This one falls under the "news to me" category: the physical arrangement of pathway-specific enzymes in order to increase metabolic productivity. The Dueber lab at UC Berkeley has been generating enzyme platforms for the last several years to circumvent rate-limiting issues such as diffusion and intermediate build-ups that suppress upstream reactions. The idea is to construct a molecular scaffold that displays tags, recruiting specific enzymes to the same subcellular location. This way, a metabolic intermediate is shuttled quickly to the next enzyme, removing some of the random walk serendipity that typically governs non-networked pathways. The initial 2007 paper can be found [here] (abstract below), and a more recent iteration builds similar scaffolds within a larger proteinaceous shell.

Engineered metabolic pathways constructed from enzymes heterologous to the production host often suffer from flux imbalances, as they typically lack the regulatory mechanisms characteristic of natural metabolism. In an attempt to increase the effective concentration of each component of a pathway of interest, we built synthetic protein scaffolds that spatially recruit metabolic enzymes in a designable manner. Scaffolds bearing interaction domains from metazoan signaling proteins specifically accrue pathway enzymes tagged with their cognate peptide ligands. The natural modularity of these domains enabled us to optimize the stoichiometry of three mevalonate biosynthetic enzymes recruited to a synthetic complex and thereby achieve 77-fold improvement in product titer with low enzyme expression and reduced metabolic load. One of the same scaffolds was used to triple the yield of glucaric acid, despite high titers (0.5 g/l) without the synthetic complex. These strategies should prove generalizeable to other metabolic pathways and programmable for fine-tuning pathway flux.