Genetically Modified Microorganisms as Components in Molecular-Scale Devices

The ability to observe and manipulate systems on the molecular scale naturally leads to speculation on technologies that allow the rational design of molecular-scale machines. Whole-cells may be the ultimate functional molecular-scale machines, and our ability to manipulate the genetic mechanisms that control these functions is relatively advanced when compared to our ability to control the synthesis and direct the assembly of man-made materials. Indeed, engineered whole cells deployed in biosensor systems can be considered one of the practical successes of molecular-scale devices. However, these devices explore only a small portion of the full functionality of the cells. Individual cells or self-organized groups of cells perform extremely complex functions that include sensing, communication, navigation, cooperation, and even fabrication of synthetic nanoscopic materials. In natural systems, these capabilities are under the control of complex genetic regulatory circuits, which are only partially understood and not readily accessible for use in engineered systems. We focus on efforts to mimic the functionality of man-made information processing devices and systems within whole cells. Placing cell capabilities under the control of these addressable and programmable devices may provide the necessary bi-directional link between cells and synthetic systems that can lead to an entirely new class of sensing, information processing, and actuating devices.

Silicon semiconductor technology is the overwhelming choice for the realization of man-made information processing devices and systems. In such devices, information is represented by voltages and currents, and is communicated through the controlled transport of electrons. Here we will describe an approach, which we have coined the “silicon mimetic” approach, where engineered genetic regulatory functions of whole cells are made to emulate the functionality of these silicon semiconductor devices.

The fundamental silicon device for information processing is the transistor, which is a 3-terminal component where the current flow between two terminals is controlled by the voltage on the third terminal (Fig. 1a).

Hiratsuka and co-workers have proposed the realization of transistor-like circuits and interconnects using enzyme-catalyzed reactions and diffusion of the products. Their proposed fundamental device would be roughly analogous to a bipolar junction transistor with current flow represented by the enzymatic conversion of a substrate into a product as controlled by effectors (Fig. 1b). These investigators propose to couple multiple enzyme transistors to form a network of biochemical reactions defined by the molecular selectivity of enzyme transistors. All of the information would be coded into molecular agents and then discriminated by the selectivity of the enzyme transistor. In direct contrast to the 10 levels of lithographically-defined interconnects called for by the year 2014 for solid-state circuits, hardwire interconnects would not be needed in this scheme.

Fig. 2 shows the biochemical reactions responsible for light production in the prokaryotic lux system that we use in the bioreporters of our bioluminescent bioreporter integrated circuit (BBIC) devices. As a careful examination of the indicated portion of Fig. 2 shows, this reaction implements the enzyme transistor with myristyl aldehyde as the substrate, FMNH2 and O2 as co-effectors, and light, FMN, and myristic acid as products.

An examination of the entire reaction shows that there are actually several interconnected enzyme transistors at work. Furthermore, since the production of the enzymes that catalyze these reactions is controlled by the expression of the lux genes, genetic regulatory functions involving inducers, regulatory proteins, and promoters are part of the circuit.

Manipulation of this complexity to produce highly functional devices is the eventual goal. However, for now the approach is to simplify all of these interconnected enzyme transistors into a single device to create whole-cell biosensors. As an example, consider the lux system once again. If the concentration of the many effectors internal to the bioluminescent biochemical reaction can be maintained at a sufficiently high level, then the entire lux genetic regulatory system and biochemical reaction can be considered a single enzyme/genetic transistor. The oxidation of myristic acid and production of light serve as the analogue of the transistor current, while the concentration of the inducer is the analogue of the controlling terminal. This enzyme/genetic transistor is the whole-cell circuit device employed in the BBIC (Fig. 3).

While the transistor-like devices described above are useful for realizing biosensors, whole-cell circuits with more utility are required to realize highly functional whole-cell sensing, information processing, and actuating devices. Staying with the silicon mimetic approach, we now describe efforts to realize logic gates within the genetic machinery of whole cells.

Logic gates are devices that compute Boolean algebraic functions. The inputs and outputs are logic levels (i.e. true/false or one/zero), and the outputs are derived from the inputs through the application of a set of simple rules. For example, consider the AND, OR, and XOR gates shown in Fig. 4.

The output of an AND gate is true only if both of its inputs are true. Likewise, the output of an OR gate is true if either of its inputs are true, while an XOR gate has a true output if one, but not both, of its inputs are true. Although each of these gates provides very modest computational power, as demonstrated by silicon integrated circuit technology, the interconnection of these devices can lead to exceptional functionality.

Our recent work has centered on constructing gene transcription modules to create biochemical devices that can be combined to create logic circuits. The transcriptional unit consists of a promoter that is inducible or controllable and a gene. The gene product may be a regulator or an enzyme. The inputs are molecular signals that control gene expression, transcription, or translation, while the output of the logic gate may be the enzyme itself or the activity of the enzyme (e.g. bioluminescence). At present we are working on the realization of AND, OR, and XOR gates. Here we present our results with the OR gate (Fig. 5).

There are at least two strategies for implementing the OR gate: 1) use two promoters that have identical gene transcriptional effects but respond to two different inducers; or 2) use a single promoter that responds in a similar manner to two different inducers. We implemented the latter strategy using a tod-lux fusion in P. putida TVA8. This operon is similarly induced by both trichloroethylene (TCE) and toluene. The result is the production of bioluminescence if either inducer is present.

While combinations of the AND, OR and XOR functions described above can implement any combinatorial logic function, these components cannot be used to implement sequential circuits that require memory of past logic states and clock signals for synchronization. However, Gardner, Cantor, and Collins have demonstrated a two-state genetic latch in E. coli that implements a one-bit memory using two repressible promoters arranged in a mutually inhibitory network. A significant challenge still remains to develop genetic circuits that make more efficient use of the memory capacity of the cell’s DNA.

Above we have described genetic circuits that rely on molecular input/output (I/O). However, for whole-cell systems that may interface with physical systems, I/O that relies on physical mechanisms would be advantageous. In particular, gene expression control with current or voltage would be ideal for hybrid whole-cell/microelectronic devices. A recent review summarizes the theoretical considerations and physiological effects of EMF and electric current pulses in living cells, but little data is available on such effects at the level of gene expression (Velizarov, S. Electro and Magnetobiology 18, 1999, 185). With the objective of identifying promoters for use in genetically engineered electrically controllable biochemical devices, we performed a preliminary search for current-inducible promoters in E. coli. In this experiment, all expressed genes in E. coli were screened for putative current inducible genes. E. coli cells were grown to log phase, placed in 14.6 mm dialysis bags and subjected to a current of 36 mA for a period of 30 minutes. An identical volume of cells was collected into a dialysis bag and placed in buffer without electric current exposure. Simultaneously, cells were diluted and plated on nutrient agar plates to ascertain if the current had any lethal effects on cells. After current exposure, the induced and untreated cells were lysed and the RNA component isolated and quantified. Both RNA preparations were reverse transcribed incorporating 33P-dATP to make the labeled cDNA which was hybridized against two identical commercial E. coli gene arrays (Genosys Panorama, The Woodlands, TX). These arrays permit the gene expression of all 4290 genes within the E. coli genome to be quantitatively assayed simultaneously. None of the cells in the three treatment groups showed signs of lethality compared with controls. Of 1521 genes initially surveyed, 50 genes were repressed over 10 fold, 40 genes were induced over 10 fold, and 5 genes were induced over 20 fold (Fig. 6).

We cannot say at this point that these genes are specifically current inducible and not, for example, affected by osmotic or oxidative reductive stress. From a pragmatic point of view, establishing specificity may not be necessary for the implementation of current inducible promoters. From this perspective it is only necessary that the physiology of the cell is not unduly compromised during current induction and the linked pathways function as designed.

The potential for whole cell biocomputing is represented by a remarkable coalescence of interdisciplinary scientific advances. Fundamental to these advances is rapidly growing insight into sensing of the cell’s immediate environment and the signal transduction cascades interconnecting both enzyme activity and gene expression. It is now not uncommon to conceptualize these interactions in the form of information or electronic logic. An excellent example is the description of the WASP protein interaction with guanosine triphosphase (Cdc42) and phosphotidylinositol 4,5 bisphosphase as a functional “AND” gate. The sequence of events associated with signal transduction can also be visualized at the level of gene expression through the power of whole genome sequence analysis and gene expression array technology. Predictive induction and repression of coordinated gene networks can be tested, as well as the elucidation of unknown interacting genes, to provide information essential to developing high fidelity biocomputing capacity. Such tools can also provide experimental evidence for mechanistic control of gene expression that may be developed using approaches, such as alteration of membrane fluidity, photoactive gene expression, or as previously discussed, electronic control of gene expression. In the broad applications of reporter gene technology, which are critical to gene expression linked biocomputing, bioluminescent reporters continue to show exceptional promise. Genome wide bioluminescent gene expression profiling has been demonstrated, and such demonstration suggests an analytical path for utilizing optical sensing of interconnectivity and large scale information processing capabilities of living cells.