It’s hard to pull off a biochemical test without controlling the flow of the liquids that comprise it. Then again, it takes a lot of work to pull one off with that control, as well. Working manually, a lab technician might have to pipette their sample, wait ten minutes, apply a rinsing buffer, wait another five minutes, wash the sample, apply something else, wait again, and so on and so forth. “There’s a protocol,” summarises Yuksel Temiz, a research staff member at IBM Research, and it changes for almost every test. The promise of using microfluidics for clinical chemistry is that it can make the whole process easier and more accessible at the point of care. So far, however, that hasn’t quite happened. “When you try to mimic or implement these protocols in microfluidics,” continues Temiz, “often you have to do it using external pumps connected to the chip with tubing. This is very standard in microfluidics, but if you talk about point of care diagnostics and ease of use, you cannot ask an untrained person to connect the tube, to connect the pump – it requires some sort of training.”
Stéphanie Descroix, a microfluidics researcher at France’s Institut Curie, agrees. “What we develop in the lab might be very interesting for us,” she says, “but it’s mainly for us. When we try to transfer the devices to biologists, then things start to change because they’re not used to all the engineering aspects.” Both realise that, for microfluidic devices to make an impact outside of physics or chemistry, developers need to make things as easy and intuitive as possible for people with very different types of expertise. “It’s not an easy task,” says Descroix. What’s more, point of care diagnostics usually need to be disposable in order to avoid patient-to- patient contamination. Microfluidic systems that rely on physical connections to a lot of off-chip technology create issues about where to draw the line between what needs to be thrown away and what can be reused, as well as requiring expert attention for every new test. “When you have a complicated system with tubing etcetera, the microfluidic cartridge needs to be disposable and the rest needs to be reusable,” notes Temiz, “but that also creates a challenge of how to connect, how to interface, and how to introduce liquids and samples.”
Past the state of the art
While microfluidic diagnostics have struggled to make the impact many forecast, lateral flow assays (LFAs) have confirmed themselves as the state of the art at the point of care, particularly in low-resource areas. Rather than using external pumps, mechanical valves and specialised manufacturing processes, these paper and plastic-based tools exploit simple capillary forces to qualitatively test liquid samples for the presence of a target substance. “I think it’s a brilliant technology,” says Temiz. “It’s cheap, easy to use and it does the job most of the time, but not every condition requires yes and no types of answer. It’s okay for a pregnancy test, for example, but there are also many health conditions that require more quantification and more precision than lateral flow tests can offer.”
Paper-based microfluidics exploit the same principle of capillary action, but without more sophisticated ways of controlling the flow, their superior sensitivity has to be balanced against the difficulty of manufacturing them. “The challenge,” as Temiz puts it, “isn’t whether we can create a flow, but how can we stop it?” When it comes time to hold the liquid for five minutes for a biochemical reaction, most capillary-driven microfluidics become useless. “We have no external control to do that,” says Temiz. ”It’s all autonomous.”
Descroix’s team, which regularly builds microfluidic devices for organ-on-a-chip applications, encountered similar issues when trying to compartmentalise different hydrogel solutions side by side in order to exchange cells between them. “To do that, people use capillary forces, which means they microfabricate lines of pillars inside the chips that will allow the fluid to stay in a given position,” she explains. “It works super well and it has been designed by very smart people, but in a real extracellular matrix, in real tissue, there’s nothing like those sorts of obstacles.” So, Temiz and Descroix were both part of teams that had clearly identified an issue with flow control in microfluidics. IBM Research wanted to develop something “generic, modular and flexible”, while Institut Curie’s biologists were crying out for a technology that would allow them to “compartmentalise fluids in a controlled and reconfigurable manner”. From there, the two groups took quite different paths.
IBM focused its attention on capillary action. Temiz joined in 2012, by which time the group had already published some papers on external mechanical tools for holding fluid in place. Since then, a number of other groups have gone further in that direction, developing traditional mechanical valves small enough to be installed on microfluidic chips, but IBM found inspiration for its new e-gates in a different type of flow: that of traffic.
“We don’t have any mechanical parts,” Temiz explains. “There’s simple physics behind it.” Indeed, capillary action brings the liquid to nothing more difficult to fabricate than a trench cut into the chip, which it is unable to cross. “So, up to now, there’s no energy needed: the liquid comes autonomously and stops at the barrier,” says Temiz, taking up the thread. “Then we apply a very small voltage for less than one second and the liquid passes over the barrier. It’s kind of a traffic light stopping and starting the flow.” These barriers, or ‘e-gates’, can be positioned anywhere on a chip to manipulate the flows and interactions of multiple different liquids by deciding when and for how long they stop and reactivate. The result is a complex microfluidic network – what IBM calls, in an analogy to controlling electrical current with transistors, a programmable liquid circuit.
What’s more, all of this is controlled via smartphone, which makes it far more intuitive and accessible for non-expert users than a dedicated lab instrument could ever be. The chip is inserted into a small peripheral, which supplies the electricity needed for the e-gates and communicates with smartphones via Bluetooth, continually updating the user about how the chip is functioning as well as allowing them to configure specific protocols.
That’s made possible by the addition of secondary electrodes that measure the presence of the liquid at different locations on the chip. By continuously monitoring the flow, the secondary architecture can give feedback to the user about whether or not the test fluids successfully traverse the programmable circuit and at what rate they are doing so. As such, it can easily indicate whether even the most complex operations happened successfully, or if some defect in the chip means the test needs to be repeated to give reliable results.
If walls could talk
Descroix’s team at Institut Curie turned to something a little less technologically advanced to compartmentalise their microfluidic chips. “We started with a fishing line,” she laughs – “exactly the same one that my father uses when he goes out to fish.” The system worked well, but the hemispherical interface created imaging issues, and the team realised that they could achieve the same functionality by microfabricating flat walls that could be slid back and forth within the chip to open and close different compartments.
“The beauty of the walls is that you can do much more than you can with a tube,” says Descroix. “You can not only compartmentalise but you can have a valving system, you can have a pumping system – you can even make windows in the walls and use them to carry some solutions or some molecules between different compartments.” That’s a particularly intriguing possibility for POC bioassays where windows made with different membranes could be used to extract and concentrate molecules from a sample before being slid into position for a particular test.
“This is completely new in terms of function,” says Descroix. “You have different channels in your system: in the first one you put your sample, you extract the molecules of interest and preconcentrate them in this membrane in the window; then you just have to slide the window and put the solution in another buffer channel. Here we’ve run electrophoresis, but you can run an immunoassay, or you can run whatever – and you can do it many times if you need to.”
On top of that, walls can be microfabricated with channels for pumping and valving, and made from a range of different materials. “You can really play with the physics of the system,” continues Descroix. “And you can adapt your mechanical production depending on the application and the question that you would like to address.”
Crash-test tests
Within Institut Curie, it’s quite simple to iterate according to the needs of the biologists across the hall. So far, they’ve 3D printed and photopolymerised plastic walls, and micro-milled stainless steel ones for different purposes. “We use 3D printing mainly, and if you change the design of the chip you can change the design of the walls in just a few minutes, so it’s very easy,” says Descroix. Collaborators can be provided with complete chips with preinstalled walls, or a set of walls that they can integrate into their own microfluidic systems. On the other side, IBM Research has developed the most generic technology possible so that industrial partners can adapt it to their own needs. Temiz refers to his team’s papers on their reconfigurable chips as proofs of concept, and stresses that so far, they’ve only worked with silicon and that they haven’t done any clinical studies. Nonetheless, the team has shown that the technology works for blood plasma, artificial urine and, with a few modifications, blood itself. “In principle, it’s a very generic technology,” he points out. “One can have a very generic chip architecture, put the samples and reagents in, and control the whole flow protocol from a smartphone app. That’s it.”
As Descroix well knows, however, the next stage is where the real work begins. “For us, when we start to transfer the technology to biologists it’s like a crash test,” she laughs. “You have very fast and very interesting feedback, and you learn very quickly whether it can be useful or if you have to work a bit more on your system.”
