Microfluidics for Synthetic Biology: From Design to Execution

1 CHAPTER FOURTEENM icrofluidics for Synthetic Biology: From Design to ExecutionM. S. Ferry,*,1I. A. Razinkov,*,1andJ. Hasty*, , Contents1. Part I: The Design of a microfluidic A parallel DAW Cell DAW hardware and software3342. Part II: Photolithography: Special notes on Soft Soft lithography: PDMS PDMS processing: Protocol3553. Part III: Experimental setup forE. Method to set up a MDAW microfluidic experiment364 Acknowledgments371 References371 AbstractWith the expanding interest in cellular responses to dynamic environments, microfluidic devices have become important experimental platforms for biologicalresearch.

2 microfluidic microchemostat devices enable precise environmentalcontrol while capturing high quality, single-cell gene expression data. For studiesof population heterogeneity and gene expression noise, these abilities are , we describe the necessary steps for experimental Microfluidics usingdevices created in our lab as examples. First, we discuss the rational designMethods in Enzymology,Volume 497#2011 Elsevier 0076-6879, DOI: rights reserved.* Department of Bioengineering, University of California, San Diego, California, USA{BioCircuits Institute, University of California, San Diego, California, USA{Molecular biology Section, Division of Biological Sciences, University of California, San Diego, California,USA1 These authors contributed equally to this microchemostats and the tools available to predict their carefully analyze the critical parts of an example device, focusing on themost important part of any microchemostat: the cell trap.}}

3 Next, we present amethod for generating on-chip dynamic environments using an integrated fluidicjunction coupled to linear actuators. Our system relies on the simple modulationof hydrostatic pressure to alter the mixing ratio between two source reservoirsand we detail the software and hardware behind it. To expand the throughput ofmicrochemostat experiments, we describe how to build larger, parallel versions ofsimpler devices. To analyze the large amounts of data, we discuss methods forautomated cell tracking, focusing on the special problems presented bySaccha-romyces cerevisiaecells.

4 The manufacturing of microchemostats is described incomplete detail: from the photolithographic processing of the wafer to the finalbonding of the PDMS chip to glass coverslip. Finally, the procedures for conduct-ingEscherichia coliandS. cerevisiaemicrochemostat experiments are Part I: IntroductionMicrofluidic technology has enjoyed considerable success and interestin recent years. microfluidic devices have been used for everything fromminiaturization of molecular biology reactions to platforms for cell growthand analysis (Bennettet al., 2008; Cooksonet al.)

5 , 2005; Daninoet al., 2010;Hersenet al., 2008; Honget al., 2004; Kurthet al., 2008; Leeet al., 2008;Rowatet al., 2009; Tayloret al., 2009; Thorsenet al., 2002). A drivingfactor for increased use of Microfluidics is the potential for more productiveexperiments, that is, accomplishing the same or more using fewer resources(primarily less reagents, consumables, and time). Furthermore, microfluidicdevices offer the unrivaled ability to precisely control and perturb theenvironment of single cells while capturing their behavior using highresolution microscopy.

6 In this report, we will concentrate on how todesign, build, operate, and analyze data from single cells growing in thechambers of high-throughput microfluidic devices. We will focus primarilyon a device built to monitor the growth ofSaccharomyces cerevisiae(yeast) in adynamically changing environment as a case study. This device is known inour lab as the MDAW or Multiple Dial-A-Wave our lab we strongly believe in the importance of acquiring single celltrajectories from our experimental runs. This requires the ability to tracksingle cells over the course of an experiment, which generally lasts 24 72 , of all technologies available in molecular biology , microfluidicsalone offers the ability to track the behavior of a large number of individualcells over the course of an experiment.

7 While other technologies, such asflow cytometry, allow the acquisition of single cell data, the experimentercannot track each individual cell in time. This leads to snap shots of howthe population as a whole changes in time, but does not capture howindividual cells progress over the course of an S. Ferryet difference between the techniques can be illuminated easily if onethinks of a population of cells containing a desynchronized genetic oscilla-tor. In this case much depends on the waveform of the oscillator. Foroscillators with sinusoidal output, the population will appear bimodal witha large portion of the cells spread between the two modes.

8 However, for anoscillator with output similar to a triangle wave, the cells will be uniformlydistributed between all phases of oscillation and therefore the populationwill have a fairly evenly distributed set of fluorescent values. Of course thebehavior of a real oscillator can be somewhere between these extremes, butthe point is that looking at the progression of a population as a whole doesnot tell you everything about its dynamics. For example, in each of the casesmentioned above, other explanations are possible, such as the transient of abistable switch, or even a genetically mixed population of cells.

9 In contrast,using a microfluidic device to follow the temporal dynamics of single cells insuch a population would allow one to easily see if any cells were Microfluidics is powerful, flow cytometry has the ability tocapture a large amount of data quickly, much more quickly than it can bedone in traditional Microfluidics . For this reason, microfluidic and flowcytometry should be thought of as complimentary, instead of competing,technologies. We often find it useful to first characterize our genetic circuitsusing flow cytometry, testing as many media or inducer concentrations aspossible, to look for behavior indicative of interesting dynamics.

10 Once theseconditions are determined we follow up with the more powerful butinvolved microfluidic in the context of this report we will be talking about microfluidicchips designed to capture single cell data over the 1 3 days of the experi-ment. Unfortunately this limits the architecture of such a chip due to thedifficulty of tracking cells. Regrettably cells such as yeast or especiallyEscherichia colihave few unique features which can be used to distinguishthem from their brethren. The full details of this will be discussed in a latersection describing cell tracking, but suffice it to say, the only truly uniquecharacteristic all cells possess visible by phase contrast microscopy is theirposition in time.