Supplementary Material

The microfluidic chips used in this study were implemented as a two-layer polydimethylsiloxane (PDMS) microfluidic chip similar to published procedures, summarized briefly as follows. The channels on the upper layer are flow microchannels used for the reagent inlets, filling chambers, and output of generated droplets, whereas the channels on the lower layer are control channels that can be pressurized or depressurized to open or close the corresponding microvalve(s). Silicon wafer molds composed of photoresist-patterned microstructures for fabrication of the fluid and control layers were created by standard photolithographic techniques. Both molds were pretreated with trimethylsilyl chloride vapor for 10 minutes to facilitate the release of final PDMS replica. Well-mixed PDMS prepolymer (GE RTV615, total weight 36 g, mixing ratio A:B 5 5:1) was poured onto the fluid layer mold to give a 6 mm–thick fluidic layer with approximately 40 mm channel depth. Another portion of PDMS prepolymer (GE RTV615, total weight 10 g, mixing ratio A:B 5 20:1) was mixed and then spin-coated onto the control layer mold at 1,500 rpm for 60 seconds. The fluidic and control layers were cured at 80uC for 15 and 18 minutes, respectively. After baking, the fluidic layer replica was peeled from the mold and aligned onto the control layer, and then the assembly was baked at 80uC for at least 6 hours to adhere the layers. The chip was peeled off the mold when the fluidic and control layers were firmly bonded together. Holes were then punched to form ports connected to the fluidic layer channels for reagent inlets and outlets and ports connected to the control layer channels for microvalve and pump actuation with hydraulic fluid (water). Adhesion of the chip to a clean glass microscope slide to seal the control channels was achieved by corona discharge treatment. The microfluidic device was baked in an oven at 80uC for 72 hours to restore the intrinsic hydrophobicity of PDMS surfaces, which is needed to minimize reagent loss and nonspecific binding of biomolecules on channel walls when moving aqueous droplets. Microvalve control lines were filled with water as hydraulic fluid. Microvalves were actuated by pressurizing the corresponding control channels (up to 60 psig) via electronic solenoid valves (Series S070, SMC, Toyko, Japan). All valves were automatically controlled through a data acquisition module (USB-4750, Advantech, Milpitas, CA) driven by a custom software program written in LabView (National Instruments, Austin, TX). The microfluidic system includes two microfluidic chips (a distribution chip of pH buffers and a DMDG chip connected by tubing). The DMDG chip is composed of three functional parts (see Figure 1B): (1) a droplet generation core, where specific quantities of reagents are measured and merged into composition-specific droplets; (2) a peristaltic pump, which produces serial compressed nitrogen pulses that can precisely deliver intact droplets to the desired location; and (3) a mixing channel. Figure S1 shows the external solution distribution chip (see Figure S1A) and the droplet generation core (see Figure S1B). The unique features of our DMDG chip (see Figure 1B) are (1) independent control of volume and composition for every droplet, enabling screening with minimal reagent consumption; (2) the ability to pause, modify, and restart the droplet generation process, for example, for replacement or change of reagents; and (3) the use of nitrogen gas rather than oil to separate droplets, eliminating the need for oil removal steps afterward. Using the microvalves to isolate each reagent inlet, the incoming reagents are not in contact with each other until the moment of droplet formation. They will be rapidly mixed and reacted thereafter while moving along the microfluidic channel. This is particularly critical because some radiolabeling tags, such as [F]SFB, are often unstable and prone to hydrolyze or decompose at a higher pH.