A 90-amino-acid peptide encoded by the circular RNA circTLL1 drives resistance to the third-generation EGFR inhibitor osimertinib in non-small cell lung cancer by promoting degradation of the metabolic enzyme NT5C2, raising GTP, and sustaining Ras/PI3K/AKT signaling. New research shows the peptide,…
Researchers have identified a 90-amino-acid peptide, designated circTLL1-90aa , that drives resistance to the third-generation EGFR tyrosine kinase inhibitor osimertinib in non-small cell lung cancer NSCLC . The peptide is encoded by a circular RNA, circTLL1 , that is upregulated in osimertinib-resistant NSCLC cells. Once translated, circTLL1-90aa binds the enzyme NT5C2 , the cytosolic 5'-nucleotidase II, and promotes its degradation. Reduced NT5C2 leads to higher GTP levels, which sustain activation of the Ras/PI3K/AKT signaling pathway. The researchers propose circTLL1-90aa as a predictive biomarker and an actionable therapeutic target for osimertinib resistance.
The clinical stakes are direct. Osimertinib is the standard first-line treatment for EGFR-mutant NSCLC, and acquired resistance eventually limits its benefit in most patients. Known resistance routes include secondary EGFR mutations, activation of parallel receptor tyrosine kinases, and histologic transformation. The newly described cascade is distinct: it begins in a class of RNA once considered non-coding, produces a small protein, and acts through a nucleotide-metabolism enzyme rather than through direct modification of the drug's target.
The study also extends the functional repertoire of circular RNAs, which have been studied mainly as microRNA sponges. Here the investigators demonstrate that circTLL1 contains an open reading frame, that the frame is translated into a 90-amino-acid protein, and that the protein interacts with a metabolic enzyme to produce drug resistance. The work links a circRNA, a micropeptide , a metabolic enzyme, and an oncogenic signaling pathway in a single resistance mechanism. The findings were reported on August 18, 2026.
The chain of evidence begins with expression. Using osimertinib-resistant and parental NSCLC cell lines, the team profiled circular RNA expression and found circTLL1 stably and significantly upregulated in the resistant cells. The change was stable, which led the investigators to treat circTLL1 as a candidate driver rather than a transient stress response.
Functional experiments established that the circRNA is not merely associated with resistance but required for it. Overexpression of circTLL1 in sensitive cells promoted osimertinib resistance. Knockdown of circTLL1 in resistant cells restored drug sensitivity, and those results held both in vitro and in the xenograft mouse models, showing that the effect is not an artifact of culture conditions.
Mechanistically, the investigators traced the phenotype to the peptide. circTLL1 carries an open reading frame that is translated into circTLL1-90aa, a protein of exactly 90 amino acids. Co-immunoprecipitation and mass spectrometry identified NT5C2 as its direct interaction partner. Binding promotes NT5C2 degradation and lowers NT5C2 protein levels. Downregulation of NT5C2 leads to higher GTP levels, and higher GTP levels support sustained activation of the Ras/PI3K/AKT pathway. That sustained signaling keeps resistant cells alive in the presence of osimertinib. Each step of the cascade, from translation of circTLL1 to activation of the pathway, was tested in the study.
The study is a laboratory investigation built around paired cell populations. The researchers established osimertinib-resistant NSCLC cell lines alongside their parental counterparts and compared their circRNA expression profiles using high-throughput RNA sequencing. They then ran functional and mechanistic experiments in vitro and in vivo: cell viability assays, apoptosis analysis, xenograft mouse models, mass spectrometry, co-immunoprecipitation, and western blotting.
The endpoints measured were:
The design is well suited to establishing mechanism. Overexpression and knockdown experiments provide direct evidence of causality: raising circTLL1 levels confers resistance, and lowering them restores sensitivity. The xenograft data extend the finding from culture dishes to a living tumor environment. The results were directionally consistent, with every manipulation moving the phenotype in the predicted direction.
What the design cannot establish is clinical scope. The experiments used cultured cell lines and mouse xenografts, not patient tumors. No patient cohort was included, and sample size and study duration were not reported. The prevalence of circTLL1 upregulation in clinical resistance, its association with time to progression, and its value as a biomarker in human tissue therefore remain untested.
Circular RNAs are products of back-splicing, in which the 3' end of an exon is joined to an upstream 5' end to form a covalently closed loop. Because these molecules lack the 5' cap and poly A tail of canonical mRNAs, they were long classified as non-coding and studied chiefly as microRNA sponges, as scaffolds for protein complexes, and as regulators of transcription. The discovery that some circular RNAs carry translatable open reading frames is recent. Translation of an uncapped circular template proceeds through cap-independent mechanisms such as internal ribosome entry sites, and the products are often small.
At 90 amino acids, circTLL1-90aa falls in the micropeptide range, the class of small proteins produced from short open reading frames in RNAs once annotated as non-coding. What distinguishes this micropeptide is its biochemical partner. NT5C2 dephosphorylates nucleoside monophosphates such as GMP, converting them to nucleosides and helping to balance the intracellular nucleotide pool. When circTLL1-90aa binds NT5C2 and sends it toward degradation, that catabolic activity is lost. With the enzyme gone, the guanine nucleotide pool shifts toward accumulation, and GTP levels rise.
The raised GTP level connects the metabolic change to signaling. Ras proteins are molecular switches that signal when bound to GTP. A higher GTP pool favors sustained Ras activation, and Ras signals through PI3K and AKT to drive survival and proliferation. In an EGFR-mutant cell treated with osimertinib, the EGFR node is blocked, but this pathway remains switched on, providing a bypass that is the functional signature of resistance. The cascade thus unites the non-coding genome, a metabolic enzyme, a nucleotide, and a classical oncogenic pathway in a single mechanism of drug resistance.
For peptide researchers, the study is a concrete case of a functional micropeptide with a defined disease phenotype. circTLL1-90aa has a known binding partner and a measurable downstream effect, which makes it tractable for biochemical and structural work. Synthetic circTLL1-90aa could serve as a reference standard for assays, as bait in interaction studies, and as an immunogen for raising antibodies. Its small size makes synthesis and modification practical, and the defined NT5C2 interaction surface offers a screening target for molecules that could block the binding and protect the enzyme.
For clinicians, the proposal is twofold. First, circTLL1 or circTLL1-90aa could serve as a predictive biomarker measured in tumor tissue at diagnosis or at progression. Biomarker claims of this kind require regulatory qualification, which begins with analytical validation of the detection assay and continues with clinical validation in prospective cohorts. Second, because the knockdown experiments that established causality also restored drug sensitivity, the circRNA and its peptide product are candidate therapeutic targets. Nucleic acid-based silencing of the circRNA, agents that block the micropeptide/NT5C2 interaction, or strategies that stabilize NT5C2 are testable directions, though none has been developed.
For the supply chain, the immediate products are research reagents: peptide reference standards, antibodies selective for circTLL1-90aa, and detection probes for the circular RNA. If the biomarker survives clinical validation, diagnostic developers will need well-characterized calibrators and controls, and the 90-amino-acid peptide becomes a specification rather than a research-only reagent. If the target survives therapeutic validation, the micropeptide/NT5C2 interface becomes a candidate drug target. Both paths run through the same bottleneck: confirmation in patient samples.
The study is internally consistent, but its external validity is unproven. Laboratory-derived resistance models, however carefully constructed, capture only a subset of the biology of patient tumors, which carry heterogeneous genetics, immune environments, and treatment histories. The frequency of circTLL1 upregulation in clinical osimertinib resistance is unknown, and no data yet link the axis to patient outcomes.
Several molecular questions remain open. The study does not explain what drives circTLL1 upregulation in resistant cells, whether transcriptional activation, RNA stabilization, or altered splicing is responsible. It identifies NT5C2 as the direct interaction partner of circTLL1-90aa, but whether NT5C2 is the only relevant substrate is untested. And although higher GTP levels are linked to sustained Ras/PI3K/AKT signaling, the precise route from a larger GTP pool to Ras activation, and whether other GTP-dependent GTPases participate, has not been mapped.
What would settle these questions is also clear. Paired tumor biopsies from NSCLC patients taken before osimertinib treatment and at progression, measured for circTLL1 and circTLL1-90aa, would establish whether the axis operates in humans and whether it correlates with outcome. Patient-derived xenograft and organoid models would test causality in clinically realistic genetic backgrounds. Structural studies of the circTLL1-90aa/NT5C2 interface would determine whether the interaction is druggable. Interventional studies of agents targeting the circRNA or the micropeptide would test whether the pathway is clinically actionable. Until then, circTLL1-90aa stands as a well-defined mechanism in model systems and a credible candidate for translation, not yet a validated clinical marker.
Related reading: GLP-1 Retinopathy Meta-Analysis Finds No Class-Wide Risk Signal, FDA's Compounded GLP-1 Warning Letters and the Embody Telehealth Model, GLP-1 therapy tied to rapid gains in mood and sleep in obesity, QCM-D analysis maps oxyntomodulin fibril stability and release kinetics.