There is a yellow-green jacket hanging behind Nobu Ueki in an old laboratory portrait. He is seated, smiling, in a matching fleece. The scene has none of biotech's familiar theatre: no glass headquarters, no dramatic machinery, no founder posed beneath a logo. It looks like what it was - a working university lab, with outerwear where outerwear goes. Yet the scientist in that modest frame would spend the next decade worrying about a central problem in drug design: how to make a dangerous molecule behave with manners.
Ueki's full name is Nobuhide. He trained as a molecular biologist, earned a PhD from the University of Tokyo, and published his earliest visible work on the curves hidden inside DNA. In the 1990s, at Mitsubishi Chemical, his research moved through enzyme expression, human genes, nuclear proteins, zinc fingers and the molecular traffic that tells proteins where to go. These sound like separate rooms. In retrospect, they share a corridor: the quiet systems that decide what happens inside a cell, and where.
After a year at Yamanouchi Pharmaceutical, Ueki joined Stony Brook University in 2000 as a postdoctoral fellow. He stayed for seventeen years, eventually becoming a research assistant professor in Michael Hayman's laboratory. The lab studied the molecular logic of cancer: transcriptional repressors, differentiation and signaling. Ueki's publication trail from those years follows a protein called Ski through its relationships with Smad proteins, retinoic acid receptors, GATA1, PU.1 and HDAC3. It is patient work about control - what turns a signal down, what keeps a protein stable, what interrupts a cell's decision.
The observation that became a lock
The consequential clue arrived while Ueki was studying a protein to understand cancer formation at the cell level. Across the cancer cells his team tested, two enzymes were unusually active: histone deacetylases, or HDACs, and cathepsin L, a protease. Plenty of researchers would have selected the more convenient enzyme and designed around it. Ueki asked whether a drug could require both.
“We wondered if we could develop drugs to target those two enzymes to make a better anti-cancer drug.”Nobuhide Ueki, 2013
The answer became a two-step chemical gate. The team attached an acetylated lysine mask to puromycin, an antibiotic that shuts down protein synthesis and is violently toxic to cells. HDAC activity removed the acetyl group. Cathepsin L then cleaved the exposed lysine. Only after both steps could the payload emerge. One enzyme was insufficient. The design behaved like an AND gate written in chemistry.
Choosing puromycin made the experiment bracingly honest. A feeble payload might conceal a leaky gate; an unforgiving one would advertise every mistake. In cultured cells, the masked compound killed a range of cancer cell lines with the relevant enzyme activities. In mice carrying human tumor xenografts, the study reported inhibited tumor growth with less effect on normal cells than unmasked puromycin. The paper appeared in Nature Communications in November 2013.
Ueki did not present the prototype as finished. He talked about structural hindrance, pharmacokinetics and the need for refinement. He also saw a second life for the mechanism. A mask built for drugs could, in principle, be attached to imaging agents. More provocatively, it might rescue compounds that pharmaceutical companies had shelved because their toxicity was impossible to contain.
“Sometimes drug companies give up on a drug because it is so toxic. If they can use this technique, they might be able to use their drug again.”Nobuhide Ueki, 2013
A 2016 paper in Theranostics reported a more efficient version of the prodrug and additional preclinical evaluation. In 2018, US Patent 9,872,919 was granted to Ueki and Hayman for the selective anticancer prodrug system. The patent remains active. The underlying logic also traveled beyond the inventors. In 2022, another research team used the two-enzyme approach to uncage a nucleoside and label RNA selectively in cancer cells in vivo. Ueki highlighted that study on LinkedIn, describing it as an extension of AcK-Lock to chemical probes as well as cytotoxic agents.
A startup waits for detachment
In 2016, the year of the follow-up prodrug paper, Ueki founded PRUNUS Therapeutics. It is a Chicagoland biotechnology startup with a small public footprint and a large biological premise. The company says its synthetic-biology platform is developing sustainable therapeutics, with a lead anti-metastatic agent in preclinical development. Its current website turns away from the two-enzyme story and toward a different transient condition: what happens when a cancer cell loses its attachment.
Normal cells are ordinarily poor nomads. Detached from the extracellular matrix or neighboring cells, they can enter anoikis, a safeguard form of cell death. Metastatic cancer cells may resist that instruction. They survive without anchorage, gather into three-dimensional spheroids and travel. PRUNUS is interested in that interval, when a cell has left one home and has not yet established another.
The company calls its lead candidate :D, pronounced “smiley,” an emoticon standing amid the sober language of preclinical oncology. PRUNUS says the candidate destroys cells growing under attachment-free, three-dimensional conditions while doing nothing to the same cells in an attached two-dimensional monolayer. It also says the candidate does not rely on familiar chemotherapeutic classes such as DNA-damaging agents or microtubule inhibitors. These are preclinical claims, not clinical results. The interesting intellectual move is the same one visible in Ueki's earlier work: do not ask only what the cell is. Ask what state it is in.
That distinction is easy to miss. Most targets are treated as nouns: a receptor, a mutation, an enzyme. Ueki's work often reads in verbs. Enzymes must act. A mask must be removed. A cell must detach. A spheroid must grow. Selectivity appears not as a permanent badge but as a sequence of events. Chemistry waits for the grammar to be correct.
The useful stubbornness of two conditions
Ueki has continued working at the bench alongside the company. His public professional history places him at Northwestern University's Feinberg School of Medicine first as a visiting scholar, then a research specialist, and since 2024 as a research associate. This double life - founder and university researcher - fits a venture whose website still speaks almost entirely in experiments. There is no celebrity-founder mythology, only mechanisms, cell models and diagrams.
His own statement of ambition is unusually direct. He wants effective, safe and affordable strategies that harness the body's defenses, and a platform that gives more people a fair chance at better treatment. The language is broader than any one compound. It explains why rescuing a discarded molecule appealed to him: invention need not always mean creating another payload. Sometimes the valuable invention is permission.
There is also a practical lesson here for anyone building a selective system outside biology. When one signal produces too many false positives, the elegant response is not always a more elaborate interpretation of that signal. It may be a second independent condition. Ueki's molecular gate was strict because it forced two pieces of evidence to agree. PRUNUS' current program adds a temporal condition: the same cell is treated differently depending on how it is growing.
Of course, strict gates impose their own price. Every added condition can exclude a true positive. Every preclinical result must survive the harder worlds of formulation, dosing, safety and human biology. Ueki understood that in 2013 when he called the first compound prototypic. The candor matters. A precise idea is not the same thing as a finished medicine, and an active patent is not a clinical milestone.
Still, the through-line is difficult to ignore. From nuclear transport in Japan to transcriptional control at Stony Brook, from two enzymes to attachment-free growth, Ueki keeps returning to cellular permission systems. He looks for the moment a cell reveals a rule about itself, then asks whether chemistry can listen without interrupting everyone else.
That may explain the unusual patience of his record. The first bent-DNA paper appeared in 1991. The selective prodrug paper arrived twenty-two years later. The company followed three years after that. Nothing in the sequence resembles a sudden pivot into biotechnology. It resembles accumulation: a researcher learns how DNA bends, how proteins enter a nucleus, how transcription is restrained, how enzymes expose a chemical bond, and how a cell behaves when its physical support disappears. A startup can begin with a pitch deck. PRUNUS appears to have begun with a long inventory of biological exceptions. Ueki's bet is that one of those exceptions can be made useful without losing the restraint that made it interesting.
The old portrait catches him smiling in that lime fleece, years before PRUNUS and its cheerfully named candidate. Behind him, the jacket hangs ready for weather. Good laboratory work has a similar modesty. It prepares for conditions. It waits for the right signals. Then, if the locks agree, it acts.