TLR4 Immune Receptor Has a Hidden Enzyme That Makes Its Own Molecular Glue

Two groups of mice got the same dose of bacterial toxin. Within 52 hours, every mouse in one group was dead. In the other group, which differed by a single amino acid in one protein, every mouse survived. That amino acid sits in TLR4, one of the best-known sentries of the immune system, and it points to a skill the textbooks never mentioned.

First, what TLR4 does. Your body runs into bacteria constantly, and innate immunity is the first line of defense. TLR4 sits on the cell surface and watches for lipopolysaccharide, also called endotoxin, a molecule on the outer coat of many bacteria. The 2011 Nobel Prize in Physiology or Medicine honored pioneering work in this area. When TLR4 spots endotoxin, a segment inside the cell called the TIR domain recruits helper proteins named TIRAP and MyD88, which pass the inflammation alarm down the line.

For decades, textbooks described the TIR domain on animal TLRs as a passive switchboard that just lets proteins dock. That always looked a little odd. In bacteria, plants and some animal proteins, TIR domains turned out to be enzymes that cut a molecule called NAD+. Animal TLRs were treated as the exception.

A paper published in Science this week overturns that exception. The work was led by Zehan Hu's team at Shanghai Jiao Tong University's School of Life Sciences and Biotechnology, with partners at ShanghaiTech University, Westlake University, the Shanghai Institute of Organic Chemistry of the Chinese Academy of Sciences and Henan Normal University.

The researchers noticed that TLR4's TIR domain keeps a glutamate that is conserved across species, the same spot that does the cutting in other TIR enzymes. When they swapped it for glutamine in mice, TLR4 barely responded to endotoxin and inflammatory signals dropped sharply. That produced the life-or-death split in the endotoxin shock test.

The bigger surprise is how the enzyme works. A typical enzyme is like scissors: it cuts one molecule, lets the pieces go and moves on to the next. TLR4's TIR cuts only once. It splits NAD+ into ADP-ribose and then holds on, with the product jamming its own active site. Earlier tests used a fluorescent stand-in for NAD+ and saw no activity at all. Using the natural molecule and sensitive mass spectrometry, the team found the enzyme does cut, but keeps what it makes.

Why keep it? Cryo-electron microscopy of TLR4, TIRAP and MyD88 together showed the ADP-ribose buried right in the seam between TLR4 and TIRAP, gluing the two together so the whole signaling hub holds. Picture a guard who spots an intruder and, instead of just blowing a whistle, cooks up a batch of glue on the spot to fix the alarm to the wall. The glue has to be made on site, too: adding ready-made ADP-ribose didn't get the complex to assemble.

That is what's new. In the classic picture of cell signaling, enzymes make messenger molecules that diffuse away and spread the news. TLR4 does the opposite. Its product stays put and becomes a structural part that holds proteins together, merging enzyme and scaffold in one domain. The team also found the active pocket is built jointly by two different proteins, TLR4 and TIRAP, unlike other known TIR enzymes that pair two copies of the same protein.

Why should anyone outside a lab care? An overactive TLR4 is tied to sepsis and to runaway inflammation in some autoimmune diseases. Sepsis is one of the deadliest conditions in intensive care. Precisely dialing down this enzyme activity might one day cool inflammation without shutting off the immune system altogether. The researchers stress this is basic science and a long way from patients. They also point out that TLR1, TLR2 and TLR6, which rely on TIRAP too, carry the same glutamate, so the trick may not be unique to TLR4.

Original source published: Science, Nov 13, 2026 Beijing time / Nov 12, 2026 US Eastern time