A three-way organelle junction controls PI(4)P metabolism and mitochondrial division

Membrane contact sites (MSC) serve critical cell physiological functions by enabling communication between organelles, for example by exchanging lipids or ions and by regulating membrane remodeling through facilitation of membrane fusion, fission, or motor protein-dependent transport processes. At MCS, the membranes of distinct organelles lie in close apposition (typically 10–30 nm) due to physical connection via tethering factors that frequently operate in conjunction with membrane lipids (1). An example of this architecture are MCS between the plasma membrane (PM) and the endoplasmic reticulum (ER) that rely in part on recognition of phosphatidylinositol 4,5-bisphosphate [PI(4,5)P₂], a signaling lipid enriched at the PM, by ER transmembrane protein tethers (2). The ER also forms MCS with essentially all other organelles of eukaryotic cells including mitochondria, endosomes, lysosomes, peroxisomes, lipid droplets, and the trans-Golgi network. In addition to MCS between two organelles, recent studies have uncovered the presence of tripartite MCS, for example between the ER, mitochondria, and the plasma membrane in yeast or between endosomes or lysosomes, mitochondria, and the ER (3–5). Recent work has further shown that MCS formation between organelles often relies on multiple, partially redundant tethers, posing experimental challenges for the functional analysis of MCS by classical genetic approaches (1).

Previous work had identified the large multidomain protein Num1 as a core component of tripartite MCS between mitochondria, the cortical ER, and the PM in yeast (5). How Num1 associates with the ER and whether this is of physiological importance for yeast cells had remained largely unknown. Casler et al. (6) not only solved the riddle as to how Num1 tethers the ER to the cell cortex to regulate mitochondrial division but also unravelled a hitherto unknown function for Num1-mediated formation of tripartite MCS in controlling the distribution and metabolism of PI(4)P on the PM of dividing yeast cells.

Capitalizing on the power of yeast genetics and the tricks of the trade of molecular biology, they identified a so-called FFAT motif within Num1 that can tie the protein to ER membranes by associating with Scs2, the yeast ortholog of mammalian VAP proteins that serve as key tethers for a plethora of ER-based MCS (1, 2). They further show that Num1, likely via its propensity to form clusters, suffices to tether the ER and, thereby, mitochondria to the PM. As the ER is known to facilitate mitochondrial fission (7), the authors probed whether genetic interference with Num1 binding to the ER or loss of Scs2 affect mitochondrial division. Indeed, they observed that cells expressing an ER-binding-defective mutant of Num1 or cells lacking Scs2 displayed comparable defects in mitochondrial division rates. A similar phenotype has been reported for mammalian cells lacking VAP proteins (3). This phenotype most likely reflects a direct role for Num1 in tethering the ER to mitochondria to facilitate membrane fission mediated by the dynamin-like protein Dnm1. A second, presumably fission-independent function of Num1 is to regulate the distribution of the mitochondrial network via tethering mitochondria to the PM. The significance of this role remains to be elucidated.

So, why do yeast cells harbor tripartite MCS between mitochondria, the ER, and the PM? To address this question, Casler et al. (6) analyzed the Num1 interactome. Surprisingly, they found Num1 to associate with the Ypp1, Efr3, and Stt4 subunits of the PM phosphatidylinositol 4-kinase and Osh3, a protein implicated in PI(4)P transport between organelles (2). Given this and the proposed function of PI(4)P in regulating mitochondrial dynamics, they monitored the distribution and turnover of PI(4)P using established sensors. Strikingly, they observed that cells lacking Num1 and, to a lesser extent, cells expressing mutant variants of Num1 selectively defective in ER tethering (i.e., lacking the FFAT motif) or mitochondria-PM tethering (i.e., lacking its pleckstrin homology or coiled-coil domains) failed to maintain the polarized distribution of PM PI(4)P in the bud of dividing yeast cells. Further elegant experiments, for example using a system that allows the conditional inactivation of PI(4)P synthesis on the PM via the targeted degradation of the PI 4-kinase complex, uncovered a selective defect of num1 KO cells in the spatiotemporally regulated turnover of PI(4)P via MCS-dependent lipid transfer to the ER, where the PI(4)P phosphatase Sac1 resides (2). These data suggest a model in which tripartite MCS between mitochondria, the ER, and the PM formed by Num1 are required to maintain PI(4)P polarization on the daughter cell PM by spatially regulating PI(4)P hydrolysis via Sac1. This mechanism appears to be of particular importance during cell division as newly emerging buds, e.g., the sites of polarized cell growth, contain little cortical ER resulting in a long half-life of PM PI(4)P. As cells approach cytokinesis, Num1-mediated formation of tripartite MCS between mitochondria, the ER, and the PM enhances the efficiency of Sac1-mediated PI(4)P turnover to restore steady-state PI(4)P levels. The function of mitochondria in this mechanism remains somewhat enigmatic but might pertain to the ability of mitochondrial Mdm36 protein to induce Num1 clustering.

PI(4)P has been found to be essential for PM homeostasis and growth. In oligodendrocytes, for example, the vast expansion of the PM required to form myelin sheaths around axons depends on local PI(4)P formation (8). Furthermore, a fundamental role of PI(4)P is to serve as the energy currency that drives the anterograde counter-transport of lipids from the ER, the central hub of lipid metabolism, towards the PM against their concentration gradient (9). In this light, the findings of Casler et al. (6) may have wider implications for the regulation of the dynamics and the nanoscale localization of plasma membrane PI(4)P also in higher eukaryotes. The relevance of PM-mitochondria contact sites in mammalian cells is unclear, and no direct ortholog of Num1 appears to exist. Intriguingly, though, delivery of PI(4)P at tripartite lysosome-ER-mitochondria contact sites (3) and from the TGN (10) has been implicated in facilitating mitochondrial division, by as of yet unidentified mechanisms. Whereas this conceivably represents a scenario in which tripartite PM-ER-mitochondria contact sites operate, further studies will have to reveal in how far such three-way junctions are conserved in higher eukaryotes and what their precise functions could be.