The functional universe of membrane contact sites

Abstract

Organelles compartmentalize eukaryotic cells, enhancing their ability to respond to environmental and developmental changes. One way in which organelles communicate and integrate their activities is by forming close contacts, often called 'membrane contact sites' (MCSs). Interest in MCSs has grown dramatically in the past decade as it is has become clear that they are ubiquitous and have a much broader range of critical roles in cells than was initially thought. Indeed, functions for MCSs in intracellular signalling (particularly calcium signalling, reactive oxygen species signalling and lipid signalling), autophagy, lipid metabolism, membrane dynamics, cellular stress responses and organelle trafficking and biogenesis have now been reported.

Fig. 1 |. Functions of membrane contact sites.

The different functions of membrane contact sites are each depicted by a representative image, with functions grouped into the four broad categories of signalling (pink circles), regulation of organelle membrane dynamics (blue circles), lipid transport (orange circle) and metabolic channelling (green circle). In some cases, specific examples of the function are shown, but for lipid transport, metabolic channelling and calcium signalling, a depiction of the general principle is given. ATG2, autophagy-related protein 2; DAG, diacylglycerol; ER, endoplasmic reticulum; GRP75, 75-kDa glucose-regulated protein; InsP₃, inositol 1,4,5-trisphosphate; IP3R, inositol 1,4,5-trisphosphate receptor; MCU, mitochondrial calcium uniporter protein; P, phosphate group; PI, phosphatidylinositol; PI4P, phosphatidylinositol 4-phosphate; PI(4,5)P₂, phosphatidylinositol 4,5-bisphosphate; PTP1B, protein-tyrosine phosphatase 1B; ROS, reactive oxygen species; RTK, receptor tyrosine kinase; VDAC, voltage-dependent anion-selective channel proteins.

Fig. 2 |. Diversity of membrane contact sites.

A | Types of unusual membrane contact sites (MCSs). Aa | Intraorganelle MCSs can form between two regions or compartments of the same organelle; an endoplasmic reticulum (ER)–ER contact, formed by atlastin, dynamin-like GTPases, is depicted. Ab | Lipid droplets form two types of contacts: bridging contacts, where the phospholipid monolayer on the surface of the droplet is continuous with the cytoplasmic leaflet of the ER, and standard lipid droplet–ER contacts. Bridging contacts are marked, and may be stabilized by the ER protein seipin. Standard ER–lipid droplet contacts can be formed by the Rab18-dependent tether NRZ (a complex of NAG, RINT1 and ZW10; known as the Dpl1 complex in Saccharomyces cerevisiae), which binds to Rab18 on the surface of lipid droplets and to Q-type soluble N-ethylmaleimide-sensitive factor attachment protein receptors in the ER. Ac | Three-way contacts can form between organelles; a contact between a lipid droplet, the ER and a vacuole in yeast is shown here. This contact is mediated by Mdm1, which contains a phosphoinositide-binding domain (PX) that binds the vacuole and a domain associated with PX domains (PXA) that binds the lipid droplet. B | Some tethering proteins move between MCSs in response to environmental stresses or changes in nutritional status. For example, in yeast, during ER stress or when ceramides accumulate in the ER, Nvj2 moves from its primary location at the nucleus–vacuole junction to ER–Golgi complex contacts, increasing the number of these contacts. Nvj2 harbours a pleckstrin homology (PH) domain, which binds phosphoinositides, and a synaptotagmin-like mitochondrial lipid-binding (SMP) domain. SMP domains can transfer lipids between membranes at MCSs.

Fig. 3 |. Phosphoinositide metabolism at membrane contact sites.

The role of membrane contact sites (MCSs) in phosphoinositide-based signalling is to generate or deplete phosphoinositides required for signalling. This regulation has been best characterized at endoplasmic reticulum (ER)–plasma membrane and ER–Golgi complex MCSs, shown here, and probably also occurs at MCSs between the ER and endosomes. For example, phosphatidylinositol (PI) is generated in the ER and transported to the plasma membrane or Golgi complex, where it is converted into phosphatidylinositol 4-phosphate (PI4P) by the enzymes phosphatidylinositol 4-kinase-α (PI4KA) on the plasma membrane and phosphatidylinositol 4-kinase-β (PI4KB) on the Golgi complex in mammals. PI transport is mediated by lipid transport proteins, including Nir2 and Sec14 in yeast. Some of these proteins use countertransport, in which they move PI4P from the plasma membrane or Golgi complex to the ER and move a second lipid, such as phosphatidylserine (PS) or cholesterol, respectively, in the other direction. When PI4P reaches the ER it can be hydrolysed by the phosphatidylinositide phosphatase Sac1, which converts PI4P into PI. Sac1 may also hydrolyse PI4P in the plasma membrane or Golgi complex; this has been termed ‘working in trans’, since Sac1 is in the ER but acting on a substrate in another membrane at an MCS. On the plasma membrane, PI4P can be converted into phosphatidylinositol 4,5-bisphosphate (PI(4,5)P₂) by the enzyme phosphatidylinositol 4-phosphate 5-kinase (PI4P5K). Oxysterol-binding protein-related protein 5/8 (ORP5/8) may also be able to travel from the plasma membrane to the ER. Enzymes are highlighted in blue. CDS1/2, cytidine diphosphate diacylglycerol synthase 1 and 2; C2CD2L, an ER-resident lipid transport protein (also known as TMEM24); OSBP, oxysterol-binding protein; PA, phosphatidic acid; PIS, phosphatidylinositol synthase; SMP, synaptotagmin-like mitochondrial lipid-binding domain; VAP, vesicle-associated membrane protein-associated protein.

Fig. 4 |. Calcium signalling and signalling in trans at membrane contact sites.

a | Examples of Ca²⁺ signalling between the plasma membrane and endoplasmic reticulum (ER) and between the ER and mitochondria. Stromal interaction molecule 1 (STIM1) senses the Ca²⁺ concentration in the ER lumen. In the resting condition, STIM1 binds Ca²⁺ and is not enriched at ER–plasma membrane membrane contact sites (MCSs) (far left). However, when the Ca²⁺ concentration in the ER decreases, STIM1 accumulates at these contact sites, where it undergoes a conformational change and activates the Ca²⁺ channel calcium release-activated calcium channel protein 1 (ORAI1) in the plasma membrane. This activation causes Ca²⁺ entry at MCSs, which increases the local concentration of Ca²⁺, activating the sarcoplasmic reticulum/ER Ca²⁺-ATPase (SERCA) Ca²⁺ channel in the ER. Thus, Ca²⁺ is channelled from outside the cell to the ER lumen at ER–plasma membrane MCSs in response to ER Ca²⁺ levels, a process known as store-operated calcium entry (SOCE). Other proteins at SOCE-mediating MCSs include the three extended synaptotagmins (E-Syt1–3) and GRAM-containing protein 2A (GRAMD2A). These proteins reside in the ER and bind the plasma membrane via C2 or GRAM domains, respectively. The E-Syts have synaptotagmin-like mitochondrial lipid-binding (SMP) domains that can transfer diacylglycerol (DAG) from the plasma membrane to the ER. The DAG is derived from phosphatidylinositol 4,5-bisphosphate (PI(4,5)P₂), which is hydrolysed by phospholipase C (PLC) when PLC is activated during SOCE. Hydrolysis also yields inositol 1,4,5-trisphosphate (InsP₃), which activates InsP₃ receptors (IP3Rs; Ca²⁺ channels in the ER). Ca²⁺ is channelled from the ER lumen to mitochondria, via an ER–mitochondria tether formed by voltage-dependent anion-selective channel protein (VDAC) in mitochondria, the ER Ca²⁺ channel IP3R and the chaperone 75-kDa glucose-regulated protein (GRP75). This tether allows Ca²⁺ entry into the space between the outer mitochondrial membrane and the inner mitochondrial membrane. The mitochondrial calcium uniporter (MCU) complex in the mitochondrial inner membrane allows Ca²⁺ entry into the mitochondrial matrix. b | Two examples of signalling in trans at MCSs, which occurs when an enzyme in one compartment operates on substrates in a second, are shown. Protein-tyrosine phosphatase 1B (PTP1B), which is in the ER, can dephosphorylate receptor tyrosine kinases (RTKs) in other organelles at ER–plasma membrane and ER–endosome MCSs. The phosphoinositide phosphatase Sac1 is a phosphatase in the ER but it can hydrolyse phosphatidylinositol 4-phosphate (PI4P) to phosphatidylinositol (PI) in the plasma membrane or Golgi complex at MCSs. In the Golgi complex, it is activated by PI4P adaptor protein 1 (FAPP1).

Fig. 5 |. Examples of autophagy at membrane contact sites.

a | Autophagosomes form at specialized domains of the endoplasmic reticulum (ER) called ‘omegasomes’ in mammalian cells and at analogous sites near ER exit sites in yeast. These specialized domains form a membrane contact site (MCS) with a nascent autophagosome, called an isolation membrane (IM). Lipid synthesis in the ER at these MCSs seems to play an important role in autophagosome growth, and lipid-synthesizing enzymes, such as phosphatidylinositol synthase (PIS), are enriched at these sites. This enzyme converts phosphatidic acid (PA) into phosphatidylinositol (PI), which is in turn used to produce phosphatidylinositol 3-phosphate (PI3P), which is required for autophagosome formation. The lipids may be transferred from the ER to the nascent autophagosomes by autophagy-related protein 2 (ATG2). Autophagosome formation is also facilitated by an ER–autophagosome tether formed by vesicle-associated membrane protein-associated proteins (VAPs), Unc-51-like autophagy activating kinase 1 (ULK1) and focal adhesion kinase family kinase-interacting protein of 200 kDa (FIP200). Vacuole membrane protein 1 (VMP1) is also enriched at these MCSs, where it seems to promote the activity of sarcoplasmic reticulum/ER Ca²⁺-ATPase (SERCA) and may regulate lipid metabolism at these sites. b | During starvation in Saccharomyces cerevisiae a specialized form of selective mitophagy, called ‘piecemeal microautophagy of the nucleus’ (PMN), degrades parts of the nuclear membrane. Pieces of the nuclear membrane are removed and internalized into the vacuole (the equivalent organelle to the mammalian lysosome), where they are degraded. The process occurs at MCSs between the nucleus and the vacuole termed the ‘nucleus–vacuolar junction’ (NVJ).

Fig. 6 |. Roles of membrane contact sites in cellular stress responses.

Membrane contact sites (MCSs) play roles in the cellular response to stress, as depicted for the main stresses discussed in this Review. Examples of the roles of MCSs for each type of stress are shown. a | Lipid stress. When ceramides accumulate in the endoplasmic reticulum (ER) of Saccharomyces cerevisiae, MCSs between the ER and the Golgi complex increase. The formation of these MCSs, which facilitate the non-vesicular movement of ceramide out of the ER to alleviate stress, requires nucleus–vacuole junction protein 2 (Nvj2), an ER-resident protein that binds the Golgi membrane via a pleckstrin homology (PH) domain. Nvj2 also contains a synaptotagmin-like mitochondrial lipid-binding (SMP) domain that may facilitate ceramide transport from the ER to the Golgi complex. b | Mechanical stress. In plants, synaptotagmin 1 (SYT1) and microtubules at ER–plasma membrane MCSs help stabilize the plasma membrane (mechanical resistance). During mechanical stress, SYT1 forms aggregates in leaf cells. In the absence of SYT1, microtubules fail to form and the plasma membrane is more susceptible to mechanical stress and rupture. c | Nutrient stress. In S. cerevisiae, the ER–mitochondria encounter structure (ERMES) tethers the ER and mitochondria. During nutrient stress such as nitrogen starvation, some mitochondria are degraded by a form of selective autophagy known as mitophagy. An isolation membrane grows from sites near the ERMES, perhaps in the ER, that will eventually engulf the mitochondria and then fuse with vacuoles (the equivalent of lysosomes in yeast), where the mitochondria will be degraded. A similar process occurs in mammalian cells, although they lack the ERMES. d | ER stress, oxidative stress and apoptosis. During reactive oxygen species (ROS)-mediated ER stress, MCSs between the ER and mitochondria are increased, the protein composition of contacts changes, and a number of signalling pathways and stress response proteins are induced at these MCSs. One protein that becomes enriched at these sites is protein kinase RNA-like ER kinase (PERK), which reduces mitochondrial motility (left panels). When stress is prolonged, unfolded proteins in the ER and other signals increase Ca²⁺ entry into mitochondria at ER–mitochondria MCSs, promoting apoptosis (right panels). Ca²⁺ entry requires a tether composed of the ER Ca²⁺ channels, inositol 1,4,5-trisphosphate receptors (IP3Rs), voltage-dependent anion-selective channel protein (VDAC) in the mitochondrial outer membrane and 75-kDa glucose-regulated protein (GRP75). The mitochondrial calcium uniporter (MCU) in the mitochondrial inner membrane allows Ca²⁺ entry into the mitochondrial matrix. In unstressed cells, antiapoptotic proteins such as B cell lymphoma 2 (BCL-2) bind and inhibit IP3Rs, reducing Ca²⁺ signalling. Removal of BCL-2 increases Ca²⁺ entry into mitochondria, promoting apoptosis. During ER stress, ER chaperones such as heat shock protein 70 family protein 5 (BiP) bind unfolded proteins and stimulate IP3Rs, increasing Ca²⁺ signalling.