APPLICATION

The goal of the project has always been clear to us: to combine the benefits of a naturally evolved organism with the benefits of human-made technology. This focus would have made a lot less sense a few years ago. Today, however, the technology is so advanced that it can sometimes even compete with the efficiency of bacterial systems. But why competition, when cooperation is possible? On the way to realizing this, we encountered a major problem: to establish a functioning and stable communication pathway between organisms and a technical application. Eventually, we managed not only to overcome this problem in theory, but also in the laboratory. To prove this, we developed , an application that uses this communication successfully. This is the most critical and important step that opens the door to a wide range of possibilities.

The Beginning

The idea of connecting biology and technology is obviously not new. Whether huge neural networks as competitors for computers analogous to the human brain, the detection of electromagnetic pulses in the brain and processing these signals, controlling a prosthesis by thought, or even to recognize the motion information in muscles directly and processing it by means of their electrochemical potentials—there are plenty of examples around. We however try to approach this idea from a different viewpoint: with microorganisms. We firmly believe that this combination is already possible in a meaningful way and will contribute significantly to the progress of all life and the increasing quality of life on this planet.

Fig. 1: A man controlling his computer with his mind.

Bacteria as Sensors

The advantages of living organisms are their variability and efficiency in terms of energy as well as the space required for their work. Both in terms of substances to be measured, as well as the measuring device, no electronic sensor currently used can even approach the performance of bacteria for detecting chemical molecules. No electronic device can be compared with the multi-stage variable design, thanks to different reading frames and the effectively used space for information storage such as genes in chromosomes. A physical measuring device is constructed invariably for a specific purpose and can hardly be designed modularly without greater effort and loss of efficiency to allow for later, unforeseen changes of function. In bacteria, the construction plan itself is directly implemented and the measuring devices are constantly produced, used, degraded and recycled. Theoretically, each substrate can be detected at minimal concentrations (of course depending on the ambient volume and cell density) and the resulting signal amplified.

The demand for such a device is high, since there is a wide range of possible applications. For example, for detecting explosives, currently, dogs must be trained expensively for each "new" olfactory compound of an explosive, while bacteria only need another cloning. Quantitative determinations of air quality in confined spaces or open cities can be achieved by recursive control of bioreactors. Even the detection of cancer cells, which today is only possible under harmful conditions, could be accomplished with a microorganism-based sensor (artificial nose).

Theoretically, the processivity can also be calculated as a counterpart to the substrate concentration on the basis of the cell density and determined for different states, thereby enabling the determination of concentrations in addition to the mere binary yes / no detection of the substance. Such conditions may include, but are not limited to, the temperature dependence we have already shown, different pH levels, the presence and absence of light in certain wavelengths, the presence and absence of various substrate molecules, and much more.

Bacteria as Processors

If the interface between bacteria and technical devices is extended not just with one gene, but many, we open the door to almost unlimited possibilities. The two interfaces we studied, temperature and pH of the environment, can be almost arbitrarily combined with others (for example, with photoreception of different wavelengths, voltage differences, or even the presence, absence or change in growth factors such as oxygen or other specific molecules). Taken together, the temperature in our experiments offers two possible outcomes: with a heat-inducible promoter, a signal can be produced and detected at elevated temperatures, or not.

Considering heat as 1 and ambient temperature as 0, we get the classical information units of the binary system. The change in pH in turn provides three detection options. In our case, at a pH ≤ 6, a specific promoter induces the production of a GFP mutant, while at neutral pH (around 7.5) no signal is generated and at pH ≥ 8, another fluorescent molecule is expressed. Accordingly, we have three signals: 0, 1 and -1. If we now combine the two differently caused signals, i.e., hot or ambient temperature and pH ≤ 6, pH around 7.5 or pH ≥ 8 we get the state space (1|1, 1|0, 1|-1, 0|1, 0|0, 0|-1), providing us with six different signal combinations. If we added a photoreceptor (detecting the presence or absence of a single wavelength), the number of states doubles to twelve (2 states for temperature times 3 states for pH times 2 states for light), suggesting that our robot could already be fed with twelve different movement patterns.

Whether these twelve different fluorescence signals are detected separately (single bacterial samples are exposed to a single stimulus) or in a shared environment (bioreactor) by means of either multiple detectors (each measuring a single frequency) or a combined detection of superposed wavelengths has to be decided on a case to case basis, since the complexity of simultaneous detection of multiple fluorescence signals increases exponentially.

The idea of processor-oriented computation stands in stark contrast to a strongly parallelized alternative: in just one milliliter of bacterial culture, with adequate bacterial density, more than 3 billion bacteria can be found. If we can get every single bacterial cell to make one calculation per second, we have a computational power similar to modern computers in a milliliter. There would also be no problem with scalability: Within a few hours, one liter of bacterial culture can be produced from one milliliter, which increases the computing power several fold.

Production and Distribution

Once the construct has been designed and, with some effort, produced by standardized work steps in the laboratory, it can be replicated almost limitlessly with very little effort and materials. Real production costs are no longer incurred, except for the power supply of the incubation chamber, the ingredients of the media and basic materials. The production of electronic processors requires significantly rarer and more expensive materials, mostly silicon, and a production facility that is immensely geared to the individual processor type. If a company has discontinued the production of a series of processors and wants to resume, the entire production chain must be re-established with considerable effort. A bacterial plasmid can be easily isolated, frozen and thus stably stored for a long time and, once needed, can be replicated within 16 hours with simple standard protocols, no matter what organisms or strains have been worked with in that facility.

Bacteria in Combination with Artificial Intelligence

The progress of artificial intelligence, ever more obscure for the developers themselves, could lead us much further into the pitch-black rabbit hole, when combined with bacteria. But right here is an incredible opportunity. When we teach the artificial intelligences to cooperatively think with living organisms and even like living organisms, we create a link and the discrepancy between living entity and "dead" machine vanishes.

The concept of a bacterial symbiosis, living in harmony with its host through reciprocal supply of essential nutrients, which we as humans have already successfully established with multiple bacterial species in our intestines (and all over our bodies), could be beneficial in convincing critics that the robots controlled by artificial intelligence act ethically and stay controllable. The symbiosis could look like this: The robot provides the environment (habitat, nutrients, ...) for a bacterial culture that in return provides the robot with cues about its own well-being and the external environment. In addition, the bacteria could also contribute to the robot's energy needs. The more intimate this symbiosis evolves, this energy contribution could become essential.

One example of how energy could be generated by microorganisms is obviously photosynthesis on a surface layer. On the other hand, if e.g., combustion engines are used to drive such machines, organisms could neutralize exhaust gases in order to improve the environment (by reducing greenhouse emissions). Filtering them for valuable compounds could provide another way to harvest energy (e.g., via naturally evolved or genetically engineered methane oxidizers). If there is no internal combustion engine, CO₂, CH₄ from the surrounding air could be used as an energy source. It is difficult to predict, where these speculations could lead us, but crossing the limits of our imagination is key to see what lies beyond.

Fig. 2: Bacteria in their natural habitat, hopefully they will be in a robot soon.

Summary

But how should such a communication interface look like and be implemented? With our experiment, the foundation has been laid. Our work can be diversified almost infinitely in the future. Similarly, we are far from knowing and understanding all the mechanisms of nature. Nearly 99% of the microorganisms on this planet are still unknown. These certainly harbor one or the other as of yet unthinkable way to generate or detect signals. The future looks bright for symbiotic life!