From plants to fish to humans, microRNAs — arguably the most sought-after molecules in genetics — work by interfering with the expression of complementary mRNAs. Now, Hervé Vaucheret and colleagues show that microRNAs (miRNAs) are not averse to biting the hand that feeds them: the Arabidopsis thaliana ARGONAUTE PROTEIN 1 (AGO1) is needed for the proper function of miRNAs and is itself regulated by a miRNA. Proving the existence of this negative feedback involved engineering artificial miRNAs, potentially spurring a new approach to targeted gene silencing.

Circumstantial evidence indicated that the AGO1 protein worked in the miRNA pathway, specifically as a component of the RNA-induced silencing complex (RISC), the ribonucleoprotein complex in which miRNAs act. Animal AGO homologues are found in the RISC, but it was the combination of genetic analysis of ago1 mutants in A. thaliana and some clever sequence manipulations that provided the proof.

Vaucheret and colleagues found that mRNAs that are normally targeted for cleavage by miRNAs accumulate in ago1 null mutants; this indicates that AGO1 could be needed for proper miRNA function and that this is the only AGO family member — out of ten — to be predominantly associated with miRNAs in the plant RISC.

Among the mRNAs that accumulate in the mutants is the AGO1 transcript itself, prompting the idea that AGO1 mRNA is negatively regulated by a miRNA. A miRNA that is complementary to AGO1 (miR168) suggested itself — correctly, as it turns out — as the negative-feedback regulator. An otherwise wild-type AGO1 gene was engineered to reduce its complementarity to miR168; this increased the levels of AGO1 mRNA and caused developmental defects that resemble those of dcl1, hen1 or hyl1 mutants that are impaired at other steps in the miRNA pathway. To be certain that AGO1 mRNA was regulated by miR168, the authors reversed the defects by generating compensatory mutations in the miR168, so that it now bound to the mutant ago1 mRNA and presumably allowed its normal degradation.

As the authors note, further experiments will be needed to prove that AGO1 is in the RISC, and, if it is, that it is the only AGO family member to be involved in miRNA function.