Every winter, a hibernating ground squirrel loses roughly a quarter of its synaptic connections. The brain cools, metabolism slows to a crawl, and the neural architecture that once supported foraging, navigation, and predator avoidance quietly dismantles itself. Every spring, synapses rebuild, circuits reconnect, and the animal wakes up functionally intact. For decades, nobody knew how. Then, in 2015, a team at the University of Cambridge led by neuroscientist Professor Giovanna Mallucci identified the molecular agent responsible: a cold shock protein called RBM3. That discovery, and the complicated story that followed, is the reason cold shock proteins have entered the vocabulary of longevity science, cold exposure practice, and neurodegenerative disease research. It’s also a story that’s frequently told badly.
What cold shock proteins are
When mammalian cells cool below roughly 35°C, a small family of RNA-binding proteins rises in concentration. These are the cold shock proteins: molecules that stabilise messenger RNA, regulate translation, and help cells maintain function under thermal stress. Several members exist , YB-1, CSDE1, and LIN28A/B among them , but two dominate the scientific and popular conversation: RBM3 (RNA-binding motif protein 3) and CIRP (cold-inducible RNA-binding protein).
Both respond to drops in temperature, both bind RNA, and both appear across a wide range of mammalian tissues. But there the resemblance thins. RBM3 has become the subject of intense neuroprotection research. CIRP has revealed a paradox that complicates the entire “cold is protective” narrative.
RBM3 and the neuroprotection story
RBM3’s fame did not begin with ice baths. It began with a puzzle in hibernation biology.
Hibernating mammals , ground squirrels, bears, hedgehogs , undergo cycles of cooling and rewarming that would be catastrophic for a human brain. During torpor, their body temperature can drop below 5°C. Their brains lose substantial synaptic density. Yet when they rewarm, those synapses regenerate fully, with no measurable cognitive deficit. Something was orchestrating the rebuild.
Mallucci’s team had been studying structural plasticity in the context of cooling — the brain’s ability to dismantle and reconstruct its connections. They found that during the rewarming phase, RBM3 levels surged. The protein appeared to be the signal that initiated synaptic reassembly. Without it, the synapses stayed lost.
Mallucci’s team then asked the question that made the research medically relevant: could RBM3 protect synapses that were being lost not to hibernation, but to disease?
In a 2015 paper published in Nature, Mallucci’s group demonstrated that it could. Using mouse models of Alzheimer’s-type pathology and prion disease, two conditions in which progressive synapse loss drives cognitive decline, they showed that early-stage cooling could stimulate RBM3 production and preserve synaptic connections. As neurodegeneration advanced, however, the brain’s ability to produce RBM3 in response to cooling diminished. The protective mechanism was failing precisely when it was most needed.
Next, they tested what happened when RBM3 levels were artificially sustained. Overexpression of RBM3 in the brains of prion-diseased mice prevented synapse loss, preserved memory-related behaviours, and significantly extended survival. When RBM3 was knocked down, neurodegeneration accelerated and disease progression sped up.
RBM3 was not merely a byproduct of cooling. It was, crucially, a mediator of structural brain repair, and its presence or absence could determine the trajectory of neurodegenerative disease, at least in mice.
But the paper’s significance comes with a boundary that must be stated plainly. Every experiment in the 2015 study was conducted in mice. The cooling protocols involved sustained hypothermia at core temperatures of approximately 33°C, maintained for extended periods under anaesthesia. This isn’t an ice bath. It’s not even close to an ice bath.
RBM3’s relevance extends beyond the brain, though that is where the evidence is strongest. A 2024 study in Communications Biology showed that RBM3 promotes mitochondrial metabolism and cellular proliferation in muscle cells, even at normal body temperature when overexpressed. In cancer biology, RBM3 has shown effects that cut both ways , in some tumour types correlating with better prognosis, in others with worse , which prevents any clean “this protein is always beneficial” conclusion.
The human evidence question
Here is where intellectual honesty requires a gear change.
The animal data for RBM3-mediated neuroprotection is clearly strong — replicated, published in Nature, and mechanistically detailed. The human data is scarce.
Mallucci’s own group provides the most frequently cited human evidence. Between roughly 2016 and 2018, her team collected blood samples from regular winter swimmers at Parliament Hill Lido in London and compared their RBM3 levels to a non-swimming control group. The swimmers showed markedly elevated RBM3 in their blood. This is encouraging and biologically plausible. It is also, as of this writing, unpublished and not peer-reviewed. The finding has been discussed in media interviews and conference presentations, but it has not passed through the formal scrutiny of scientific publication. Mallucci is the world’s foremost authority on RBM3 and neurodegeneration, and blood-level measurements are a reasonable if imperfect proxy for tissue expression. But an unpublished finding is not proof that regular cold plunging activates meaningful neuroprotective CSP expression in the brain.
Dose-response uncertainty compounds the problem. Cell culture studies suggest that CSP expression is maximised at sustained mild hypothermia , core temperatures in the 32–33°C range held for several hours. A person submerging in water at 6°C for three minutes will experience rapid skin cooling and a significant drop in peripheral tissue temperature, but their core temperature may fall only modestly, and for a fraction of the duration used in animal studies. Nobody has mapped the relationship between practical cold exposure parameters , water temperature, duration, frequency, body composition , and CSP expression in human tissues. It may be that cold plunging does produce some degree of RBM3 upregulation. It may be that the stimulus is too brief or too shallow for the effect to matter clinically. The honest answer is that we do not yet know.
This does not erase the independent, well-documented effects of cold exposure on catecholamine release, mood, and autonomic regulation. But the specific claim , that your ice bath is producing the same neuroprotective protein response seen in the mouse models , outruns the evidence.
CIRP: the protein with a double life
If RBM3 is the hopeful story in cold shock protein biology, CIRP is the cautionary one.
Like RBM3, CIRP rises in concentration when cells cool. Inside the cell, it performs a stabilising role: binding messenger RNA, protecting transcripts from degradation, and helping maintain protein production under thermal stress. At this level, CIRP is part of the cell’s cold-adaptation toolkit, and its intracellular functions are broadly consistent with the “cold builds resilience” narrative that dominates popular coverage.
Trouble begins when CIRP leaves the cell.
In 2013, a team led by Dr Ping Wang at the Feinstein Institutes for Medical Research published a finding in Nature Medicine that reframed CIRP’s biology. They showed that when CIRP is released into the extracellular space , as happens during haemorrhagic shock, sepsis, and other conditions involving severe tissue stress , it behaves as a damage-associated molecular pattern, or DAMP. Extracellular CIRP (eCIRP) binds to TLR4-MD2 receptors on immune cells, triggering a cascade of proinflammatory cytokine release. In animal models, this contributed to organ damage and worsened outcomes in sepsis.
A protein that protects cells from the inside can trigger dangerous inflammation from the outside. Controlled cold exposure, which keeps cells intact and functioning, presumably keeps CIRP where it belongs , inside the cell, doing useful work. Pathological stress, tissue injury, or overwhelming systemic crisis can liberate it into the extracellular space, where it becomes an inflammatory alarm. For anyone trying to understand the molecular logic of cold, CIRP is the clearest reminder that context , not just the molecule , determines the outcome.
Heat shock proteins vs cold shock proteins: the trade-off and the case for contrast therapy
Cells also respond to heat, and the relationship between their heat response and their cold response is more complicated than simple addition.
Heat shock proteins (HSPs), particularly HSP70 and HSP27, are among the most studied stress-response molecules in biology. They act as molecular chaperones: refolding damaged proteins, preventing aggregation, stabilising cellular structures under thermal and oxidative stress. Regular sauna use, hot water immersion, and exercise all upregulate HSPs. The evidence linking HSP expression to cardiovascular protection, inflammatory regulation, and cellular repair is extensive and, in several areas, more developed than the CSP literature.
Intuitively, contrast therapy — alternating between heat and cold — should activate both systems. Heat produces HSPs, cold produces CSPs, and combining them gives you the full spectrum.
The molecular reality has a catch. A 2019 study in the Journal of Applied Physiology by Fyfe and colleagues showed that cold water immersion after resistance training attenuated training-induced increases in HSP27 and reduced HSP72 protein content in skeletal muscle. Cold did not just activate its own proteins. It suppressed the heat-activated ones.
Timing, then, is not incidental , it is a design parameter with molecular consequences. Apply cold immediately after a heat stimulus that would normally drive HSP production, and you may dampen the adaptation you wanted. Contrast therapy’s case, then, is not “pile up stress signals.” It is a sequencing argument. Heat drives HSPs: chaperone activity, protein repair, inflammatory regulation. Cold drives CSPs: RNA stabilisation, translational control, and in the case of RBM3, synaptic protection. But the heat session needs enough time to establish its molecular response before cold exposure potentially attenuates it. Contrast therapy is the only common practice that engages both cellular defence systems , provided the stimuli are separated enough in time for each to do its work.
No study has directly measured HSP and CSP expression in a controlled human contrast therapy protocol with optimised timing. The framework is supported by the individual findings, but it remains a molecular hypothesis. What it offers is a more precise reason to practise contrast therapy than the assumption that hot and cold are both good, so doing both must be better.
For anyone practising contrast therapy, the molecular logic suggests separating heat and cold stimuli by enough time for each response to establish , minutes, not seconds , rather than rapid alternation. Nobody has tested this timing question directly, but the direction of the evidence favours sequence over speed.
The therapeutic horizon
One critical shift in cold shock protein research rarely appears in popular coverage. For the first decade of RBM3 research, the question was: can cooling protect the brain? In animals, the answer was yes. But the clinical problem was obvious. You cannot cool a person with Alzheimer’s disease to 33°C for hours and call it a therapy. Sustained hypothermia carries its own risks , cardiac arrhythmia, infection, coagulopathy , and is impractical for a chronic progressive disease.
Mallucci’s team changed the question. In a 2021 paper in Life Science Alliance, they identified the signalling pathway through which cooling activates RBM3: TrkB receptor stimulation, leading through PLCγ1 to pCREB, which switches on RBM3 transcription. More importantly, they showed that drugs activating TrkB , small-molecule agonists already being studied in other contexts , could induce RBM3 production in mice without any temperature drop at all. The mice were warm. RBM3 levels rose. Synapses were protected.
A separate approach, reported in EMBO Molecular Medicine by Preussner and colleagues in 2023, used antisense oligonucleotides to boost RBM3 expression through gene-level intervention, again without cooling. Both approaches produced significant neuroprotective effects in animal models.
“What RBM3 gives us is a way of intervening without cooling,” Mallucci has said. The point is not that ice baths are medicine. The point is that studying what cold does to cells has revealed a protein, a pathway, and a therapeutic target that might one day protect human brains from neurodegeneration , delivered as a drug, not as a dip in freezing water.
Neither approach has reached human clinical trials for neurodegeneration. Both are still in animal models. But the trajectory is real, the molecular logic clearly sound, and the research programmes active.
Beyond the ice bath
What makes cold shock proteins exciting may not be what they mean for your morning plunge. It may be what they mean for medicine. A protein discovered through hibernation research, validated in neurodegeneration models, and now being pursued as a drug target that works at normal body temperature. The ice bath was never the destination. It was the observation that opened the door.
