Cells let nearly half of knot-prone misfolded proteins slip past degradation

The mechanism is topological rather than chemical. A stretch of the amino acid chain loops back on itself, a free end threads through the loop, and the resulting entanglement is either part of the protein's working shape or a mistake ^[3]. Yang Jiang's group chose proteins where the entanglement belongs, on the grounds that their earlier work identified those as the ones most likely to get it wrong ^[4].

The study did no new bench work. The team took an existing dataset of proteins carrying a degradation marker in human fibroblast cells and cross-referenced it against a structural database ^[5], one of four datasets originally gathered for other purposes and reused under the NSF synthesis center O'Brien directs at Penn State ^[6][2]. That design buys genome-scale coverage and gives up experimental control, which is the trade worth keeping in mind when reading the numbers.

The 93% figure is an association between a structural feature and a tag ^[7], or 1.93 times the tagging rate of unentangled proteins ^[12]. The simulation result moves two variables at once: it compares tagged proteins with entanglements against untagged proteins without them, and finds the former four times more likely to misfold ^[9]. That comparison shows the tagged-entangled population is enriched for misfolding, though it doesn't isolate what the entanglement alone contributes.

Whether an untagged protein was actually left alone is a separate question the data can't answer. Absence of one degradation marker, in one cell type, in a dataset built for another question, is not the same as persistence in a living cell, and aggregation itself isn't measured here. The step from an untagged remainder to disrupted protein turnover and a contribution to aging is the researchers' extrapolation, and they present it as one ^[10].

There is also an arithmetic check worth running, with its assumptions on the table. If roughly half of entangled proteins go untagged, the tagging rate for that class is near 50%; divide by 1.93 and the unentangled rate lands near 26%, leaving about three-quarters of ordinary proteins untagged as well ^[13]. Read that way, entanglement raises your odds of being caught, and the leak is a property of the system rather than a quirk of knots. The account as supplied does not state the denominators, so treat this as a constraint on interpretation rather than a finding. Phys.org's headline renders the result as "as much as half," while the text says nearly half ^[14], and the gap between those two phrasings is roughly the precision the underlying estimate can support.

One detail deserves more attention than the percentages. Young proteins were already marked, some while still being synthesised ^[8], which puts part of the quality-control decision upstream of the finished molecule and closer to the folding process itself. That is a mechanistic lead, and it is testable at the ribosome.

My read, with conditions: if the escape fraction holds up in a second cell type, the useful target becomes fold topology rather than any single aggregation-prone disease protein ^[11]. If it does not, this is a careful fibroblast result about a newly described misfolding class, which is still worth having.