10.2021-Sense.nano-Paul-Blainey

PAUL BLAINEY: It's good to be with you today. I'm Paul Blainey, I'm an associate professor of biological engineering at MIT and my lab is located at the Broad Institute. I'm going to talk about scalable combinatorial screening and a microfluidic platform my group has developed to enable this.

I want to start with just a few disclosures before I get into the talk so that everyone can see those. And I'll start by noting how complex biology is. And it's a real triumph of modern biological research that for so many biological systems we can now identify most of the parts that comprise. However, we don't yet understand what all those parts do or how they work together in order to support the function of a cell or an organism made up of those cells.

And that extends to the influence of environmental factors and drug treatments. Now, we have some really powerful biotechnologies at our fingertips already. But these are typically at their strongest when we're testing one experimental variable at a time, not the large sets of variables that are needed to untangle all these interactions I was just mentioning. And so we really do need new experimental approaches that enable us to manipulate many experimental variables at once.

And so that's the purpose of the technology platform based on droplet microfluidics that my group has developed and has been applying to a number of different projects in recent years. Now, I want to start by introducing microfluidics. Microfluidics is simply the manipulation of fluids at very small scales. So, how small? Well, about a billion or a trillion times smaller than the everyday scale that we're familiar with at home.

So your water bottle, if you have one, that's going to be roughly a liter in volume. And that's the kind of volume scale that we're used to working with in our kitchens at home. If you fill up a pot of water to boil some pasta, you're going to be talking about a couple liters of water. So microfluidics is a billion or trillion times smaller than that. Now, if we go in the other direction thinking about fluid scales larger than that every day scale, if we go about the same distance in the other direction, we get up to the scale of oceans, which is a mind bogglingly large volume scale. We're talking about cubic kilometers in oceans.

And so that is to say that the microfluidic scale is much smaller than the every day as oceans are larger than our everyday experience with liquids. Now, microfluidics also enter that common daily experience. So for example, an inkjet printer is a pretty sophisticated microfluidic system that can eject nano liter smaller droplets of ink to form a printout. We have capillary flows and materials like textiles are another example of nanoscopic microfluidic flows in our experience.

And of course as human beings we, ourselves, are microfluidic systems. And our cardiovascular circulation takes place certainly at a microfluidic scale. Now, I'll introduce your three different classes of microfluidic systems. The first on the left, lab on chip microfluidics. And so these are highly engineered microfluidic devices which are an analog of micro electronic devices where chambers and channels and active elements like pumps and valves are microfabricated to form a lab on chip system that can replicate quite complex laboratory processes manipulating molecules and cells.

In the middle I'm showing [AUDIO OUT] microfluidic droplets, a distinctive class of microfluidics where small aqueous droplets are created in a hydrophobic continuous phase. And it's these droplets that serve as the containers for biochemical processes. One of the distinguishing features of droplet microfluidics is scale. We can create thousands, even millions, of these microfluidic droplets to carry out a lot of processes in parallel.

And finally, I want to mention hydrogel microfluidics at the top on the right here. This is an emerging class of microfluidics that's conceptually different. So the idea here is to use an organic hydrogel to constrain reaction processes so that we can separate many different reactions in a single phase system. So back to the complexity of biology for a moment. The more we come to understand about how these complex biological networks of molecules and cells work, the stronger our hypotheses for how to intervene chemically in those to achieve desired therapeutic outcomes.

But there's a challenge with discovering combinatorial drug treatments. Chemical space is big to begin with. And when we start talking about combinations of chemicals drawn from those spaces, this really becomes an astronomically large and intractable screening problem. So like I mentioned, the droplet microfluidics is about the best we can do for scale. And so is there room to apply that technology to make a dent in this combinatorial chemical screening problem.

Now there's one major challenge to overcome on our way to achieving that. And that's what you see here, which is that drug molecules leak out of one droplet and go into another. And so this basically makes it impossible to maintain a chemical library in droplet form, since we would require having a different compound stably associated with each droplet. And so you see the problem here as over the course of this movie the concentration of a dye standing in for a drug molecule equilibrates across each set of seven droplets.

But you also see the solution here, as well. Now this group of seven droplets in the middle, which is contained in a sealed micro-well patterned in the substrate, is not picking up dye from the adjacent micro-wells. And so what you're seeing here is now chemical containment not at the droplet level but at the micro-well level. And so we leveraged that to carry out our combinatorial drug screening assays micro-well by micro-well.

Now, another requirement for combinatorial screening is that we need to manage the compound consumption. And that's where the miniaturized nature of droplet microfluidics is really handy. In combinatorial screening, each compound is tested not just by itself but together with many others. And so we need to run a lot of different assays with each compound. And those assays need to be microscale in order for the total compound consumption to be manageable. And finally, we really like this microfluidic array approach because the droplets organize themselves into the micro-wells in order to formulate the combinations we want to test. And so there's no need to program a robot to pick out exactly the combinations that we want to test in each assay.

So, this movie summarizes the workflow that we've applied for common trial drug screening in the droplet micro-well system. Each droplet gets a color code identifying its contents. Those are loaded into the micro-well array, as you saw. And we can visualize these under the microscope to read the color codes and re-identify each compound. Finally, we merge each set of droplets to formulate the combinatorial assay and then read out cell growth or other cell functional endpoint in order to assess the impact of each set of compounds on the cells.

And so this is not only the most efficient combinatorial compound screening system available today, it's also one that's potentially applicable at the point of care for personalized drug screening. Because it's miniaturized and it runs pretty quickly and does not require a large number of cells, either, it's potentially suitable for taking cells from an individual patient, running a small drug screen in order to determine which drug therapy is optimal for that individual patient.

Now, we've pushed this technology into other areas, as well. Combinations are important not just for drugs but also for microbes. So in another study, a group of MIT students isolated microbes from soil on campus and elsewhere and then tested the pairwise interactions of 20 of those isolates across a broad range of culture conditions. So this large scale data collection, together with a new continuous quantitative framework for characterizing the interactions between microbes, gave us a bird's eye view of these microbial interactions that was more comprehensive than anyone had carried out before.

And so we learned a lot about how these microbes interact. And one of the takeaway lessons was that positive interactions between microbes-- specifically these were one microbe is benefiting in terms of its growth from the presence of another-- are much more common than has really been recognized before. Now, in our next study we up the ante in terms of these microbial interactions, studying not interactions between a simple pair of microbes, but now looking at interactions among a larger group of microbes and how those contribute to community level functions.

So for this work we built arrays that contained larger micro-wells able to accommodate a bigger set of droplets. And here we set up a toy problem where the community level function we were interested in was the support of the growth of a particular microbe. In this case, herbis beryllium, a model plant symbiont. So this is the kind of thing you might want to spray on seeds before you plant them in an agricultural application. And so indeed, we did find microbial community compositions that supported the growth of herbis beryllium robustly across a variety of different chemical environments.

So that was what we were looking for, was nice to find it. And even better, we learned a few things about how to build such a community that can support a desired function. Now, the team was using these large arrays. But a postdoc in the group, Sherry Ackerman, decided these arrays weren't even quite big enough. And so she skilled them up further to make what we call mChip, our biggest array to date, which contains almost 200,000 micro-wells as well as a high density and coding system so that we could encode as many as 1,000 inputs to each screening batch.

And so Sherry partnered with Pardis Sabeti's group to put this scaled up version of the micro-well array platform to work in viral diagnostic testing. And so the project focused on scaling up the Sherlock CRISPR-based nucleic acid detections in chemistry such that we could probe not only for one viral sequence but viral sequences representing all human associated viruses, and to do that not just for one clinical sample from a patient but from clinical samples collected from many different patients potentially all in one screening batch.

And so this work came out fantastically well. I think it's a testament to the power of the Sherlock chemistry in terms of its specificity and the ease with which new assays can be designed. And earlier this year, this combinatorial testing concept was translated onto a commercial microfluidic device and a smaller panel focused on respiratory viruses was submitted in an emergency use authorization application to FDA.

And so we're hoping this technology can make a positive impact as we're all working against the COVID pandemic going into the fall. So I'll stop there. Thanks for listening.