CFPS Reaction Miniaturization Below 10 µL

Cell-free protein synthesis has conventionally run at volumes suited to manual pipetting, not because the chemistry needs that much room. Push the same reaction below 10 microliters, down to 2 or even 0.5 microliters, and the biochemistry holds up fine. What breaks is everything around it: how you dispense the liquid, how you stop the well from drying out, how you read the result afterward. Teams that treat miniaturization as a pipetting problem lose months to it. The ones who treat it as a surface chemistry and fluid mechanics problem tend to get it right the first time, and that distinction is the whole argument of this piece.

An application note run on the Echo 525 acoustic liquid handler put a number on this. CFPS reactions at 0.5, 1.0, and 2.5 microliters produced protein yield that scaled linearly with volume, for both a lysate-based system and PURExpress, using linear DNA template. No drop-off in activity, no penalty for shrinking the reaction. That's the number worth holding onto for the rest of this piece: a 50 to 100-fold jump in throughput isn't a hypothetical anymore, it's been shown, and it's the difference between screening 200 variants and screening 10,000. Protein engineers running large libraries, anyone using PURE-system reagents (which get expensive fast at standard volumes), and labs trying to grow throughput without growing their reagent budget at the same rate are the ones this matters to most. Sepia Biosciences, a cell-free protein synthesis reagent company whose OpenCFPS™ line is priced and formulated specifically for plate-based screening of large variant libraries at small reaction volumes, is one supplier targeting exactly this gap. Sepia Biosciences offers OpenCFPS™ for this kind of high-throughput screening work, and running large numbers of small-volume reactions is a core use case for the platform.

What changes physically when a CFPS reaction drops below 10 µL

Surface area doesn't shrink at the same rate volume does. Cut a reaction's volume by ten-fold and the surface-to-volume ratio doesn't just tick up, it jumps, because area scales with the square of a linear dimension while volume scales with the cube. At sub-10 µL scale, that's not a rounding error. It's the whole story.

Three consequences follow, and none of them are subtle. Evaporation comes first: a small reaction loses a real fraction of its mass to the air during a normal 2 to 4-hour incubation, and as water leaves, everything else concentrates, so Mg²⁺, PEG, and amino acid levels drift out of range without anyone touching a pipette. Adsorption comes second: proteins and nucleic acids stick to plastic walls, and when total protein or DNA mass in a well is measured in nanograms, losing even a small fraction to the wall matters in a way it doesn't at 50 µL. Gas exchange comes third: oxygen and CO₂ move across the liquid surface, and since that surface is now large relative to the volume beneath it, exchange rates shift enough to bend energy-regeneration kinetics.

Published work settles the vessel-shape question outright, and it's worth being blunt about what it shows: geometry decides yield just as much as volume does. Running CFPS reactions in microcentrifuge tubes and increasing volume cut production efficiency significantly, while flat-bottom plate formats run at comparable volumes showed no such drop. Same chemistry, same reagents, different container, different outcome. Anyone still optimizing a miniaturized CFPS reaction in a round-bottom tube because that's what happened to be on the shelf is fighting the container before the biochemistry even gets a chance to work.

There's a mixing consequence too, and it's easy to miss. Convective mixing, the kind you get from shaking a tube, does most of the homogenizing work at 50 µL. Below 10 µL, diffusion takes over, and diffusion is slow. Components don't distribute the way they would in a shaken tube, so parameters dialed in at 25 to 50 µL (Mg²⁺ concentration, template amount, energy source levels) often need re-optimizing once the reaction runs at 2 to 5 µL, because the local concentration a ribosome or polymerase actually experiences has shifted. Continuous-exchange formats, which feed nutrients in and pull byproducts out through a membrane, blunt some of this, which is a big part of why CECF designs keep resurfacing in miniaturized work.

Liquid handling precision as the rate-limiting step in sub-10 µL CFPS

Below roughly 1 microliter, ordinary tip-based pipettors stop being trustworthy, and no amount of careful technique fixes that. Dead volume in the tip, surface tension pulling liquid back up, a coefficient of variation that climbs as volume shrinks: none of that is an edge case at 2 to 5 µL. It's the default failure mode. A 10% pipetting error on a 2 µL reaction is 200 nanoliters, and 200 nL is plenty to push Mg²⁺ or template concentration clean out of the range where the reaction works at all.

Acoustic liquid handling solves this by never touching the liquid. The Echo 525 uses sound waves to eject 25-nanoliter droplets from a source well straight into a destination well: no tip, no contact, none of the surface-tension trouble that comes with syringe-based dispensing at these volumes. The Beckman Coulter application note showed this working reliably at 0.5, 1.0, and 2.5 µL, with yield tracking volume linearly across both a lysate-based system and PURExpress. That linearity is the important part: protein yield per microliter stayed constant, which means the smaller reaction isn't a lesser version of the bigger one. It's the same reaction, just less of it.

Acoustic dispensing does carry a setup cost: source reagents need to sit in Echo-compatible, low-dead-volume plates, a workflow decision made before the screen starts, not bolted on afterward. Labs without an acoustic handler aren't stuck, though. Nanoliter dispensing valves, piezoelectric dispensers, and microfluidic droplet generators can all get liquid into sub-microliter wells, each trading off throughput, format flexibility, or cost differently. Still, acoustic dispensing is the most thoroughly validated route for plate-based CFPS miniaturization, and anyone building a screening pipeline from scratch should default to it, not treat it as one option among several.

Evaporation control and vessel format choices at sub-10 µL scale

At 2 to 5 microliters, evaporation isn't a slow background process. It's fast enough to noticeably concentrate a reaction within a standard incubation window. Mg²⁺, PEG, and amino acids all creep upward as water leaves, and none of that drift is intentional or easy to predict without measuring it directly.

A handful of fixes address this, and most labs end up combining more than one rather than betting on a single favorite. A mineral oil overlay sits on top of the aqueous reaction and blocks evaporation without mixing in, and it's simple to set up in a standard plate. Humidity-controlled incubation chambers keep the surrounding air close to saturated, which removes the driving force for evaporation in the first place. Sealed plates, using adhesive optical film or heat-seal foil rated for the incubation temperature, work well too, though if the plan is fluorescence readout, the seal has to actually transmit light at the right wavelengths, or the sealing step undoes its own benefit at the readout stage. And 384-well or 1536-well plates, with their deep, narrow wells, simply expose less surface area per unit of volume than a shallow 96-well plate does.

Plate material isn't a footnote here either. Polypropylene and polystyrene bind protein differently, and low-binding surface treatments, BSA blocking, PEG coating, cut down on wall losses that matter far more at small volumes than at 50 µL. Whatever format gets picked has to work with the readout instrument: a sealed film that blocks the excitation or emission wavelength of a fluorescent reporter defeats the entire point of using that reporter. The Vertical-II CFPS array device, described by Jackson and Fan in the Journal of Laboratory Automation in 2014, was built specifically to fit a standard 96-well plate holder, which let it slot into existing commercial reagent dispensers and plate readers without custom hardware. That's the same principle worth applying when choosing an off-the-shelf plate: fit the tools already on the bench, don't force new ones in.

Plate formats and reactor geometries that have been validated for miniaturized CFPS

Standard 96-well plates running at 2 to 10 µL are the easiest entry point, since they work with almost any liquid handler or plate reader already sitting in a lab. Evaporation is the main thing to manage there. Move to 384-well plates and per-well reagent use drops further, the well geometry itself limits evaporation exposure, and acoustic dispensing handles the format without complaint.

Continuous-exchange cell-free, or CECF, devices push further still, and the yield gap isn't small. A 96-unit microfluidic CECF device described in the literature produced yields up to 87 times higher than a conventional batch reaction, by continuously feeding nutrients in and pulling inhibitory byproducts out through a dialysis membrane. Its follow-up, the Vertical-II device, increased protein expression 406% over the horizontal baseline while using 5 times less reagent than a commercial device, which is about as strong a published case as exists for CECF in a plate-compatible, miniaturized format. Setting one up isn't as simple as a straightforward batch reaction, but when the priority is squeezing the most protein out of every microliter of reagent, CECF earns the added complexity. Choosing batch over CECF purely to save setup time leaves yield on the table, and that's a real cost, not a rounding error.

Further out past the 10 µL threshold, droplet microfluidics compartmentalizes CFPS reactions in water-in-oil emulsions at very small scales, enabling very high-throughput variant screening in a single run. That's beyond the sub-10 µL range this piece focuses on, but it marks the outer edge of what compartmentalized CFPS can do.

Picking a format comes down to the goal, and the three goals don't share one. Comparing relative yield across a variant library points toward 384-well plates with acoustic dispensing, since that combination is the most practical to run at scale. Chasing absolute yield at small volume points toward a CECF device. Ultrahigh-throughput directed evolution, screening well past what a plate can hold, points toward droplet microfluidics. Trying to force one format to do all three jobs is the most common way these projects go sideways.

Template and reagent preparation considerations specific to small volumes

Linear PCR templates skip the cloning step entirely, which makes them the obvious choice for variant screening where speed matters. The tradeoff: linear DNA is exposed to exonucleases in lysate-based systems, so a nuclease inhibitor like GamS (or an equivalent) has to go into the mix, and that additive needs its own line item in the volume budget, not an afterthought tacked on at the end.

Template concentration deserves more attention at small volumes than it gets at large ones. If the template stock is dilute, hitting the right template amount in the reaction can eat up a large share of the total volume, crowding out everything else that needs to fit in the well. The fix is straightforward: keep template stocks concentrated enough that the volume added stays at or below 10% of the total reaction.

Order of operations matters too. At 2 to 5 µL total, adding each component separately stacks pipetting error on pipetting error. Building a master mix with everything except the variable, usually the template, and dispensing that variable last keeps the error contained to a single addition instead of five or six.

Reagent stability is worth checking directly rather than assuming it holds. Source plates can sit on an acoustic handler's deck for the length of a long run, and energy-regeneration components in particular should be confirmed stable at room temperature over that window before a large screen depends on it. A 2024 study in the Beilstein Journal of Organic Chemistry (Bartsch, Lütz, and Rosenthal) found that E. coli extract-based CFPS held up against a fairly wide range of technical additives, including polymers, organic solvents, and salts. That widens the space available for optimizing small-volume reactions without necessarily destabilizing them in the process.

Lot-to-lot reagent variation gets amplified at small scale in a way that's easy to underestimate. A shift in reagent lot that would barely register at 50 µL can show up as a measurable yield swing at 2 µL. Sepia Biosciences publishes lot-level QC data for OpenCFPS™, which gives researchers a way to catch that kind of shift before it quietly wrecks a large miniaturized screen.

Reading out results from sub-10 µL CFPS reactions reliably

A 2 µL reaction producing the same per-microliter yield as a 50 µL reaction still ends up with far less total protein in absolute terms, simply because there's less reaction to begin with. Any readout method has to be sensitive enough to catch that smaller mass, and not every method clears that bar.

Fluorescent reporters are the most practical answer for screening, and they should be the default rather than the fallback. Signal tracks protein mass directly, standard plate readers handle 96- and 384-well formats without trouble, and there's no purification step standing between the reaction and the number. This is the exact approach the Beckman Coulter application note used to demonstrate linear scalability at 0.5, 1.0, and 2.5 µL. Gel-based detection doesn't really work at sub-10 µL per lane without a concentration step first, so it belongs in confirmatory runs at somewhat larger volumes, not the primary screen.

For enzymes, coupling the CFPS reaction directly to a substrate assay in the same well sidesteps the protein-mass problem entirely and pushes throughput even higher; the Beilstein study mentions this approach in its introduction, even though its main contribution lies elsewhere. AlphaScreen, TR-FRET, and other bead-based proximity assays are sensitive enough to handle small protein masses too, and they earn their place when the target isn't a natural reporter or an enzyme with a convenient substrate.

Well-to-well noise is unavoidable at this scale: evaporation differences, edge wells running slightly warmer or cooler than center wells, small pipetting variation, all of it adds up. Including a constitutive reporter, a second fluorophore expressed from an internal standard in every well, gives a way to normalize for that noise and separate real biological signal from technical drift. Plate reader settings, path length correction, gain, whether the read comes from the top or bottom of the well, shift the apparent signal too. Fix those settings before the screen starts. Changing them mid-run makes the data impossible to compare, full stop.

Translating miniaturized screening results to larger CFPS production runs

Miniaturized reactions are built for throughput and for comparing variants against each other. Milligram-scale production reactions are built for squeezing out the most absolute yield. Both run on the same underlying biochemistry, but the conditions that suit one don't automatically suit the other, and assuming they do is where scale-up projects lose weeks they didn't need to lose.

Some things do carry over cleanly. Relative yield rankings, which variant beats which, generally hold from 2 to 5 µL up to 25 to 50 µL, and the linear scalability data from the Beckman Coulter application note backs that up for the systems it tested. Other things don't transfer, and need a second look every time a reaction scales up: Mg²⁺ and PEG concentration, template amount per unit volume, and the level of energy source, whether that's creatine phosphate, phosphoenolpyruvate, or glucose. These are the parameters most sensitive to volume, and they're the first place to check when a scaled-up reaction underperforms what its small-scale data predicted.

CECF isn't limited to miniaturized array devices either. The same continuous-exchange principle that lifts yield in those small formats applies just as well to bench-scale dialysis-bag CFPS, so a researcher who hits a yield ceiling with batch reactions at larger scale can move to CECF without switching reagent systems.

OpenCFPS™ is built to support both ends of this pipeline, rapid screening of large variant libraries at small volume and milligram-scale production, using the same reagent formulation and the same QC standards throughout. That consistency means a researcher moving from a 2 µL screen to a 50 µL production run isn't also switching suppliers or requalifying a new reagent system halfway through the project. Keeping a written record of the exact conditions used in the miniaturized screen (volumes, plate format, incubation time, reagent lot) is what turns that transition into a matter of adjustment rather than a rebuild of the whole optimization from scratch.