History of Cryoprotectant Design
We're building technology to extend the time window we have to match a donor organ to the person whose life it can save. To do that, we need to cool the organ past the danger zone of ice formation, the temperature range where water snaps into ice crystals that expand and tear cells apart. For decades, scientists have searched for a biocompatible cryoprotectant that can stop those crystals from forming.
In 1940, Basile J. Luyet, a Swiss-born physicist and priest, published the first scientific attempt at vitrifying a biological sample in Life and Death at Low Temperatures (Luyet, 1940). He tried, unsuccessfully, to vitrify red blood cells by rapidly cooling them without chemical additives. We now know this doesn't work because the nucleation rate of ice in supercooled water is staggeringly fast. To beat it by speed alone, you'd have to drop the temperature by more than a million degrees per second, a rate you can achieve in a microscopic droplet, but not a whole organ.
The field turned to chemistry for help.
Two important advances took Luyet’s work on low-temperature biology and brought vitrification within reach. In 1949, Christopher Polge, Audrey Smith, and Alan Parkes found that glycerol protected fowl sperm through several freeze-thaw cycles (Parkes, 1949), and a year later, Smith extended this work to red blood cells (Smith, 1950).
Ten years later, James Lovelock and Mervyn Bishop put forward dimethyl sulfoxide (DMSO) as a substantial improvement over glycerol (Lovelock and Bishop, 1959).
Beyond slowing ice crystal growth, one of the key advantages they quickly recognized was that DMSO rapidly permeates through biological membranes. While glycerol needed 2 hours of equilibration to protect bovine red blood cells, DMSO needed only 30 seconds. Fast permeation, it turned out, was key for cryoprotectant efficacy in larger biological systems.
Glycerol and DMSO were the proof of concept: the right small molecule really can carry a cell through a freeze-thaw cycle intact. In 1969, Alan Karow framed cryoprotectants as a new class of drugs, calling for them to be optimized and discovered just like pharmaceuticals rather than stumbling onto them by accident (Karow, 1969). Yet the vision never scaled. Today, still only a few dozen cryoprotectants are commonly used.
Greg Fahy and collaborators have been instrumental in optimizing cryoprotectant formulations based on this small set of molecules, culminating in his cornerstone work developing the M22 cryoprotectant formulation (Fahy, 2004). Even with this development, biocompatibility remains the wall the field keeps hitting.
A cryoprotectant can harm a cell in several ways at once, including, but not limited to: (1) pulling water out (osmotic shock), (2) unfolding proteins, (3) binding where it shouldn't, and (4) disrupting metabolic pathways
Recent work has begun to make progress here, notably Adam Higgins screens 21 individual cryoprotectants and hundreds of mixtures (Higgins, 2025). But that's a small corner of an enormous search space, and biocompatibility has to hold across every cell type in an organ, each with its own tolerances.
These many faces of cryoprotectant biocompatibility are also a clue. If no single molecule has to carry the full protective dose, the dose can be divided across components and mechanisms so that no one channel of toxicity overwhelms the cell. With over 400 unique cell types in the human body, each with its own tolerances, finding those combinations is a search problem; we’ve taken it on.
We screen molecules at a scale the cryobiology field hasn't seen, then learn which ones can share the load. Our pipeline runs in five stages: in silico screening, efficacy testing, permeation testing, biocompatibility screening, and whole-organ assessment. We'll touch on all five, but our focus here is the Biocompatibility Screen.
Until’s Screening Platform
We begin in silico. For over 250,000 candidate molecules, we run physics simulations of how each interacts with water and breaks ice nucleation and extension. This is done at two levels of detail: classical models that treat atoms as abstract balls-and-springs and quantum models that more accurately simulate how a molecule's electron cloud interacts with water’s. We then combine each molecule’s simulation results with a set of its measured chemical properties (size, shape, charge, and how readily it forms hydrogen bonds) into a single numerical fingerprint that captures the molecule in a form our model can read. Our house-trained machine learning model then reads that fingerprint and predicts how well the molecule will work.
Hits move down the pipeline to Efficacy Testing. The model predicts which molecules will vitrify and efficacy testing is where we find out if it's right. We take a diverse set of predicted hits and misses into the lab and measure the concentration each one actually needs to vitrify. We’re keeping an eye out for molecules that vitrify at low concentration. The confirmed results feed back to retrain the model, and strong performers advance to permeation testing.
A molecule that vitrifies is only useful if it can get inside the cell quickly. With permeation testing, we measure how quickly each candidate crosses the cell membrane. This is the DMSO-versus-glycerol distinction made quantitative: two molecules can be similarly good at stopping ice, but the one that permeates in seconds rather than hours is the one we need. Molecules that both vitrify and permeate quickly advance in our pipeline; the slow ones drop out.
In Vitro Biocompatibility Screening. All the molecules that have made it to this point in the pipeline vitrify and permeate well, but they still must be tolerated by the cell. Here, we assess if the cryoprotectant is biocompatible.
In Q2 alone, we ran over 10,000 wells across 2,903 formulations, spanning both standard lab cell lines and primary cells taken from tissue, which behave more similarly to the cells in a living organ. A translationally-relevant assay needs a panel of diverse cell types, so we screen across a wide panel. This allows us to get the closest possible representation of an organ using simple cell models. Here we show just one as an example. Every result feeds back into our biocompatibility model, sharpening its ability to spot combinations of cryoprotectants that work.
Recall that a cryoprotectant can harm a cell through several independent mechanisms? This turns out to work in our favor. Spread the required dose across components that damage the cell in different ways, and no single mechanism ever builds up enough to trouble it, even as the mixture as a whole reaches a vitrifying concentration. We call this our Biocompatibility Budget, and it is the key to formulations that are both effective and gentle to the organ.
Try the interactive below: increase your vitrification dose across four illustrative mechanisms. Real biological systems are nowhere near this simple, since mechanisms overlap and interact, so treat this as purely conceptual.
For example, we can compare the biocompatibility of several commonly used cryoprotectants. In Figure 1a, we plot the cell viability as a function of dose in vitrification units. On this scale the dashed line at 1.0 marks the dose a cryoprotectant actually needs to vitrify. So each curve answers the question: how many cells are still alive at the dose that gets you to vitrification? For solutions made of a single cryoprotectant, the answer is not many.
But if we combine single components into one multicomponent cryoprotectant (CPA-1042), we can reach the vitrification threshold without crossing the biocompatibility budget. In Figure 1b, CPA-002, CPA-069, CPA-081, and CPA-313 are all single components of CPA-1042. At vitrification concentration, most single components fall to half viability or below, but lower doses of each combined into CPA-1042 successfully achieves a biocompatible vitrification dose.
The real test, however, is against the field's best. Figure 2 tests CPA-1042 alongside the field standard, VMP, in the same primary human vascular cell model. At the dose needed to vitrify, both standards have fallen to roughly half viability, CPA-1042 is still holding near 85%.
Whole-Organ Assessment. These cell-based tests are vital, but the real endpoint is the ability to hold up in a whole organ. Here the challenge shifts from permeation to perfusion: a cryoprotectant no longer has to just cross into cells, it has to travel the entire vascular network and reach every cell of an entire organ. A cryoprotectant that appears to be gentle in a dish but is too viscous or too impermeant to perfuse the microvasculature will fail here. Only a whole-organ test can catch it.
We evaluate formulations at organ-scale in two ways. First we perfuse the cryoprotectant into an organ and wash it out, measuring perfusability and biocompatibility directly. The second is a trial run of the entire journey an organ would take after donation: load, vitrify, rewarm, and unload (read more about this journey in Glass, Not Ice, and the rewarming process in Rewarming).
Whole-organ assessment is the hardest test in the pipeline and the one that matters most, keep an eye out for our future explorations on this!
The Engine, Running
The pipeline is now fully operational and continuously expanding. It has already surfaced cryoprotectants chemically distinct from anything previously reported and multicomponent formulations that exploit those differences to protect cells where the field's standards fall short. By the end of 2026 we expect to have screened over 2,000 individual components and 10,000 formulations for biocompatibility.
For fifty years, cryoprotectant discovery has moved a few compounds at a time; we have built a pipeline that screens, stress-tests, and learns from formulations at scale, and feeds everything it learns back into the design of the next generation. Cryoprotectant biocompatibility will be a cornerstone of organ banking and medical hibernation, and it will be solved the way modern drug discovery solves hard problems: by searching, systematically and at scale.
Glossary
Biocompatibility How well a cryoprotectant is tolerated by living cells at the concentrations needed to protect them. Cryoprotectant biocompatibility is the property that decides whether a formulation is usable for organ cryopreservation.
Biocompatibility Budget The idea that each way a cryoprotectant can damage a cell has its own tolerance limit, so the protective dose can be spread across mechanisms without any single one exceeding its budget. Also called sub-additive toxicity, it is what makes a biocompatible vitrification of organs possible.
Cryoprotectant A chemical added before cooling that protects a sample from freezing damage, usually by promoting vitrification of organs and tissues instead of ice. Modern organ cryopreservation depends on finding cryoprotectants with high enough biocompatibility. Also known as a Cryoprotective Agent or CPA
Dimethyl Sulfoxide (DMSO) An organosulfur compound, (CH₃)₂SO, and one of the most widely used cryoprotectants in cryobiology. It slows ice growth and permeates cells quickly, which made it a workhorse of cryopreservation.
Glycerol A sugar alcohol with the formula C₃H₅(OH)₃, and the first cryoprotectant shown to work. Its slow permeation limits its use at the scale of organ cryopreservation.
M22 A concentrated, multi-component vitrification solution from Greg Fahy's group, named for the roughly -22°C temperature at which it is introduced. It is a benchmark for the vitrification of organs, though its biocompatibility limits leave room for gentler formulations.
VMP A concentrated vitrification solution (around 8.4 M) from Fahy's group, used in kidney vitrification research and was the cryoprotectant used in the landmark Han et al. paper on rat kidney vitrification. It serves as a field standard for measuring newer cryoprotectants in organ cryopreservation.
Perfusion Delivering a cryoprotectant through an organ's own blood vessels so it reaches every cell. A formulation too viscous to perfuse the microvasculature fails at organ scale even if it looks gentle in a dish.
Permeation How readily a cryoprotectant crosses the cell membrane to reach the inside of the cell. Fast permeation gets the protectant in before it can do harm; slow permeation leaves the cell exposed.
Vitrification Cooling a liquid into a glassy, ice-free solid instead of letting it crystallize. The vitrification of organs preserves structure that ice would destroy, but reaching it takes a high cryoprotectant dose, which is why biocompatibility is the limiting factor.
Have something to say?
Email us at