Making proteins without cells: speed, costs and limits


Cell extracts or purified molecular components let researchers test protein designs without growing modified cells for each one.
Cell-free protein production lets researchers screen hundreds or thousands of designs without modifying and growing cells for each one. That can shorten the cycle of designing, making and testing proteins from days or weeks to hours, Boston University synthetic biologist Wilson Wong told Nature in its October 6 report.
Living cells have their own priorities: they can deactivate inserted genetic instructions or use resources for survival and reproduction rather than the desired product. Cell-free systems instead use the molecular machinery that copies DNA instructions into messenger RNA and reads that RNA to build proteins. The approach is not new; it helped researchers decipher the genetic code in the 1960s.
There are two main ways to supply that machinery. Researchers can break cells open and use their contents, an extract called a lysate, or assemble a system from purified components. The purified approach is called Protein Synthesis Using Recombinant Elements, or PURE. Emily Chen, a production scientist at biotechnology company New England Biolabs, told Nature that lysates often produce more protein at lower cost than PURE systems.
“Cell-free” describes the protein-making reaction, not its entire preparation. For sensor tests, Stanford University bioengineer Michael Jewett’s team grew Escherichia coli bacteria, broke them open and used their contents. Lysates also need added ingredients, including glucose and nucleotides, to supply energy and raw materials.
Water sensors illustrate why a fast, short-lived reaction can be useful. Living-cell sensors must stay alive, and the substance being measured must pass through the cell’s outer barriers. Cell-free sensors avoid those requirements and can produce a measurable signal in tens of minutes, Nature reports.
In 2020, Northwestern University synthetic biologist Julius Lucks’s group described ROSALIND, a chemically defined sensing system. It uses transcription factors—proteins that control whether DNA is copied into RNA—to block RNA production until they encounter a target substance. The protein then releases the DNA, allowing RNA production to generate a signal. As a proof of principle, the team used freeze-dried ROSALIND systems to detect copper and zinc in municipal water samples contaminated by a 2018 California wildfire.
For a separate lead sensor, Lucks, Jewett and colleagues faced a selectivity problem: many proteins that recognize lead also bind zinc. The researchers fed measurements from many altered versions of lead-detecting proteins into a machine-learning model. They used its results to design another set of variants, then tested those proteins in a cell-free system.
After a few rounds, the researchers obtained a sensor capable of detecting lead at concentrations as low as 5.7 parts per billion, Nature reports. That is an experimental detection result, not a validated public-use water-test specification.
The advantages do not make cell-free systems a cheap replacement for large-scale cell-based production. Nature reports that reagents for a litre-scale reaction could cost more than US$4,000, while most studies report producing less than two grams of protein. The article puts cell-free protein synthesis at roughly ten times the cost of cell-based methods. Some researchers keep screening affordable by running reactions at volumes below one microlitre.
Custom recipes can reduce that expense. Jewett’s laboratory tested 1,231 formulations using its own lysates and narrowed the recipe to 12 reagents. That work reduced costs by 95%, to less than $100 per gram of protein produced. The result applies to the team’s optimized recipe, rather than every cell-free system.
Energy supplies impose another limit. Bacterial or yeast cultures can keep producing material as researchers add fresh growth medium and remove products. Cell-free systems quickly exhaust their energy supplies; Wong says their reactions need to finish within a few hours. That suits a sensor that signals in minutes better than a process intended to run continuously.
A lysate that makes proteins successfully is not necessarily a good model of the biological process a researcher wants to understand. Evan Karousis, an independent researcher and lecturer in Bern, warns that extracts from rabbit reticulocytes—immature red blood cells—have unusual protein-production machinery. They can begin reading RNA at non-standard sites and do not require the RNA “cap” that normally helps protect the message and start protein production.
Karousis stopped a three-year project after discovering that rabbit reticulocyte extracts used an atypical version of the RNA-breakdown process he was investigating. He recommends matching the lysate to the biology under study. His former University of Bern group developed extracts from several human cell lines as alternatives for studying how cells read RNA to make proteins.
Researchers are developing freeze-dried formulations for portable water tests, diagnostics used where patients receive care, and on-demand production of biological medicines such as vaccines. These remain development aims, not established medical applications.
For laboratory work, some of Lucks’s sensors are available through the nonprofit reagent repository Addgene, and his ROSALIND papers include Excel files with reaction instructions. University of Minnesota biophysicist Vincent Noireaux recommends that newcomers begin with standardized commercial systems before making their own lysates for larger-scale, customized experiments.

Unite has announced October and November walkouts at seven depots that supply hospitals across England.
2 min

Flu and COVID-19 remain at low levels, the agency says. It plans to track respiratory viruses throughout winter.

Researchers found smaller ant colonies in wasp-occupied trees and that trees in oil palm plantations were more likely to contain wasps than those in logged forest.
2 min