Scientists Create Proteins That Switch On Only Where Inflammation Burns

Inflamed tissue speaks a distinctive chemical language, and one of its loudest words is nitric oxide. This small, reactive molecule is produced in abundance by immune cells when they mount an inflammatory response, yet biologists have never been able to harness it directly to control the activity of folded, working proteins. A team of researchers at Peking University, led by Tao Liu, has now changed that. In a study published in Nature Biomedical Engineering, they describe a protein engineering strategy called NOCAGE, which renders proteins temporarily inactive until they encounter nitric oxide, at which point their function is chemically restored. The approach, demonstrated in mice, allows therapeutic proteins to wake up only in inflamed environments, redirects viral gene delivery toward inflamed tissue, and turns engineered probiotics into living diagnostics for intestinal inflammation.

The central trick of NOCAGE lies in a single, carefully chosen amino acid. Many proteins depend on a glutamate residue at a catalytic or structural hotspot; remove or alter that glutamate and the protein loses its activity. The researchers substituted this essential glutamate with a synthetic, non-canonical amino acid whose side chain is chemically masked by a caging group. In this caged state, the protein folds normally but remains functionally silent, because the residue cannot perform its usual role. When the masked amino acid encounters nitric oxide, a chemical reaction strips away the cage, regenerating the native glutamate and switching the protein back on. In effect, the researchers have built a molecular tripwire: the protein stays dormant everywhere except where inflammation has raised nitric oxide levels.

Getting this synthetic amino acid into proteins required overcoming a classic challenge in chemical biology. Non-canonical amino acids are incorporated into proteins through genetic code expansion, a technique that repurposes an orthogonal aminoacyl-tRNA synthetase and tRNA pair to read a stop codon as an instruction to insert the unnatural residue. The team engineered a nitric oxide-responsive synthetase, which they named NorERS, through directed evolution from a pyrrolysyl-tRNA synthetase scaffold. Using fluorescence-activated cell sorting to screen large libraries, they selected variants that efficiently and specifically incorporated the caged glutamate, which they call NorE, into proteins in both bacterial and mammalian cells. Intact protein mass spectrometry confirmed that the incorporation was faithful, and iterative rounds of error-prone PCR mutagenesis and screening pushed the incorporation efficiency to levels comparable with well-established systems.

The chemistry of the caging group itself was equally deliberate. The researchers evaluated several ortho-phthalaldehyde-based caging groups and found that a methoxy-substituted variant reacted with nitric oxide with remarkable efficiency, achieving roughly 95 percent decaging conversion in the presence of a nitric oxide donor. Crucially, the caged amino acid proved resistant to a battery of other reactive oxygen and nitrogen species, physiologically relevant ions, metabolites, serum, and even liver homogenates. This selectivity matters enormously: an inflammation-triggered drug would be useless if it were accidentally activated by hydrogen peroxide or superoxide elsewhere in the body. The team also showed that activation works across a physiologically relevant range of pH values and in the presence of abundant biological nucleophiles such as glutathione, underscoring the robustness of the design.

To demonstrate the breadth of the platform, the researchers applied NOCAGE to an unusually diverse set of proteins. They engineered a nitric oxide-responsive antibody fragment, a luciferase enzyme, the anti-inflammatory cytokine interleukin-10, a bacterial toxin, and viral capsids from adeno-associated virus. In each case, the same logic applied: identify a glutamate or similar residue essential for binding or catalysis, replace it with the caged amino acid, and watch function collapse until nitric oxide restores it. An engineered anti-GFP nanobody lost its binding ability entirely until nitric oxide treatment restored it, and an engineered version of the drug antibody adalimumab, which neutralizes tumor necrosis factor-alpha, was similarly silenced and then reactivated. Even proteins lacking a suitable glutamate could be adapted, because the team extended the approach to caged aspartic acid and showed that substitutions at serine or other positions could also confer nitric oxide responsiveness.

The in vivo experiments are where the strategy reveals its therapeutic ambition. In mouse models of acute liver inflammation induced by lipopolysaccharide, an engineered nitric oxide-responsive luciferase called NO-Rluc lit up specifically in inflamed livers while remaining dark in healthy tissue, providing a real-time readout of nitric oxide production in living animals. Longitudinal imaging tracked the activation of the probe over hours following injection, and blocking nitric oxide synthase with the inhibitor L-NMMA suppressed the signal, confirming that the bioluminescence genuinely reported endogenous nitric oxide rather than an off-target effect. Such inflammation-localized imaging could eventually allow clinicians to visualize where inflammatory activity is concentrated in a patient’s body without invasive biopsies.

The therapeutic implications go beyond imaging. An engineered version of interleukin-10, a cytokine with powerful anti-inflammatory effects that has historically caused systemic side effects in clinical trials, was silenced by NOCAGE and reactivated only in inflamed tissue. In mice with lipopolysaccharide-induced systemic inflammation, the caged cytokine reduced tumor necrosis factor-alpha and interleukin-6 levels, and in a model of chemically induced colitis it ameliorated disease while limiting off-target activity. The team also engineered a nitric oxide-responsive version of Pseudomonas exotoxin A fused to an anti-HER2 nanobody, a construct relevant to cancer therapy, in which toxicity was suppressed until nitric oxide unmasked the catalytic glutamate. Safety evaluations, including cell viability assays across multiple cell lines and four-week dosing studies in mice with serum chemistry and histopathology of major organs, showed no detectable toxicity attributable to the caged proteins themselves.

Gene delivery represents another striking application. Adeno-associated virus vectors are the workhorses of gene therapy, but they transduce cells indiscriminately, which can cause unwanted expression in healthy tissue. By caging a conserved glutamate at position 563 in the AAV2 capsid, the researchers created viral particles that could only infect cells in nitric oxide-rich environments. In mice with inflamed livers, the engineered capsids delivered their genetic payload selectively to inflamed tissue, while wild-type capsids transduced without such discrimination. Sensitivity assays showed dose-dependent activation across a range of nitric oxide concentrations, and the modified vectors showed no increase in neutralizing antibody responses in treated animals. A capsid-based trigger of this kind could one day allow gene therapies to be aimed precisely at diseased, inflamed tissue while sparing the rest of the body.

Perhaps the most imaginative application is diagnostic. The researchers equipped the probiotic bacterium Escherichia coli Nissle, a strain with a long history of safe human use, with a gene encoding the nitric oxide-responsive luciferase. When these engineered biosensors were administered to mice, they produced a luminescent signal only upon encountering nitric oxide in the gut, enabling non-invasive detection of intestinal inflammation in a model of inflammatory bowel disease. Because luminescence from deep tissue is difficult to detect directly, the system’s sensitivity, demonstrated as several-hundred-fold signal increases upon nitric oxide exposure in bacterial cultures, offers a path toward swallowable or implantable living diagnostics that report the chemical state of the gut from the inside.

The authors, including co-first authors Wenkang Cai, Junhao Cui and Zhiying Zeng, describe NOCAGE as a generalizable method for post-translational control of protein function, and the evidence supports that claim. By converting a hallmark inflammatory molecule into a universal activation key, the platform unifies protein therapeutics, gene delivery and biosensing under a single chemical logic. Patent applications covering the technology have been filed by Peking University, and the work was funded by the National Natural Science Foundation of China and related programs. Considerable work remains before caged proteins reach the clinic, including studies of pharmacokinetics, immunogenicity and the nitric oxide thresholds of human disease tissue. But the conceptual advance is clear: proteins can now be engineered to listen for inflammation and respond on cue, opening a route to medicines that act only where and when the body signals that they are needed.

Subject of Research: Engineering nitric oxide-responsive proteins using caged non-canonical amino acids for inflammation-targeted activation

Article Title: Engineering inflammation-responsive proteins through nitric oxide-caged amino acids

Article References: Cai, W., Cui, J., Zeng, Z., Xiang, Z., Xie, Y., Su, Y., Zuo, Y., Liu, Y., Wang, H., Chang, L., Wang, X., Wang, J., Ma, J.-A., & Liu, T. (2026). Engineering inflammation-responsive proteins through nitric oxide-caged amino acids. Nature Biomedical Engineering. https://doi.org/10.1038/s41551-026-01782-9

Image Credits: AI Generated

DOI: 10.1038/s41551-026-01782-9

Keywords: nitric oxide, protein engineering, genetic code expansion, non-canonical amino acids, inflammation, NOCAGE, gene delivery, AAV capsids, biosensors, interleukin-10, inflammatory bowel disease, post-translational control