But there are important considerations when modifying a decoy to engineer novel pathogen recognition specificities. First, for an immune response to be effective and timely, all components involved in perception should localize to the same subcellular compartment. Indeed, whereas the RPS5/PBS1 complex localizes to the plasma membrane (PM), the TuMV-Nla protease mainly resides in the nucleus [5]. Hence, though cleavage of PBS1TuMV by TuMV-NIa activates RPS5, the induction of the immune response is delayed to a late stage of infection, when enough TuMV-Nla protein has accumulated in the cytoplasm and the virus has spread. As a result of this, Arabidopsis lines overexpressing PBS1TuMV suffered from trailing necrosis upon infection with TuMV, a phenomenon wherein HR trails the spread of the virus. Kim et al. suggest targeting RPS5/PBS1 to the same subcellular location as TuMV-NIa protease to circumvent this issue. However, several studies have shown that subcellular localization is important for the proper activation of multiple NLR proteins, including RPS5 5 and 6. Targeting RPS5 to a different subcellular compartment could impair its downstream signaling. Second, there must be constraints as to how much PBS1 can be altered without triggering auto-immune responses. For example, engineering PBS1 to recognize a pathogen-derived protease with a similar substrate specificity as a host protease would inevitably trigger auto-immunity. In addition, a slight alteration of the PBS1 sequence may be enough to trigger HR. For instance, inserting five amino acids in the AvrPphB cleavage site triggers auto-immune responses [4]. Finally, in case an effector target has an important physiological function, modifications of such an effector target should not impair its original function or interfere with the wild-type effector target when both alleles are present.