Biochimica et Biophysica Acta - ETH Z · Biochimica et Biophysica Acta xxx (2013) xxx–xxx ☆ This article is part of a Special Issue entitled: Functional and structural diversity

Biochimica et Biophysica Acta xxx (2013) xxx–xxx

BBAMCR-16874; No. of pages: 16; 4C: 2, 5, 7

Contents lists available at SciVerse ScienceDirect

Biochimica et Biophysica Acta

j ourna l homepage: www.e lsev ie r .com/ locate /bbamcr

Review

Organization and function of membrane contact sites☆

Sebastian C.J. Helle 1, Gil Kanfer 1, Katja Kolar 1, Alexander Lang 1, Agnès H. Michel 1, Benoît Kornmann ⁎Institute of Biochemistry, ETH Zürich, HPM G16 Schafmattstrasse, 18 8093 Zürich Switzerland

☆ This article is part of a Special Issue entitled: Functiendoplasmic reticulum.⁎ Corresponding author. Tel.: +41 44 633 65 75.

E-mail address: [email protected] (B1 Equal participation.

0167-4889/$ – see front matter © 2013 Elsevier B.V. Allhttp://dx.doi.org/10.1016/j.bbamcr.2013.01.028

Please cite this article as: S.C.J. Helle, et al., Or10.1016/j.bbamcr.2013.01.028

a b s t r a c t
a r t i c l e i n f o
Article history:Received 22 November 2012Received in revised form 18 January 2013Accepted 21 January 2013Available online xxxx

Keywords:Membrane contact siteOrganelleEndoplasmic reticulumCalciumLipidInterorganelle communication

Membrane-bound organelles are a wonderful evolutionary acquisition of the eukaryotic cell, allowing thesegregation of sometimes incompatible biochemical reactions into specific compartments with tailored mi-croenvironments. On the flip side, these isolating membranes that crowd the interior of the cell, constitutea hindrance to the diffusion of metabolites and information to all corners of the cell. To ensure coordinationof cellular activities, cells use a network of contact sites between the membranes of different organelles.These membrane contact sites (MCSs) are domains where two membranes come to close proximity, typicallyless than 30 nm. Such contacts create microdomains that favor exchange between two organelles. MCSs areestablished and maintained in durable or transient states by tethering structures, which keep the twomembranes in proximity, but fusion between the membranes does not take place. Since the endoplasmicreticulum (ER) is the most extensive cellular membrane network, it is thus not surprising to find the ERinvolved in most MCSs within the cell. The ER contacts diverse compartments such as mitochondria,lysosomes, lipid droplets, the Golgi apparatus, endosomes and the plasma membrane. In this review, wewill focus on the common organizing principles underlying the many MCSs found between the ER and virtu-ally all compartments of the cell, and on how the ER establishes a network of MCSs for the trafficking of vitalmetabolites and information. This article is part of a Special Issue entitled: Functional and structural diversityof endoplasmic reticulum.

© 2013 Elsevier B.V. All rights reserved.

1. Functions of MCSs

MCSs display a wide variety of functions including, but not limitedto regulation of immune response, apoptosis, organelle dynamics andtrafficking. At the molecular level, the function of MCSs most ofteninvolves calcium (Ca2+) and/or lipid exchange between two compart-ments. Trafficking of both Ca2+ and lipids must be governed by strictrules. In the case of Ca2+, tight coordination is required because it isa very potent signaling molecule that acts locally, and which can betoxic when deregulated (reviewed in [1]), and in the case of lipidsbecause their variety contributes to determining the identity andphysicochemical properties of all membrane-bound compartments(reviewed in [2]). Ca2+ and lipids also have in common diffusion-limited properties. For Ca2+, diffusion is limited by the abundance ofCa2+-binding activities in all compartments of the cell, while diffusionof lipids is limited by their insolubility in most of the aqueous cellularvolume. MCSs allow localized and targeted exchange of both Ca2+

and lipids to coordinate homeostasis across the many compartmentsof the cell. Since the ER is both the main site of lipid synthesis andthe main cellular Ca2+ store, it is not surprising that MCSs involving

onal and structural diversity of

. Kornmann).

rights reserved.

ganization and function ofme

the ER provide an important function in both of these exchangereactions.

1.1. Calcium signaling

The ER [Ca2+] is in the millimolar range, while that of the cytosolis nanomolar (reviewed in [1]). This difference is established bySarcoplasmic/Endoplasmic Reticulum Ca2+ ATPases (SERCA), whichpump Ca2+ into the ER lumen, and by Plasma Membrane Ca2+

ATPases (PMCA), which pump it out of the cell (Fig. 1B). This gradientis used in signaling events whereby the opening of ER Ca2+ channels(the Ryanodine Receptors, RyR and the Inositol-1,4,5-triphosphateReceptor, IP3R) causes a transient surge in cytosolic [Ca2+], triggeringa number of downstream events by activating Ca2+-binding proteins.Ca2+ signaling governs processes as diverse as memory, vision,fertilization, muscle contraction, proliferation, cell migration, immuneresponse and transcription. Ca2+ does not only signal to the cytosol,but also to other compartments. Intracellular Ca2+ diffusion is slowedby the presence of many Ca2+-binding activities. This slow diffusionmaintains steep local [Ca2+] gradients and non-equilibrium conditions,which are best illustrated by such spectacular phenomena as Ca2+

oscillations and Ca2+waves [1]. On the small scale, slow Ca2+ diffusionpromotes the formation of cytosolic Ca2+ microdomains, where the[Ca2+] differs from that of the bulk cytosol. Microdomains of thiskind are often created at MCSs, which coordinate vast movements ofthe Ca2+ pool across different compartments.

mbrane contact sites, Biochim. Biophys. Acta (2013), http://dx.doi.org/

http://dx.doi.org/10.1016/j.bbamcr.2013.01.028

mailto:[email protected]


http://www.sciencedirect.com/science/journal/01674889



Fig 1. MCSs and Ca2+ homeostasis. A) Organization of the junctional membrane complexes in striated cardiac and skeletal muscles. DHPR (yellow) is found in T-tubules, whichinvaginate from the cell surface into the cell interior. DHPR gates the opening of RyR (red) in cardiac cells by increasing the local [Ca2+] (Ca2+-gated Ca2+ release), in skeletalcells by physically contacting them across the cytosolic space between the SR and the PM (allosteric coupling). B) Organization of Ca2+ microdomains at MCSs. The blue color rep-resents the local [Ca2+]. A steep [Ca2+] gradient is established by constitutive Ca2+ pumping performed by SERCA (red) towards the ER lumen, and PMCA (yellow) towards cellexterior. ER–PM MCSs allow direct ER Ca2+ refilling via STIM/Orai-dependent (purple/green) SOCE with minimal Ca2+ escape towards the bulk cytosol. ER–mitochondria MCSsallow direct Ca2+ exchange from the ER channel IP3R (tan) to the mitochondrial uptake machinery VDAC (blue) and MCU (green).

2 S.C.J. Helle et al. / Biochimica et Biophysica Acta xxx (2013) xxx–xxx

1.1.1. Excitation–contraction in muscle cellsA classical contact site example is the spatial arrangement of the

muscle ER (called Sarcoplasmic Reticulum, SR) and the Plasma-Membrane (PM), and its role in generating the signal that results inmuscle contraction [3]. Muscle contraction is triggered by a massiveinflux of Ca2+ into the cytosol, which activates myosin movementalong actin filaments in the sarcomeres (Fig. 1A). This Ca2+ influxis the result of the synergistic activation of two mechanisms:1) the opening of voltage-gated PM Ca2+ channels, such as thedihydropyridine receptor (DHPR), which respond to a change in PMpotential originating at the neuromuscular synapse, and 2) the syn-chronous opening of the main SR Ca2+ channel RyR (reviewed in[4]). The synchronization of both events is ensured by their physicalcoupling at MCSs between the SR and the PM. The myocyte PM isinvaginated in between sarcomeres. These invaginations create trans-verse tubules (T-tubules), which project into the center of the cell.These T-tubules are closely bracketed by SR tubules, forming charac-teristic junctional membrane complexes (JMC) [5], referred to asdyads, where one T-tubule associates with one tubule of SR, or triads,where one T-tubule is bracketed by two tubules of SR (Fig. 1A). RyRsare found on the SR tubules while DHPRs are on the opposingT-tubules. It is this proximity between RyR and DHPR that allowscoupled gating of both channels. Interestingly, although the spatialproximity between both types of Ca2+ channels is conserved, theway DHPR “talks” to RyR depends on the muscle type. In cardiac mus-cle, myocardial RyR (RyR2) senses the Ca2+ emanating from DHPR.Ensuing RyR opening is therefore part of a positive feedback loop[6]. In non-pulsatile skeletal muscles, however, DHPR is thought to di-rectly open skeletal muscle RyR (RyR1) through a physical connectionbetween both channels [7,8]. According to this model, a voltage-dependent change in DHPR conformation is transmitted to RyR1 viathe C-terminus of DHPR cytoplasmic β1a subunit. This subunit ofDHPR bridges the cytosolic gap in between T-tubule and SR to directlybind RyR1 and gate its opening [9]. In both cases, close apposition ofSR and PM at MCSs allows the coordination of cellular events, whichpermits rapid and precise muscular contraction.

1.1.2. The store-operated Ca2+ entry

1.1.2.1. Function. Ca2+ signaling is not restricted to excitable musclecells. Non-excitable cells also use Ca2+ signaling in important signal-ing cascades, and they also use MCSs to coordinate these cascadesacross the many compartments of the cell.

Please cite this article as: S.C.J. Helle, et al., Organization and function ofme10.1016/j.bbamcr.2013.01.028

The ER lumen is the site of numerous fundamental calcium-dependent processes, such as protein folding and quality control[10]. After a discharge caused by opening of IP3R – the main Ca2+

channel in the ER of non-excitable cells – it is essential that the lumi-nal [Ca2+] returns rapidly to its normal level. This is ensured by animportant cellular mechanism, which triggers the entry of Ca2+

ions into the cytosol from the extracellular space in conditionswhere the ER Ca2+ store is depleted.

This Store-Operated Ca2+ Entry (SOCE) is not only important forER function, but also for the proper execution of Ca2+-dependentsignaling cascades. By constantly refilling the ER, SOCE allows to sus-tain elevated cytosolic [Ca2+] during extended periods of times. Forinstance, SOCE is required for many immune functions, such asT-lymphocyte activation and proliferation (reviewed in [11]).

1.1.2.2. SOCE mechanism. How does a Ca2+ channel situated in the PMknow that the [Ca2+] in the ER lumen is low? This question has keptresearchers busy for decades, until RNAi screens revealed an astound-ingly simple device for membrane–membrane communication, madeof only two central components: 1) the sensor Stromal InteractionMolecule (STIM) [12,13] and 2) the effector PM Ca2+ channel Orai[14–18]. STIM is a type I transmembrane ER protein. It senses luminal[Ca2+] via its luminal N-terminus, which bears Ca2+-binding EF-handmotifs [12]. Ca2+ depletion in the lumen causes the unfolding of theEF-hands and allows the formation of STIM oligomers [19–21]. Oligo-merization is accompanied by a conformational change in the C-terminal cytosolic moiety [22], exposing a polybasic tract of aminoacidsnear the C-terminus. The polybasic tract can bind to acidic phos-phoinositides found in the PM [19,22–24]. As a result, the localizationof STIM, which is rather homogenous in the ER in Ca2+ replete condi-tion [12,19,25], becomes restricted to a few structures (“puncta”) uponER Ca2+ depletion. Examination using TIRF microscopy (see Box 3)indicates that these puncta are MCSs between the ER and the PM[12,25–30]. At these ER–PM MCSs, STIM can recruit the Ca2+-specificchannel Orai and gate it by direct physical interaction [31–35]. Thissystem is so simple that it can be recapitulated in vitro. RecombinantSTIM cytosolic fragments bind to and gate Orai channels produced inyeast [36]. As yeast does not have any of the SOCE components, it isvery likely that this gating is direct.

1.1.2.3. SOCE and MCSs: how?.What recruits STIM to the MCSs? A like-ly model is that in resting conditions, STIM diffuses passively in theplane of the ER membrane, and that upon stimulation, the exposure




Box 1Static or dynamic?

Very dynamic processes happen at membrane contact sites,begging the question of whether these are themselves static ordynamic structures. In a growing yeast cell, the vast majorityof the cortical ER (i.e. the part of the ER that is not the nuclear en-velope) is found in direct proximity with the plasma membranethroughout cell life, suggesting that tethering is durable. ER–mitochondria connections also appear to be stable structures.ERMES complexes can be imaged for minutes in yeast [107]and the dynamics of ER and mitochondria are coupled in thelamellipodia of Cos-7cells [120], suggesting that contact ismaintained during the movements of these organelles. Thisshows that some contact sites can be pretty stable. Howeverit is not the rule. Many interorganelle structures display very dy-namic behaviors, such as the STIM–ORAI interaction (1.1.2.2),or the ORP1L-mediated ER–endosome contact [94,95]. STIMand ORAI only interact upon ER-Ca2+ depletion and ORP1L isrecruited to endosomes upon endosome sterol repletion. Whatis less clear is whether these dynamic proteins are recruited tostatic contact sites, or whether these sites are newly formedby the recruitment of these proteins. When SOCE is stimulatedrepeatedly, STIM can be recruited to the same locations, sug-gesting that these locations could be contact sites that aremaintained between SOCE stimulations [37]. On the other hand,the association between the ER and PM, as detected by trans-mission electron microscopy, is increased upon SOCE, or bySTIM overexpression, suggesting that STIM also has the abilityto induce tethering [40]. This tethering could operate byextending preexisting contact sites, or by creating contact sitesde novo. The nucleus–vacuole junction (NVJ) also appears to bea dynamic structure. Yeast cells in exponential phase do nothave much of a NVJ, but entry into stationary phase, nitrogenlimitation or rapamycin treatment triggers the expansion of NVJs[116]. The mechanisms underlying this induction are unclear, andcould include transcriptional induction of the proteins involved intethering. Indeed, the tethering protein Nvj1 is induced by a varie-ty of stresses dealing with starvation response, including entry in-to stationary phase [158], and its artificial overexpression alonecan induce the formation and spreading of NVJs [115]. Yet tran-scriptional induction of Nvj1 is likely not the only player, as cellsconstitutively overexpressing Nvj1 are still able to respond to star-vation by making even more NVJs [117].

3S.C.J. Helle et al. / Biochimica et Biophysica Acta xxx (2013) xxx–xxx

of its phosphoinositide-binding polybasic domain causes STIM to betrapped at preexisting ER–PM MCSs [25]. The examination of cyclesof SOCE activation/deactivation by TIRF microscopy reveals thatSTIM puncta form repeatedly at the same location, suggesting thatthe MCSs are indeed preexisting and are maintained between activa-tion cycle [37]. Moreover, consistent with passive diffusion, formationand dissociation of STIM oligomers, as well as its translocation to andfrom puncta, do not require ATP [38]. The structures that establishand maintain these preexisting MCSs are not known. However, theextent of ER–PM connection is affected by SOCE activation or STIMoverexpression, suggesting that, on the contrary, STIM creates theMCS that it will use to gate Orai [25,39,40]. It is possible that theinteraction of STIM with PM lipids and with Orai adds a tetheringforce to preexisting MCSs, thereby enlarging them (see Section 2.3and Box 1).

1.1.2.4. SOCE and MCSs: why?. Why is SOCE integrated at ER–PMMCSs? By allowing the sensor in the ER to directly interact with theeffector in the PM, MCSs reduce the number of necessary players tothe strict minimum. But the extraordinary complexity of most cellularsignaling events shows that simplicity is hardly a selecting force. In-stead, it is possible that integrating both SOCE sensor and effector atMCSs fulfills other functions. Having PM Ca2+ channels opened atsites of contact with the ER may allow Ca2+ to flow directly fromthe extracellular space to the ER lumen, bypassing the bulk cytosol(Fig. 1B), much like in the case of ER–mitochondrial Ca2+ exchange(see Section 1.1.3). This model finds supporting evidence: SERCA in-hibitors are usually used to induce a full-blown SOCE that is not mit-igated by ER refilling. In such conditions, SOCE activation causes amassive increase in cytosolic [Ca2+], but when SOCE is activatedwithout affecting SERCA, only a transient modest rise in cytosolic[Ca2+] is observed [41]. ER refilling can even take place without anydetectable cytosolic [Ca2+] rise, provided that SOCE is slowed by amild knockdown of STIM or by limiting extracellular [Ca2+]. This sug-gests that the tight coupling between PM opening and ER Ca2+

pumping allows Ca2+ to directly transfer from the extracellularspace to the ER lumen before it can diffuse into the bulk cytosol(Fig. 1B) [42]. To achieve this tight coupling, the SERCA may directlyinteract with STIM, a notion supported by co-immunoprecipitationexperiments [43].

Thus, integrating the ER [Ca2+] signal at ER–PM MCSs may notonly be a way to shorten the signal transduction pathway, but itmay also be a way to minimize cytosolic [Ca2+] perturbations whilerefilling the ER pool.

1.1.3. The ER–mitochondrial Ca2+ crosstalk

1.1.3.1. The role of contact sites. Back in the 1960s, isolated mitochon-dria were shown to take up Ca2+ when externally applied in vitro (foran historical review, see [44]). This import is dependent on a mito-chondrial Ca2+ uniport (MCU) [45,46] on the inner mitochondrialmembrane (IMM) and on porins, which are non-selective OMMpores for small molecules. This phenomenon also requires an intactmembrane potential across the IMM. However, the physiological rel-evance of these findings was soon questioned as the [Ca2+] necessaryto activate the uniport were orders of magnitude higher than thephysiological [Ca2+] found in the cytosol. The conclusion at thetime was that mitochondrial [Ca2+] import should never take placein a healthy cell. This view changed with the advent of methodsallowing in vivo organellar [Ca2+] monitoring, which showed that mi-tochondrial [Ca2+] increased dramatically upon opening of the IP3Rin live cells [47]. Yet, in the same conditions, the cytosolic [Ca2+]did not increase to sufficient levels to activate the MCU, raising a par-adox. This paradox was resolved by the following hypothesis: due tothe slow diffusion of Ca2+, the local [Ca2+] around the mouth ofthe IP3R may be much higher than in the rest of the cytosol. A close

Please cite this article as: S.C.J. Helle, et al., Organization and function ofmembrane contact sites, Biochim. Biophys. Acta (2013), http://dx.doi.org/10.1016/j.bbamcr.2013.01.028

proximity between the ER and the mitochondria may thus allow acti-vation of the MCU [48]. This hypothesis established the frameworkfor an abundance of studies aimed at revealing the physiologicalimportance of ER–mitochondria connections in Ca2+ signaling.Measuring [Ca2+] at ER–mitochondria contact sites using geneticallyencoded [Ca2+] reporters artificially targeted to these sites (seeBox 3) showed that they indeed represent microdomains of the cyto-sol, where the [Ca2+] can raise to the micromolar range, compatiblewith the activation of the MCU [49]. Moreover, expressing artificialER–mitochondria tethers made of a protein that bears both ER andmitochondrial targeting sequences (see Box 3), can increase the cou-pling between the ER-efflux and mitochondria-influx, again under-lining the importance of ER–mitochondria MCSs in Ca2+ exchange[49–51].

Thus, mitochondria take up Ca2+ owing to ER–mitochondriaMCSs. What is the physiological function of this uptake? Three roleshave been proposed. First, Ca2+-dependent enzymes are found inthe mitochondrial matrix and may require mitochondrial Ca2+




uptake for their proper function. Second, as Ca2+ is toxic in the cyto-sol, mitochondria could act as Ca2+ buffers to limit the cytosolic Ca2+

increase following IP3R stimulation. Third, mitochondrial Ca2+ couldact as an important trigger for mitochondrial apoptosis.

1.1.3.2. Ca2+-dependent respiration. Pyruvate dehydrogenase, NAD+-isocitrate dehydrogenase and 2-oxoglutarate dehydrogenase arethree mitochondrial enzymes centrally located in the Krebs cycle thatare activated by Ca2+ [52]. Other Ca2+-regulated mitochondrial activ-ities include ATP/ADP exchange, ATP/Pi exchange and ADP phosphory-lation by the F0/F1 ATPase [53]. Thus, mitochondrial energy productionis coupled to mitochondrial Ca2+ uptake. How important is that forcellular bioenergetics? This question should optimally be addressedby directly manipulating the players of mitochondrial Ca2+ influx,which was not possible until the recent discovery of MCU [45,46].Interesting leads, nonetheless, came from manipulating a process up-stream of mitochondria Ca2+ influx, i.e. ER efflux. When knocked-outof all isoforms of IP3R, chicken cells react by increasing autophagy[54]. Autophagy is a cellular mechanism usually induced in responseto nutrient limitations, suggesting that cells devoid of IP3R areexperiencing starvation. The reduced or abrogated ER–mitochondrialCa2+ exchange in these cells could thus lead to an under-stimulationof the above-mentioned enzymes, leading to decreased ATP produc-tion, to which cells respond by mounting a starvation response. Insupport of this model, the AMPK pathway is activated in these cells,indicating a low ATP/ADP ratio [54].

1.1.3.3. Cytosolic Ca2+ buffering. During IP3R or RyR activation, all theCa2+ that ends up in mitochondria does not end up in the cytosol.This could imply that mitochondria have a buffering role by avoidingthat excess Ca2+ reaches the cytosol. Stimulating IP3R or RyR inconditions that inhibit mitochondrial Ca2+ uptake (for example inpresence of a mitochondrial uncoupler that dissipates mitochondrialmembrane potential) results in a much bigger cytosolic [Ca2+] in-crease [55,56]. As cytosolic Ca2+ is a potent trigger for necrosis, thisbuffering may have cytoprotective effects.

1.1.3.4. Induction of apoptosis. Beside its potential cytoprotective ac-tion, mitochondrial Ca2+ may also have a proapoptotic role. Isolatedmitochondria treated with high [Ca2+] in vitro undergo opening ofthe so-called permeability transition pore (PTP). Although originalstudies describe the PTP as a pore opening in both mitochondrialmembranes independent of MCU activity [57], the current consensusis that PTP is a pore in the OMM, which is opened by Ca2+ overload inthe matrix (for review see [58]). Opening of this pore causes the re-lease of small inter-membrane space proteins, such as CytochromeC, which triggers a cascade of events leading to programmed celldeath.

Could ER–mitochondria Ca2+ exchange similarly lead to PTP open-ing? Several lines of evidence suggest that this is the case. Antiapoptoticproteins such as Bcl-2 were shown to lower ER [Ca2+] when over-expressed, which in this context, could explain at least part of theirantiapoptotic function [59]. Similarly, deletion of proapoptotic proteinsled to decreased ER [Ca2+], in turn leading to decreased apoptoticsensitivity [60]. More directly involving Ca2+ homeostasis, a mutationin the ER Ca2+ pump SERCA1 was shown to act as an oncogene byinhibiting apoptosis, suggesting again that high ER [Ca2+] is necessaryto activate the PTP in the mitochondria [61]. Similarly, downregulatingIP3R renders cells resistant to a variety of apoptotic stimuli [62,63]. Thus,direct ER–mitochondria Ca2+ exchange is a likely trigger of the mito-chondrial pathway of apoptosis.

The experiments described in Sections 1.1.3.2, 1.1.3.3 and 1.1.3.4assessed the importance of mitochondrial Ca2+ influx by manipulatingER efflux or mitochondrial membrane potential, which are bound to af-fect more than just mitochondrial Ca2+ influx. The recent identificationof themolecular players of mitochondrial Ca2+ influx [45,46] will bring


the power of genetics to assess the physiological importance of thisphenomenon. It is notable that mice could be treated with MCU RNAifor weeks without noticeable effects on their health. Moreover, mito-chondria isolated from these animals showed normal bioenergeticsparameter despite their near complete loss in Ca2+ uptake [45]. Closerinspection will be required to clarify the role of mitochondrial Ca2+

uptake in the above processes, and indeed the tools necessary for thisexamination are now available.

1.2. Interorganelle lipid exchange

Cellular lipids fulfill a wide range of functions. They can act asenergy storage in the form of triglycerides, as important parts ofintra- and intercellular signaling cascades and as main building blocksof bio-membranes. Nowadays, literature describes more than 1000different lipid species in eukaryotic cells [64], distributed asymmetri-cally between organelles and even between the two leaflets of one bi-layer [65]. Phosphatidylcholine (PC), for instance, is not only the mostcommon lipid in the cell but also most abundant in the endoplasmicreticulum (ER). Phosphatidylethanolamine (PE) and cardiolipin(CL), in contrast, are significantly enriched in mitochondria. CLconcentrations are furthermore up to four times higher in the IMMcompared to the OMM [66,67]. These asymmetric distributions havefunctional consequences. For instance, the compact packing ofsphingolipids and sterols in the plasma-membrane, where theselipids are enriched allows this membrane to resist increased levelsof mechanical stress [66]. The asymmetric distribution of cellularlipids is partly explained by their asymmetric synthesis and break-down. For instance a complex network of phosphatidylinositol (PI)kinases and phosphatases establishes and maintains a gradient of dif-ferent phosphoinositides on the different cellular membranes. How-ever, most lipids are synthesized in one place – the ER – suggestingthe existence of an active and regulated lipid sorting system. The ve-sicular transport system shuttles lipids together with proteins acrossthe endomembrane, but this mode of transport is far from the wholestory. PE, cholesterol and ceramide can for instance reach the GOLGIand PM even when vesicular transport is inhibited [68–70]. More im-portantly, mitochondria are completely disconnected from vesiculartransport routes although they are centrally involved in cellularlipid homeostasis. Indeed, the de novo aminoglycerophospholipid bio-synthesis pathway is spatially segregated between ER and mitochon-dria. Phosphatidylserine (PS) is made in the ER membrane by the PSsynthase (PSS). PS is then decarboxylated to PE by a decarboxylase(PSD) that is integral to the IMM. PE is further processed to PC bythree sequential methylation steps performed by ER-resident pro-teins (Fig. 2D). This convoluted biosynthesis pathway involves aback-and-forth journey from the ER to the IMM. Moreover, PC,which is abundant in both mitochondrial membranes, needs to travelback to the mitochondria (for review, see [71]).

In this context, MCSs between pairs of organelles have long beenthought to facilitate lipid exchange by ensuring the specificity andefficacy of the exchange reaction [72].

1.2.1. Lipid transport proteinsNon-vesicular lipid transport is catalyzed by lipid transport pro-

teins (LTPs), which are able to extract one lipid molecule from amembrane and deliver it to another. This type of transport involvesspecial lipid-transport domains (LTD) that can shelter the hydropho-bic moieties of lipids from the aqueous environment (Fig. 2A and B).The mode of action of these proteins has been extensively studiedin vitro and corroborated by structural analyses (for reviews see[73,74]). Different types of LTD bind different lipids. In addition totheir LTDs, LTPs often contain different combinations of targetingdomains (Fig. 2A), which direct them to special compartments.For example, the diphenylalanine-in-an-acidic-tract (FFAT) motifbinds to the ER protein VAP (VAMP-associated-protein) [75–77], the




A

BC

D

Fig. 2. MCSs and lipid homeostasis. A) Non-exhaustive list of LTPs and their domain organization. Many of these LTPs are found at MCSs. ORPs bind to sterols and to PI4P. The ste-roidogenic acute regulatory protein-related lipid transfer (START) domain can bind either sterols, phospholipids or ceramides. PITP and Sec14-like domains bind PC and PE.Lipid-binding by SMP-containing proteins is not experimentally established, but SMP is remotely homologous to the TULIP domain. B) LTPs often show affinity for two or morelipid species. Here are shown: PITPα bound to a PC and a PI molecule, Sph1 (a Sec14-paralog) bound to a PI and PC molecule. CERT (START-containing) bound to a ceramideand a DAG molecule and Kes1 (ORP) bound to an oxysterol and a PI4P molecule. While the polar heads can adopt different conformations (see Sph1 and Kes1), the hydrophobictails bind in the same pocket, showing that only one molecule can be transported at a time. The crystal structures used in this figure have the following pdb IDs: 1t27 and 1uw5for PITPα, 3b7q and 3b7N for Sph1, 2e3q and 2z9y for CERT and 3spw and 1zhw for Kes1. C) Model of how a gradient in sterol and PC is generated by the interconnected cycleof ORPs (blue) and PITP/Sec14 proteins (yellow). ORPs exchange sterols for PIPs. PIPs being abundant in the late secretory organelles and rare in the ER, due to the action ofPI4P phosphatases (red), this creates a net transport of sterols toward the late-secretory organelles. PITP/Sec14-like proteins exchange PI for PC. PI being actively removed fromthe Golgi by phosphorylation by PI kinase (green), the PITP exchange reaction creates a net transport of PC from the Golgi to the ER. Combining these two exchange reactions resultsin a net PC-sterol exchange that is fueled by ATP hydrolysis by the Golgi PI-kinase (green). D) de novo PC biosynthesis requires two ER–mitochondria exchange reactions.Phosphatidylserine (PS) is synthesized in the ER. It is converted to phosphatidylethanolamine (PE) in the IMM by the PS-decarboxylase (Psd). PE is further converted to phospha-tidylcholine (PC) by three methylation steps, which take place in the ER.


Pleckstrin-Homology (PH) domains bind to specific PIPs found in theGolgi and the plasma membrane [78–81], and the ArfGAP1-Lipid-Packing-Sensor (ALPS-) motifs, which bind to highly curved mem-branes [82]. LTPs bearing two targeting motifs for two different mem-branes are thus naturally directed to MCSs between these twomembranes. One intriguing feature of many LTDs is their ability tobind two lipid species in a competitive manner. Many LTPs havebeen crystallized in complex with different lipids. This is the case ofthe PI Transporter Protein (PITP) with PI and PC [83,84], the ceramidetransporter CERT with ceramide and diacylglycerol (DAG) [85], andthe Oxysterol-binding-protein-Related-Domain (ORD) of Kes1 withsterols and PI4P [86,87] (Fig. 2B). In these examples, whereas thepolar head of lipids contact the protein at sites that are specific for


each lipid species, the aliphatic tails occupy the same hydrophobicpocket. This renders the binding of two lipids mutually exclusive.This feature suggests that many of these proteins function as lipidexchanger rather than unidirectional transporter.

1.2.2. Non-vesicular lipid exchange in lipid homeostasisCurrent models suggest that specific lipid exchange reactions

may happen at MCSs. For example, a large subfamily of LTPs has anOxysterol-binding-protein-Related-Domain (ORD) domain, and istherefore called ORP (oxysterol-binding protein related proteins).The ORD domain is highly conserved and binds a variety of sterols.It has recently been found to also bind PI4P competitively with sterols[87], and to promote the trading of one lipid molecule against the




Box 2Contact-sites and lipid rafts.

Lipid rafts are thought to be assemblages of lipids and proteins that are laterally segregated from the rest of the membrane through a com-bination of protein–protein and protein–lipid interactions, lipid phase immiscibility and association with the cytoskeleton [159]. They havelong been operationally defined as detergent-resistant membranes (DRM). Indeed, they resist extraction with 1% Triton X100 at 4 °C, aproperty that has been conveniently used for their purification and analysis [160]. However, due to their small size, lipid rafts have neverbeen directly observed in live cells and indirect methods are used to infer their existence, which has thus long remained controversial[161]. Rafts are enriched in select proteins and in lipids, such as sterols and sphingolipids. Several lines of evidence make a connection be-tween membrane contact sites and lipid rafts. Mitochondria-associated membranes (MAMs) are operationally defined as an ER fraction thatcosediments with mitochondria during subcellular fractionation and are therefore thought to represent the subfraction of the ER that is at-tached to mitochondria in vivo (see Box 3). The MAM fraction has characteristics of lipid rafts. First, several analyses show that the MAMfraction is detergent resistant [162,163]. Second, this fraction is enriched with sterols and simple sphingolipids [162,164,165]. ManyMAM-enriched proteins have been shown to associate with DRMs, including the Sigma-1 receptor, the presenilin 1 and 2, PEMT, Vdac1and mitofusin-2 [162,163]. ER–PM contact sites may also have raft characteristics. When STIM is recruited to ER–PM contact sites, it con-comitantly relocates to DRMs, suggesting that ER–PM contact sites are lipid rafts [166]. Finally, the nuclear vacuole junction (NVJ) sharescommon points with rafts, as select vacuolar proteins are enriched or excluded from NVJs [117]. While enrichment of a protein at NVJ canbe explained by direct interactions of these proteins with the NVJ structural components Vac8 and Nvj1, selective exclusion of vacuolarproteins is more difficult to explain. One model is that the lipids in the vacuole may also be laterally segregated, leading to enrichmentand exclusion of select proteins that have differential affinities for different lipid phases, akin to canonical lipid rafts. In support of this no-tion, pharmacological or genetic inhibition of very long-chain fatty acid synthesis affects NVJ morphology and function, suggesting a rolefor specific lipids in the assembly of these contact sites [119].It is surprising to find ER proteins on lipid rafts. The lipid phase-separation required to assemble rafts is supposed to take place only on mem-branes with sufficient sterols and sphingolipids, and therefore the sterol- and sphingolipid-poor ER membrane should be devoid of canonicalrafts [66]. Yet, since these DRM-associated ER proteins are found at contact-sites, it may be that the ER membrane in which they insert isindeed detergent soluble, but that the membrane to which they associate in trans is a DRM. For example, activated STIM interacts withnegatively-charged lipids on the plasma membrane through a polybasic domain on its cytosolic domain. Therefore, even if the ER membraneis totally soluble, STIM may remain associated with detergent-resistant bits of the PM. This explanation cannot be invoked for MAMs, asboth ER and OMM are supposed to be devoid of canonical lipid rafts. As MAM-derived DRMs contain both ER- and OMM-integral proteins, itis not clear which membrane is detergent-resistant. It is notable, however that drugs that disturb rafts by affecting sterol concentration(such as methyl-beta-cyclodextrin, which sequesters sterols out of membranes), cause relocalization of MAM proteins. This relocalizationcan be observed both at the level of the biochemical MAM fractionation and at the level of the subcellular localization [162].


other in an in vitro assay. Therefore, although the in vivo relevance oftheir LTP activity is questioned [88], ORPs have been proposed totransport sterols from the ER to the Golgi and PM. Indeed, manyORPs localize at MCSs between these compartments (see Section1.2.3). Because PI4P is immediately dephosphorylated in the ER byspecific phosphatases [89], this exchange reaction should only goone way. Sec14 on the other hand can transport both PC and PI be-tween the ER and the Golgi [90]. In this model, the ATP dependentconversion of PI to PI4P by Golgi resident PI kinases provides thedirectionality for the reaction. Thus, coupling both cycles of ORPsand Sec14 could provide a neat device that uses the power of ATP topump sterol into the late secretory pathway (Fig. 2C, [87]).

The genetic dissection of these pathways is complicated by thepleiotropic roles of lipids in membrane biology. PI4P for instance is acentral regulator of vesicle trafficking. Another lipid transportertightly linked to vesicular traffic is CERT (Fig. 2A–B), which transportsceramide from the ER to the Golgi [75], where it is used for the pro-duction of DAG and sphingomyelin [91]. DAG is important for vesiclebudding from the trans-Golgi [92], showing again the entanglementof lipid transport with other aspects of membrane biology.

1.2.3. MCSs in lipid exchangeWhile energy-dependent processes provide directionality to lipid

exchange, MCSs may provide specificity. Many LTPs are targeted toMCSs. For example, both CERT and several ORPs have two targeting do-mains: one PH-domain and one FFAT motif. The FFAT-motif anchorsthese LTPs to the ER by interacting with the ER-protein VAP, whilethe PH domain binds to PIPs on closely apposed membranes [74]. Forinstance, the yeast ORP Osh1 is found at a contact site between theperinuclear ER (the nuclear envelope), where it interacts with yeast


VAPs (Scs2 and Scs22 [76]), and the vacuole (the yeast lysosome),which it contacts via its PH domain [93] (Fig. 3). This Nuclear-ERVacuole Junction (NVJ) is a contact site formed dynamically witha function that is only partly understood (see Section 1.3.1). The mam-malian ortholog Oxysterol-binding-protein-related-protein 1L (ORP1L)is also involved in dynamic contact sites between the ER and late-endosomes/lysosomes. In this case, ER-endo-/lysosomes contact sitesare formed in response to low cholesterol, which is sensed by ORP1L'sORD. Formation of the ER/endosome contact site by binding of ORP1L'sFFAT domain to VAP causes dissociation of dynein-binding proteinsfrom the late endosome, resulting in a cellular redistribution of theseendosomes [94–96]. Thus LTPs cannot only exchange and transportspecific lipids across the aqueous phase of the cell, but they can also es-tablish specific contact sites and serve as sensors of lipid levels betweentwo compartments.

As stated before, ORD domains are found in many eukaryote pro-teins. For example, yeast has seven ORPs. Four of them (Osh2, Osh3,Osh6 and Osh7) localize to contact sites between the ER and theplasma-membrane [76,89,97]. Two of these four (Osh2 and Osh3)harbor a FFAT and a PH domain necessary for their localization(Fig. 2A). The other two are targeted to contact sites by unknownmechanisms [98]. Moreover, sterol–PI4P exchange activity appearsto be one of the most conserved features of ORDs [87].

1.2.4. Lipid exchange at ER–mitochondria MCSs and the SMP domainAlthough the prime example for MCS-dependent non-vesicular

lipid exchange is the ER–mitochondria connection (see Section 1.2),it is unsettling to note that the LTPs involved are mostly unknown.In yeast, the best candidates are the members of the ER–MitochondriaEncounter Structure (ERMES). This protein complex is made of




Fig. 3. Established and potential tethers at various MCSs. 1) Potential tethers between the ER and the PM in metazoan, yeast or both. For explanations on individual proteins andcomplexes, see Section 2.3. 2) Potential tethers between the ER and the mitochondria. See Section 2.2. 3) Potential tethers between the ER and the vacuole at the nucleus–vacuolejunction (NVJ) in yeast. The interaction of Vac8 and Nvj1 constitutes the main tether (see Section 2.1), but Osh1 interaction with Scs2 and Scs22 may also participate in tethering.4–6) Potential tethers between the ER and the late endosome, the Golgi apparatus and lipid droplets. See Section 2.4.


five subunits. One of the subunits (Mmm1) is integral to the ERmembrane, while two others (Mdm10 and Mdm34) are inserted inthe OMM [99–101]. This complex was identified based on its ER–mitochondrial tethering activity: its function can be bypassed by theexpression of an artificial ER–mitochondria tether (see Section 2.2.1and Box 3). Deletion of any member of the ERMES complex causespleiotropic phenotypes that are difficult to decipher [102]. Yet, inter-esting insight came from an unbiased genetic interaction screen thatdetected a strong correlation in the patterns of genetic interactionof ERMES mutants and that of strains lacking the mitochondrial PSdecarboxylase (Psd1, Fig. 2D). That Psd1 substrate and product haveto respectively come from and return to the ER (Fig. 2D), suggests arole for ERMES in lipid transport. Whether ERMES may itself act asan LTP for this transport reaction is a matter of debate. On the onehand, recent bioinformatic analyses suggested that ERMES proteinsmay directly bind lipids. Three ERMES components (the ER proteinMmm1, the cytosolic Mdm12 and the OMM protein Mdm34) sharea protein domain, called SMP, which is common to a group of con-served eukaryotic Synaptotagmin-like, Mitochondrial and PH-domaincontaining proteins (SMP, Fig. 2A) [103,104]. Homologies betweenthese proteins are difficult to see at the level of the primary sequence,but are more obvious using hidden Markov models that take the pre-dicted secondary structure into account. Additional homology predic-tion shows that the SMP domain is part of a larger protein family withstructurally characterized members. Members of this family harboran interesting structure made of a very elongated hydrophobicgroove, which binds various lipids and hydrophobic molecules. Forthis reason, the family was named TULIP (for tubular-lipid-binding)[104]. Phospholipids can be bound by TULIP proteins. Interestingly,in this case, the hydrophilic heads “snorkel” out of the groove anddo not make any contact with the protein. Thus, phospholipid bindingby TULIP proteins appears more or less independent of the lipid spe-cies. Whether SMP domains are also able to bind lipids in a hydropho-bic groove is unclear, but all other yeast SMP-containing proteins(Nvj2, Tcb1, Tcb2 and Tcb3) were also found to localize at various


MCSs [105]. SMP must therefore have a common function at MCSs,and it is tempting to speculate that it is lipid exchange.

But on the other hand, some data conflict with the idea thatERMES members act as LTPs. First, ERMES deletion only has a limitedimpact on ER–mitochondria lipid exchange [99,106] – whether it hasany effect at all is even controversial [107] – showing that ERMES-independent lipid exchange can take place. Second, ERMES was iden-tified as a tether because its absence could be compensated for byan artificial ER–mitochondria tether. Two of the three SMP-bearingERMES components Mdm12 and Mdm34 are dispensable as longas this artificial tether is present. Since this artificial tether is notexpected to work as an LTP, it seems that the LTP activity of Mdm12and Mdm34, if any, is not required under laboratory conditions [99].

In mammals, where ERMES is not conserved, the identity of ER–mitochondria LTPs is even more mysterious.

1.2.5. Lipid storageFinally, one important function of lipids is their use as energy stor-

age. This storage takes place mainly in the form of neutral lipids, suchas triglycerides stored into intracellular lipid droplets. Lipid droplets(LDs) are thought to emanate from the ER by triglyceride and sterylesters accumulation in between the two leaflets of the ER membrane(reviewed in [108] and [109]). As a result, LDs have a core of neutrallipids surrounded by a single leaflet of amphipathic lipids. It isunclear, whether continuity between the ER membrane and the sin-gle leaflet surrounding LDs is maintained during the whole lifetimeof the LDs [110]. It is clear, however, that LDs are covered with closelyapposed ER cisternae on most of their surface [111–113]. These ER–LD contact sites likely serve in translocating newly synthesized neu-tral lipids from the ER to the core of the LDs [114].

1.3. Other functions of MCSs

Besides lipid and Ca2+ exchange, a few example MCSs performadditional functions in organelle biology.




Box 3Methodologies to study contact sites.

The complete toolbox of modern molecular and systems biology is available for the study of MCS, but a few methods have proven partic-ularly useful for this purpose and will be discussed here.

• Biochemical purification of MCSs. Classical subcellular fractionation can separate different organelles according to their densities orsedimentation coefficient using isodensity or velocity centrifugation, respectively. But could organelles that are physically tetheredhave special physical characteristics that allow purification? This is the rationale behind purification of Mitochondria-associatedmembranes (MAM) [167]. Purification of mitochondria commonly involves a step of differential velocity centrifugation duringwhich mitochondria sediment at a lower speed than the bulk ER. However, following this step, mitochondria are found heavily con-taminated with ER membranes. Thus, mitochondria have to be separated from the contaminating ER by a step of isodensity centri-fugation across a continuous or step gradient of sucrose, Percoll, Optiprep™ or Nycodenz®. The ER contaminant that is retrievedfrom the gradient has a different composition from that of bulk ER. This ER fraction, originally dubbed “fraction X” [72] is thereforethought to represent a sub-domain of the ER physically attached to mitochondria. This strategy was emulated in the search forother fractions such as PM-associated membranes (PAM) [168,169] or plastid associated membranes (PLAM) in plants andalgae [170]. The MAM fraction is enriched for many lipid-related enzymes and Calcium signaling proteins, consistent with a roleof ER–mitochondria connection in these two processes (for review see [171]). Yet the definition of MAM as the “fraction of theER that cosediment with mitochondria” is an operational definition, just as “detergent-resistant membranes” is an operational def-inition for lipid rafts (Box 2). The nature of the ER found in the MAM fraction is poorly defined. Is it only the lipid and proteins directlyattached to mitochondria? Is it the mitochondria-attached ER plus just enough ER to seal a vesicle around it? Is it an undeterminedamount of ER that stays connected to the mitochondria-attached ER? Moreover, as any subcellular fraction, MAM is not completelypure. Indeed, most proteins retrieved in the MAM fraction are not ER proteins [172]. Other methods, such as microscopy, aimed atlocalizing MAM-associated factors with respect to ER and mitochondria do not necessarily clarify the picture, as the subcellular lo-calization of MAM-enriched proteins is not evident. The canonical MAM protein FACL4 is found at the whole mitochondrial surface[173]. The same is true for AMFR/GP78 [174], or the presenilins 1 and 2 [173]. By contrast, other highly MAM-enriched proteinssuch as the Sigma-1 receptor [175] or Ero1alpha [176], show a staining that overlaps only with select regions of mitochondria,with most of the signal nowhere close to them. To complicate things, different studies disagree on the extent of MAM enrichmentof different factors. For example, the amount of IP3R3 found in the MAM differs between studies [143,144]. Even in a single study,when virtually all IP3R3 is found in the MAM by subcellular fractionation, only a fraction of it is found close to mitochondria by im-munofluorescence [144]. Thus, akin to lipid rafts ten years ago, much needs to be done to take MAMs from an operational defini-tion to a solidly-grounded biological reality.

• Microscopy is of course a method of choice to visualize the encounter of two organelles. Organelle-targeted fluorescent proteinsare perfect markers for live-cell microscopy. However, contact sites per se are happening at distances (typically less than 30 nm)far below the resolution limit of light microscopy, making it difficult to discriminate genuine contact sites from mere colocalization.For instance, a recent study shows that a cell line that displays reduced ER–mitochondria colocalization as assessed by confocallight microscopy, actually shows increased ER–mitochondria contacts by electron microscopy [177]. Thus, care must be takenwhen analyzing colocalization of organelles at light-microscopy resolution, as changes in colocalization may reflect changes in or-ganelles distribution rather than changes in connection. These limitations can be circumvented by choosing proper imaging tech-niques. For example, TIRF microscopy is a method of choice to visualize ER–PM connections in adherent cells (for examples see[12,19,25,28]). Since the evanescent field of total internal reflection is narrow, the ER only enters this field when in a rangeof 100 nm from the cover-slip-bound PM. Although the 100 nm-wide evanescent field of TIRF microscopy is still quite largerthan the range of interorganelle connection, TIRF performs much better than confocal microscopy that displays a resolution inZ in the order of the micrometer. Another way to circumvent resolution issues is to image areas of the cells where organellesare sufficiently scarce to be well resolved. The lamellipodia of Cos-7 cells, for instance, are very thin (typically 100–200 nmthick) [178], abrogating the problem of Z-resolution. The confinement of the cytoplasm in these lamellipodia has two addition-al effects: First it causes the spreading of the cellular components, which allows resolution of individual organelles such asmitochondria, ER cisternae and ER tubules [120]. Second, it forces intersecting organelles against each other (mitochondria,for instance, are typically 300–500 nm thick and must thus flatten to fit into lamellipodia [179]). As a consequence, everyobserved organelle intersection represents a contact site de facto [134]. Electron microscopy does not suffer from resolutionlimitations, at the expense of versatility and the possibility to study live cells. New light microscopy techniques that are notlimited by Abbe's law, approach the resolution needed to study contact sites and will likely provide invaluable tools for thispurpose.

• Genetic screens have proven extremely useful to discover molecular players of organelle contact sites. RNAi screens have beenat the root of the discovery of the components of SOCE [12–16]. In the cases of both STIM and ORAI, screens were designedto identify RNAi targets that affect SOCE. Identification of these factors later led to the discovery of the contact-sites whereER–PM communication was taking place. In this case, the function led to the structure. A difficulty in analyzing the resultsof such screens is that they are prone to hit functionally quite far from their targets. And much work may remain to bridgethe function to the structure. For example, an RNAi screen has identified LETM1 as a protein necessary for Ca2+ import intomitochondria [180]. Yet, other studies indicate that LETM1 acts as a K+/H+ exchanger [181,182]. It is thus still controversialwhether a defect in K+/H+ exchange causes the Ca2+-import defect or if LETM1 can also transport Ca2+. But screens canalso be designed to seek for mutants with reduced or abolished tethering activity. The tethering activity of the ERMES complexwas discovered following this principle [99]. Mutant yeast cells were selected based on their ability to be complemented by anartificial ER–mitochondria tether (see below). Such mutants were part of a single complex found at the interface of the ER andthe mitochondria, which is involved in a plethora of processes dealing with virtually all aspects of mitochondrial biology. Thus





in this case, the structure leads to the function. The drawback of this reversed approach is that the real function of these con-tact sites remains not fully understood [102,107].

• Artificial tethers represent a powerful way to approach MCSs. Pioneered by Hajnóczky and colleagues, this approach consists inencoding targeting signals for different organelles into the same protein, or into different proteins that will later be joined togetherthrough dimerization motifs [49,51]. Such synthetic proteins can act as artificial tethers, thereby increasing the connection betweentwo organelles. This feature was used to show that increased ER–mitochondria connection caused increased Ca2+ exchange be-tween both organelles [51], and also to find mutants of natural tethers that relied on an artificial tether for growth (see above) [99].Additionally, due to their dual affinity, such synthetic proteins can bring reporters to defined contact sites. For instance, artificialtetherswith fluorescentmoieties allow localization of contact siteswithin the cell. Similarly, artificial tethers fused toCa2+ biosensorsshowed that local [Ca2+] at ER–mitochondria connections were ~25-fold higher than in the rest of the cytosol [49]. However thesetethers have the potential to generate non-physiological situations. Indeed, it is virtually impossible to target an artificial protein to acontact site using two targeting sequences without concomitantly increasing the tethering forces between both organelles involved.For instance, Csordás et al. designed a FRET-based reporter that targets the ER–mitochondria interface thanks to a rapamycin-induceddimerization between the ER-targeted FRET donor and the mitochondria-targeted acceptor. Upon rapamycin treatment, FRET signal,which is originally found at few discrete foci, progressively expands to cover almost all the mitochondrial surface, due to the intrinsictethering activity of the reporter [49]. Care should thus be taken when using such synthetic proteins.

Box 3 (continued)


1.3.1. Piecemeal microautophagy of the nucleusIn yeast, under nutrient limiting conditions, the vacuole becomes

the largest cellular organelle and is used for the autophagic recyclingof parts of the cytoplasm. With these conditions, a MCS between thenuclear ER and the vacuole (nuclear–vacuole junction, NVJ) [115]becomes the site of a special type of autophagy, whereby bits of the nu-cleus are delivered to the vacuole. The NVJ grows during starvation to apoint where it covers a significant fraction of the nucleus. This MCSthen invaginates into the vacuole. At the final stage, scission of the ERand fusion of the vacuolar membrane release a vesicle containingnuclear cargo into the vacuole lumen, where degradation occurs.This phenomenon is therefore called piecemeal microautophagyof the nucleus (PMN) [116]. Many aspects of PMN remain unclear.(1) While the machineries that are responsible for the formation ofthe NVJ are well-characterized (see Section 2.1), those involved inthe scission of the ER and abscission of the vacuolar vesicle are un-known. (2) The lipids in the interior face of the vacuole are resistantto degradation by vacuolar lipases, which is important to maintainthe integrity of the vacuolar membrane. It is thus not clear how thevesicle, once detached from the vacuolar membrane, becomes a sub-strate for degradation while the same membrane, when still connectedto the vacuolar membrane is not. This distinction could be related to adiffusion barrier established at the NVJ, whichmay exclude select lipidsand proteins from the forming vesicle [117] (Box 2). Alternatively, re-cruitment of lipid modifying enzymes to the NVJ, and thereby to thePMN-vesicle membrane, could locally change the lipid composition ofthis membrane [118,119], rendering it selectively sensitive to the vac-uolar lipases. (3) The function of PMN is unclear, especially since muta-tions disabling this process do not yield notable growth defects [116],but roles can be speculated. For instance PMN could regulate the sizeof NVJ by deleting overgrown interfaces. Alternatively, since the nucle-olus is usually the nuclear region adjacent to the NVJ, PMN could servein the recycling of ribosome biogenesis factories [116].

1.3.2. Mitochondrial dynamicsMitochondria are highly dynamic organelles. They represent, after

the ER, the second largest intracellular membrane network. This net-work is constantly remodeled by fusion and fission events, and by cy-toskeletal driven movements that allow faithful inheritance duringasymmetric cell division and allow mitochondria to reach distantplaces in the cell, such as the neuronal synapse. ER–mitochondriacontact sites have recently been shown to play a role in all of theseaspects of mitochondrial dynamics.


1.3.2.1. Mitochondrial movement. Pioneering work in the lab of GiaVoeltz showed that ER and mitochondrial dynamics are coupled[120]. By imaging very flat lamellipodia, Voeltz and colleagues couldresolve individual mitochondria and individual ER tubules. Both organ-elles move in a coordinated manner on a subset of acetylated microtu-bules. Mitochondrial movement along microtubules is a well-studiedphenomenon and one of the best-characterized adapter proteins inthis process is the GTPase Miro [121]. Miro is a mitochondria-tail-anchored protein that contains two GTPase and two Ca2+-bindingdomains. Miro, anchored to the OMM binds to kinesin heavy-chain viaa second adaptor protein. This binding is necessary for mitochondrialmovement along neuronal axons. But the surprising feature of Miro isits ability to stop mitochondrial movement upon high [Ca2+] [122].This phenomenon is dependent on both Miro's Ca2+-binding domainsand occurs by sequestration of kinesin heavy chain motor head awayfrom microtubules. One of the two mammalian paralogs of Miro,Miro-1, is found at ER–mitochondria interfaces in the lamellipodia ofCos-7cells, linking mitochondrial movement and ER–mitochondriaconnection [100]. Moreover, Gem1, the unique yeast Miro, is a memberof the ERMES complex [100,101]. These findings may partly explain thecoupling between ER and mitochondrial dynamics. The proximity ofthe ER may provide the Ca2+ necessary to regulate Miro. In supportof this notion, expression of a mutant form of VAPB, involved intype-8 amyotrophic lateral sclerosis (see Section 2.2.4), increasesER–mitochondria Ca2+ coupling and inhibits mitochondrial movementin a way that depends on Miro Ca2+-binding domains [123]. Thefunctional role of the coupling of ER and mitochondrial movement is,however, unclear.

1.3.2.2. Mitochondrial fusion and fission. Both mitochondria fusion andfission involve GTPases [124–127]. During fission, the dynamin-related protein 1 (Drp1) assembles into helical oligomers that cir-cumscribe the mitochondrial tubule. GTP hydrolysis then providesthe energy to squeeze the mitochondrion down to a point of fission[125,126,128–130]. In fusion, Mitofusins (MFNs) on opposite mito-chondria interact with each other while GTP hydrolysis forces thefusion of the two membranes [127,131]. In mammals, two MFNs(Mfn1 and Mfn2) are necessary for this function, therefore deletionsof either or both result in fragmented mitochondria [131]. Mfn2may have acquired an additional function besides its role in mito-chondrial fusion, namely, ER–mitochondria tethering. Mfn2 KO cellshave a fragmented ER, which can be rescued by expressing anER-targeted version of Mfn2 [132]. A fraction of endogenous Mfn2 isfound in the ER membrane, more precisely at MAMs (see Box 3),





where it is thought to participate in ER–mitochondria tethering byinteracting in trans with Mfn1 and mitochondrial Mfn2. Why thesame proteins are used for both mitochondrial fusion and tetheringto the ER is unclear, as is the fact that Mfn2 can promote ER–mitochondria tethering without provoking fusion of both organelles[132]. As tethering is a necessary first step in membrane fusion, atethering activity is encoded within mitofusins [133]. Evolution maythus have reused this property for ER–mitochondria tethering, whilefinding a way to prevent unwanted ER–mitochondria fusion.

Whether ER–mitochondria MCSs influence mitochondrial fusion isnot known. In the contrary, fission appears to happen preferentially atER–mitochondria MCSs [134]. In a study using both live cell microsco-py and electron tomography, the group of Gia Voeltz observed thatevents of mitochondrial fission were most frequent at sites of contactwith the ER. In yeast, EM-tomography revealed that MCSs correlatedwith sites of slight constriction of the mitochondria. In live mamma-lian cells, they could see a correlation between intersections ofthe ER and the mitochondrial networks and mitochondrial fissionevents [134]. Drp1 is recruited to mitochondrial constriction sites bya mitochondrial-fission-factor (Mff) [135]. Yet a characteristic mito-chondrial constriction could also be observed at ER–mitochondriaintersection in the absence of Mff or Drp1, showing that ER–mitochondria contact is acting upstream of the classical machineryof mitochondrial fission [134]. What could be the purpose of the cou-pling between mitochondrial fission and ER–mitochondrial MCSs?One possibility is that, due to the thinness of the lamellipodia usedin the live imaging experiments [134] (see Box 3), an ER tubule cross-ing a mitochondrion will mechanically generate a constriction on themitochondrion. This mechanical constriction could be the triggeringfactor that recruits the mitochondrial fission machinery. Moregenerally, mechanically induced fission could play a role in resolvingintracellular membrane entanglements. Alternatively, ER contactcould be required to provide factors needed for fission of the mito-chondrial membrane. These factors could not only be proteins butalso lipids (see Section 1.2), which may be necessary for the mem-brane remodeling that underlies fission.

1.3.2.3. Mitochondrial inheritance. During asymmetric cell division,organelles that cannot be created de novomust be faithfully inherited.In budding yeast, mitochondrial inheritance is partially dependent ona Mitochondrial Myo2 Receptor-related protein 1 (Mmr1) [136]. As itbinds the myosin motor Myo2, Mmr1 is thought to promote thetransport of mitochondria along actin cables from the mother to thebud [136–138]. Yet, MMR1 is also one of a few genes whose messen-ger RNA is transported into the bud [139]. This suggests that Mmr1 isrequired in the bud rather than in the mother, which is inconsistentwith a role in transporting mitochondria from the mother to thebud. To accommodate this observation, Liza Pon's lab proposes anoth-er model: Mmr1 is not required to transport mitochondria to the budbut instead is required to anchor them to the ER at the bud tip [140].In such a model, the Myo2-binding activity is only required to localizeMmr1 to the bud tip, but not to transport mitochondria. Supportingthis model, MMR1 deletion causes a defect in mitochondrial anchor-ing to the bud tip and an increase in retrograde mitochondrialmovement. This ER–mitochondria contact at the bud tip could beestablished by interaction with a fraction of Mmr1 attached to theER, substantiated by the presence of a small fraction of Mmr1 in theMAM fraction [140] (see Box 3).

2. Tethering complexes

The establishment of a functional MCS can be subdivided in anumber of functional steps. The first one is tethering of two mem-branes. In a second step, effector proteins are recruited to the MCSto perform Ca2+ or lipid exchange, or any other function of that par-ticular MCS. However effectors may participate in tethering as much


as tethers could also be effector proteins, which complicates thesearch for protein tethers responsible for the establishment of MCSs.Bona fide tethers should in principle satisfy the following criteria:

1) They should be physically present at MCSs.2) They should provide a tethering force between adjacent organ-

elles, for example through the physical association of proteinsinserted on opposite membranes.

3) Deletion of a tether should abolish the contacts between twoorganelles.

4) Deletion of a tether should affect all physiological processes takingplace at this MCS.

5) Hyperactivation of this tether should increase the number and/orsize of MCSs.

As we will see, very few of the many proposed tethers fulfill all ofthese criteria, probably due to the fact that several redundant tethersexist between a pair of organelles (Table 1), or because MCSs act innetworks where the loss of one MCS can be compensated for by an-other type of MCS.

2.1. The nuclear—vacuole junction

The yeast NVJ is probably the simplest of all known MCS, sinceonly two proteins appear to be necessary and sufficient for its estab-lishment. The NVJ protein 1 (NVJ1) on the outer nuclear membraneinteracts with vacuole-related 8 protein (Vac8) on the vacuolar mem-brane [115,141]. These two proteins alone satisfy all of the criteriadescribed above for a tether. Deletion of either of these proteins issufficient to abolish the NVJ completely, while overexpression in-creases their extent [115,116]. Finally, PMN, the only well-describedprocess happening at the NVJ, is crucially dependent on the presenceof both proteins [116,117]. Other proteins are recruited to the NVJ butdo not appear to play a role in tethering per se, such as the ORP Osh1[93,142], the SMP-containing Nvj2 [105] and the enoyl-CoA reductaseTsc13 [118,119]. These proteins are thus likely effectors of physiolog-ical functions, such as lipid exchange. Interestingly, none of the teth-ering proteins is conserved, while all of the effectors are. Themammalian orthologs of at least two of the effectors are likely to bepresent at MCSs. ORP1L, a metazoan ortholog of Osh1, harbors bothPH and FFAT domains that direct the protein to ER–late endosomecontact sites, and Testis-EXpressed protein 2 (Tex2), the metazoanortholog of Nvj2, has an SMP domain, which has so far only beenfound in MCS-proteins [95,105]. Both proteins may thus be found atMCSs that are analogous to the NVJ, but established and maintainedby other sets of tethers.

2.2. ER–mitochondria tethers

ER–mitochondria contact sites are probably the best understoodMCSs in both lower and higher eukaryotes in terms of the tetheringfactors.

2.2.1. The ERMES complexIn budding yeast, a protein complex that tethers the ER and mito-

chondria was discovered in a forward genetic screen aimed at findingmutants that can only grow in the presence of exogenous artificialER–mitochondria tethering activity [99]. This artificial tether – a GFPmolecule bracketed by mitochondrial and ER targeting sequencesat the N- and C-termini, respectively [51] (see Box 3) – is dispensablein wildtype cells but becomes indispensable for the respiratorygrowth of certain mutants, suggesting that their endogenous tether-ing capacity has been lost. These mutants are found in genes encodingmembers of the ERMES complex, a protein complex made of ERand mitochondrial proteins, thus satisfying our second criterion. Inaccordance to the first criterion, ERMES localizes to few punctate




Table 1Summary of the structural and functional players of membrane contact sites.

ER–PM contact sitesPotential tethers

Orai–Stim STIM (ER calcium sensor) and Orai (PM calcium channel) form a transient complex upon ER calciumstore depletion. This tether may contribute to tethering the PM to the ER membrane.

Section 2.3

Ist2 ER–membrane protein, can bind PM through a PIP-binding C-terminal domain. Section 2.3VAP–ORPs Many ORPs can bind simultaneously to the ER protein VAP, and to PIPs in the PM. Section 2.3Tricalbins The ER-resident tricalbins 1–3 contain two distinct membrane-binding domains, which can bind the PM.

They localize to ER–PM MCSs.Section 2.3

Junctophilins In muscle cells, SR-resident junctophilins connect the SR to the PM, which they contact through theirlipid-binding MORN-domains.

Section 2.3.1

EffectorsMuscle contraction DHPR DHPR is a voltage-gated PM Ca2+ channel that gates RyR at SR–PM contact sites. RyR is a SR Ca2+ channel.

RyR gating by DHPR amplifies the cytosolic Ca2+ influx during muscle contraction.Section 1.1.1

RyRSOCE Orai STIM is an ER Ca2+ sensor and Orai is a PM Ca2+ channel. Upon Ca2+ depletion in the ER, Stim binds to and

gates Orai at ER–PM MCSs.Section 1.1.2

STIMLipid exchange ORPs ORPs exchange sterols for PI4P. They are often localized at various MCSs. Section 1.2.3

Tricalbins Tricalbins harbor an SMP domain, suggesting these proteins could act as LTPs. Section 1.2.4

ER–mitochondria contact sitesPotential tethers

ERMES Complex made of integral ER and OMM proteins. Members of the ERMES complex can be functionallysubstituted by an artificial ER–mitochondria tether.

Section 2.2.1

VDAC–Grp75–IP3Rcomplex

Complex formed by the ER-resident Ca2+ channel IP3R, the cytosolic chaperone Grp75, and thenon-selective channel of the outer membrane VDAC.

Section 2.2.2

Mfn2 An ER-localized subfraction of Mfn2 binds Mfn1/2 in trans on the OMM, thereby tethering the ER tomitochondria. Mfn2 depletion causes ER fragmentation and Ca2+ overload.

Section 2.2.3

VAPB–PTPIP51B Revealed by a yeast to hybrid screen, this complex consists of the ER protein VAPB and the mitochondrialprotein PTPIP51B. Mutations in VAPB are linked to inheritable forms of multiple sclerosis.

Section 2.2.4

Bap31–Fis1 The ER protein Fis1 and the OMM protein Bap31 interact with each other, thereby possibly forming a MCSbetween both organelles. The complex seems to function in apoptosis regulation.

Section 2.2.5

EffectorsMitochondrialCa2+ influx

IP3R IP3Rs are ER-resident ligand-gated channels, which mediate ER Ca2+-release. Section 1.1.3VDAC OMM general diffusion pore for small hydrophilic molecules. Section 1.1.3MCU Mitochondrial Ca2+ uniport responsible for mitochondrial matrix Ca2+ uptake. Section 1.1.3

Lipid exchange ERMES Three members of the ERMES complex harbor an SMP domain, suggesting these proteins could act as LTPs. Section 1.2.4Mitochondrialdynamics

Miro GTPase involved in mitochondrial movement, found at ER–mitochondria MCSs. Yeast homolog is part ofERMES complex.

Section 1.3.2.1

DRP1 GTPase involved in fission of mitochondrial membranes. Drp1 and mitochondrial fission events areenriched at ER–mitochondria contact sites.

Section 1.3.2.2

Mmr1 Yeast myosin binding protein, required for proper transmission of mitochondria to the bud. May berequired to anchor mitochondria to the ER at the bud tip.

Section 1.3.2.3

ER–vacuole/lysosome contact sitesPotential tethers

Vac8–Nvj1 complex Complex composed of vacuolar membrane protein Vac8 and outer nuclear membrane protein Nvj1,connecting both organelles at nucleus–vacuole junctions (NVJ). Its presence is crucial for establishmentof the NVJ and for downstream processes.

Section 2.1

EffectorsPMN V-ATPase Involved in formation of the electrochemical gradient across the vacuolar membrane at the NVJ sites

where the PMN takes place.Section 1.3.1

Lipid exchange Osh1 Enriched at the NVJ. ORPs like Osh1 exchange sterols for PI4P. Section 1.2.3Nvj2 Enriched at the NVJ. Harbors an SMP domain, suggesting it could act as LTPs. Section 1.2.4

Fatty-acidsynthesis

Tsc13 A polytopic ER-resident protein enriched at NVJ and required for very-long-chain fatty acids biosynthesisin yeast.

Section 1.3.1

ER–Golgi contact sitesPotential tethers

VAP–ORPs ORPs, CERTs and PITPs can bind both the ER (via association with VAP) and PIPs in late secretoryorganelles, including the Golgi.

Section 2.4VAP–CERTVAP–PITP

EffectorsLipid exchange ORPs ORPs exchange sterols for PI4P. They are often localized at various MCSs. Section 1.2.3

CERT CERT extracts ceramide from the ER and transfers it to the Golgi. Section 1.2.3PITPs PITPs exchange PI for PC between various membranes. Section 1.2.3

ER–lipid droplets contact sitesPotential tethers

FATP1–DGAT2 The diacylglycerol acyltransferase 2 (DGAT2) in lipid droplets may interact with the acyl-CoA synthase(FATP1) in the ER membrane to couple triglycerides synthesis and their deposition in lipid droplets.

Sections 1.2.5and 2.4

ER–late endosome contact sitesPotential tethers

VAP–ORP1L Late-endosome-bound-ORP1L contacts the ER protein VAP in condition of cholesterol depletion in theendosome.

Section 2.4

PTP1B–EGFR PTP1B is an ER-resident phosphatase. It dephosphorylates internalized EGFR. Expression of a phosphatase mutantincreases ER–late-endosome association.

Section 2.4






structures at the ER–mitochondria interface where the artificial tetheris also enriched.

Whether ERMES is strictly a tether is a matter of debate (seeSection 1.2.4). On the one hand, it can be substituted by an artificialtether, suggesting that its function is limited to tethering. On theother hand, only two of the four essential components of ERMES –

Mdm12 and Mdm34 – can be efficiently replaced by the artificialtether, showing that the other ERMES components have functions be-yond tethering [99]. Yet, deletion of Mdm12 or Mdm34 causes a com-plete collapse of the ERMES complex, suggesting that these additionalfunctions are carried out by the remaining ERMES components out-side of the complex, therefore outside of MCSs. Alternatively, it isalso possible that the artificial tether helps hold together an other-wise crippled ERMES complex lacking Mdm12 or Mdm34.

2.2.2. The VDAC–GRP75–IP3R complexIn mammalian cells, Ca2+ transfers mouth-to-mouth from the ER

IP3R to the mitochondrial porin VDAC (voltage dependent anionchannel, see Section 1.1.3.1). Both proteins were actually found tointeract via a cytosolic Hsp70, the glucose-regulated-protein 75(GRP75) [143]. This indirect interaction of IP3R and VDAC could pro-vide a tethering force between the ER and mitochondria on top oftheir role in Ca2+ homeostasis. In accordance, knockdown of GRP75decreases ER–mitochondria Ca2+ exchange.

2.2.3. Mitofusin-2Mitofusin-2 (Mfn2) is a well-characterized mitochondrial fusion

factor and Mfn2 knockout (KO) cells have fragmented mitochondria.However, unexpectedly, they also display many ER-related problems.Their ER is fragmented and has a resting [Ca2+] nearly twice that ofwildtype. Having an ER phenotype in a mitochondrial mutant mayhint towards a lack of ER–mitochondria tethering. A small fractionof Mfn2 is found in the MAM fraction, suggesting that it is insertedin the ER membrane, where it can interact with both Mfn2 andMfn1 in the mitochondrial membrane, thereby tethering the twoorganelles [132]. Indeed, the ER fragmentation but not the mitochon-drial fragmentation of the Mfn2 KO cells can be rescued by expressinga Mfn2 transgene fused to an ER targeting sequence, suggesting thatthis transgene rescues tethering, but not the mitochondrial fusionfunction of Mfn2. The nature and the composition of the MAMfraction (see Box 3) are affected in Mfn2 KO cells, consistent with aloosening of the ER–mitochondria connection. ER–mitochondriaCa2+ transfer is also decreased by ~25% [132].

2.2.4. The VAPB–PTPIP51 complexThe ER protein VAP interacts with the FFAT motif found in a

plethora of proteins localized at MCSs between the ER and other mem-branes, such as the Golgi, the PM and the endo-/lysosome. The VAPparalog VAPB was further used as bait in a yeast two-hybrid screen,which led to the identification of a binding partner, the protein-tyrosine-phosphatase-interacting protein (PTPIP) 51 [144]. SincePTPIP51 is an OMM integral protein, PTPIP51–VAPB interaction shouldonly take place at ER–mitochondria MCSs. In support of this notion, afair fraction of VAPB is found in the MAM fraction (see Box 3).PTPIP51 does not harbor a FFAT motif, thus the interaction between itand VAPB may happen via other mechanisms. Knocking-down eitherVAPB or PTPIP51 causes a ~10% decrease in ER–mitochondrial Ca2+ ex-change [144]. Interestingly, a mutation in VAPB (P56S) known to causetype-8 amyotrophic lateral sclerosis (ALS), shows an increased affinityfor PTPIP51, enrichment in the MAM fraction and mitochondrialcolocalization [144]. Overexpression of this mutant but not of thewildtype form of VAPB causes a ~1.4 fold increase in ER–mitochondriaCa2+ exchange. VAPBP56S also decreases the amount of motile mito-chondria in the axon by decreasing the association of Miro with micro-tubules. This decrease can be reversed by expressing a mutant of Mirothat cannot bind Ca2+, suggesting that VAPBP56S acts upon Miro by


increasing [Ca2+] at the ER–mitochondria MCS [123]. These findingscould bring a new twist to the study of the etiology of ALS.

Although PTPIP51 does not have an FFATmotif, the involvement ofVAPB suggests that VAPs might constitute a hub for several MCSs ofthe ER with other cellular organelles.

2.2.5. The Bap31–Fis1 complexB-cell-receptor-associated protein 31 (Bap31) is an ER protein

that is cleaved by caspases during apoptosis. Fission 1 homologue(Fis1) is the metazoan ortholog of a yeast OMM protein importantfor recruiting dynamin-related protein to mitochondria during mito-chondrial fission. While this original role has been assumed by Mff(see Section 1.3.2.2), metazoan Fis1 acquired a new role in the regu-lation of apoptosis. Overexpression of Fis1 leads to caspase processingof Bap31, while knockdown of Fis1 inhibits it. Both proteins interactphysically and constitutively, and they recruit procaspase-8 upon ap-optosis induction [145]. This constitutive interaction suggests thatFis1 and Bap31 could provide a tethering force between the ER andthe mitochondria. However, the function of this complex has notbeen studied beyond its role in apoptosis promotion.

2.2.6. Mmr1As stated in Section 1.3.2.3, yeast Mmr1 may establish an MCS in

the daughter cell to anchor mitochondria to the bud-tip [140]. Themolecular details of this tethering are unclear; in particular, Mmr1'spartner(s) responsible for targeting it to the ER have still to beidentified.

MMR1 shows synthetic genetic interactions with members of theERMES complex (2.2.1)[99], suggesting that both contact sites mayhave partially overlapping roles.

2.3. ER–plasma-membrane tethers

While many proteins are known to localize either constitutively orinducibly to ER–PM MCSs, much less is known about the tetheringcomplexes that establish them. Many of the ER–PM MCS-localizedproteins have a capacity to bind both membranes. Thus many can in-crease ER–PM association when overexpressed, however, no knownloss-of-function leads to a complete loss of ER–PM tethering.

Lipid exchangers could act as tethers. In yeast, many ORPs aretargeted to ER–PM MCSs (reviewed in [146]). Two of them – Osh2and Osh3 – have both a FFAT and a PH domain, which could providea tethering force by binding to ER VAP proteins (Scs2 and Scs22 inyeast) and to PM phosphoinositides simultaneously. Another potentialtether comes from the SMP-containing extended-synaptotagmin/tricalbins proteins (Esyt1-3 in humans, Tcb1-3 in yeast) [105].Yeast tricalbins accumulate at ER–PM MCS. They are targeted to theER via their N-terminal transmembrane domain and interact withthe PM via their C-termini composed of one SMP and repetitions ofC2-domains [147]. They may thus participate in ER–PM tethering.

STIM is engaged in the ER membrane with its transmembranesequence. A conformation switch following ER Ca2+ depletion ex-poses a C-terminal polybasic stretch that binds to negatively chargedphospholipids in the PM, providing the basis for a transient tether[23,24]. Indeed, the extent of ER–PM contact is increased upon SOCEactivation [25,40]. Moreover, overexpressing a constitutively activeSTIM induces the formation of tethering structures not only betweenthe ER and the PM, but also between adjacent ER sheets, suggestingthat STIM has an intrinsic tethering activity [40]. Yet, activated STIMmay just extend preexisting structures, instead of creating newones. Indeed, successive SOCE activations recruit STIM repeatedly tothe same sites, suggesting that these sites remain in the absence ofSTIM [37].

The yeast protein Ist2 is an ER putative ion channel, which, muchlike STIM, harbors a C-terminal polybasic domain [148–150]. Whenoverexpressed in yeast or mammalian cells, Ist2 can increase the





extent of ER–PM tethering [148,151]. This is particularly spectacularin yeast where Ist2 overexpression can literally velcro the nuclear en-velope against the cell cortex. However, IST2 deletion only has a littleinfluence on the fact that a large fraction of the ER is in direct contactwith the plasma membrane, suggesting that other tethers must exist.

2.3.1. In muscle cellsDHPR and RyR1 physically interact in the membrane junction

complexes. However, this interaction is unlikely to provide the struc-tural basis of these complexes as overexpressing both does not in-crease the extent of SR–PM MCSs [152]. In an attempt to identifystructural components of junctional membrane complexes (JMCs,see Section 1.1.1), the group of Hiroshi Takeshima created a libraryof monoclonal antibodies raised against muscular membrane pro-teins, and screened this library for JMC specific epitopes using Elec-tron Microscopy (EM). This approach led to the identification ofjunctophilins, a family of proteins with similar architectures [153].Junctophilins are anchored in the SR via their C-terminal transmem-brane domain, while their N-termini consist of repetitions of mem-brane occupation and recognition nexuses (MORN). The MORNshows a specific affinity for the PM, therefore providing the basis fora tether. Mutants of some of the junctophilins show disorganizationof the JMC, while ectopic expression of junctophilins in non-musclecells leads to the formation of new ER–PM MCSs. Thus, junctophilinsare likely to contribute to the establishment of SR–PM tethers inmuscles.

2.4. Other MCSs

Tethering factors involving other cellular structures are much lesscharacterized. The interaction of VAP proteins with the FFAT domainsof many proteins, such as CERT or ORPs proteins, could contribute tothe tethering of the ER to various organelles [76,77,81]. We alreadymentioned the existence of a contact site between the ER and thelate endosome. This contact site may be established by ORP1L via itsFFAT and PH domains in an inducible fashion, depending on the endo-some cholesterol content [94,95]. Whether ORP1L is the actual tether isnot clear. Moreover it is not the only candidate for an ER–late endo-some tether. Like most plasma-membrane receptor tyrosine kinases,the epidermal growth factor receptor (EGFR) is endocytosed upon acti-vation and subsequently targeted for degradation in the late endo-some/lysosome. On the late endosome, EGFR is dephosphorylated bythe protein tyrosine phosphatase 1B (PTP1B; not to be confused withPTPIP51, mentioned in Section 2.2.4, nor with PITPs, mentioned inSection 1.2.1), which is an ER protein [154]. Therefore, the topologyof this dephosphorylation reaction in trans implies the formation ofan ER–late-endosome MCS. When a catalytically inactive form ofPTP1B is expressed, the enzyme–substrate complex cannot be resolved.Therefore their normally transient interaction becomes durable. As aresult, EM-observable contact sites are created and maintained be-tween the ER and late endosomes, indicating that these two proteinscould be sufficient for the establishment of new MCSs [155]. The inter-action of EGFR with wildtype PTP1B may, however, be too transient tocreate a real tethering force, thus additional tethers may be involved(Table 1).

What tethers the ER closely around lipid droplets is unclear. Arecent study proposed that the fatty-acyl Coenzyme A ligase FATP1,an ER protein that catalyzes the first step of triglyceride biosynthesis,interacts with the DAG acyltransferase DGAT2 on the surface of LDs,providing a potential tethering force [114]. This model not only im-plies that DGAT2 localizes to LDs, but also that it is physically insertedin the monolayer of the LD. DGAT2 harbors two closely-spaced trans-membrane domains with no hydrophilic aminoacid in between them,thus perfectly fitted to insert into the LD outer-layer. However, DGAT2is not always colocalized with LDs. It can be seen throughout the ERin certain conditions and it even cofractionates with mitochondria-


associated membranes [156] (see Box 3). How DGAT2 translocates fromthe ER membrane to the LD outer-layer is completely unknown andthe relative importance of lipid trafficking by this MCS versus lipid traf-ficking by direct fusion of the ER-membrane with the LD outer-layerneeds to be clarified.

3. Conclusions and perspectives

Biochemistry aims to understand biology in terms of chemicalreactions. Yet the cell is far from being a homogenous chemical reac-tor. Microdomains and subcompartments promote certain reactionsand inhibit others, and organelles play a prime role in this compart-mentalization [157]. MCSs are emerging as a new layer of complexitythat allows specific and directional exchange of metabolites andinformation between subcompartments. Long confined to phenome-nological observations, the biology of MCS has undergone a drasticspeed-up in the recent yearswith the identification ofmany factors in-volved in their structure and/or function. But many questions are stillleft unanswered. Are there other molecules besides lipids and Ca2+

that are specifically exchanged at MCSs? Do the different kinds oftethers between the same organelles collaborate in establishing oneMCS or does each one establish a specialized MCS linked to a specificfunction? How important is non-vesicular lipid transport within theendomembrane system as compared to vesicular lipid transport?What LTPs are operating between the ER and the mitochondria?What is really the MAM fraction? Finally, although the functionsof most contact sites in lipid and Ca2+ exchange are similar, the fac-tors involved seem unrelated between various MCSs and betweendifferent clades, begging the question of the origin of these struc-tures and of how they have been maintained throughout eukaryoticevolution.

We can expect that upcoming research will answer these ques-tions and reveal the common organizing principles of this privilegedtype of interorganellar communication.

Acknowledgements

We would like to warmly thank Alicia E. Smith for critical com-ments on the manuscript. This work was supported by grants fromthe Swiss National Science Foundation (PP00P3_13365) and fromthe ETH (ETH 21 11-2).

References

[1] D.E. Clapham, Calcium signaling, Cell 131 (2007) 1047–1058.[2] W. Dowhan, Molecular basis for membrane phospholipid diversity: why are

there so many lipids? Annu. Rev. Biochem. 66 (1997) 199–232.[3] A. Sandow, Excitation–contraction coupling in muscular response, Yale J. Biol.

Med. 25 (1952) 176–201.[4] F. Brette, C. Orchard, Resurgence of cardiac t-tubule research, Physiology

(Bethesda) 22 (2007) 167–173.[5] B.A. Block, T. Imagawa, K.P. Campbell, C. Franzini-Armstrong, Structural evi-

dence for direct interaction between the molecular components of the trans-verse tubule/sarcoplasmic reticulum junction in skeletal muscle, J. Cell Biol.107 (1988) 2587–2600.

[6] A. Fabiato, Calcium-induced release of calcium from the cardiac sarcoplasmicreticulum, Am. J. Physiol. 245 (1983) C1–C14.

[7] R.A. Bannister, Bridging the myoplasmic gap: recent developments in skeletalmuscle excitation–contraction coupling, J. Muscle Res. Cell Motil. 28 (2007)275–283.

[8] K.G. Beam, R.A. Bannister, Looking for answers to EC coupling's persistent ques-tions, J. Gen. Physiol. 136 (2010) 7–12.

[9] R.T. Rebbeck, Y. Karunasekara, E.M. Gallant, P.G. Board, N.A. Beard, M.G.Casarotto, A.F. Dulhunty, The β(1a) subunit of the skeletal DHPR binds to skele-tal RyR1 and activates the channel via its 35-residue C-terminal tail, Biophys. J.100 (2011) 922–930.

[10] H. Coe, M. Michalak, Calcium binding chaperones of the endoplasmic reticulum,Gen. Physiol. Biophys. 28 Spec No (2009) F96–F103.

[11] S. Feske, Calcium signalling in lymphocyte activation and disease, Nat. Rev.Immunol. 7 (2007) 690–702.

[12] J. Liou, M.L. Kim, W. Do Heo, J.T. Jones, J.W. Myers, J.E. Ferrell, T. Meyer, STIM is aCa2+ sensor essential for Ca2+-store-depletion-triggered Ca2+ influx, Curr. Biol.15 (2005) 1235–1241.





[13] J. Roos, P.J. DiGregorio, A.V. Yeromin, K. Ohlsen, M. Lioudyno, S. Zhang, O. Safrina,J.A. Kozak, S.L. Wagner, M.D. Cahalan, G. Veliçelebi, K.A. Stauderman, STIM1, anessential and conserved component of store-operated Ca2+ channel function, J.Cell Biol. 169 (2005) 435–445.

[14] M. Vig, C. Peinelt, A. Beck, D.L. Koomoa, D. Rabah, M. Koblan-Huberson, S. Kraft,H. Turner, A. Fleig, R. Penner, J.-P. Kinet, CRACM1 is a plasma membrane proteinessential for store-operated Ca2+ entry, Science 312 (2006) 1220–1223.

[15] S. Feske, Y. Gwack, M. Prakriya, S. Srikanth, S.-H. Puppel, B. Tanasa, P.G. Hogan,R.S. Lewis, M. Daly, A. Rao, A mutation in Orai1 causes immune deficiency byabrogating CRAC channel function, Nature 441 (2006) 179–185.

[16] S.L. Zhang, A.V. Yeromin, X.H.-F. Zhang, Y. Yu, O. Safrina, A. Penna, J. Roos, K.aStauderman, M.D. Cahalan, Genome-wide RNAi screen of Ca(2+) influxidentifies genes that regulate Ca(2+) release-activated Ca(2+) channelactivity, Proc. Natl. Acad. Sci. U. S. A. 103 (2006) 9357–9362.

[17] A.V. Yeromin, S.L. Zhang, W. Jiang, Y. Yu, O. Safrina, M.D. Cahalan, Molecularidentification of the CRAC channel by altered ion selectivity in a mutant ofOrai, Nature 443 (2006) 226–229.

[18] M. Prakriya, S. Feske, Y. Gwack, S. Srikanth, A. Rao, P.G. Hogan, Orai1 is an essentialpore subunit of the CRAC channel, Nature 443 (2006) 230–233.

[19] J. Liou, M. Fivaz, T. Inoue, T. Meyer, Live-cell imaging reveals sequential oligo-merization and local plasma membrane targeting of stromal interaction mole-cule 1 after Ca2+ store depletion, Proc. Natl. Acad. Sci. U. S. A. 104 (2007)9301–9306.

[20] P.B. Stathopulos, G.-Y. Li, M.J. Plevin, J.B. Ames, M. Ikura, Stored Ca2+

depletion-induced oligomerization of stromal interaction molecule 1 (STIM1)via the EF–SAM region: an initiation mechanism for capacitive Ca2+ entry,J. Biol. Chem. 281 (2006) 35855–35862.

[21] P.B. Stathopulos, L. Zheng, G.-Y. Li, M.J. Plevin, M. Ikura, Structural and mecha-nistic insights into STIM1-mediated initiation of store-operated calcium entry,Cell 135 (2008) 110–122.

[22] M. Muik, M. Fahrner, R. Schindl, P. Stathopulos, I. Frischauf, I. Derler, P. Plenk, B.Lackner, K. Groschner, M. Ikura, C. Romanin, STIM1 couples to ORAI1 via an in-tramolecular transition into an extended conformation, EMBO J. 30 (2011)1678–1689.

[23] C.M. Walsh, M. Chvanov, L.P. Haynes, O.H. Petersen, A.V. Tepikin, R.D. Burgoyne,Role of phosphoinositides in STIM1 dynamics and store-operated calcium entry,Biochem. J. 425 (2010) 159–168.

[24] M.K. Korzeniowski, M.A. Popovic, Z. Szentpetery, P. Varnai, S.S. Stojilkovic, T.Balla, Dependence of STIM1/Orai1-mediated calcium entry on plasma mem-brane phosphoinositides, J. Biol. Chem. 284 (2009) 21027–21035.

[25] M.M. Wu, J. Buchanan, R.M. Luik, R.S. Lewis, Ca2+ store depletion causes STIM1to accumulate in ER regions closely associated with the plasma membrane,J. Cell Biol. 174 (2006) 803–813.

[26] S.L. Zhang, Y. Yu, J. Roos, J.A. Kozak, T.J. Deerinck, M.H. Ellisman, K.A. Stauderman,M.D. Cahalan, STIM1 is a Ca2+ sensor that activates CRAC channels and migratesfrom the Ca2+ store to the plasma membrane, Nature 437 (2005) 902–905.

[27] Y. Baba, K. Hayashi, Y. Fujii, A. Mizushima, H. Watarai, M. Wakamori, T. Numaga,Y. Mori, M. Iino, M. Hikida, T. Kurosaki, Coupling of STIM1 to store-operatedCa2+ entry through its constitutive and inducible movement in the endoplasmicreticulum, Proc. Natl. Acad. Sci. U. S. A. 103 (2006) 16704–16709.

[28] R.M. Luik, M.M. Wu, J. Buchanan, R.S. Lewis, The elementary unit ofstore-operated Ca2+ entry: local activation of CRAC channels by STIM1 at ER–plasma membrane junctions, J. Cell Biol. 174 (2006) 815–825.

[29] J.C. Mercer, W.I. Dehaven, J.T. Smyth, B. Wedel, R.R. Boyles, G.S. Bird, J.W. Putney,Large store-operated calcium selective currents due to co-expression of Orai1 orOrai2 with the intracellular calcium sensor, Stim1, J. Biol. Chem. 281 (2006)24979–24990.

[30] P. Xu, J. Lu, Z. Li, X. Yu, L. Chen, T. Xu, Aggregation of STIM1 underneath theplasma membrane induces clustering of Orai1, Biochem. Biophys. Res. Commun.350 (2006) 969–976.

[31] T. Kawasaki, I. Lange, S. Feske, A minimal regulatory domain in the C terminus ofSTIM1 binds to and activates ORAI1 CRAC channels, Biochem. Biophys. Res.Commun. 385 (2009) 49–54.

[32] C.Y. Park, P.J. Hoover, F.M. Mullins, P. Bachhawat, E.D. Covington, S. Raunser, T.Walz, K.C. Garcia, R.E. Dolmetsch, R.S. Lewis, STIM1 clusters and activatesCRAC channels via direct binding of a cytosolic domain to Orai1, Cell 136(2009) 876–890.

[33] J.P. Yuan, W. Zeng, M.R. Dorwart, Y.-J. Choi, P.F. Worley, S. Muallem, SOAR andthe polybasic STIM1 domains gate and regulate Orai channels, Nat. Cell Biol.11 (2009) 337–343.

[34] Y. Wang, X. Deng, Y. Zhou, E. Hendron, S. Mancarella, M.F. Ritchie, X.D. Tang, Y.Baba, T. Kurosaki, Y. Mori, J. Soboloff, D.L. Gill, STIM protein coupling in the acti-vation of Orai channels, Proc. Natl. Acad. Sci. U. S. A. 106 (2009) 7391–7396.

[35] M. Muik, M. Fahrner, I. Derler, R. Schindl, J. Bergsmann, I. Frischauf, K. Groschner,C. Romanin, A cytosolic homomerization and a modulatory domain withinSTIM1 C terminus determine coupling to ORAI1 channels, J. Biol. Chem. 284(2009) 8421–8426.

[36] Y. Zhou, P. Meraner, H.T. Kwon, D. Machnes, M. Oh-hora, J. Zimmer, Y. Huang, A.Stura, A. Rao, P.G. Hogan, STIM1 gates the store-operated calcium channel ORAI1in vitro, Nat. Struct. Mol. Biol. 17 (2010) 112–116.

[37] J.T. Smyth, W.I. Dehaven, G.S. Bird, J.W. Putney, Ca2+-store-dependent and-independent reversal of Stim1 localization and function, J. Cell Sci. 121(2008) 762–772.

[38] M. Chvanov, C.M. Walsh, L.P. Haynes, S.G. Voronina, G. Lur, O.V. Gerasimenko, R.Barraclough, P.S. Rudland, O.H. Petersen, R.D. Burgoyne, A.V. Tepikin, ATP deple-tion induces translocation of STIM1 to puncta and formation of STIM1-ORAI1


clusters: translocation and re-translocation of STIM1 does not require ATP,Pflügers Archiv/: European, J. Physiol. 457 (2008) 505–517.

[39] G. Lur, L.P. Haynes, I.A. Prior, O.V. Gerasimenko, S. Feske, O.H. Petersen, R.D.Burgoyne, A.V. Tepikin, Ribosome-free terminals of rough ER allow formationof STIM1 puncta and segregation of STIM1 from IP(3) receptors, Curr. Biol. 19(2009) 1648–1653.

[40] L. Orci, M. Ravazzola, M. Le Coadic, W.-W. Shen, N. Demaurex, P. Cosson, From thecover: STIM1-induced precortical and cortical subdomains of the endoplasmicreticulum, Proc. Natl. Acad. Sci. U. S. A. 106 (2009) 19358–19362.

[41] H. Jousset, M. Frieden, N. Demaurex, STIM1 knockdown reveals thatstore-operated Ca2+ channels located close to sarco/endoplasmic Ca2+ ATPases(SERCA) pumps silently refill the endoplasmic reticulum, J. Biol. Chem. 282(2007) 11456–11464.

[42] I.M. Manjarrés, M.T. Alonso, J. García-Sancho, Calcium entry-calcium refilling(CECR) coupling between store-operated Ca(2+) entry and sarco/endoplasmicreticulum Ca(2+)-ATPase, Cell Calcium 49 (2011) 153–161.

[43] A. Sampieri, A. Zepeda, A. Asanov, L. Vaca, Visualizing the store-operatedchannel complex assembly in real time: identification of SERCA2 as a newmember, Cell Calcium 45 (2009) 439–446.

[44] E. Carafoli, The interplay of mitochondria with calcium: an historical appraisal,Cell Calcium 52 (2012) 1–8.

[45] J.M. Baughman, F. Perocchi, H.S. Girgis, M. Plovanich, C.A. Belcher-Timme, Y.Sancak, X.R. Bao, L. Strittmatter, O. Goldberger, R.L. Bogorad, V. Koteliansky,V.K. Mootha, Integrative genomics identifies MCU as an essential componentof the mitochondrial calcium uniporter, Nature 476 (2011) 341–345.

[46] D. De Stefani, A. Raffaello, E. Teardo, I. Szabò, R. Rizzuto, A forty-kilodalton pro-tein of the inner membrane is the mitochondrial calcium uniporter, Nature 476(2011) 336–340.

[47] R. Rizzuto, M. Brini, M. Murgia, T. Pozzan, Microdomains with high Ca2+ close toIP3-sensitive channels that are sensed by neighboring mitochondria, Science262 (1993) 744–747.

[48] R. Rizzuto, P. Pinton, W. Carrington, F.S. Fay, K.E. Fogarty, L.M. Lifshitz, R.A. Tuft,T. Pozzan, Close contacts with the endoplasmic reticulum as determinants ofmitochondrial Ca2+ responses, Science 280 (1998) 1763–1766.

[49] G. Csordás, P. Várnai, T. Golenár, S. Roy, G. Purkins, T.G. Schneider, T. Balla, G.Hajnóczky, Imaging interorganelle contacts and local calcium dynamics at theER–mitochondrial interface, Mol. Cell 39 (2010) 121–132.

[50] M. Giacomello, I. Drago, M. Bortolozzi, M. Scorzeto, A. Gianelle, P. Pizzo, T.Pozzan, Ca2+ hot spots on the mitochondrial surface are generated by Ca2+ mo-bilization from stores, but not by activation of store-operated Ca2+ channels,Mol. Cell 38 (2010) 280–290.

[51] G. Csordás, C. Renken, P. Várnai, L. Walter, D. Weaver, K.F. Buttle, T. Balla, C.A.Mannella, G. Hajnóczky, Structural and functional features and significance ofthe physical linkage between ER and mitochondria, J. Cell Biol. 174 (2006)915–921.

[52] R.M. Denton, Regulation of mitochondrial dehydrogenases by calcium ions,Biochim. Biophys. Acta 1787 (2009) 1309–1316.

[53] A.I. Tarasov, E.J. Griffiths, G.A. Rutter, Regulation of ATP production by mitochon-drial Ca(2+), Cell Calcium 52 (2012) 28–35.

[54] C. Cárdenas, R.A. Miller, I. Smith, T. Bui, J. Molgó, M. Müller, H. Vais, K.-H. Cheung,J. Yang, I. Parker, C.B. Thompson, M.J. Birnbaum, K.R. Hallows, J.K. Foskett, Essen-tial regulation of cell bioenergetics by constitutive InsP3 receptor Ca2+ transferto mitochondria, Cell 142 (2010) 270–283.

[55] G. Hajnóczky, R. Hager, A.P. Thomas, Mitochondria suppress local feedbackactivation of inositol 1,4,5-trisphosphate receptors by Ca2+, J. Biol. Chem. 274(1999) 14157–14162.

[56] E. Boitier, R. Rea, M.R. Duchen, Mitochondria exert a negative feedback on thepropagation of intracellular Ca2+ waves in rat cortical astrocytes, J. Cell Biol.145 (1999) 795–808.

[57] R.A. Haworth, D.R. Hunter, The Ca2+-induced membrane transition in mito-chondria II Nature of the Ca2+ trigger site, Arch. Biochem. Biophys. 195 (1979)460–467.

[58] P. Bernardi, A. Rasola, Calcium and cell death: the mitochondrial connection,Subcell. Biochem. 45 (2007) 481–506.

[59] P. Pinton, D. Ferrari, P. Magalhães, K. Schulze-Osthoff, F. Di Virgilio, T. Pozzan, R.Rizzuto, Reduced loading of intracellular Ca(2+) stores and downregulation ofcapacitative Ca(2+) influx in Bcl-2-overexpressing cells, J. Cell Biol. 148 (2000)857–862.

[60] L. Scorrano, S.A. Oakes, J.T. Opferman, E.H. Cheng, M.D. Sorcinelli, T. Pozzan, S.J.Korsmeyer, BAX and BAK regulation of endoplasmic reticulum Ca2+: a controlpoint for apoptosis, Science 300 (2003) 135–139.

[61] M. Chami, B. Oulès, G. Szabadkai, R. Tacine, R. Rizzuto, P. Paterlini-Bréchot, Roleof SERCA1 truncated isoform in the proapoptotic calcium transfer from ER tomitochondria during ER stress, Mol. Cell 32 (2008) 641–651.

[62] T. Jayaraman, A. Marks, T cells deficient in inositol 1,4,5-trisphosphate receptorare resistant to apoptosis, Mol. Cell. Biol. 17 (1997) 3005–3012.

[63] H. Sugawara, M. Kurosaki, M. Takata, T. Kurosaki, Genetic evidence for involve-ment of type 1, type 2 and type 3 inositol 1,4,5-trisphosphate receptors in sig-nal transduction through the B-cell antigen receptor, EMBO J. 16 (1997)3078–3088.

[64] C.R.H. Raetz, in: J.N.H., G.B.A.B.T.-N.C. (Eds.), Chapter 11 Genetic Control ofPhospholipid Bilayer Assembly, BiochemistryElsevier, 1982, pp. 435–477.

[65] R.G. Sleight, Intracellular lipid transport in eukaryotes, Annu. Rev. Physiol. 49(1987) 193–208.

[66] G. vanMeer, D.R. Voelker, G.W. Feigenson, Membrane lipids: where they are andhow they behave, Nat. Rev. Mol. Cell Biol. 9 (2008) 112–124.





[67] G. van Meer, A.I.P.M. de Kroon, Lipid map of the mammalian cell, J. Cell Sci. 124(2011) 5–8.

[68] J.E. Vance, E.J. Aasman, R. Szarka, Brefeldin A does not inhibit the movement ofphosphatidylethanolamine from its sites for synthesis to the cell surface,J. Biol. Chem. 266 (1991) 8241–8247.

[69] J.W. Kok, T. Babia, K. Klappe, G. Egea, D. Hoekstra, Ceramide transport from en-doplasmic reticulum to Golgi apparatus is not vesicle-mediated, Biochem. J. 333(Pt 3) (1998) 779–786.

[70] L. Urbani, R.D. Simoni, Cholesterol and vesicular stomatitis virus G protein takeseparate routes from the endoplasmic reticulum to the plasma membrane,J. Biol. Chem. 265 (1990) 1919–1923.

[71] C. Osman, D.R. Voelker, T. Langer, Making heads or tails of phospholipids inmitochondria, J. Cell Biol. 192 (2011) 7–16.

[72] J.E. Vance, Phospholipid synthesis in a membrane fraction associated withmitochondria, J. Biol. Chem. 265 (1990) 7248–7256.

[73] S. Lev, Non-vesicular lipid transport by lipid-transfer proteins and beyond, Nat.Rev. Mol. Cell Biol. 11 (2010) 739–750.

[74] A. Toulmay, W.A. Prinz, Lipid transfer and signaling at organelle contact sites:the tip of the iceberg, Curr. Opin. Cell Biol. 23 (2011) 458–463.

[75] K. Hanada, Discovery of the molecular machinery CERT for endoplasmicreticulum-to-Golgi trafficking of ceramide, Mol. Cell. Biochem. 286 (2006) 23–31.

[76] C.J.R. Loewen, A. Roy, T.P. Levine, A conserved ER targeting motif in three fami-lies of lipid binding proteins and in Opi1p binds VAP, EMBO J. 22 (2003)2025–2035.

[77] D. Peretti, N. Dahan, E. Shimoni, K. Hirschberg, S. Lev, Coordinated lipid transfer be-tween the endoplasmic reticulum and theGolgi complex requires theVAP proteinsand is essential for Golgi-mediated transport, Mol. Biol. Cell 19 (2008) 3871–3884.

[78] R.J. Haslam, H.B. Koide, B.A. Hemmings, Pleckstrin domain homology, Nature363 (1993) 309–310.

[79] M.A. Lemmon, K.M. Ferguson, Molecular determinants in pleckstrin homologydomains that allow specific recognition of phosphoinositides, Biochem. Soc.Trans. 29 (2001) 377–384.

[80] B.J. Mayer, R. Ren, K.L. Clark, D. Baltimore, A putative modular domain present indiverse signalling molecules, Cell 73 (1993) 629–630.

[81] M.J. Rebecchi, S. Scarlata, Pleckstrin homology domains: a common fold withdiverse functions, Annu. Rev. Biophys. Biomol. Struct. 27 (1998) 503–528.

[82] G. Drin, J.-F. Casella, R. Gautier, T. Boehmer, T.U. Schwartz, B. Antonny, A generalamphipathic alpha-helical motif for sensing membrane curvature, Nat. Struct.Mol. Biol. 14 (2007) 138–146.

[83] M.D. Yoder, L.M. Thomas, J.M. Tremblay, R.L. Oliver, L.R. Yarbrough, G.M.Helmkamp, Structure of a multifunctional protein mammalian phos-phatidylinositol transfer protein complexed with phosphatidylcholine, J. Biol.Chem. 276 (2001) 9246–9252.

[84] S.J. Tilley, A. Skippen, J. Murray-Rust, P.M. Swigart, A. Stewart, C.P. Morgan, S.Cockcroft, N.Q. McDonald, Structure-function analysis of human [corrected]phosphatidylinositol transfer protein alpha bound to phosphatidylinositol,Structure 12 (2004) 317–326.

[85] N. Kudo, K. Kumagai, N. Tomishige, T. Yamaji, S. Wakatsuki, M. Nishijima, K.Hanada, R. Kato, Structural basis for specific lipid recognition by CERT responsi-ble for nonvesicular trafficking of ceramide, Proc. Natl. Acad. Sci. U. S. A. 105(2008) 488–493.

[86] Y.J. Im, S. Raychaudhuri, W.A. Prinz, J.H. Hurley, Structural mechanism for sterolsensing and transport by OSBP-related proteins, Nature 437 (2005) 154–158.

[87] M. de Saint-Jean, V. Delfosse, D. Douguet, G. Chicanne, B. Payrastre, W. Bourguet,B. Antonny, G. Drin, Osh4p exchanges sterols for phosphatidylinositol4-phosphate between lipid bilayers, J. Cell Biol. 195 (2011) 965–978.

[88] C.T. Beh, C.R. McMaster, K.G. Kozminski, A.K. Menon, A detour for yeast oxysterolbinding proteins, J. Biol. Chem. 287 (2012) 11481–11488.

[89] C.J. Stefan, A.G. Manford, D. Baird, J. Yamada-Hanff, Y. Mao, S.D. Emr, Osh pro-teins regulate phosphoinositide metabolism at ER–plasma membrane contactsites, Cell 144 (2011) 389–401.

[90] V.A. Bankaitis, J.R. Aitken, A.E. Cleves, W. Dowhan, An essential role for aphospholipid transfer protein in yeast Golgi function, Nature 347 (1990)561–562.

[91] K. Huitema, J. van den Dikkenberg, J.F.H.M. Brouwers, J.C.M. Holthuis, Identifica-tion of a family of animal sphingomyelin synthases, EMBO J. 23 (2004) 33–44.

[92] C.L. Baron, V. Malhotra, Role of diacylglycerol in PKD recruitment to the TGN andprotein transport to the plasma membrane, Science 295 (2002) 325–328.

[93] T.P. Levine, S. Munro, Dual targeting of Osh1p, a yeast homologue ofoxysterol-binding protein, to both the Golgi and the nucleus–vacuole junction,Mol. Biol. Cell 12 (2001) 1633–1644.

[94] M. Johansson, N. Rocha, W. Zwart, I. Jordens, L. Janssen, C. Kuijl, V.M. Olkkonen, J.Neefjes, Activation of endosomal dynein motors by stepwise assembly ofRab7-RILP-p150Glued, ORP1L, and the receptor betalll spectrin, J. Cell Biol. 176(2007) 459–471.

[95] M. Johansson, M. Lehto, K. Tanhuanpa, T.L. Cover, V.M. Olkkonen, Theoxysterol-binding protein homologue ORP1L interacts with Rab7 and altersfunctional properties of late endocytic compartments, Biol. Cell 16 (2005)5480–5492.

[96] N. Rocha, C. Kuijl, R. van der Kant, L. Janssen, D. Houben, H. Janssen, W. Zwart, J.Neefjes, Cholesterol sensor ORP1L contacts the ER protein VAP to controlRab7-RILP-p150 Glued and late endosome positioning, J. Cell Biol. 185 (2009)1209–1225.

[97] P. Wang, W. Duan, A.L. Munn, H. Yang, Molecular characterization of Osh6p, anoxysterol binding protein homolog in the yeast Saccharomyces cerevisiae, FEBS J.272 (2005) 4703–4715.


[98] T.A. Schulz, M.-G. Choi, S. Raychaudhuri, J.A. Mears, R. Ghirlando, J.E. Hinshaw,W.A. Prinz, Lipid-regulated sterol transfer between closely apposed membranesby oxysterol-binding protein homologues, J. Cell Biol. 187 (2009) 889–903.

[99] B. Kornmann, E. Currie, S.R. Collins, M. Schuldiner, J. Nunnari, J.S. Weissman, P.Walter, An ER–mitochondria tethering complex revealed by a synthetic biologyscreen, Science 325 (2009) 477–481.

[100] B. Kornmann, C. Osman, P. Walter, The conserved GTPase Gem1 regulates endo-plasmic reticulum–mitochondria connections, Proc. Natl. Acad. Sci. U. S. A. 108(2011) 14151–14156.

[101] D.A. Stroud, S. Oeljeklaus, S. Wiese, M. Bohnert, U. Lewandrowski, A. Sickmann,B. Guiard, M. van der Laan, B. Warscheid, N. Wiedemann, Composition and to-pology of the endoplasmic reticulum–mitochondria encounter structure, J.Mol. Biol. 413 (2011) 743–750.

[102] B. Kornmann, P. Walter, ERMES-mediated ER–mitochondria contacts: molecularhubs for the regulation of mitochondrial biology, J. Cell Sci. 123 (2010)1389–1393.

[103] I. Lee, W. Hong, Diverse membrane-associated proteins contain a novel SMPdomain, FASEB J. 20 (2006) 202–206.

[104] K.O. Kopec, V. Alva, A.N. Lupas, Homology of SMP domains to the TULIP super-family of lipid-binding proteins provides a structural basis for lipid exchangebetween ER and mitochondria, Bioinformatics 26 (2010) 1927–1931.

[105] A. Toulmay, W.A. Prinz, A conserved membrane-binding domain targets pro-teins to organelle contact sites, J. Cell Sci. 125 (2012) 49–58.

[106] C. Voss, S. Lahiri, B.P. Young, C.J. Loewen, W.A. Prinz, ER-shaping proteins facili-tate lipid exchange between the ER and mitochondria in S. cerevisiae, J. Cell Sci.125 (2012) 4791–4799.

[107] T.T. Nguyen, A. Lewandowska, J.-Y. Choi, D.F. Markgraf, M. Junker, M. Bilgin, C.S.Ejsing, D.R. Voelker, T.A. Rapoport, J.M. Shaw, Gem1 and ERMES do not directlyaffect phosphatidylserine transport from ER to mitochondria or mitochondrialinheritance, Traffic 13 (2012) 880–890.

[108] T.C. Walther, R.V. Farese, Lipid droplets and cellular lipid metabolism, Annu. Rev.Biochem. 81 (2012) 687–714.

[109] S.L. Sturley, M.M. Hussain, Lipid droplet formation on opposing sides of theendoplasmic reticulum, J. Lipid Res. 53 (2012) 1800–1810.

[110] N. Jacquier, V. Choudhary, M. Mari, A. Toulmay, F. Reggiori, R. Schneiter, Lipiddroplets are functionally connected to the endoplasmic reticulum in Saccharo-myces cerevisiae, J. Cell Sci. 124 (2011) 2424–2437.

[111] E.J. Blanchette-Mackie, N.K. Dwyer, T. Barber, R.A. Coxey, T. Takeda, C.M.Rondinone, J.L. Theodorakis, A.S. Greenberg, C. Londos, Perilipin is located onthe surface layer of intracellular lipid droplets in adipocytes, J. Lipid Res. 36(1995) 1211–1226.

[112] H. Wolinski, D. Kolb, S. Hermann, R.I. Koning, S.D. Kohlwein, A role for seipin inlipid droplet dynamics and inheritance in yeast, J. Cell Sci. 124 (2011)3894–3904.

[113] H. Robenek, O. Hofnagel, I. Buers, M.J. Robenek, D. Troyer, N.J. Severs,Adipophilin-enriched domains in the ER membrane are sites of lipid dropletbiogenesis, J. Cell Sci. 119 (2006) 4215–4224.

[114] N. Xu, S.O. Zhang, R.A. Cole, S.A. McKinney, F. Guo, J.T. Haas, S. Bobba, R.V. Farese,H.Y. Mak, The FATP1–DGAT2 complex facilitates lipid droplet expansion at theER–lipid droplet interface, J. Cell Biol. 198 (2012) 895–911.

[115] X. Pan, P. Roberts, Y. Chen, E. Kvam, N. Shulga, K. Huang, S. Lemmon, D.S.Goldfarb, Nucleus–vacuole junctions in Saccharomyces cerevisiae are formedthrough the direct interaction of Vac8p with Nvj1p, Mol. Biol. Cell 11 (2000)2445–2457.

[116] P. Roberts, S. Moshitch-Moshkovitz, E. Kvam, E. O'Toole, M. Winey, D.S. Goldfarb,Piecemeal microautophagy of nucleus in Saccharomyces cerevisiae, Mol. Biol. Cell14 (2003) 129–141.

[117] R. Dawaliby, A. Mayer, Microautophagy of the nucleus coincides with a vacuolardiffusion barrier at nuclear–vacuolar junctions, Mol. Biol. Cell 21 (2010)4173–4183.

[118] S.D. Kohlwein, S. Eder, C.S. Oh, C.E. Martin, K. Gable, D. Bacikova, T. Dunn, Tsc13pis required for fatty acid elongation and localizes to a novel structure at thenuclear-vacuolar interface in Saccharomyces cerevisiae, Mol. Cell. Biol. 21(2001) 109–125.

[119] E. Kvam, K. Gable, T.M. Dunn, D.S. Goldfarb, Targeting of Tsc13p to nucleus–vacuole junctions: a role for very-long-chain fatty acids in the biogenesis ofmicroautophagic vesicles, Mol. Biol. Cell 16 (2005) 3987–3998.

[120] J.R. Friedman, B.M. Webster, D.N. Mastronarde, K.J. Verhey, G.K. Voeltz, ER slid-ing dynamics and ER–mitochondrial contacts occur on acetylated microtubules,J. Cell Biol. 190 (2010) 363–375.

[121] X. Guo, G.T. Macleod, A. Wellington, F. Hu, S. Panchumarthi, M. Schoenfield, L.Marin, M.P. Charlton, H.L. Atwood, K.E. Zinsmaier, The GTPase dMiro is requiredfor axonal transport of mitochondria to Drosophila synapses, Neuron 47 (2005)379–393.

[122] X. Wang, T.L. Schwarz, The mechanism of Ca2+-dependent regulation ofkinesin-mediated mitochondrial motility, Cell 136 (2009) 163–174.

[123] G.M. Mórotz, K.J. De Vos, A. Vagnoni, S. Ackerley, C.E. Shaw, C.C.J. Miller,Amyotrophic lateral sclerosis-associated mutant VAPBP56S perturbs calciumhomeostasis to disrupt axonal transport of mitochondria, Hum. Mol. Genet. 21(2012) 1979–1988.

[124] G.J. Hermann, J.W. Thatcher, J.P. Mills, K.G. Hales, M.T. Fuller, J. Nunnari, J.M.Shaw, Mitochondrial fusion in yeast requires the transmembrane GTPaseFzo1p, J. Cell Biol. 143 (1998) 359–373.

[125] W. Bleazard, J.M. McCaffery, E.J. King, S. Bale, A. Mozdy, Q. Tieu, J. Nunnari, J.M.Shaw, The dynamin-related GTPase Dnm1 regulates mitochondrial fission inyeast, Nat. Cell Biol. 1 (1999) 298–304.





[126] E. Smirnova, L. Griparic, D.L. Shurland, A.M. van der Bliek, Dynamin-relatedprotein Drp1 is required for mitochondrial division in mammalian cells, Mol.Biol. Cell 12 (2001) 2245–2256.

[127] N. Ishihara, Y. Eura, K. Mihara, Mitofusin 1 and 2 play distinct roles in mitochon-drial fusion reactions via GTPase activity, J. Cell Sci. 117 (2004) 6535–6546.

[128] E. Ingerman, E.M. Perkins, M. Marino, J.A. Mears, J.M. McCaffery, J.E. Hinshaw, J.Nunnari, Dnm1 forms spirals that are structurally tailored to fit mitochondria,J. Cell Biol. 170 (2005) 1021–1027.

[129] D. Otsuga, B.R. Keegan, E. Brisch, J.W. Thatcher, G.J. Hermann, W. Bleazard, J.M.Shaw, The dynamin-related GTPase, Dnm1p, controls mitochondrial morpholo-gy in yeast, J. Cell Biol. 143 (1998) 333–349.

[130] A.M. Labrousse, M.D. Zappaterra, D.A. Rube, A.M. van der Bliek, C elegansdynamin-related protein DRP-1 controls severing of the mitochondrial outermembrane, Mol. Cell 4 (1999) 815–826.

[131] H. Chen, S.A. Detmer, A.J. Ewald, E.E. Griffin, S.E. Fraser, D.C. Chan, MitofusinsMfn1 and Mfn2 coordinately regulate mitochondrial fusion and are essentialfor embryonic development, J. Cell Biol. 160 (2003) 189–200.

[132] O.M. de Brito, L. Scorrano, Mitofusin 2 tethers endoplasmic reticulum to mito-chondria, Nature 456 (2008) 605–610.

[133] T. Koshiba, S.A. Detmer, J.T. Kaiser, H. Chen, J.M. McCaffery, D.C. Chan, Structural basisof mitochondrial tethering by mitofusin complexes, Science 305 (2004) 858–862.

[134] J.R. Friedman, L.L. Lackner, M. West, J.R. DiBenedetto, J. Nunnari, G.K. Voeltz, ERtubules mark sites of mitochondrial division, Science 334 (2011) 358–362.

[135] H. Otera, C. Wang, M.M. Cleland, K. Setoguchi, S. Yokota, R.J. Youle, K. Mihara,Mff is an essential factor for mitochondrial recruitment of Drp1 during mito-chondrial fission in mammalian cells, J. Cell Biol. 191 (2010) 1141–1158.

[136] T. Itoh, A. Toh-E, Y. Matsui, Mmr1p is a mitochondrial factor for Myo2p-dependentinheritance of mitochondria in the budding yeast, EMBO J. 23 (2004) 2520–2530.

[137] R.L. Frederick, K. Okamoto, J.M. Shaw, Multiple pathways influence mitochon-drial inheritance in budding yeast, Genetics 178 (2008) 825–837.

[138] V.R. Simon, T.C. Swayne, L.A. Pon, Actin-dependent mitochondrial motility inmitotic yeast and cell-free systems: identification of a motor activity on themitochondrial surface, J. Cell Biol. 130 (1995) 345–354.

[139] K.A. Shepard, A.P. Gerber, A. Jambhekar, P.A. Takizawa, P.O. Brown, D. Herschlag,J.L. DeRisi, R.D. Vale, Widespread cytoplasmic mRNA transport in yeast: identifi-cation of 22 bud-localized transcripts using DNA microarray analysis, Proc. Natl.Acad. Sci. U. S. A. 100 (2003) 11429–11434.

[140] T.C. Swayne, C. Zhou, I.R. Boldogh, J.K. Charalel, J.R. McFaline-Figueroa, S. Thoms,C. Yang, G. Leung, J. McInnes, R. Erdmann, L.a Pon, I. Boldogh, L.A. Pon, Role forcER and Mmr1p in anchorage of mitochondria at sites of polarized surfacegrowth in budding yeast, Curr. Biol. 21 (2011) 1994–1999.

[141] J.I. Millen, J. Pierson, E. Kvam, L.J. Olsen, D.S. Goldfarb, The luminal N-terminus ofyeast Nvj1 is an inner nuclear membrane anchor, Traffic 9 (2008) 1653–1664.

[142] E. Kvam, D.S. Goldfarb, Nvj1p is the outer-nuclear-membrane receptor foroxysterol-binding protein homolog Osh1p in Saccharomyces cerevisiae, J. CellSci. 117 (2004) 4959–4968.

[143] G. Szabadkai, K. Bianchi, P. Várnai, D. De Stefani, M.R. Wieckowski, D. Cavagna,A.I. Nagy, T. Balla, R. Rizzuto, Chaperone-mediated coupling of endoplasmic re-ticulum and mitochondrial Ca2+ channels, J. Cell Biol. 175 (2006) 901–911.

[144] K.J. De Vos, G.M. Mórotz, R. Stoica, E.L. Tudor, K.-F. Lau, S. Ackerley, A. Warley,C.E. Shaw, C.C.J. Miller, VAPB interacts with the mitochondrial protein PTPIP51to regulate calcium homeostasis, Hum. Mol. Genet. 21 (2012) 1299–1311.

[145] R. Iwasawa, A.-L. Mahul-Mellier, C. Datler, E. Pazarentzos, S. Grimm, Fis1 andBap31 bridge the mitochondria–ER interface to establish a platform for apopto-sis induction, EMBO J. 30 (2011) 556–568.

[146] S. Raychaudhuri, W.A. Prinz, The diverse functions of oxysterol-binding pro-teins, Annu. Rev. Cell Dev. Biol. 26 (2010) 157–177.

[147] C.E. Creutz, S.L. Snyder, T.A. Schulz, Characterization of the yeast tricalbins:membrane-bound multi-C2-domain proteins that form complexes involved inmembrane trafficking, Cell. Mol. Life Sci. 61 (2004) 1208–1220.

[148] W. Wolf, A. Kilic, B. Schrul, H. Lorenz, B. Schwappach, M. Seedorf, Yeast Ist2 re-cruits the endoplasmic reticulum to the plasma membrane and creates aribosome-free membrane microcompartment, PLoS One 7 (2012) e39703.

[149] K. Maass, M.A. Fischer, M. Seiler, K. Temmerman,W. Nickel, M. Seedorf, A signal com-prising a basic cluster and an amphipathic alpha–helix interacts with lipids and is re-quired for the transport of Ist2 to the yeast cortical ER, J. Cell Sci. 122 (2009) 625–635.

[150] E. Ercan, F. Momburg, U. Engel, K. Temmerman, W. Nickel, M. Seedorf, A con-served, lipid-mediated sorting mechanism of yeast Ist2 and mammalian STIMproteins to the peripheral ER, Traffic 10 (2009) 1802–1818.

[151] G. Lavieu, L. Orci, L. Shi, M. Geiling, M. Ravazzola, F. Wieland, P. Cosson, J.E.Rothman, Induction of cortical endoplasmic reticulum by dimerization of acoatomer-binding peptide anchored to endoplasmic reticulum membranes,Proc. Natl. Acad. Sci. U. S. A. 107 (2010) 6876–6881.

[152] H. Takekura, H. Takeshima, S. Nishimura, M. Takahashi, T. Tanabe, V. Flockerzi, F.Hofmann, C. Franzini-Armstrong, Co-expression in CHO cells of two muscle pro-teins involved in excitation–contraction coupling, J. Muscle Res. Cell Motil. 16(1995) 465–480.

[153] H. Takeshima, S. Komazaki, M. Nishi, M. Iino, K. Kangawa, Junctophilins: a novelfamily of junctional membrane complex proteins, Mol. Cell 6 (2000) 11–22.

[154] F.G. Haj, P.J. Verveer, A. Squire, B.G. Neel, P.I.H. Bastiaens, Imaging sites of recep-tor dephosphorylation by PTP1B on the surface of the endoplasmic reticulum,Science 295 (2002) 1708–1711.

[155] E.R. Eden, I.J. White, A. Tsapara, C.E. Futter, Membrane contacts betweenendosomes and ER provide sites for PTP1B-epidermal growth factor receptorinteraction, Nat. Cell Biol. 12 (2010) 267–272.


[156] S.J. Stone, M.C. Levin, P. Zhou, J. Han, T.C. Walther, R.V. Farese, The endoplasmicreticulum enzyme DGAT2 is found in mitochondria-associated membranes andhas a mitochondrial targeting signal that promotes its association with mito-chondria, J. Biol. Chem. 284 (2009) 5352–5361.

[157] J.C.M. Holthuis, C. Ungermann, Cellular microcompartments constitute generalsub-organellar functional units in cells, Biol. Chem. 394 (2012) 151–161.

[158] A.P. Gasch, P.T. Spellman, C.M. Kao, O. Carmel-Harel, M.B. Eisen, G. Storz, D.Botstein, P.O. Brown, Genomic expression programs in the response of yeastcells to environmental changes, Mol. Biol. Cell 11 (2000) 4241–4257.

[159] K. Simons, E. Ikonen, Functional rafts in cell membranes, Nature 387 (1997)569–572.

[160] D. Lingwood, K. Simons, Detergent resistance as a tool in membrane research,Nat. Protoc. 2 (2007) 2159–2165.

[161] S. Munro, Lipid rafts: elusive or illusive? Cell 115 (2003) 377–388.[162] T. Hayashi, M. Fujimoto, Detergent-resistant microdomains determine the

localization of sigma-1 receptors to the endoplasmic reticulum–mitochondriajunction, Mol. Pharmacol. 77 (2010) 517–528.

[163] E. Area-Gomez, M. Del Carmen Lara Castillo, M.D. Tambini, C. Guardia-Laguarta,A.J.C. de Groof, M. Madra, J. Ikenouchi, M. Umeda, T.D. Bird, S.L. Sturley, E.A.Schon, Upregulated function of mitochondria-associated ER membranes inAlzheimer disease, EMBO J. 31 (2012) 4106–4123.

[164] M. Fujimoto, T. Hayashi, T.-P. Su, The role of cholesterol in the association ofendoplasmic reticulum membranes with mitochondria, Biochem. Biophys. Res.Commun. 417 (2012) 635–639.

[165] T. Hayashi, T.-P. Su, Sigma-1 receptors at galactosylceramide-enriched lipidmicrodomains regulate oligodendrocyte differentiation, Proc. Natl. Acad. Sci. U.S. A. 101 (2004) 14949–14954.

[166] B. Pani, H.L. Ong, X. Liu, K. Rauser, I.S. Ambudkar, B.B. Singh, Lipid rafts deter-mine clustering of STIM1 in endoplasmic reticulum–plasma membrane junc-tions and regulation of store-operated Ca2+ entry (SOCE), J. Biol. Chem. 283(2008) 17333–17340.

[167] A.E. Rusiñol, Z. Cui, M.H. Chen, J.E. Vance, A unique mitochondria-associatedmembrane fraction from rat liver has a high capacity for lipid synthesis and con-tains pre-Golgi secretory proteins including nascent lipoproteins, J. Biol. Chem.269 (1994) 27494–27502.

[168] H. Pichler, B. Gaigg, C. Hrastnik, G. Achleitner, S.D. Kohlwein, G. Zellnig, A.Perktold, G. Daum, A subfraction of the yeast endoplasmic reticulum associateswith the plasma membrane and has a high capacity to synthesize lipids, Eur. J.Biochem. 268 (2001) 2351–2361.

[169] K. Kozieł, M. Lebiedzinska, G. Szabadkai, M. Onopiuk, W. Brutkowski, K.Wierzbicka, G. Wilczyński, P. Pinton, J. Duszyński, K. Zabłocki, M.R.Wieckowski, Plasma membrane associated membranes (PAM) from Jurkatcells contain STIM1 protein: is PAM involved in the capacitative calciumentry? Int. J. Biochem. Cell Biol. 41 (2009) 2440–2449.

[170] M.X. Andersson, M. Goksör, A.S. Sandelius, Optical manipulation reveals strongattracting forces at membrane contact sites between endoplasmic reticulumand chloroplasts, J. Biol. Chem. 282 (2007) 1170–1174.

[171] A. Raturi, T. Simmen, Where the endoplasmic reticulum and the mitochondriontie the knot: the mitochondria-associated membrane (MAM), Biochim. Biophys.Acta 1883 (2012) 213–224.

[172] A. Zhang, C.D. Williamson, D.S. Wong, M.D. Bullough, K.J. Brown, Y. Hathout, A.M.Colberg-Poley, Quantitative proteomic analyses of human cytomegalovirus-induced restructuring of endoplasmic reticulum–mitochondrial contacts at latetimes of infection, Mol. Cell. Proteomics 10 (2011), (M111.009936).

[173] E. Area-Gomez, A.J.C. de Groof, I. Boldogh, T.D. Bird, G.E. Gibson, C.M. Koehler,W.H. Yu, K.E. Duff, M.P. Yaffe, L.A. Pon, E.A. Schon, Presenilins are enriched inendoplasmic reticulum membranes associated with mitochondria, Am. J. Pathol.175 (2009) 1810–1816.

[174] H.J. Wang, G. Guay, L. Pogan, R. Sauvé, I.R. Nabi, Calcium regulates the associationbetween mitochondria and a smooth subdomain of the endoplasmic reticulum,J. Cell Biol. 150 (2000) 1489–1498.

[175] T. Hayashi, T.-P. Su, Sigma-1 receptor chaperones at the ER–mitochondrioninterface regulate Ca(2+) signaling and cell survival, Cell 131 (2007)596–610.

[176] S.Y. Gilady, M. Bui, E.M. Lynes, M.D. Benson, R. Watts, J.E. Vance, T. Simmen,Ero1alpha requires oxidizing and normoxic conditions to localize to themitochondria-associated membrane (MAM), Cell Stress Chaperones 15 (2010)619–629.

[177] P. Cosson, A. Marchetti, M. Ravazzola, L. Orci, Mitofusin-2 independent juxtapo-sition of endoplasmic reticulum and mitochondria: an ultrastructural study,PLoS One 7 (2012) e46293.

[178] K. Xu, H.P. Babcock, X. Zhuang, Dual-objective STORM reveals three-dimensionalfilament organization in the actin cytoskeleton, Nat. Methods 9 (2012) 185–188.

[179] B. Huang, S.A. Jones, B. Brandenburg, X. Zhuang, Whole-cell 3D STORM revealsinteractions between cellular structures with nanometer-scale resolution, Nat.Methods 5 (2008) 1047–1052.

[180] D. Jiang, L. Zhao, D.E. Clapham, Genome-wide RNAi screen identifies Letm1 as amitochondrial Ca2+/H+ antiporter, Science 326 (2009) 144–147.

[181] E. Froschauer, K. Nowikovsky, R.J. Schweyen, Electroneutral K+/H+exchangein mitochondrial membrane vesicles involves Yol027/Letm1 proteins, Biochim.Biophys. Acta 1711 (2005) 41–48.

[182] K.S. Dimmer, F. Navoni, A. Casarin, E. Trevisson, S. Endele, A. Winterpacht, L.Salviati, L. Scorrano, LETM1, deleted in Wolf–Hirschhorn syndrome is requiredfor normal mitochondrial morphology and cellular viability, Hum. Mol. Genet.17 (2008) 201–214.




Documents

Biochimica et Biophysica Acta - ETH Z · Biochimica et Biophysica Acta xxx (2013) xxx–xxx ☆ This article is part of a Special Issue entitled: Functional and structural diversity