The morning of January 16th, 2026, was cold and damp. Johan Elf, the supervisor of soon-to-be PhD graduate Jinwen Yuan, shivered, but not from the cold. He had just learned that the opponent for the defence, which was planned to start at 9:15, was stuck on a broken train between Stockholm and Uppsala. Luckily, the delay was not severe, and at 9:30, a slightly flushed Björn Högberg opened the show by introducing and utterly demolishing the central dogma of biology as a completely useless model for anything, really. When Jinwen entered the stage, she did so as the lady of the rings she had ultimately conquered, proving her place as the legitimate queen of LacI. The following discussion between Jinwen and Björn was intense, interesting, and extremely long. Jinwen, however, was not deterred, delivering eloquent and insightful answers until the opponent finally had to surrender. When the questioning of a PhD candidate by the opponent is particularly lengthy, the committée sometimes compensates by going easy. This committée did not get that memo, and when the session was finally over at around 2 pm, it was a group time record by a large margin. The slightly starved audience was, however, generously compensated. Two gigantic “Smörgåstårtor” were waiting in the Öbring room, much to the delight of Johan Elf. Jinwen passed, of course, and the celebration continued well into the night at restaurant Logen. Jinwen sure knows how to throw a party 🥳

Top left: Both the opponent and the defendant seemed to enjoy themselves during the questioning. Top right: Mingle with bubbles and smöråstårta! Bottom left: Jinwen receiving a bound copy of her thesis from the committée, Erik Marklund, Anniina Vihervaara, and Erik Aurell. Bottom right: Jinwen’s thesis cover art.

In her thesis, Jinwen used the classic lac repressor to understand how transcription factors recognize the right target and how their microscopic motions shape that process.

By combining modeling with large‑scale measurements, she found that LacI’s ability to tell DNA sequences apart depends mainly on how efficiently it associates with them, rather than how long it stays bound. Recognition, it turns out, is all about making the right first move. This insight held true across thousands of DNA variants measured on a high‑throughput microarray.

To see what structural features of LacI control this association step, she focused on the “hinge” region, which can switch between folded and unfolded conformations. Based on MD simulations, she engineered two versions of LacI with subtly different hinge behaviors, and these tiny tweaks produced big effects: one mutant became a strong but overly indiscriminate repressor, while the other became a weaker but more selective one. Surprisingly, both searched for their targets inside living cells just as fast as the original protein. This suggests that the search process is robust, even when other properties shift.

Finally, she applied a new method, Optical Pooled Screening, that uses genetic barcodes and single‑molecule imaging to study many LacI variants at once. A pilot test showed that the approach works, but a few design improvements are needed before scaling up.

Read Jinwen’s thesis here