By Monica McNerney and Mark Styczynski  

Biosensors are a “big business” for bioengineers: they have been developed for all kinds of applications, from dynamically controllable microbial cell factories to enabling high-throughput screening. For a while now, our group has been convinced that bacterial biosensors are a good candidate for low-cost, point-of-care diagnostics since they are cheap and (comparatively) easy to engineer, store, and grow.  However, since the lab equipment – freezers, shaking incubators, plate readers, etc. – utilized by most bioengineers to develop, test, store, and ultimately use their biosensors is not available away from the wet bench, there is not tons of precedent or best practices for people trying to actually deploy them outside of lab settings.  And if you do want to deploy them outside of lab settings, there is quite a bit of work involved to make them robust, reliable, and portable.  

In our recent Nature Communications paper, we demonstrate the potential to further bridge the gap between sensors that work in the lab and ones that could actually be used in the field. We made a field-deployable biosensor for zinc deficiency, which causes over 100,000 childhood deaths each year and is most prevalent in remote, low-income regions that have no well-equipped laboratories or clinics, and in some places not even electricity.  A low-cost, minimal-equipment test for zinc could help improve assessment and treatment of this problem.  

In developing this zinc assay, we first spent lots of time cloning, evaluating, and tuning constructs, followed by sensor redesign, more cloning, and so on, in the seemingly-endless dance of the synthetic biology “DBTL” cycle.  We ultimately created a sensor that worked the way we wanted it to: cells produced one of three different visible colors based on whether serum added to the test had low, borderline, or healthy zinc levels. We improved the assay time in the lab by an order of magnitude, and moved one of the response thresholds an order of magnitude, making it clinically relevant. We took some steps towards addressing issues that would be important for field deployability – some successfully, and others less so – with a number of results supporting the potential of these sensors as the basis for a diagnostic, and with a vague vision (Figure 1) of how this could work in the field without electricity or running water.  And we were proud of all of this. But the reviewers, they pushed us harder; they said that all of our molecular biology was well and good and interesting, but how close could we actually get this to real field deployability? This was potentially overwhelming. With one of us trying to graduate soon, we were tempted to look for a different place to publish our work as it was, but we ultimately decided to dig in, because the only way we might ever get our sensors into the field helping people was to tackle these obstacles.   

Figure 1: Our initial vision for test manufacture and minimal-equipment interpretation, with uncertainties in implementation indicated.  

First, we went back to the things that had previously failed. For example, years prior we had tried to freeze-dry a previous version of our sensor cells, but found issues with viability upon rehydration that we had chosen not to pursue.  This time, when we freeze-dried more concentrated cells in a different protecting media, they rehydrated quite well!  So, that was a first step towards making the sensors field-deployable, in this case by eliminating the need for -80°C storage of cells by enabling them to be stored at room temperature.    

But there was more to be done: we needed to show that we could replace standard lab equipment with devices available to heath workers, such as fixed volume pipettors, hand-powered centrifuges, and a smart phone.  Some of our efforts worked better than others. For example, we tried to use a recently-published approach to centrifuging mL volumes of liquids without any electricity, but found that the induced gravitational force from this approach was insufficient to separate red blood cells from serum. The principle of electricity-free separation remains valid, though; it just was a problem we felt someone else would be better suited to solving. Instead, we focused on eliminating the need for a shaking incubator by taking advantage of a 37°C incubator available to any person: the human body. We added serum to lyophilized cells, taped the tubes to one of our bodies, and proceeded with a normal work day.  Excitingly, this body incubator approach worked: the cells produced different colors based on serum zinc concentrations. However, we encountered a bit of a reproducibility problem: the degree and intensity of color change varied based on whether the user was moving around lab all day or spent the day reading on the lab futon.  To overcome this variation in effective incubator rpm, we ran tests with known zinc concentrations in parallel with tests with unknown zinc so that the user could compare the color of the test to standards that were run with the exact same amount of motion. This used a bit more serum for each test, but provided much more reliable results.  

We also realized that we wanted to minimize interpretation challenges associated with color readouts. Grad students who spend years differentiating between bacterial colors become exceptionally good at classifying varying degrees of purple, red, and orange, but to the untrained eye, intermediate colors can be hard to interpret. We had focused on a purely color-based test output because we had consistently heard (in the seven years we’ve been working on aspects of this project) that equipment-free means equipment-free (no machines or electronics allowed).  But based on some recent conversations with people in the field, we found out that things had changed as technology has become more pervasive and robust: aid workers are now doing nutritional surveys on tablets or smartphones. And this opened up a great opportunity for us. We teamed up with tech-savvy high schoolers to build a smartphone app that automatically processes the colors of our sensor cells, removing the human element from interpretation.  The user needs only take a picture of the cells, and the app reports the level of zinc in the serum sample, accurately classifying serum zinc as “low”, “borderline”, or “healthy”. 

Figure 2: Our refined, more proven vision for test manufacture and minimal-equipment interpretation. We have now demonstrated almost the whole end-to-end process, though more refinement on the electricity-free centrifuge is needed.

Together, the steps we took can enable a health worker equipped with just a small pack of components to assess zinc concentration in remote, minimally-equipped regions, which could dramatically improve the capability of public health and nutrition programs (Figure 2).  We’re also excited that this helps prove the potential scope of field-deployable diagnostics. Portability is a huge benefit of recently developed “cell-free” sensors, but whole cells are still better at sensitively detecting select compounds, such as those that have membrane receptors or those that are actively transported into cells.  We hope that our approach for making clinically relevant, field deployable sensors can help to bring a new class of bacterial diagnostics into the field, where they enable point-of-care assessment and improved public health programs.

Related reading: 

McNerney, M.P., Michel, C.L., Kishore, K., Standeven, J., Styczynski, M.P. Dynamic and tunable metabolite control for robust minimal-equipment assessment of serum zinc. Nat Commun 10, 5514 (2019). doi:10.1038/s41467-019-13454-1