The intricate intracellular infrastructure of all known life forms is based on proteins. The folded shape of a protein determines both the protein's function and the set of molecules it will bind to. This tight coupling between a protein's function and its interconnections in the molecular interaction network has consequences for the molecular course of evolution. It is also counter to human engineering approaches. Here we report on a simulation study investigating the impact of random errors in an abstract metabolic network of 500 enzymes. Tight coupling between function and interconnectivity of nodes is compared to the case where these two properties are independent. Our results show that the model system under consideration is more robust if function and interconnection are intertwined. These findings are discussed in the context of nanosystems engineering.