Biosensing of disease biomarkers has been acknowledged as a key technology for efficient monitoring and treatment strategies, possibly even disease eradication. Existing systems are usually pre-determined, single-use and incapable of introducing a range of sensibilities that can be harnessed by employing bioengineered organisms or systems, such as toehold switches or protein receptors with high affinity to the ligand of interest.

While our microbial fuel-cell (MFC) is a proof of concept for cancer biosensing, our goal was to finally reach a device that can be widely-used at the level of an end user, just like commercial rapid tests at a pharmacy.
Such a device would have a replaceable sensing module, depending on the needs of the user.

Based on all our human practice work, however, we arrived at the conclusion that this end-user cannot be a regular patient, at least not in France.
Indeed, according to the French law, only medical analysis laboratories and a practitioner who has received a permission from the general director of the biomedecine agency can process a genetic test to detect a genetic disease such as cancer. So, when fully developed, our test would be used in those kind of laboratories and not in pharmacies or by the patient themselves as an auto-test.

Moreover, as our test makes use of a genetically engineered organism, considering current biohazard regulations, end-user implementation of bioengineered systems seems unachievable at a general public level. This must not come as a roadblock, since our goal is to finally reach cell-free testing. Before the cell-free MFC system is being developed, our system requires Biosafety Level 1 labs or scientific field-stations, where necessary biohazard measures are readily available.
The simplest use would be setting up the MFC efficiently, connecting it to the network and registering information that can be immediately assessed, streamed and studied live even at a remote location. All these by implementing hardware-based safety features, such as leakage sensing and online inlet control, along with on-board alarm systems which will set-off if MFC is ruptured.

Out of these safety systems, our current device already contains leakage sensing and the hardware for upgrading inlet/ outlet control. Safety is also maximized by using non-toxic, compostable PLA material for the MFC, as well as by introducing non-toxic bacteria.
Although, in the final system that we envisaged, miniaturized microfluidics-based cell-free methods should be implemented. Miniaturizing the system will give us the capacity to speed up sensing as suggested in various papers [1] and developing multiple microfluidic pools will facilitate sensing of multiple targets at the same time.
At this point, if sensing modules are made of non-toxic material, they could be disposable in a highly sustainable manner, and detoxification of the sample is also guaranteed. In this regard, introducing high-current generating hardware can become more advantageous, because it can be used for multiple purposes. Specifically, it can be used for permeabilizing cells for the incoming RNA sample and as a neutralizing feature, when used for longer periods.

If our vision is fully materialized, it will lead to a versatile, modular, sustainable internet-of-things sensing toolkit, built on an equity-mindset such that mass-producing it and delivering the luxury of marker-biosensing to the larger public may become possible in sustainable ways. At a versatility of this level, society could readily adapt new marker capacity on the go, without waiting for entirely new device manufacturing. Because this time instead of re-engineering the hardware, one would need to replace the toehold sensing system and deliver it to the market.

Having a versatile device available in a next-door medical analysis laboratory, could bring a real improvement to early-stage cancer detection. Indeed, according to the testimony we received from a cancer survivor, today many cancer detection methods rely on results interpretation of the doctor thanks to the different medical imaging technologies (radiology, mammography).

To achieve this final goal, further development and analysis are required, both at the biological and at computational levels.

At a biological level, taking into account our human practices work, we have 4 different choices to implement our test.

• Continue our studies with the MtrCAB and the toehold switch controlling the expression of CymA in an engineered E. coli, which has a drawback of being time consuming according to researchers we interviewed about the insertion of Shewanella’s MTR nano-conduct complex between the periplasm and the outer-membrane in an engineered inE. coli. Indeed, even if the Mtr complex remains very unspecific for the different bacteria, the proper incorporation of the MtrB subunit in E. coli’s outer-membrane is really challenging. However, this system would give us a better signal compared to the next solution imagined.
• Incorporate only the toehold switch controlling the expression of CymA in an engineered E. coli, but not together with the MtrCAB operon. Indeed, according to the researchers we talked to, as we just want to get an ON/OFF switch with an electric output, we would not need the Mtr complex to get a proper signal. The toehold unfolding itself, by the only expression of the Cytochrome A (CymA) nanoconduction may provide a sufficient electric output to distinguish both the OFF and ON states of the switch. This is probably the most convenient and rapid way to get proper experimental results, however the electric output might be too low to be detected in the presence of the lncRNA PANTR1. So, false negative output might be a problem to make the difference between samples from sick and healthy people.
• Replace the E. coli chassis by an engineered Shewanella oneidensis MR-1 strain in which the wild-type cymA gene would be knocked-down and replaced by this same gene under the control of the PANTR1 toehold switch. S. oneidensis is a natural electroactive organism and has a way higher external electron transfer capacity than any engineered E. coli. However, engineering S. oneidensis is not as straightforward as E. coli and the amount of electricity generated with Shewanella may be quite variable.
• We consider using a cell-free system that is known to be more efficient and convenient for RNA detection through a toehold switch structure. However, this is possible if, and only if, the gene expressed in the presence of the lncRNA is an easily monitored marker/ protein (like GFP or luciferase that are commonly used due to their fluorescent or luminescent properties, respectively). However, our project relies on the opposite principles because we need a structure to allow nano-conduct insertion displayin a cell and then to regulate the external electron transfer through a structural protein (as CymA) expressed just in case of the lncRNA PANTR1 detection.

At computational levels, multiple improvement objectives can be summarized:

• Develop de novo board for current sensing, which will be optimized for lower current registering.
• Minimize the MFC and re-design it for SLA-3D printing for improving detail of leakage-susceptible regions.
• Develop and train machine learning algorithms, designed to detect the same biomarker even when MFC-related signal relativity is observed.
• Streamline hardware for meeting highest safety standards, to make its public-use relevant.

Finally, we will have to link the embedded Arduino chip system (or the Raspberry nanocomputer) to the cloud server and then to the real-time measurement application on a smartphone or tablet.
On the cloud server we would require a database called “relational” as the data to collect are linked to each other. Moreover, for data treatment, as we are working with quantitative values, we will have to use an SQL database and work with a server software ASP.NET. Further, we will then need display technologies to get data on smartphones. The mobile client allows us to display all the data received from the cloud. By launching the application, it will harvest all the data from the cloud and update them in real-time. This data can either be harvested by the mobile client itself or by the cloud (it requires an internet connection). Today, technologies to develop an application are variable depending on the support we want to put it on: Microsoft Xamarin technology for both Android and IOS applications, Google Flutter technology for UX-UI with a strong Android application degree, React native technology allows to rapidly transform website developed with React into a smartphone application, Apple Swift technology for IOS applications , Kotlin for Android applications, and more may be available in the near future.

References

[1] Cui Y, Lai B, Tang X. Microbial Fuel Cell-Based Biosensors. Biosensors (2019) 9: E92.