxins to an external electrode: (a) The mediator should easily penetrate through the bacterial membrane to reach the thioredoxins inside the bacteria. (b) The redox-potential of the mediator should fit the potential of the thioredoxins (e.g., the mediator potential should be positive enough to provide fast electron transfer from the enzymes, but it should not be too positive as to prevent significant loss of potential). (c) Neither oxidation state of the mediator should interfere with other metabolic processes (should not inhibit them or be decomposed by them). (d) The mediator should easily escape from the cell through the bacterial membrane. (e) The oxidation states of the mediator should be chemically stable in the electrolyte solution, they should be well soluble, and they should not adsorb onto the bacterial cells or electrode surface. (f) The electrochemical kinetics of the redox process of the mediator at the electrode should be fast (electrochemically reversible) (Wilkinson, 2000).

Luckily enough, one can take advantage of the extended screening activity on these kinds of redox mediators, conducted in studies dedicated to design and implementation of bacterial fuel cells (Katz et al., 2003).

Many different organic and organometallic compounds have been tested in combination with bacteria to evaluate the efficiency of mediated electron transport from the internal bacterial metabolites to the anode of a biofuel cell. Thionine has been used extensively as a mediator of the electron transport from Proteus vulgaris (Bennetto et al., 1985; Thurston et al., 1985; Delaney et al., 1984; Kim et al., 2000) and from E. coli (Roller et al., 1984; Bennetto et al., 1983). Other organic dyes that have been used include benzylviologen, 2,6-dichlorophenolindophenol, 2-hydroxy-1,4-naphthoquinone, phenazines (phenazine ethosulfate, safranine), phenothiazines (alizarine brilliant blue, N,N-dimethyl disulfonated thionine, methylene blue, phenothiazine, toluidine blue) and phenoxazines (brilliant cresyl blue, gallocyanine, resorufin) (Delaney et al., 1984; Roller et al., 1984; Bennetto et al., 1983; Park & Zeikus, 2000; Ardeleanu et al., 1983; Davis & Yarbrough, 1962; Patchett et al., 1988; Kreysa et al., 1990). These organic dyes were tested with Alcaligenes eutrophus, Anacystis nidulans, Azotobacter chroococcum, Bacillus subtilis, Clostridium butyricum, E. coli, Proteus vulgaris, Pseudomonas aeruginosa, Pseudomonas putida and Staphylococcus aureus bacteria, usually using glucose and succinate as substrates. Among the evaluated dyes, phenoxazine, phenothiazine, phenazine, indophenol, bipyridilium derivatives, thionine and 2-hydroxy-1,4-naphthoquinone were found to be very efficient in maintaining relatively high cell voltage output when current was drawn from the biofuel cell (Delaney et al. 1984). Some other dyes do not function as effective mediators because they are not rapidly reduced by the microorganisms, or they lack sufficient negative potential. Ferric chelate complexes (e.g., Fe(III)EDTA) were successfully used with Lactobacillus plantarum, Streptococcus lactis and Erwinia dissolvens, oxidizing glucose (Vega & Fernández, 1987).

Since thionine has frequently been used as a mediator in microbial fuel cells, mono- and disulfonated derivatives of thionine have been applied to determine the effect of hydrophilic substituents on mediation of electron transfer from E. coli to an anode (Lithgow et al., 1986). Changing from thionine to 2-sulfonated thionine and 2,6-disulfonated thionine results in an increased efficiency of mediated electron transport. The low efficiencies of the biofuel cells operating with thionine and 2-sulfonated thionine were attributed to interference with electron transfer by adsorption of the mediator on the microbial membrane. It should be noted that the overall efficiency of the electron-transfer mediators depends also on many other parameters, and in particular on the electrochemical rate constant of mediator re-oxidation, which depend, in turn, on the electrode material.

Since an electron-transfer mediator needs to meet many requirements, some of which are mutually exclusive, it is not possible to reach perfect conditions for electron transport from a bacterial cell to an electrode. A mixture of two mediators can be useful in optimizing efficiency. A solution containing thionine and Fe(III)EDTA was applied to mediate electron transport from E. coli, oxidizing glucose as a primary substrate to an anode (Tanaka et al., 1983). Although both mediators can be reduced by E. coli, thionine is reduced over 100 times faster than Fe(III)EDTA. The electrochemical oxidation of the reduced thionine is much slower than oxidation of Fe(II)EDTA, however. Therefore, electrons obtained from the oxidation of glucose in the presence of E. coli are transferred mainly to thionine under the operational conditions of the cell. The reduced thionine is rapidly re-oxidized by Fe(III)EDTA, the rate of which has been shown to be very fast, k_et = 4.8×10⁴ M⁻¹s⁻¹. Finally, the reduced chelate complex, Fe(II)EDTA, transfers electrons to the anode by the electrode reaction of a Fe(III)EDTA/Fe(II)EDTA couple with a sufficiently large rate constant. One more example of enhanced electron transport in the presence of a mixture of mediators was shown for Bacillus strains oxidizing glucose as a primary substrate. The biofuel cell was operated in the presence of methylviologen (MV²⁺) and 2-hydroxy-1,4-naphthoquinone or Fe(III)EDTA (Akiba et al., 1985). Methylviologen can efficiently accept electrons from the bacterial cells, but its reduced state (MV^•+) is highly toxic for the bacteria and immediately inhibits the fermentation process. In the presence of a secondary mediator that has a more positive potential, MV^•+ is efficiently re-oxidized to MV²⁺. The reduced secondary mediator (quinone or Fe(II)EDTA) then transports the electrons to the anode.

In Figure 7.1 the relative typical redox potentials of thioredoxin (−0.47 V vs SCE) (Watson et al., 2003) along with those of safranin (−0.53 mV) and thionine (0.18 mV) are reported with the structure of both redox mediators. These two dyes appear to be suitable for affecting the redox state of thioredoxin. Particularly, whereas safranin can reduce the enzyme, thionine can oxidize it. Therefore, these two molecules seem to be a suitable choice for controlling the redox state of the disulfides of thioredoxins. We know, from section 6.2, that the redox state of thioredoxin’s S-S bridges determines whether the molecule is able to bind gyrase, determining eventually the fate of the gene expression in Rhodobacter.

Such an all-or-none attempt has indeed been done using Rhodobacter sphaeroides. The experiment compared the absorption spectra of two different bacterial suspensions in the bacteriochlorophyll spectral range. One of these was left to grow for 24 h in normal aerobic conditions, whereas the second was incubated with 1 mM thionine. The flask also contained three electrodes, a Au wire as WE, a reference (SCE) and a counter (Pt). The working electrode was kept at a potential of 0 V for 24 h. Such a potential helped maintain thionine in an oxidized state. Oxidized thionine, once diffusing inside bacteria, could reduce the disulfides of thioredoxins, providing an increase in the supercoiling activity of gyrases on the bacterial chromosome that should cause a decrease in puc and puf operon transcription (see section 6.2), hence a reduced amount of bacteriochlorophyll present in the bacteria. After 24 h, the amount of bacteria in the two flasks was evaluated by optical absorption spectroscopy; subsequently, bacterial concentration was adjusted to a value suitable for absorption spectroscopy in the bacteriochlorophyll spectral range (700–900 nm) and the corresponding spectra were acquired in a comparative fashion. Figure 7.2 shows the spectra obtained, which show a decrease in the normalized bacteriochlorophyll absorption of the sample under potentiostatic control, whereas the control, without the redox mediator, does not show any appreciable difference due to the application of a potential.

These data are just preliminary and, as such, they do not allow one to draw any robust conclusions on the validity of the approach used. Of course, the questions still to be answered are numerous and even the positive identification of the mechanisms at play can be questioned (e.g., whether the redox mediator operates according to the described mechanism, see section 6.2, or whether it directly affects the bacteriochlorophyll oxidation state).

We believe that to go any further in the mechanistic understanding of the phenomena taking place during redox control of the cytoplasmic space by direct electrochemistry, one needs a more systematic approach probably involving the dissection of the problem into its main constituent parts. At any rate, these preliminary results appear to be promising and deserve further attention.

It is now useful to think of possible ways to analyze the process by subdividing it into a number of steps that allow one to achieve full control of the phenomenon. One can think that the first step to overcome is to demonstrate that the proposed approach is really able to control the oxidation state of the inner space of a model system mimicking a cell envelope. It is conceivable to have an artificial lipid envelope filled with a suitable redox molecule whose oxidation state could be affected by the use of the selected redox mediator. A possibility could be to fill the envelope with potassium ferricyanide, which possesses a redox potential ≈ 436 mV at pH 7, and which could be easily reduced to ferrocyanide by thionine or safranin according to: [Fe(CN)₆]⁴⁻ ⇌ [Fe(CN)₆]³⁻ + e⁻. This reaction can be followed spectroscopically (at 420 nm, molar extinction coefficient = 1040 M⁻¹cm⁻¹) or optically (under a microscope and with suitable concentrations, the reduction reaction makes the dye change its color from brown to pale yellow). Furthermore both ferro- and ferricyanide are known to be impermeable through the plasma membrane, thus ensuring their confinement inside the envelope.

Quite natural candidates for this kind of test appear to be liposomes. They are one of the several possible self-assembling structures that amphiphilic molecules such as phospholipids can form when dispersed in an aqueous solution. For a comprehensive exposition of these and other self-assembling structures the reader can refer to any of the numerous reviews and books present in the literature (for a quantitative description see Israelachvili, 2005). These aggregates can be formed in various sizes in the range from a few nm to several tens of micrometers. According to their typical size, they are called SUVs (for Small Unilamellar Vesicles, from a few to tens of nanometers), LUVs (for Large Unilamellar Vesicles, from tens to hundreds of nanometers), or GUVs (for Giant Unilamellar Vesicles, in the micron to tens of microns range). They consist of a single phospholipid bilayer that, for thermodynamic reasons connected to the minimization of the free energy associated with the unsaturated boundaries, bends and forms a closed envelope that mimics a biological membrane. Considering that it is possible to assemble liposomes from lipid mixtures that are present in bacterial membranes (e.g., PE:PG in 3:1 molar ratio is found in the inner E. coli membrane), these structures appear to be suitable candidates for mimicking their behavior with regard to the main aspect of our interest, which is the possibility of affecting their inner redox state by redox mediators.