How Nobel Prize science, AI, and small molecules are converging on the biggest opportunity in medicine — and what we’re building at CellRep.
By Alfonso Amat, CEO & Co-Founder, CellRep
Some of you know Markov chains from finance. Others from biology. I’ve spent the last two years learning they’re the same thing — and that understanding this equivalence is the key to unlocking cellular rejuvenation.
The core insight is disarmingly simple: aging is not hardware failure. Your DNA — the hardware — is almost perfectly intact in aged cells. What degrades is the operating system: the epigenetic program that tells each cell what to be and how to behave. Fix the software, and the hardware runs like new.
This isn’t science fiction. It’s the logical conclusion of seven decades of Nobel Prize–winning research. And at CellRep, we’re building the tools to do it.
A Frog, a Samurai, and a Sheep
The story begins in 1962, when a British biologist named John Gurdon did something the scientific establishment considered impossible. He took the nucleus from a fully differentiated intestinal cell of a frog and transplanted it into an enucleated egg. The result: a complete, functional tadpole. A mature cell — supposedly locked into its fate forever — had been reprogrammed to create an entire organism.
Gurdon proved that differentiation doesn’t destroy genetic information. Every cell in your body carries the complete blueprint. Aging doesn’t erase it — it just buries it under layers of accumulated noise.
Three decades later, Dolly the sheep confirmed the principle in mammals. Then in 2006, Shinya Yamanaka at Kyoto University made the discovery that would reshape modern biology. He found that just four transcription factors — Oct4, Sox2, Klf4, and c-Myc — could reprogram adult mouse cells back to an embryonic-like state. Mature cells could become pluripotent again. Yamanaka shared the 2012 Nobel Prize with Gurdon for this work.
But full reprogramming had a problem: cells that go all the way back to a stem-like state forget what they are. A neuron that becomes a stem cell is no longer a neuron. Worse, one of the four factors — c-Myc — is an oncogene.
The breakthrough came from an unexpected direction.
The Information Theory of Aging
David Sinclair at Harvard Medical School has spent over two decades building a framework that connects all of these observations. His work began in the late 1990s with yeast sirtuins — proteins that regulate chromatin and are diverted from their posts to repair DNA damage. In 2008, his team showed this mechanism operates in mammals: SIRT1 redistributes across the genome during aging, stabilizing DNA but disrupting gene expression (Oberdoerffer et al., Cell, 2008). The core of what he calls the Information Theory of Aging was already there: chromatin-modifying proteins leave their regulatory positions to fix DNA breaks, and after repeated cycles, not all return to their original locations. The epigenetic landscape gradually degrades. Young cells run clean code; aged cells run corrupted code on the same hardware.
He articulated this fully in his 2019 book Lifespan: Why We Age — and Why We Don’t Have To, and formalized it as a review in Nature Aging (2023). But the critical prediction — that cells retain a “backup copy” of their youthful epigenetic program, and that it should be possible to restore it — was tested directly in 2020.
Sinclair’s lab delivered three of Yamanaka’s four factors — Oct4, Sox2, and Klf4, deliberately excluding the oncogenic c-Myc — to retinal ganglion cells in aged mice. The treatment restored youthful DNA methylation patterns, promoted axon regeneration, and reversed vision loss in both a glaucoma model and naturally aged animals. No tumors. No loss of cell identity. The results, published in Nature (Lu et al., 2020), demonstrated something remarkable: partial reprogramming could rejuvenate complex tissues in vivo.
Then in January 2023, his team provided the most direct experimental proof. Using the ICE (Inducible Changes to the Epigenome) mouse model, they showed that non-mutagenic DNA damage — breaks that are faithfully repaired with zero increase in mutations — still caused accelerated aging at physiological, cognitive, and molecular levels. The hardware was fine. The software had degraded. And OSK reprogramming reversed age-associated changes, reducing epigenetic age by up to 57% across multiple clocks (Yang et al., Cell, 2023).
The implication was profound: aging is, at its core, an information problem. And information problems have information solutions.
From Theory to Therapeutics: The Landscape in 2026
This is no longer a theoretical conversation. As of early 2026, the field has crossed a decisive threshold.
In January 2026, Life Biosciences received FDA clearance for the first-ever human clinical trial of partial epigenetic reprogramming — their ER-100 gene therapy targeting optic neuropathies. This is widely considered the most significant milestone in the reprogramming field since Yamanaka’s original discovery.
Altos Labs, founded with $3 billion and staffed by scientists including Nobel laureate Yamanaka himself, published a landmark Cell paper in 2025 identifying a pervasive “mesenchymal drift” across more than 40 human tissue types during aging — reversed by partial reprogramming. NewLimit, backed by $130 million in Series B and a $45 million investment from Eli Lilly, demonstrated that partial reprogramming can restore youthful killing activity in aged human CD8+ T cells — the first such demonstration in human immune cells. And in June 2025, AbbVie acquired Capstan Therapeutics for up to $2.1 billion, validating in vivo cell engineering as a major therapeutic modality.
Longevity biotech financing reached $8.49 billion in 2024 — a 220% increase from the prior year — with reprogramming platforms alone capturing over $5 billion in cumulative investment. This is no longer a niche. It is the next frontier of medicine.
Why Small Molecules — Not Gene Editing
Here is where our path diverges from most of the field.
The dominant approaches to cellular reprogramming — AAV-delivered gene therapy, CRISPR-based editing, mRNA delivery — share a fundamental limitation. They are, to borrow an analogy I use often, like debugging software with an ice pick to the motherboard. Precise in theory, but structurally mismatched to the problem.
Cells are not simple switches to be flipped. They are more like supercomputers running complex programs — Markov chains with strong feedback loops, 30 trillion of them working in coordinated harmony. Changing Markov chains with robust feedback loops requires more than toggling a single switch. You need to modify the switch and the thousands of downstream molecules it influences.
Gene editing changes one element permanently. RNA therapies deliver transient signals through complex delivery systems. Neither naturally captures the coordinated, multi-pathway modulation that biological systems actually require.
Small molecules do.
Consider: HIV treatment required triple therapy — three drugs creating pressure on different parts of the viral lifecycle simultaneously. Single-target interventions failed. The combinatorial approach succeeded. The same logic applies to cellular reprogramming. A 2019 study in PNAS showed that combining rapamycin, trametinib, and lithium in Drosophila produced a 48% lifespan extension — far exceeding what any single drug achieved alone. And a 2025 Aging Cell paper demonstrated that AI-designed polypharmacological compounds extended lifespan in over 70% of cases, compared to a 0.07% hit rate from unbiased screening — a thousand-fold improvement.
Small molecules are reversible, dose-controllable, combinable, scalable, non-immunogenic, and free of genomic integration risk. They don’t force cells into new states — they guide cells through their own natural restoration pathways. This is what we call epigenetic steering: not correcting code one byte at a time, but writing a software patch that lets the system restore itself.
This approach has deep scientific validation. In 2013, Deng Hongkui’s group at Peking University achieved something most thought impossible: reprogramming mouse fibroblasts to pluripotency using only small molecule cocktails — zero genetic manipulation. By 2022, they extended this to human cells (Nature). And by 2025, they demonstrated chemical reprogramming of human blood cells from a single fingerstick (Cell Stem Cell). The trajectory is clear: small molecules can do everything transcription factors can, with better safety and scalability.
Where We Started: An Immunologist’s Discovery
CellRep’s origin story doesn’t begin with longevity. It begins with a clinical observation.
My co-founder, Dr. Federico Perdomo-Celis — MD/PhD trained at Institut Pasteur — spent years studying HIV Elite Controllers, the roughly 1% of HIV-infected individuals whose immune systems naturally suppress the virus without medication. What he discovered was striking: their CD8+ T cells weren’t simply “younger” by chronological measures. They occupied an earlier, more potent position in the differentiation pathway — a memory-like state with superior survival, proliferative capacity, and killing function.
In Markov chain terms, their cells had higher transition probabilities toward stemness and lower probabilities toward terminal differentiation. They were, functionally, stuck in a better neighborhood of the state space.
Federico’s next question changed everything: what if we could pharmacologically guide any T cell backward to this state?
He did it. Using small molecules, he shifted the distribution — reshaping the entire bell curve of a T cell population toward younger, more functional phenotypes. The transition probabilities changed. The biology followed.
What We’re Building
CellRep is an AI-powered small molecule platform company. Our AI system — 8+ models trained to navigate molecular state spaces — identifies compounds and combinations that optimize backward movement through the differentiation chain. We don’t guess. We model millions of transitions and predict which molecular cocktails will produce the desired epigenetic shift.
Our immediate application: enhancing engineered T cell therapies, including CAR-T, for solid tumors.
This is not a speculative target. It is the single largest unmet need in immuno-oncology. Seven FDA-approved CAR-T therapies exist today — every one of them for blood cancers. Not a single CAR-T has been approved for solid tumors, which account for approximately 90% of all cancers. (Two non-CAR-T cell therapies — a TIL therapy for melanoma and a TCR therapy for synovial sarcoma — have received FDA approval for solid indications, but the core CAR-T barrier remains unsolved.) The reason is well-established: T cell exhaustion. Inside the hostile microenvironment of a solid tumor, CAR-T cells rapidly lose function through a process that is fundamentally epigenetic — a progressive remodeling of chromatin toward a dysfunctional state. Checkpoint inhibitors like anti-PD-1 can transiently reactivate these cells, but they do not reverse the underlying epigenetic program (Pauken et al., Science, 2016).
Our lead compound demonstrates a 6.6-fold improvement in T cell polyfunctionality and a 90.5% antitumor capacity in HER2-targeting CAR-T cells versus 62.6% for untreated controls. These results have been validated across three independent labs: Baylor College of Medicine in Houston, IDCBIS in Bogotá, and our own facilities. We hold 2 provisional patents covering our methods and AI platform, with 2 additional filings underway.
Critically, our small molecules are used ex vivo — during the CAR-T manufacturing process — which means they function as manufacturing reagents under 21 CFR Part 1271, not as standalone drugs requiring a separate IND. They are incorporated under the clinical institution’s existing IND. This compresses our path to clinic from the typical 7–10 year drug development timeline to something dramatically shorter.
Our near-term goal: a first-in-human CAR-T trial at Baylor College of Medicine in 2026.
The Longer Vision — With Both Feet on the Ground
I want to be transparent about what is proven and what is speculative.
What is proven: small molecules can epigenetically rejuvenate T cells ex vivo, improving their function against solid tumors. We have the data. We have the patents. We have the clinical pathway. This is what we’re executing on today.
What is speculative — but scientifically grounded: the same epigenetic steering principles that rejuvenate exhausted T cells may extend to broader cellular rejuvenation. T cell exhaustion and aging share the same class of root cause — progressive epigenetic corruption. The molecular machinery is conserved. If you can guide an immune cell backward in the Markov chain, the question becomes: what other cells can you guide?
The longevity market is projected to reach $63 billion by 2035. Economic analysis suggests that slowing aging enough to extend healthy lifespan by just one year would be worth $38 trillion globally (Scott, Ellison & Sinclair, Nature Aging, 2021). With 2.1 billion people over 60 expected by 2050, the demographic pressure is relentless. The science, the market, and the need are converging.
Our ex vivo data today serves as proof-of-mechanism for a future in vivo drug candidate pipeline. Today we enhance CAR-T cells in a dish. Tomorrow, the same platform could identify compounds that rejuvenate cells inside the body. That’s the bridge we’re building — one validated step at a time.
Execution, Not Just Vision
We’ve raised approximately $1M to date. Our team of 11 spans the US, Colombia, and Argentina — including 5 PhDs in immunology (Institut Pasteur, Stanford), AI, and computational biology, with two dedicated researchers embedded at Baylor College of Medicine. Our CCO departed a tenured university position to dedicate 100% to CellRep. We are currently raising a $4.5 million Series Seed to fund our path through first-in-human studies.
We don’t have a billion-dollar war chest. What we have is a platform that works, data that’s been independently validated, a regulatory pathway that’s faster than anyone expects, and a team that’s building this the hard way — from the science up.
Gurdon proved cells can be reprogrammed. Yamanaka gave us the factors. Sinclair explained the information theory. Perdomo found that immune cells can be guided to earlier states. We built the navigation system.
Others attack DNA with ice picks. We write software patches.
We’re not claiming to reverse all aging. We’re helping cells recover lost function — starting with the cells that fight cancer, and building from there.
The question isn’t if we’ll extend healthy lifespan. It’s how — and who will build the tools that make it possible.
Alfonso Amat is CEO and Co-Founder of CellRep Corporation, an AI-powered small molecule platform company developing compounds that enhance engineered T cell therapies for solid tumors. CellRep operates across the US (Houston), Colombia (Bogotá), and Argentina.For more information, visit cellrep.bio
