* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Download Mechanisms Shaping the Membranes of Cellular Organelles
Cell encapsulation wikipedia , lookup
Protein moonlighting wikipedia , lookup
Protein phosphorylation wikipedia , lookup
Mechanosensitive channels wikipedia , lookup
G protein–coupled receptor wikipedia , lookup
Membrane potential wikipedia , lookup
Magnesium transporter wikipedia , lookup
Cell nucleus wikipedia , lookup
Organ-on-a-chip wikipedia , lookup
Ethanol-induced non-lamellar phases in phospholipids wikipedia , lookup
Theories of general anaesthetic action wikipedia , lookup
Lipid bilayer wikipedia , lookup
Cytokinesis wikipedia , lookup
Signal transduction wikipedia , lookup
Model lipid bilayer wikipedia , lookup
SNARE (protein) wikipedia , lookup
Cell membrane wikipedia , lookup
ANRV389-CB25-14 ARI 29 June 2009 17:48 V I E W A N I N C E S R E Review in Advance first posted online on August 13, 2009. (Minor changes may still occur before final publication online and in print.) D V A Annu. Rev. Cell Dev. Biol. 2009.25. Downloaded from arjournals.annualreviews.org by TEL-AVIV UNIVERSITY on 08/24/09. For personal use only. Mechanisms Shaping the Membranes of Cellular Organelles Yoko Shibata,1 Junjie Hu,2 Michael M. Kozlov,3 and Tom A. Rapoport1 1 Howard Hughes Medical Institute and Department of Cell Biology, Harvard Medical School, Boston, Massachusetts 02115; email: yoko [email protected], tom [email protected] 2 College of Life Sciences, Nankai University, 300071 Tianjin, China; email: [email protected] 3 Department of Physiology and Pharmacology, Sackler Faculty of Medicine, Tel Aviv University, 699978 Tel Aviv, Israel; email: [email protected] Annu. Rev. Cell Dev. Biol. 2009. 25:14.1–14.26 Key Words The Annual Review of Cell and Developmental Biology is online at cellbio.annualreviews.org endoplasmic reticulum, mitochondria, caveolae, tubules, reticulons, dynamins This article’s doi: 10.1146/annurev.cellbio.042308.113324 c 2009 by Annual Reviews. Copyright All rights reserved 1081-0706/09/1110-0001$20.00 Abstract Cellular organelles have characteristic morphologies that arise as a result of different local membrane curvatures. A striking example is the endoplasmic reticulum (ER), which consists of ER tubules with high curvature in cross-section, peripheral ER sheets with little curvature except at their edges, and the nuclear envelope with low curvature except where the nuclear pores are inserted. The ER may be shaped by several mechanisms. ER tubules are often generated through their association with the cytoskeleton and stabilized by two families of integral membrane proteins, the reticulons and DP1/Yop1p. Similar to how curvature is generated in budding vesicles, these proteins may use scaffolding and hydrophobic insertion mechanisms to shape the lipid bilayer into tubules. In addition, proteins of the dynamin family may deform the ER membrane to generate a tubular network. Mechanisms affecting local membrane curvature may also shape peripheral ER sheets and the nuclear envelope, as well as mitochondria and caveoli. 14.1 ANRV389-CB25-14 ARI 29 June 2009 17:48 Contents Annu. Rev. Cell Dev. Biol. 2009.25. Downloaded from arjournals.annualreviews.org by TEL-AVIV UNIVERSITY on 08/24/09. For personal use only. INTRODUCTION . . . . . . . . . . . . . . . . . . 14.2 GENERATION OF CURVATURE DURING VESICLE BUDDING . . 14.6 GENERATION OF THE TUBULAR ER NETWORK . . . . . . . . . . . . . . . . . . 14.8 GENERATION OF MEMBRANE SHEETS IN THE ER . . . . . . . . . . . . .14.12 SHAPING MITOCHONDRIA. . . . . . .14.14 GENERATING THE CURVATURE OF CAVEOLAE . . . . . . . . . . . . . . . . . .14.17 CONCLUSIONS . . . . . . . . . . . . . . . . . . . .14.19 INTRODUCTION Membrane-bound organelles in eukaryotic cells have characteristic shapes. Some organelles, such as lysosomes and peroxisomes, are relatively spherical, but others are more complex. For example, mitochondria form a tubular network, the endoplasmic reticulum (ER) is an interconnected system of sheets and tubules, and the Golgi apparatus consists of a stack of perforated sheets. Cellular membrane shapes are generally dynamic. They often change during processes such as cell movement, division, and growth, but also under steady state conditions. For example, many membrane-bound compartments have the ability to bud small vesicles and to fuse with vesicles; that is, they communicate with one another by vesicular trafficking. Membrane shaping relates to the generation of membrane curvature (McMahon & Gallop 2005, Zimmerberg & Kozlov 2006). For example, high curvature is seen in cross-sections of narrow tubules, at the edges of closely spaced membrane sheets, or with small transport vesicles. Conversely, low curvature membranes include the sheet structures of the ER or Golgi. Intracellular membranes can change their shapes without altering their basic topology. For example, a flat membrane may undergo bending, or a tubule may be drawn out of a flat membrane. Alternatively, the membrane shape can change ER: endoplasmic R I E V reticulum E W S 14.2 C E I N A D V A N Shibata et al. topologically by remodeling, a process that occurs most often by fusion or fission. During fusion two separate membranes merge into one, whereas during fission a continuous membrane is divided into two separate ones. For example, ER tubules can fuse with an existing tubule to form a three-way junction. How the characteristic shape of an organelle is achieved and how membrane curvature is generated are still only poorly understood, but these questions are now central to a new research field. Here, we will concentrate on the mechanisms of ER morphogenesis but point out that other organelles are shaped by similar principles. The mechanisms by which biological membranes are shaped and remodeled can be described by physical terms. The shape of a membrane at a given point of its surface can be described by two principle curvatures, c1 and c2 , which are the inverse of the radii of two perpendicular arcs lying on the membrane plane, R1 and R2 (c1 = 1/R1 and c2 = 1/R2 ; see Figure 1, which also gives a more detailed definition of curvature). The curvature is considered to be small if the radii are larger than the thickness of the membrane d, R1 d, R2 d. It is considered to be large when one of the radii is comparable with or somewhat larger than the membrane thickness, R1 or R2 ≥ d. Taking into account that the lipid bilayer thickness is d ≈ 4 nm, intracellular tubules or vesicles with radii of ∼25– 30 nm have relatively high curvatures. Lipid bilayers tend to remain flat because the generation of curvature requires energy. The energy cost for bending a bilayer can be determined from the elastic model of lipid membranes (Helfrich 1973, 1990) (see Figure 1). Generation of a fragment of a lipid tubule with a diameter of 60 nm and a length of 60 nm requires ≈70 kcal/mol, whereas formation of a spherical vesicle of the same diameter requires ≈300 kcal/mol (see Figure 1 legend). Both energies are much larger than the characteristic thermal energy (≈0.6 kcal/mol), which means these shapes cannot be generated simply by thermal fluctuations. However, these bending energies can be generated by the hydrolysis of a few tens of adenosine triphosphate (ATP) ANRV389-CB25-14 ARI 29 June 2009 17:48 Curvature Curved shapes Cylinder Saddle COPII-coated vesicles Yop1p-generated membrane tubules Neck of budding vesicle R2 R1 20 nm c1 = R c2 = R–2 –1 c1 = c2 = R J = 2(R–1) K = R–2 –1 100 nm c1 = R J = R–1 –1 c2 = 0 K=0 c1 = –c2 = R–1 J=0 K = –R–2 Figure 1 Curvature and shapes of membranes. A membrane is represented by the mid surface between the two lipid monolayers (left panel ). The shape of every small element of the mid surface is described by two perpendicular arcs of radii R1 and R2 . The curvatures of the arcs, c 1 = R1 and c 2 = R1 are referred to as principle curvatures of the surface and fully characterize the local shape of the membrane. More 1 2 commonly, two combinations of the principle curvatures are used: the total curvature, J = c 1 + c 2 , and the Gaussian curvature, K = c 1 c 2 (Helfrich 1973). The other panels show three basic shapes adopted by biological membranes and electron microscopy images of actual examples (Image of COPII vesicle, courtesy of L. Orci and R. Schekman; image of neck of a budding vesicle, from Gallop and McMahon 2005). The formulas show the relationship between the curvatures and the radii. Generation of membrane curvature requires energy, which can be computed using the Helfrich theory of membrane elasticity (Helfrich 1973, 1990). The bending energy 2 per unit area of the membrane, fB , is expressed through the total, J, and Gaussian, K, curvatures by f B = 12 · κ B · J + κ̄ · K , where κ B ≈ −19 Joule ≈ 10 kcal/mol is the lipid bilayer bending modulus, and κ̄ is the modulus of the Gaussian curvature (the latter has not been 10 measured directly but can be estimated to be in the range −κ B < κ̄ < 0). To compute the total bending energy, Ftot , the energy per unit area has to be integrated over the entire membrane surface, Ftot = f B d A. It should be noted that the total curvature J determines the energy of both membrane shaping and remodeling, whereas the energy of the Gaussian curvature K is only important for remodeling. For example, the energy of the Gaussian curvature of a spherical vesicle does not change when the vesicle is arbitrarily deformed, but it changes by 4π κ̄ for splitting a vesicle into two and by −4π κ̄ for fusion of a vesicle with a planar membrane. Based on the Helfrich cyl model, the bending energy of a cylindrical membrane of radius R and length L equals Ftot = π κ B RL , and for a spherical membrane it is = 8π κ B + 4π κ̄. molecules (each ≈ 10 kcal/mol), consistent with the notion that metabolically generated energy can be used to shape biological membranes. These considerations indicate that high membrane curvature requires molecular mechanisms that generate and stabilize these energetically unfavorable states. These mechanisms are based on the interplay between lipids and proteins and require regulation to achieve the dynamic behavior of cellular membranes. In principle, lipids alone can generate membrane curvature if the lipid composition of the membrane monolayers is asymmetric. For example, a nonbilayer lipid species could be E V I E W S 14.3 I N E www.annualreviews.org • Mechanisms Shaping the Membranes of Cellular Organelles A C s phere Fto t R Annu. Rev. Cell Dev. Biol. 2009.25. Downloaded from arjournals.annualreviews.org by TEL-AVIV UNIVERSITY on 08/24/09. For personal use only. Midsurface element Sphere D V A N ANRV389-CB25-14 ARI 29 June 2009 Membrane mid surface: surface between the two monolayers of a membrane- Annu. Rev. Cell Dev. Biol. 2009.25. Downloaded from arjournals.annualreviews.org by TEL-AVIV UNIVERSITY on 08/24/09. For personal use only. Molecular motors: ATPases that move associated cargo along microtubules (includes kinesins and cytoplasmic dynein) or actin filaments (myosins) Amphipathic helix: has distinct hydrophobic and hydrophilic surfaces V I E W concentrated in one of the membrane monolayers. Nonbilayer lipids, such as diacylglycerol or lysophospholipids, tend to self-assemble into extremely bent monolayers with curvature radii close to the lipid monolayer thickness, R ≈ d/2 (for review, see Zimmerberg & Kozlov 2006). When mixed with regular lipids in a membrane and asymmetrically distributed between the membrane monolayers, these lipids can produce high curvature of the whole bilayer. Lipid asymmetry can also lead to a difference in membrane area between the two monolayers. In this case, curvature is generated because the two monolayers are practically nonstretchable and noncompressible, but are coupled to one another along the membrane mid surface [bilayer-coupling mechanism (Sheetz & Singer 1974)]. In agreement with these predictions, it has been found experimentally that high mole fractions of some lipids can deform liposomes into tubules (Stowell et al. 1999), and certain lipids can segregate and lead to buds and domains of moderate curvature (Bacia et al. 2005, Baumgart et al. 2003). However, theoretical calculations show that the generation of strongly bent membranes with curvature radii of a few tens of nanometers requires fairly strong monolayer asymmetry (Zimmerberg & Kozlov 2006), which is not expected based on the lipid composition of cell membranes (van Meer et al. 2008). Therefore, the current consensus in the field is that high membrane curvature in intracellular membranes is generated by proteins. Nevertheless, lipids play an important role, modifying and facilitating the function of curvature-generating proteins in decisive ways (Devaux et al. 2008). Proteins could cause curvature indirectly by generating lipid asymmetry. For example, they could produce nonbilayer lipids in one of the membrane monolayers, or they could transfer lipids from one bilayer leaflet to another (flippases). However, most of the evidence to date suggests that proteins generate membrane curvature in a direct manner, rather than indirectly through lipids. Proteins appear to use three major, but not mutually exclusive, mechanisms to directly R E 17:48 S 14.4 C E I N A D V A N Shibata et al. generate high membrane curvature (McMahon & Gallop 2005, Zimmerberg & Kozlov 2006) (Figure 2). The first mechanism involves proteins that exert mechanical force on a lipid bilayer (Figure 2a). Thin membrane tubules can be pulled out of low-curvature membrane reservoirs by molecular motors as they move along microtubules or actin filaments. Alternatively, membrane tubules can be pulled when they attach to polymerizing microtubules or actin filaments. According to theoretical estimates, tubules of 30 nm radii require a pulling force of 20 pN, which can be generated by about 10 molecular motors (Zimmerberg & Kozlov 2006). The second mechanism by which proteins bend membranes is scaffolding (Figure 2b). Here, a protein or protein network has a highly curved surface with a strong affinity for the lipid polar head groups. The scaffolding proteins interact with a lipid membrane along the bent surface and force the bilayer to adopt the same curvature. This requires that the rigidity of the protein scaffold is high compared to the membrane bending rigidity, and that the energy of the membrane-protein attachment exceeds the membrane-bending energy. The third mechanism of membrane bending is hydrophobic insertion (wedging) (Figure 2c). Here, proteins insert hydrophobic domains, such as the hydrophobic surface of an amphipathic helix, into the upper part of a membrane monolayer. This results in the perturbation of the packing of the lipid head groups and generates local membrane curvature. According to theoretical calculations, maximal curvature is produced if the hydrophobic insertion is shallow, penetrating the membrane to about one third of the lipid monolayer thickness, that is, to approximately the interface between the polar head groups and the hydrocarbon chains (Campelo et al. 2008). As discussed below, some integral membrane proteins have also been proposed to generate curvature by a combination of scaffolding and hydrophobic insertion (wedging). Protein embedding alone could cause membrane curvature by the bilayer coupling effect ARI 29 June 2009 17:48 if a membrane protein displaces lipids primarily in one monolayer. Because the effect would be averaged over the total membrane surface, a tubule with a curvature radius of 30 nm would require that these proteins generate an overall area difference of 13% between the monolayers. While this suggests that most integral membrane proteins can only cause high curvature when they are extremely abundant, it is conceivable that some membrane proteins displace lipids specifically at a certain depth of the lipid bilayer and thus generate local curvature by the hydrophobic insertion mechanism, similar to how amphipathic helices act, or that they combine hydrophobic insertion (wedging) and scaffolding effects in a cooperative manner. Membrane remodeling by fusion and fission is tightly related to membrane shaping (Chernomordik & Kozlov 2005, 2008; Chernomordik et al. 2006). During fusion two lipid bilayers come in close contact with one another, leading initially to the fusion of the approaching monolayers (hemi-fusion), followed by fusion of the outer monolayers (Figure 3). The first step in a fusion reaction is to bring the two membranes into close proximity. For example, the fusion of transport vesicles with their target membrane is caused by the association of SNARE proteins; three helices of SNARE proteins in one membrane, all with their Ntermini in the cytosol, zipper together with one helix of the same orientation in the other membrane to force the bilayers into close proximity (Sudhof & Rothman 2009). The actual fusion process may be initiated by the generation of a a le u Tub ule tub t lls amen u r p fil oto ton ar mskele l u lec yto Mo ong c al Tubule binds to polymerizing cytoskeleton filament Tubule −−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−→ Figure 2 Three major mechanisms by which proteins generate curvature. (a) Membrane deformation by proteins that exert mechanical force. In the cases shown, tubules are pulled out of a flat membrane by either being attached to molecular motors (blue) that move along cytoskeleton fibers, or by binding to the tip of a polymerizing cytoskeleton filament. The directions of movement of the molecular motor, the polymerizing microtubule, and the membrane tubules are indicated by arrows. (b) Curvature generation by scaffolding proteins. Scaffolding proteins have a rigid curved shape, whose concave surface interacts with the membrane, forcing the bilayer to adopt the same curvature. (c) Curvature generation by the hydrophobic insertion (wedging) mechanism. Soluble proteins insert hydrophobic regions into a membrane monolayer. Shown in blue and red are the hydrophilic and hydrophobic parts of an amphipathic helix, respectively, in cross section. Note that shallow insertion of the hydrophobic surface of the helix generates maximal curvature. Integral membrane proteins that occupy more space in one leaflet of the membrane than the other could also generate membrane curvature by the bilayer-coupling mechanism. b Scaffolding proteins c Amphipathic helix Integral membrane protein E V I E W R S 14.5 I N E www.annualreviews.org • Mechanisms Shaping the Membranes of Cellular Organelles A C Annu. Rev. Cell Dev. Biol. 2009.25. Downloaded from arjournals.annualreviews.org by TEL-AVIV UNIVERSITY on 08/24/09. For personal use only. ANRV389-CB25-14 D V A N Annu. Rev. Cell Dev. Biol. 2009.25. Downloaded from arjournals.annualreviews.org by TEL-AVIV UNIVERSITY on 08/24/09. For personal use only. ANRV389-CB25-14 ARI 29 June 2009 a Membrane fusion pathway b Membrane fission pathway 17:48 Figure 3 Pathways of membrane remodeling by fusion and fission. (a) A fusion pathway starts with the generation of a strong local deformation of one membrane. A protrusion or dimple is formed, which establishes a point-like contact with the target membrane. Relaxation of the protrusion drives fusion stalk formation and subsequent hemifusion. Decay of the hemifusion intermediate into a fusion pore completes membrane fusion. (b) Membrane fission begins with the splitting of the inner monolayer of a membrane neck, which results in a hemi-fission structure similar to that of a fusion stalk. Decay of this structure drives separation of the membrane neck into two nonconnected membranes. V I E W R E point-like protrusion or a strongly curved dimple in one of the membranes (Efrat et al. 2007, Kozlov & Chernomordik 1998, Martens et al. 2007) (Figure 3). The generated dimples would point to the adjacent membrane (Figure 3) and promote close membrane contact. Fusion of the dimple with the target membrane would be favored by the accompanying relaxation of the strongly curved shape of the dimple. Similarly, the first step of fission is hemifission, where only one monolayer divides, which is then followed by splitting of the second monolayer (Kozlov & Chernomordik 2002, Kozlovsky & Kozlov 2003). Membrane fission may be driven by the relaxation of a highly curved funnel-like shape, generated in the thin necks that connect the dividing membranes (Kozlovsky & Kozlov 2003) (Figure 3). The generation of protrusions and dimples in fusion and of narrow necks in fission requires energies of tens of kilocalories per mole (Efrat et al. 2007, Kozlov & Chernomordik 1998, Martens et al. 2007), which must be provided by proteins. S 14.6 C E I N A D V A N Shibata et al. These considerations show that both membrane shaping and remodeling employ proteins that generate high membrane curvature. GENERATION OF CURVATURE DURING VESICLE BUDDING Most of the principles by which membranes are shaped and remodeled were discovered in studies on clathrin-mediated endocytosis, where a small vesicle with a highly curved membrane is generated from the essentially flat plasma membrane (for review, see Brodsky et al. 2001, Higgins & McMahon 2002). Clathrin forms cages around a budding vesicle, but it does not form a rigid scaffold and does not interact directly with the membrane (Kirchhausen 2000). Therefore, several other components generate most of the curvature required for vesicle budding, employing hydrophobic insertion (wedging) and scaffolding mechanisms. Hydrophobic insertion is often mediated by amphipathic helices. Examples include the 17:48 N-terminal helix of epsin (helix zero of the ENTH domain) (Ford et al. 2002), the amphipathic N-terminal helices of the BAR domain– containing proteins amphiphysin (Peter et al. 2004) and endophilin (Gallop et al. 2006, Masuda et al. 2006), and the amphipathic helix between the PX and BAR domains of Sorting Nexin 9 (Pylypenko et al. 2007). The lipidinserted polypeptide segments are often flexible and unstructured in solution and adopt a helical conformation only upon binding to a lipid bilayer (Gallop et al. 2006). The amphipathic helices of the ENTH and N-BAR domains penetrate the lipid monolayers to a depth that is most effective in generating local curvature (Gallop et al. 2006); calculations show that with only a few mole percent of these proteins, a curvature of a few tens of nanometers radius can be generated (Campelo et al. 2008). The scaffolding mechanism is exemplified by N-BAR domains. These form bananashaped dimers, whose concave surface contains positive charges that interact with the head groups of the phospholipids (Li et al. 2007, Peter et al. 2004, Pylypenko et al. 2007, Weissenhorn 2005, Zhu et al. 2007). They can form curvature radii of 11–14 nm. Another type of BAR domain, the F-BARs, uses the scaffolding mechanism to generate membrane structures of lower curvature (∼30 nm radius) (Henne et al. 2007, Shimada et al. 2007). They can form oligomeric spirals of different pitch through tip-to-tip and lateral interactions to generate tubules of variable diameters (Frost et al. 2008). The BAR domains are rigid, a prerequisite to impose their curvature onto the interacting bilayer. Interestingly, while all these BAR domains generate positive curvature, yet another type of BAR domains, the I-BARs, have convex surfaces and induce negative curvature in cellular protrusions (Saarikangas et al. 2009). The fission of a vesicle from the plasma membrane is mediated by the dynamin-1 protein (Kosaka & Ikeda 1983). Dynamin-1 is a GTPase that also contains a lipid-interacting PH domain and a proline-rich domain with which it interacts with other curvatureinducing proteins, such as amphiphysin and endophilin (Praefcke & McMahon 2004). Dynamin-1 forms short spirals around the neck of a budding vesicle that continuously form and disassemble during the GTP hydrolysis cycle (Bashkirov et al. 2008, Pucadyil & Schmid 2008, Ramachandran & Schmid 2008). It was suggested that upon GTP hydrolysis, the dynamin helix is released from the lipid surface but transiently maintains a rigid frame around the membrane neck, preventing it from swelling (Bashkirov et al. 2008); the elastic stress accumulated within the membrane neck would then relax by hemi-fission followed by fission of the neck. Most curvature-inducing proteins use a combination of hydrophobic insertion (wedging) and scaffolding mechanisms. For example, N-BAR proteins insert the N-terminal amphipathic helix into the lipid bilayer and use their banana shape to scaffold the curved membrane. Similarly, dynamin-1 inserts a loop of its PH domain into the bilayer and forms a spiral scaffold around the neck of the budding vesicle (Hinshaw & Schmid 1995, Ramachandran & Schmid 2008, Takei et al. 1995). Endocytosis appears to be facilitated by the actin cytoskeleton. In fact, some curvaturegenerating proteins, such as the BAR proteins, interact with actin (Itoh et al. 2005, Tsujita et al. 2006), suggesting that actin may facilitate membrane deformation or provide mechanical force for vesicle fission (Engqvist-Goldstein & Drubin 2003). Taken together, the number of proteins involved in endocytosis is bewildering, and it is still unclear how they cooperate to generate curvature during vesicle formation and fission. The same principles for generating membrane curvature, hydrophobic insertion (wedging) and scaffolding, apply to the budding of vesicles from the ER and Golgi. However, compared to the formation of an endocytic vesicle, only a small number of different proteins is required. The small GTP binding proteins Sar1 and Arf1 first use the hydrophobic insertion mechanism to generate local curvature by inserting an N-terminal amphipathic helix into the lipid bilayer (Beck et al. 2008, Bielli et al. 2005, Krauss et al. 2008, Lee et al. 2005, PX and PH domains: lipid binding domains often specific for certain phosphoinositides E V I E W S www.annualreviews.org • Mechanisms Shaping the Membranes of Cellular Organelles 14.7 I N E 29 June 2009 A C ARI R Annu. Rev. Cell Dev. Biol. 2009.25. Downloaded from arjournals.annualreviews.org by TEL-AVIV UNIVERSITY on 08/24/09. For personal use only. ANRV389-CB25-14 D V A N ANRV389-CB25-14 ARI 29 June 2009 17:48 Lundmark et al. 2008). In the next stage of vesicle budding, coat proteins serve as scaffolds. In the case of vesicles budding from the ER, the COPII coat is assembled in two steps. First, the a Annu. Rev. Cell Dev. Biol. 2009.25. Downloaded from arjournals.annualreviews.org by TEL-AVIV UNIVERSITY on 08/24/09. For personal use only. Sheets Nuclear envelope Tubules b bowtie-shaped Sec23/24 complex forms the inner shell of the coat by binding to the membrane through its concave surface (Bi et al. 2002). In the next step, the Sec13/31 heterodimers bind, generating the cage-like outer shell of the COPII coat (Fath et al. 2007, Stagg et al. 2008). In the case of vesicles budding from the Golgi, Arf1 recruits the coatamer complex in a single step to form the COPI coat (Donaldson et al. 1992, Zhao et al. 1997). As will be discussed below, several of the paradigms of curvature generation during vesicle budding apply to organelle morphogenesis. However, some differences should be pointed out. First, a vesicle is not a permanent structure; it generally fuses quickly with another membrane, dissipating its high membrane curvature. By contrast, cellular organelles, though dynamic, maintain their curvature over long time periods. Second, all curvature-generating proteins involved in vesicle formation are soluble proteins that associate only transiently with the lipid bilayer, whereas at least some proteins that determine organelle curvature are integral membrane proteins. Finally, the budding of a small vesicle only requires the generation of local curvature, whereas organelles often need mechanisms that shape large membrane areas. GENERATION OF THE TUBULAR ER NETWORK Figure 4 R E V Morphology of the endoplasmic reticulum (ER). (a) The different subdomains of the ER, the nuclear envelope, the peripheral sheets, and the tubular network, are highlighted in a COS-7 cell expressing a general ER marker fused to green fluorescent protein (GFP). (b) Tubule-forming proteins localize to the tubular ER, as seen for a GFP fused version of Rtn4a (green) in a BSC1 cell. By I E contrast, W the luminal ER protein calreticulin (red ) is also found in the nuclear envelope and peripheral ER sheets. S 14.8 C E I N A D V A N Shibata et al. The ER is composed of sheet-like structures and a polygonal array of tubules connected by three-way junctions (Baumann & Walz 2001, Shibata et al. 2006, Voeltz et al. 2002) (Figure 4a). ER sheets are relatively flat areas, often extending over several micrometers with little membrane curvature. ER sheets include the nuclear envelope whose curvature is negligible owing to the large size of the nucleus it surrounds. In addition, in higher eukaryotes ER sheets are found in the peripheral ER (the ER outside the nuclear envelope), usually close to the nucleus. In contrast, ER tubules are cylindrical units with high membrane curvature in cross-section. In Saccharomyces cerevisiae, ER tubules have a diameter of about 30 nm and are located close to the cell cortex. They are 17:48 connected by a few tubules with the nuclear envelope. In higher eukaryotes, ER tubules have a diameter of about 60 nm, and the peripheral ER extends throughout the entire cytoplasm. The whole ER is a continuous membrane system with a common luminal space. How the morphologically distinct domains are set up is still poorly understood, but significant progress has been made in elucidating the mechanism by which the tubular ER network is generated and maintained. In all eukaryotes the ER network is extremely dynamic, with tubules undergoing continuous fusion and fission reactions (Lee & Chen 1988, Prinz et al. 2000). In mammalian cells, ER tubules are often close to microtubules and can be generated by the force exerted by either microtubule-associated molecular motors or by polymerizing microtubule tips (Terasaki et al. 1986, Waterman-Storer & Salmon 1998). The proteins that mediate the interaction with motors or microtubules have not yet been identified. In S. cerevisiae, ER tubules are generated along actin filaments, and some evidence suggests that the myosin Myo4p is involved (Estrada et al. 2003). Although the cytoskeleton is required for the generation and distribution of the ER network, ultimately the alignment of ER tubules with the cytoskeleton is not perfect. In addition, the ER network does not collapse upon depolymerization of actin filaments in yeast (Prinz et al. 2000), and it retracts only with some delay upon depolymerization of microtubules in mammalian cells (Terasaki et al. 1986). Thus, mechanisms other then those based on the cytoskeleton stabilize and maintain the tubules. In vitro, an ER network can be generated by the fusion of small vesicles in the absence of an intact cytoskeleton (Dreier & Rapoport 2000), suggesting that these additional mechanisms are sufficient to generate and stabilize ER tubules. Proteins involved in ER tubule formation were discovered by the use of an in vitro system, in which the formation of an ER network could be recapitulated, starting with small ER-enriched vesicles from Xenopus laevis eggs (Dreier & Rapoport 2000, Voeltz et al. 2006). Network formation is inhibited by reagents that modify cysteine residues in membrane proteins. The target of these reagents was identified as reticulon 4a (Rtn4a) and confirmed by the inhibitory effect of Rtn4a antibodies on network formation (Voeltz et al. 2006). Rtn4a belongs to the reticulon family, whose members exist in most, if not all, eukaryotic cells (Oertle & Schwab 2003). There are four reticulon genes in mammals, each with different splice forms, and two reticulon genes in S. cerevisiae (Rtn1p and Rtn2p). The conserved region is the Cterminal reticulon homology domain of approximately 200 amino acids, which contains two long hydrophobic segments that each sit in the membrane as a hairpin and expose no significant part of themselves on the luminal side of the membrane. The deletion of RTN1 and RTN2 in S. cerevisiae does not result in ER morphology defects, unless the cells are placed in a high osmolarity medium, indicating that additional proteins are involved in ER tubule formation (Voeltz et al. 2006). A candidate protein, DP1 (also called REEP5), was found by pull-down experiments with Xenopus Rtn4a. Similarly, the yeast homolog of DP1, Yop1p, was found to interact with both Rtn1p and Rtn2p. The DP1/Yop1p protein family is again ubiquitous and includes six mammalian DP1 and REEP proteins. The DP1/Yop1p proteins do not share any primary sequence homology with the reticulons, but they contain a conserved domain of about 200 amino acids, which includes two long hydrophobic segments that each form a hairpin in the membrane. The deletion of RTN1, RTN2, and YOP1 in S. cerevisiae leads to a drastic loss of tubular ER (Voeltz et al. 2006). Most of the peripheral ER converts into sheets, although some tubules are still present. The depletion of three reticulons in mammalian cells has similar effects (Anderson & Hetzer 2008). These data suggest that the reticulons and DP1/Yop1p are major components that shape the tubular ER. Interestingly, the triple-deletion mutant in yeast exhibits only a moderate growth defect, indicating that in yeast much of the tubular ER Coatamer: a complex of proteins that serves as the building block for COPI coated vesicles E V I E W S www.annualreviews.org • Mechanisms Shaping the Membranes of Cellular Organelles 14.9 I N E 29 June 2009 A C ARI R Annu. Rev. Cell Dev. Biol. 2009.25. Downloaded from arjournals.annualreviews.org by TEL-AVIV UNIVERSITY on 08/24/09. For personal use only. ANRV389-CB25-14 D V A N ANRV389-CB25-14 ARI 29 June 2009 is dispensable (Voeltz et al. 2006). However, the simultaneous depletion of Rtn1 and Yop1 in Caenorhabditis elegans causes a 60% decrease in embryonic viability (Audhya et al. 2007), suggesting that an intact tubular ER is important in higher eukaryotes. Consistent with their proposed role in ER tubule formation, the reticulons and DP1/Yop1p localize almost exclusively to the tubular ER; they avoid the low-curvature areas of the nuclear envelope and the sheets of the peripheral ER even when highly overexpressed (Voeltz et al. 2006) (Figure 4b). In mammalian cells, the overexpression of some isoforms of the reticulons leads to long, unbranched tubules and to the disappearance of peripheral ER sheets. In addition, the overexpression of the reticulons or DP1 makes the ER network resistant to the collapse that normally follows the depolymerization of microtubules, indicating that these proteins can stabilize ER tubules (Shibata et al. 2008). In yeast, the overexpression of Rtn1p or Yop1p also leads to long tubules (Voeltz et al. 2006). The reticulons and DP1/Yop1p are not only necessary, but also sufficient for tubule formation (Hu et al. 2008). Purified yeast Yop1p or Rtn1p, reconstituted with pure lipids into proteoliposomes, deform the lipid bilayer into narrow tubules. The diameter of these structures (15–17 nm) is smaller than that of normal ER tubules, owing to a higher concentration of the tubule-inducing proteins. When the reticulons or DP1/Yop1p are overexpressed in vivo, the ER tubules also become narrower, and luminal proteins are displaced. The reticulons and DP1/Yop1p might shape ER tubules by utilizing two cooperating mechanisms, hydrophobic insertion (wedging) and scaffolding (Hu et al. 2008, Shibata et al. 2008) (Figure 5). The proposed hydrophobic insertion mechanism is based on the presence of the two hydrophobic hairpins in the tubule-shaping proteins. Although the structure of the hairpins is unknown, they might displace the lipids preferentially in the upper part of the outer monolayer, causing local curvature in an analogous way to the insertion of amphipathic helices. It is Annu. Rev. Cell Dev. Biol. 2009.25. Downloaded from arjournals.annualreviews.org by TEL-AVIV UNIVERSITY on 08/24/09. For personal use only. V I E W R E 17:48 S 14.10 C E I N A D V A N Shibata et al. a Hairpin structure of reticulons and DP1/Yop1p in membrane b Figure 5 Wedging and scaffolding by the reticulons and DP1/ Yop1p. (a) The conserved region of the reticulons and DP1/Yop1p contains two long hydrophobic segments that each form a hairpin in the membrane. The double-hairpin structure might adopt a wedge shape that curves the membrane. (b) The reticulons and DP1/Yop1p also form homo- and heterooligomers that could serve as arc-like scaffolds (orange) around a membrane tubule ( gray). also conceivable that they span the outer monolayer and shallowly insert into the inner monolayer, close to the membrane mid surface, which would bend the bilayer in a similar way. Finally, they could simply form a cone shape across the entire bilayer and bend the membrane by the bilayer-coupling mechanism (Figure 5a), although this alone would likely be insufficient to generate enough curvature. A scaffolding mechanism for the reticulons and DP1/Yop1p is suggested by the observation that these proteins form homo- and heterooligomers and are relatively immobile in the plane of the membrane (Shibata et al. 2008). Mutants of yeast Rtn1p, which no longer localize exclusively to the tubular ER or are inactive 17:48 in inducing ER tubules, oligomerize less extensively and are more mobile. The conserved domains of the reticulons and DP1/Yop1p, which contain the two hairpin-shaped membrane segments, are responsible for the localization of the proteins and their oligomerization. One possibility is that the oligomers form arc-like scaffolds that would mold the bilayer into tubules (Figure 5a). Calculations show that these arcs or rings around the tubules would need to occupy only about 10% of the total membrane surface to generate near perfect tubules (Hu et al. 2008). Rtn1p and Yop1p are among the most abundant membrane proteins in yeast and could together occupy 10% of the total surface of the peripheral ER. Assuming that the reticulons and DP1/Yop1p form arcs rather than rings around the tubules could explain why other ER membrane proteins can diffuse in the plane of the membrane over long distances. The oligomers formed by the reticulons and DP1/Yop1p must be of relatively small size and undergo continuous assembly and disassembly, because otherwise all molecules would eventually localize to one site of the ER and no molecules would be available for the rest of the ER. The disassembly of the oligomers may indeed be an active process because the depletion of ATP in yeast or mammalian cells decreases the diffusional mobilities of the reticulons and DP1/Yop1p in the plane of the membrane (Shibata et al. 2008). The mechanism by which the oligomers are disassembled remains to be clarified. The reticulons and DP1/Yop1p are not the only proteins required for the formation of the tubular ER network. Recent work indicates that a class of membrane-bound, dynamin-like proteins, which includes the atlastins in mammals and Sey1p in yeast, play an important role (Hu et al. 2009). Although the atlastins and Sey1p do not share sequence homology, they belong to the same family of dynamin-like GTPases and have the same membrane topology; that is, they possess at their C-termini two hydrophobic membrane anchors with only 3– 4 amino acids between them on the luminal side of the membrane. These GTPases localize predominantly to the tubular ER and interact with the reticulons and DP1/Yop1p. Depletion of the atlastins or expression of dominantnegative versions leads to long ER tubules with little branching. Antibodies to the atlastins inhibit ER network formation in vitro, and yeast cells lacking Sey1p and either Rtn1p or Yop1p have strong defects in the morphology of the tubular ER. Together, these data indicate that the atlastins and Sey1p function in ER network formation, likely by affecting the generation of tubule branches. A possible mechanism by which the atlastins and Sey1p function in ER network formation is based on the analogy to dynamin-1 (Hu et al. 2009). Assuming a similar mechanism for the atlastins and Sey1p, these proteins could be involved in the fission of ER tubules. This process would be facilitated if the ER tubules were under tension by the cytoskeleton, either by direct association or through molecular motors. Alternatively, the atlastins and Sey1p could function in fusion. In fact, they are actually more similar to the dynamin-like mitofusion/Fzo1p proteins, which are anchored in the outer mitochondrial membrane by two closely spaced hydrophobic segments and are implicated in the fusion of the tubular mitochondria (see below). Given that Sey1p is much less abundant than Rtn1p or Yop1p in yeast, it seems likely that the reticulons and DP1/Yop1p are the main structural components that shape the tubules of the ER, whereas the atlastins and Sey1p associate with them at discrete points along the tubules. During GTP hydrolysis, the GTPases would destabilize the lipid bilayer, which in turn would lead to fission or fusion of the ER tubules. Interestingly, mutations in one atlastin isoform cause the most frequent early-onset hereditary spastic paraplegia (HSP), a disease affecting the long axons of corticospinal motor neurons (Salinas et al. 2008). Given the role of the atlastins in ER network formation, the disease may be caused by ER morphology defects (Hu et al. 2009). It is plausible that mutants of atlastin would affect most severely the longest axons, in which the ER network Hereditary spastic paraplegia: a group of neuronal disorders marked by progressive spasticity and weakness of the lower limbs E V I E W S www.annualreviews.org • Mechanisms Shaping the Membranes of Cellular Organelles 14.11 I N E 29 June 2009 A C ARI R Annu. Rev. Cell Dev. Biol. 2009.25. Downloaded from arjournals.annualreviews.org by TEL-AVIV UNIVERSITY on 08/24/09. For personal use only. ANRV389-CB25-14 D V A N ANRV389-CB25-14 ARI 29 June 2009 needs to extend a long distance from the cell body to the axon terminus. Two other proteins, spastin and REEP1, mutant alleles of which are known to frequently cause HSP, may also affect ER network formation. A long isoform of the microtubule-severing adenosine triphosphatase (ATPase) spastin is membrane-bound and interacts with the atlastins and the reticulons (Evans et al. 2006, Mannan et al. 2006, Sanderson et al. 2006); it could affect microtubule dynamics and thus ER tubule formation. REEP1, a member of the DP1/Yop1p family of tubule-forming proteins could affect the shape of ER tubules directly. These three proteins collectively account for well over 50% of all HSP cases, suggesting that ER morphology defects may be a predominant mechanism underlying these neuropathogenic disorders. In summary, the generation of the tubular ER network appears to involve several principles: (a) the cytoskeleton-based pulling of tubules, (b) the shaping of tubules by the reticulons and DP1/Yop1p, which might utilize both hydrophobic insertion (wedging) and scaffolding mechanisms, and (c) the formation of an interconnected network, which involves the atlastin and Sey1p GTPases. These principles are still rather speculative, but they are consistent with how membranes are shaped and remodeled in other systems. Clearly, the mechanism of ER network formation requires further studies, particularly in systems composed of purified components. Structural studies are needed, but will be challenging, given that most of the players are integral membrane proteins that form heterogeneous oligomers. The regulation of the components is another unresolved issue. For example, there are major changes of ER morphology during mitosis in higher eukaryotes, and it is therefore possible that the components involved in ER network formation are modified, perhaps by phosphorylation. Finally, it is possible that at least some of the components involved in network formation have other functions as well. For example, some of the reticulon isoforms have long segments in the cytosol, which presumably interact with other proteins. It will be interesting to see whether Annu. Rev. Cell Dev. Biol. 2009.25. Downloaded from arjournals.annualreviews.org by TEL-AVIV UNIVERSITY on 08/24/09. For personal use only. V I E W R E 17:48 S 14.12 C E I N A D V A N Shibata et al. these proteins link the tubule-forming proteins to other cellular processes. GENERATION OF MEMBRANE SHEETS IN THE ER Peripheral ER sheets have little overall curvature along the plane of the two membranes. The abundance of sheets varies among different cell types, but is particularly prominent in cells that are highly specialized in secretion. Here, the sheets are generally stacked on top of each other, and both the cytosolic and luminal spacings are constant (∼50–100 nm). Because the luminal width of sheets is about the same as the diameter of ER tubules, the edges of the sheets must have a similarly high curvature as the tubules. The mechanisms used to form peripheral ER sheets are largely unknown. However, one possibility is that the same proteins that stabilize ER tubules, the reticulons and DP1/Yop1p, stabilize the high curvature of the sheet edges by forming arc-like scaffolds along the curved membrane (Figure 6a). Scaffolding the edges could generate even large areas of sheets, although the width between the membranes would likely fluctuate. This mechanism may explain why the observed luminal width of ER sheets is similar to the diameter of ER tubules, and why the reticulons are upregulated when ER sheets proliferate (Benyamini et al. 2009). Indeed, endogenous and overexpressed forms of the reticulons, as well as the atlastins, localize to the edges of ER sheets (Hu et al 2009 and Y. Shibata, unpublished data). The sheetedge localization would also allow for tubules to readily emerge from or fuse into the edges of peripheral ER sheets. An alternative or additional mechanism by which peripheral ER sheets could be formed is intraluminal bridges (Figure 6b). Such bridges have been observed by deep-etch electron microscopy and could prevent excessive fluctuations of the luminal width (Senda & Yoshinaga-Hirabayashi 1998). A candidate bridging protein may be CLIMP-63, an abundant integral membrane protein with a large 29 June 2009 17:48 b c Figure 6 Three possible mechanisms to generate the sheets of the peripheral endoplasmic reticulum (ER). The cross-section of an ER membrane sheet is shown in gray. (a) The tubule-shaping proteins (the reticulons and DP1/Yop1p) could form arc-like scaffolds (orange) around the edges of the sheet and stabilize their high curvature. (b) Transmembrane proteins (blue) that are localized in both membranes could interact through their luminal domains to keep the two membranes flat and at a fixed distance from one another. (c) Transmembrane proteins ( yellow) could form a flat scaffold on the cytosolic side of the ER membrane. coiled-coil luminal domain that can form 90nm rod-like oligomers (Klopfenstein et al. 2001). At endogenous levels, this protein localizes to peripheral sheets and, when overexpressed, it induces the proliferation of sheets with a constant luminal width (Y. Shibata, unpublished data). However, depletion of this protein by RNAi does not noticeably affect the abundance of peripheral sheets (unpublished data), suggesting that other functionally equivalent proteins may exist. Peripheral ER sheets could also be generated by proteins that form a flat scaffold on the cytoplasmic side of the ER membrane (Figure 6c). A candidate scaffolding protein is p180, an abundant integral membrane protein with a large cytosolic coiled-coil domain. At endogenous levels, this protein also localizes predominantly to ER sheets (Y. Shibata, unpublished data). Overexpression of p180 in yeast or in tissue culture cells leads to the proliferation of ER sheets, whereas depletion of this protein in a monocyte cell line prevents ER proliferation during differentiation (Becker et al. 1999, Benyamini et al. 2009). p180 may also be involved in stacking sheets by bridging the cytosolic surfaces of ER membranes. The functions proposed for p180 are analogous to those of the golgins, which are large coiled-coil proteins that are thought to flatten and link the stacks of the Golgi apparatus (Short et al. 2005). Because CLIMP-63 and p180 are found only in vertebrates, it is possible that additional players are involved in sheet formation. Interestingly, several observations suggest a tug-of-war between peripheral ER sheets and tubules. The overexpression of Rtn4a in mammalian cells leads to the disappearance of ER sheets (Voeltz et al. 2006), whereas the overexpression of CLIMP-63 reduces the abundance of ER tubules (Y. Shibata, unpublished data). The reduction or absence of the tubule-forming proteins in mammalian or yeast cells converts the tubules into sheets (Anderson & Hetzer 2008, Voeltz et al. 2006). These data indicate that the relative abundance of ER sheets and tubules is dictated by the expression levels of sheet- and tubule-shaping proteins. The other sheet-like ER domain is the nuclear envelope. The sheets formed by the inner and outer nuclear membranes are again separated by a distance of ∼50 nm, which means that at the sites where they are connected, at the nuclear pores, the membrane has a high curvature (Antonin & Mattaj 2005). The nuclear envelope might be shaped in large part by a E V I E W R Annu. Rev. Cell Dev. Biol. 2009.25. Downloaded from arjournals.annualreviews.org by TEL-AVIV UNIVERSITY on 08/24/09. For personal use only. a S www.annualreviews.org • Mechanisms Shaping the Membranes of Cellular Organelles 14.13 I N E ARI A C ANRV389-CB25-14 D V A N ANRV389-CB25-14 ARI 29 June 2009 luminal bridging mechanism involving the SUN and nesprin proteins (Worman & Gundersen 2006). The SUN proteins are integral membrane proteins that localize to the inner nuclear membrane through their interaction with the nuclear lamins, whereas the nesprins span the outer nuclear membrane. These proteins interact with each other in the nuclear envelope lumen through nesprin’s conserved KASH domain (Padmakumar et al. 2005). Depletion of either the SUNs or the nesprins results in a nuclear envelope with large bulges (Crisp et al. 2006). Homologs of both the SUNs and nesprins also exist in yeast and have been shown to affect nuclear envelope morphology (King et al. 2008). However, these proteins appear to be involved in other processes as well, such as nuclear positioning and spindle pole body duplication. The high curvature of the nuclear membranes around the nuclear pores is likely stabilized by a nuclear pore protein subcomplex (Nup84 complex). This complex contains the Sec13 protein, which is also a component of the COPII coat required for vesicle budding from the ER. In addition, it contains a nuclear porespecific Sec13 homolog, Seh1, as well as five other proteins with structural motifs that are also found in vesicle coats (Devos et al. 2004). In one model, these proteins form rods along the curved nuclear pore membrane vertical to the plane of the envelope (Debler et al. 2008, Hsia et al. 2007). Alternatively, they could form a membrane-associated lattice, similar to that seen in the COPII coats (Brohawn et al. 2008). In either case, they would serve as a scaffold that stabilizes the high curvature. Additionally, several nuclear pore proteins contain amphipathic sequence motifs, which at least in the case of a component of the Nup84 complex can insert into the lipid bilayer and stabilize local curvature (Drin et al. 2007). The nuclear pores could contribute to maintain the flat nuclear envelope, similar to how the reticulons and DP1/Yop1p might generate peripheral ER sheets. However, the number of nuclear pores is variable in different cell types (Hetzer et al. 2005), and nuclear envelopes without pores can IM: inner mitochondrial membrane Annu. Rev. Cell Dev. Biol. 2009.25. Downloaded from arjournals.annualreviews.org by TEL-AVIV UNIVERSITY on 08/24/09. For personal use only. OM: outer mitochondrial membrane V I E W R E 17:48 S 14.14 C E I N A D V A N Shibata et al. be formed in Xenopus egg extracts (Goldberg et al. 1997, Hetzer et al. 2005). Recent results suggest that reticulons and Yop1p are involved in early stages of nuclear pore assembly (Dawson et al. 2009), perhaps by transiently stabilizing high curvature at the site of pore insertion. However, these proteins must eventually leave the nuclear envelope and are not essential for nuclear pore insertion. SHAPING MITOCHONDRIA Several aspects discussed for the morphogenesis of the ER apply to mitochondria as well. However, mitochondria are more complex than the ER in that they have two membranes, the outer mitochondrial membrane (OM) and the inner mitochondrial membrane (IM) (Figure 7a). The IM separates the innermost compartment, the matrix, from the intermembrane space; it folds and undulates to form deep invaginations, called cristae. Although mitochondrial shape can vary greatly among different species and cell types and change under varying physiological conditions, these architectural features are conserved in all eukaryotes. Like the ER, mitochondria form an interconnected tubular network (Figure 7b) that continuously fuses and divides, and they use the cytoskeleton to move and distribute within the cell. In budding yeast and plants, this movement occurs mostly along actin filaments via myosin motors (Boldogh & Pon 2007, Frederick & Shaw 2007). In other eukaryotes, including fission yeast, mitochondria move on microtubules using kinesin or dynein motors. Depolymerization of actin or microtubules inhibits mitochondrial movement. While prolonged perturbation leads to clustered and/or coalesced mitochondria, acute cytoskeletal depolymerization has little effect (for review, see Griparic & van der Bliek 2001), indicating that mitochondrial morphology may not be as highly dependent on the cytoskeleton as the ER. In higher eukaryotes, the interaction between mitochondria and the kinesin motor may be mediated by the large GTPase Miro, a protein that is anchored to the OM via a C-terminal ANRV389-CB25-14 ARI 29 June 2009 17:48 a Outer membrane Inner membrane b c Outer membrane Cristae Inner membrane Figure 7 The morphology of mitochondria. (a) Thin-section electron micrograph of a mitochondrion from a bat pancreatic cell shows the different membrane structures of the outer mitochondrial membrane (OM), inner nuclear membrane (IM), and cristae. (Adapted from Fawcett 1981.) (b) Mitochondria form an interconnected tubular network, as visualized by Mitotracker staining in a C2C12 cell (Griparic & van der Bliek 2001). (c) A reconstructed three-dimensional view, using electron microscopic tomography, of the cristae of a mitochondrion isolated from rat liver (Mannella 2000). Arrows denote the tubular cristae junctions. transmembrane segment, as well as the cytosolic Milton protein (Glater et al. 2006, Guo et al. 2005, Wang & Schwarz 2009). In budding yeast, the Miro-related protein Gem1p also plays a role in mitochondrial morphology (Frederick et al. 2004), but its precise function is unclear because mitochondrial movement is actin-based; Gem1p does not associate with the myosin Myo2p, and functions independently of known myosin adaptors (Frederick et al. 2008). The overall morphology of the mitochondrial network depends on balanced rates of fission and fusion. Perturbing the fusion machinery leads to more fragmented mitochondria, E V I E W R S 14.15 I N E www.annualreviews.org • Mechanisms Shaping the Membranes of Cellular Organelles A C Annu. Rev. Cell Dev. Biol. 2009.25. Downloaded from arjournals.annualreviews.org by TEL-AVIV UNIVERSITY on 08/24/09. For personal use only. Cristae D V A N ANRV389-CB25-14 ARI 29 June 2009 whereas defects in the fission apparatus lead to a netlike, interconnected morphology (for review, see Detmer & Chan 2007, Hoppins et al. 2007, Okamoto & Shaw 2005). Interestingly, perturbing both processes restores normal morphology in some cells. Both fission and fusion events involve dynamin-like proteins. For mitochondrial fission, the soluble dynamin-related protein Drp1/Dnm1p is first recruited to certain sites of the mitochondrial OM by the soluble adaptor Cav4/Mdv1p and the membrane protein Fis1. In vitro experiments and the analogy to classical dynamin-1 suggest that these proteins form oligomeric spirals around the mitochondria (Ingerman et al. 2005); fission of both the OM and IM would then occur through some nucleotidedependent conformational change of these oligomers. Fusion of the OM is mediated by the dynamin-like protein mitofusin/Fzo1p (fuzzy onion). The protein is anchored in the OM via two transmembrane segments that form a hairpin within the membrane, a topology that is similar to that of the atlastins and Sey1p; the GTPase domain faces the cytosol. Mitofusin/Fzo1p proteins interact in trans with each other via a C-terminal coiled-coil domain, which could tether together opposing OMs (Koshiba et al. 2004). However, because the interacting proteins form an antiparallel coiledcoil, the association would be expected to keep the membranes at some distance, rather than force them into close apposition for fusion as the zippering of SNARE helices does. It appears more likely that mitofusin/Fzo1p, like other dynamin-like GTPases, forms spirals around membrane tubules; the highly curved free tips of the tubules would be fusogenic (Praefcke & McMahon 2004). A mitofusin-related, bacterial protein indeed forms spirals around lipid tubes (Low & Lowe 2006). Interestingly, the X-ray structure of this protein indicates that in the nucleotide-free and GDP-bound states the hydrophobic segments are buried (Low & Lowe 2006). Although the bacterial dynamins’ function is unknown, they might offer a simplified system to study the mechanism of the mitofusin/Fzo1p. Annu. Rev. Cell Dev. Biol. 2009.25. Downloaded from arjournals.annualreviews.org by TEL-AVIV UNIVERSITY on 08/24/09. For personal use only. V I E W R E 17:48 S 14.16 C E I N A D V A N Shibata et al. Fusion of the IM is mediated by another dynamin-like protein, OPA1/Mgm1p, which associates with the IM and exposes its GTPase domain to the intermembrane space. This protein is proteolytically processed; in yeast, both the shorter, soluble form and the longer, membrane-bound form are necessary for IM fusion (Herlan et al. 2003), whereas in mammals, only the membrane-bound version is required (Ishihara et al. 2006). Mechanistic details of how these proteins mediate fusion are unclear, but perhaps they also create fusogenic areas by promoting high curvature. In yeast, IM and OM fusion is coupled by a physical link between Mgm1p and Fzo1p through the adaptor protein Ugo1p (Sesaki & Jensen 2004). How the overall tubular structure of mitochondria is generated is not well understood. Whereas mitochondrial tubules have a typical diameter of 0.5–1 um and are therefore less curved than ER tubules (30–60 nm), their overall cylindrical shape suggests that some active mechanism is required to generate them. Are dynamin-mediated fusion and fission and cytoskeletal tethering sufficient for shaping mitochondrial tubules? Most likely not, because disruption of the fusion and fission machinery or acute depolymerization of the cytoskeleton does not discernibly affect the diameter of the tubules. These results indicate that some other structural components are involved in directly shaping mitochondrial tubules. In yeast, a number of proteins have been identified whose absence or mutation cause tubulation defects (for review, see Okamoto & Shaw 2005); the mitochondria are condensed into spherical, giant structures with very large diameter. Some of these proteins—(Mmm1p, Mdm10p, and Mdm12p) were initially proposed to mediate mitochondria-actin attachments by recruiting the Arp2/3 complex (Boldogh et al. 2003). However, these proteins have also been implicated in beta-barrel protein insertion into the OM (Meisinger et al. 2004, 2007), suggesting that some unknown beta-barrel protein may be involved in the tubulation of mitochondria. The possibility also exists that more classical membrane-bending proteins that 17:48 employ hydrophobic insertion and scaffolding mechanisms are involved in shaping mitochondrial tubules. For example, endophilin B1, a BAR domain-containing protein, is involved in maintaining mitochondrial morphology in mammalian cells (Karbowski et al. 2004), although it has been implicated in fission. In addition, still unidentified proteins may exist, similar to the reticulons/DP1/Yop1p involved in ER tubule formation. How the cristae are shaped is another interesting issue. These membrane structures originate at narrow tubular openings of the IM called cristae junctions (Figure 7c). The cristae themselves are tubular, sheet-like, or even crystalloid (Figure 7c). They can be interconnected by three-way junctions and can span the entire width of a mitochondrion (for review, see Frey & Mannella 2000, Mannella 2006). Whereas cristae adapt their shapes depending on varying physiological conditions, the shape of cristae junction tubules remains constant (a cross-sectional diameter of ∼28 nm), suggesting that a tight protein scaffold maintains this structure. At least two independent mechanisms that involve wedging and/or scaffolding have emerged that contribute to the formation of the high curvature areas of cristae. The first involves the ATP synthase machinery, which is an abundant component of cristae membranes. Mutant yeast cells lacking the e and g subunits of the ATP synthase have mitochondria with onion-like sheets of IM (Paumard et al. 2002). These subunits are nonessential for the enzymatic activity, but are required for the dimerization and oligomerization of the ATP synthase. Structural data indicate that the ATP synthase dimer has a wedge shape, where the dimer complex takes up more space on the matrix side of the membrane (MinauroSanmiguel et al. 2005). Additionally, electron microscopy shows that ATP synthase dimers form 1-um oligomeric ribbons around the highcurvature tips of the cristae (Strauss et al. 2008). Thus, like the reticulons and DP1/Yop1p, the ATPase complex may cause membrane curvature by both wedging and scaffolding. Recent results suggest the e and g subunits of the yeast ATP synthase cooperate with Fcj1p, a homolog of the mammalian mitofilin protein, to form cristae junctions (Rabl et al. 2009). The overexpression of Fcj1p increases the number of cristae junctions and its deletion results in the loss of junctions. Fcj1p may stabilize the negative curvature at branch points of the IM and cristae. The second mechanism involves OPA1/ Mgm1p, the IM dynamin-like proteins implicated in fusion, and the prohibitins. Mutation or depletion of OPA1/Mgm1p or of the prohibitins leads to more fragmented mitochondria with aberrantly large, bulging cristae, which sometimes appear disconnected (Meeusen et al. 2006, Merkwirth et al. 2008). Prohibitins themselves oligomerize into large, ring-shaped scaffold-like structures (Tatsuta et al. 2005), but seem to indirectly affect cristae structure by influencing the proteolytic processing of OPA1 (Merkwirth et al. 2008); expression of the OPA1 long isoform containing the transmembrane segment rescues the cristae morphology defect in cells lacking prohibitin. How exactly these proteins affect cristae morphology remains to be elucidated. Lipid rafts: cholesterol-enriched microdomains in cell membranes GENERATING THE CURVATURE OF CAVEOLAE Caveolae can be considered to be distinct, relatively stable organelles. They are flask-like invaginations located on the surface of many vertebrate cells (Figure 8); the largest diameter of the bulb is 60–80 nm (for review, see Parton et al. 2006, Parton & Simons 2007, Stan 2005). Although discovered more than half a century ago, the function of these structures is not entirely clear; they have been implicated in lipid uptake and regulation, signal transduction, endocytosis, and virus entry. Caveolae are specialized forms of lipid rafts, enriched in cholesterol and sphingolipids (Simons & Toomre 2000). The major structural component of caveolae is the integral membrane protein caveolin (Rothberg et al. 1992). In mammals there are three caveolin genes. Caveolin-1 E V I E W S www.annualreviews.org • Mechanisms Shaping the Membranes of Cellular Organelles 14.17 I N E 29 June 2009 A C ARI R Annu. Rev. Cell Dev. Biol. 2009.25. Downloaded from arjournals.annualreviews.org by TEL-AVIV UNIVERSITY on 08/24/09. For personal use only. ANRV389-CB25-14 D V A N ANRV389-CB25-14 ARI 29 June 2009 17:48 a Annu. Rev. Cell Dev. Biol. 2009.25. Downloaded from arjournals.annualreviews.org by TEL-AVIV UNIVERSITY on 08/24/09. For personal use only. 100 nm b Figure 8 V R E The morphology of caveaolae. (a) The flask-like invaginations of caveolae in an endothelial cell are visualized by thin-section electron microscopy (Stan 2002). (b) Caveolin forms filaments on the surface of caveola, as viewed from the cytoplasm I E using rapid-freeze, deep-etch electron microscopy in a human fibroblast W lower panels show individual caveolae. Image courtesy of J. Heuser. cell. The S 14.18 C E I N A D V A N Shibata et al. is the main member of the family and is expressed in many cell types. Its deletion in mice leads to the complete absence of caveolae (Drab et al. 2001). Caveolin may use both wedging and scaffolding to shape caveolae, similar to the mechanisms by which the reticulons and DP1/Yop1p are thought to shape ER tubules. Caveolin is extremely abundant, with about 145 molecules per typical caveola (Pelkmans & Zerial 2005). The protein contains an amphipathic helix (α2a helix) that is inserted into the bilayer (Le Lan et al. 2006, Parton et al. 2006). This helix also binds cholesterol and could cause the raftlike lipid composition of caveolae (Epand et al. 2005). Calculations show that the curvature of caveolae could be generated by the bilayer coupling mechanism if the amphipathic helix was inserted into the outer leaflet of the bilayer and each helix sequestered 13 cholesterol molecules (Parton et al. 2006). In addition, curvature might be generated by caveolin’s membrane anchor. The protein contains a hydrophobic segment of 33 residues that forms a hairpin in the membrane, with no residue on the external side of the membrane (Glenney & Soppet 1992). This is supported by the fact that caveolin can move into lipid bodies (Fujimoto et al. 2001, Ostermeyer et al. 2001, Pol et al. 2001), in which the protein must sit in a lipid monolayer that surrounds an entirely hydrophobic core. The hairpin structure could cause membrane bending in a similar way as discussed for the tubule-shaping ER proteins. The scaffolding mechanism is supported by electron microscopy studies, which indicate that caveolae are coated by filaments of caveolin (Rothberg et al. 1992) (Figure 8b). In sucrose gradient centrifugation, detergent-solubilized caveolin migrates at approximately 350 kDa (Sargiacomo et al. 1995). Caveolin appears to oligomerize into 10-nm particles (about 10– 15 molecules), which then interact with each other to form filaments. The oligomerization of caveolin requires several regions of the protein (Das et al. 1999, Schlegel & Lisanti 2000, Song et al. 1997), but the exact mechanism of assembly remains unclear. ARI 29 June 2009 17:48 Whether caveolin is sufficient to form caveolae is unknown. Recent studies have identified a conserved cytoplasmic protein, cavin (also known as PTRF), as a putative, additional coat protein (Hill et al. 2008, Liu & Pilch 2008). The cellular level of cavin correlates with that of caveolin, cavin interacts with caveolin in mature caveolae, and the deletion of cavin results in the loss of caveolae (Hill et al. 2008, Liu et al. 2008). What precisely cavin is doing and how it cooperates with caveolin remains to be elucidated. In some endothelial cell types a specialized structure is seen at the neck of caveolae (stomatal diaphragm), whose major component is the PV1 protein (Stan 2005). Several molecules of this single-spanning membrane protein may interact through their extracellular domains to bridge the neck. CONCLUSIONS The regulation of membrane curvature is now recognized as a major factor that determines the shape of membrane-bound organelles. It appears that high-curvature membranes are mainly formed by proteins that utilize hydrophobic insertion (wedging) and scaffolding mechanisms, similar to how curvature is generated in budding vesicles. These curvature-generating mechanisms cooperate with cytoskeleton-based force generation. In addition, membrane remodeling by fusion and fission often involves GTPases of the dynamin family. These principles of membrane shaping and remodeling appear to be quite general, although we have focused only on a few organelles. However, in all cases the mechanistic details are still unclear. It is also generally unknown how flat membrane sheets are generated. Although this does not necessarily require active mechanisms, it is likely that proteins help keep lipid bilayers equidistant and stacked on top of each other. Finally, a major question is how membrane morphology is linked to function. For example, are ER sheets and ER tubules functionally distinct? It has been proposed that ER sheets correspond to rough ER, where membrane-bound polysomes synthesize secretory and membrane proteins; ER tubules may correspond to a protected membrane area where the high membrane curvature precludes the binding of large polysomes, allowing other proteins to gain access to the membrane surface to carry out processes such as lipid synthesis and Ca2+ signaling (Shibata et al. 2006). Again, these proposals are speculative and require experimental testing. Although our discussion of the shaping of organelles demonstrates how little we know, the recent advances in the field give confidence that rapid progress can be made on this fundamental problem in cell biology. Lipid bodies: Particles consisting of a lipid monolayer surrounding a hydrophobic interior made of cholesterol esters and triglycerides SUMMARY POINTS 1. Membrane-bound organelles have characteristic shapes that require mechanisms that regulate membrane curvature. 2. High membrane curvature in organelles is generated by two major mechanisms that are similar to those involved in forming small intracellular vesicles: hydrophobic insertion (wedging) and scaffolding. In addition, cytoskeleton-based force generation plays an important role in shaping organelles. 3. Fusion and fission is a major mechanism of membrane remodeling. It appears to be generally catalyzed by proteins of the dynamin family of GTPases. 4. The tubules of the ER are shaped by the reticulons and DP1/Yop1, which might use hydrophobic insertion (wedging) and scaffolding mechanisms for membrane deformation. The dynamin-like proteins atlastin and Sey1p appear to be required for network formation. E V I E W R S 14.19 I N E www.annualreviews.org • Mechanisms Shaping the Membranes of Cellular Organelles A C Annu. Rev. Cell Dev. Biol. 2009.25. Downloaded from arjournals.annualreviews.org by TEL-AVIV UNIVERSITY on 08/24/09. For personal use only. ANRV389-CB25-14 D V A N ANRV389-CB25-14 ARI 29 June 2009 17:48 5. Similar principles of membrane shaping and remodeling may apply to the morphogenesis of mitochondria and caveolae. FUTURE ISSUES 1. What are the actual structures of the reticulons and DP1/Yop1p? How deeply do the hydrophobic segments insert into the lipid bilayer? Do they form arc-like scaffolds around tubules? Annu. Rev. Cell Dev. Biol. 2009.25. Downloaded from arjournals.annualreviews.org by TEL-AVIV UNIVERSITY on 08/24/09. For personal use only. 2. Are the reticulons and DP1/Yop1p regulated? Do they interact with other proteins that link them to additional cellular functions? 3. Are the atlastins and Sey1p directly involved in fusion or fission of ER tubules? 4. How are ER tubules associated with molecular motors or the tips of growing microtubule or actin filaments? 5. How are the sheets of the peripheral ER and the nuclear envelope generated and maintained? 6. What is the advantage for a cell to have different ER morphologies? Do sheets correspond to sites where membrane-bound polysomes are located, and how are these membrane areas kept distinct? 7. How are the tubules of mitochondria generated? How are the cristae shaped? How do the dynamin-like proteins cause fusion and fission, and how is their activity coupled to cause coordinated changes of the inner and outer membranes? 8. How exactly do caveolin and other proteins cause the formation of flask-like membrane structures? Can the formation of caveolae be reproduced with pure proteins and lipids? DISCLOSURE STATEMENT The authors are not aware of any biases that might be perceived as affecting the objectivity of this review. ACKNOWLEDGMENTS We thank Janet Iwasa for help with the figures and Adrian Salic for critical reading of the manuscript. Y.S. is supported by the American Heart Association. T.A.R. is a Howard Hughes Medical Institute Investigator. LITERATURE CITED V I E W R E Anderson DJ, Hetzer MW. 2008. Reshaping of the endoplasmic reticulum limits the rate for nuclear envelope formation. J. Cell Biol. 182:911–24 Antonin W, Mattaj IW. 2005. Nuclear pore complexes: round the bend? Nat. Cell Biol. 7:10–12 Audhya A, Desai A, Oegema K. 2007. A role for Rab5 in structuring the endoplasmic reticulum. J. Cell Biol. 178:43–56 Bacia K, Schwille P, Kurzchalia T. 2005. Sterol structure determines the separation of phases and the curvature of the liquid-ordered phase in model membranes. Proc. Natl. Acad. Sci. USA 102:3272–77 S 14.20 C E I N A D V A N Shibata et al. 17:48 Bashkirov PV, Akimov SA, Evseev AI, Schmid SL, Zimmerberg J, Frolov VA. 2008. GTPase cycle of dynamin is coupled to membrane squeeze and release, leading to spontaneous fission. Cell 135:1276–86 Baumann O, Walz B. 2001. Endoplasmic reticulum of animal cells and its organization into structural and functional domains. Int. Rev. Cytol. 205:149–214 Baumgart T, Hess ST, Webb WW. 2003. Imaging coexisting fluid domains in biomembrane models coupling curvature and line tension. Nature 425:821–24 Beck R, Sun Z, Adolf F, Rutz C, Bassler J, et al. 2008. Membrane curvature induced by Arf1-GTP is essential for vesicle formation. Proc. Natl. Acad. Sci. USA 105:11731–36 Becker F, Block-Alper L, Nakamura G, Harada J, Wittrup KD, Meyer DI. 1999. Expression of the 180-kD ribosome receptor induces membrane proliferation and increased secretory activity in yeast. J. Cell Biol. 146:273–84 Benyamini P, Webster P, Meyer DI. 2009. Knockdown of p180 eliminates the terminal differentiation of a secretory cell line. Mol. Biol. Cell 20:732–44 Bi X, Corpina RA, Goldberg J. 2002. Structure of the Sec23/24-Sar1 pre-budding complex of the COPII vesicle coat. Nature 419:271–77 Bielli A, Haney CJ, Gabreski G, Watkins SC, Bannykh SI, Aridor M. 2005. Regulation of Sar1 NH2 terminus by GTP binding and hydrolysis promotes membrane deformation to control COPII vesicle fission. J. Cell Biol. 171:919–24 Boldogh IR, Nowakowski DW, Yang HC, Chung H, Karmon S, et al. 2003. A protein complex containing Mdm10p, Mdm12p, and Mmm1p links mitochondrial membranes and DNA to the cytoskeleton-based segregation machinery. Mol. Biol. Cell 14:4618–27 Boldogh IR, Pon LA. 2007. Mitochondria on the move. Trends Cell Biol. 17:502–10 Brodsky FM, Chen CY, Knuehl C, Towler MC, Wakeham DE. 2001. Biological basket weaving: formation and function of clathrin-coated vesicles. Annu. Rev. Cell Dev. Biol. 17:517–68 Brohawn SG, Leksa NC, Spear ED, Rajashankar KR, Schwartz TU. 2008. Structural evidence for common ancestry of the nuclear pore complex and vesicle coats. Science 322:1369–73 Campelo F, McMahon HT, Kozlov MM. 2008. The hydrophobic insertion mechanism of membrane curvature generation by proteins. Biophys. J. 95:2325–39 Chernomordik LV, Kozlov MM. 2005. Membrane hemifusion: crossing a chasm in two leaps. Cell 123:375–82 Chernomordik LV, Kozlov MM. 2008. Mechanics of membrane fusion. Nat. Struct. Mol. Biol. 15:675–83 Chernomordik LV, Zimmerberg J, Kozlov MM. 2006. Membranes of the world unite! J. Cell Biol. 175:201–7 Crisp M, Liu Q, Roux K, Rattner JB, Shanahan C, et al. 2006. Coupling of the nucleus and cytoplasm: role of the LINC complex. J. Cell Biol. 172:41–53 Das K, Lewis RY, Scherer PE, Lisanti MP. 1999. The membrane-spanning domains of caveolins-1 and -2 mediate the formation of caveolin hetero-oligomers. Implications for the assembly of caveolae membranes in vivo. J. Biol. Chem. 274:18721–28 Dawson TR, Lazarus MD, Hetzer MW, Wente SR. 2009. ER membrane-bending proteins are necessary for de novo nuclear pore formation. J. Cell Biol. 184:659–75 Debler EW, Ma Y, Seo HS, Hsia KC, Noriega TR, et al. 2008. A fence-like coat for the nuclear pore membrane. Mol. Cell 32:815–26 Detmer SA, Chan DC. 2007. Functions and dysfunctions of mitochondrial dynamics. Nat. Rev. Mol. Cell Biol. 8:870–79 Devaux PF, Herrmann A, Ohlwein N, Kozlov MM. 2008. How lipid flippases can modulate membrane structure. Biochim. Biophys. Acta 1778:1591–600 Devos D, Dokudovskaya S, Alber F, Williams R, Chait BT, et al. 2004. Components of coated vesicles and nuclear pore complexes share a common molecular architecture. PLoS Biol. 2:e380 Donaldson JG, Cassel D, Kahn RA, Klausner RD. 1992. ADP-ribosylation factor, a small GTP-binding protein, is required for binding of the coatomer protein beta-COP to Golgi membranes. Proc. Natl. Acad. Sci. USA 89:6408–12 Drab M, Verkade P, Elger M, Kasper M, Lohn M, et al. 2001. Loss of caveolae, vascular dysfunction, and pulmonary defects in caveolin-1 gene-disrupted mice. Science 293:2449–52 Dreier L, Rapoport TA. 2000. In vitro formation of the endoplasmic reticulum occurs independently of microtubules by a controlled fusion reaction. J. Cell Biol. 148:883–98 E V I E W S www.annualreviews.org • Mechanisms Shaping the Membranes of Cellular Organelles 14.21 I N E 29 June 2009 A C ARI R Annu. Rev. Cell Dev. Biol. 2009.25. Downloaded from arjournals.annualreviews.org by TEL-AVIV UNIVERSITY on 08/24/09. For personal use only. ANRV389-CB25-14 D V A N ANRV389-CB25-14 ARI 29 June 2009 Drin G, Casella JF, Gautier R, Boehmer T, Schwartz TU, Antonny B. 2007. A general amphipathic alphahelical motif for sensing membrane curvature. Nat. Struct. Mol. Biol. 14:138–46 Efrat A, Chernomordik LV, Kozlov MM. 2007. Point-like protrusion as a prestalk intermediate in membrane fusion pathway. Biophys. J. 92:L61–63 Engqvist-Goldstein AE, Drubin DG. 2003. Actin assembly and endocytosis: from yeast to mammals. Annu. Rev. Cell Dev. Biol. 19:287–332 Epand RM, Sayer BG, Epand RF. 2005. Caveolin scaffolding region and cholesterol-rich domains in membranes. J. Mol. Biol. 345:339–50 Estrada P, Kim J, Coleman J, Walker L, Dunn B, et al. 2003. Myo4p and She3p are required for cortical ER inheritance in Saccharomyces cerevisiae. J. Cell Biol. 163:1255–66 Evans K, Keller C, Pavur K, Glasgow K, Conn B, Lauring B. 2006. Interaction of two hereditary spastic paraplegia gene products, spastin and atlastin, suggests a common pathway for axonal maintenance. Proc. Natl. Acad. Sci. USA 103:10666–71 Fath S, Mancias JD, Bi X, Goldberg J. 2007. Structure and organization of coat proteins in the COPII cage. Cell 129:1325–36 Fawcett DW. 1981. The Cell. Philadelphia: W.B. Saunders Ford MG, Mills IG, Peter BJ, Vallis Y, Praefcke GJ, et al. 2002. Curvature of clathrin-coated pits driven by epsin. Nature 419:361–66 Frederick RL, McCaffery JM, Cunningham KW, Okamoto K, Shaw JM. 2004. Yeast Miro GTPase, Gem1p, regulates mitochondrial morphology via a novel pathway. J. Cell Biol. 167:87–98 Frederick RL, Okamoto K, Shaw JM. 2008. Multiple pathways influence mitochondrial inheritance in budding yeast. Genetics 178:825–37 Frederick RL, Shaw JM. 2007. Moving mitochondria: establishing distribution of an essential organelle. Traffic 8:1668–75 Frey TG, Mannella CA. 2000. The internal structure of mitochondria. Trends Biochem. Sci. 25:319–24 Frost A, Perera R, Roux A, Spasov K, Destaing O, et al. 2008. Structural basis of membrane invagination by F-BAR domains. Cell 132:807–17 Fujimoto T, Kogo H, Ishiguro K, Tauchi K, Nomura R. 2001. Caveolin-2 is targeted to lipid droplets, a new “membrane domain” in the cell. J. Cell Biol. 152:1079–85 Gallop JL, Jao CC, Kent HM, Butler PJ, Evans PR, et al. 2006. Mechanism of endophilin N-BAR domainmediated membrane curvature. EMBO J. 25:2898–910 Glater EE, Megeath LJ, Stowers RS, Schwarz TL. 2006. Axonal transport of mitochondria requires milton to recruit kinesin heavy chain and is light chain independent. J. Cell Biol. 173:545–57 Glenney JR Jr, Soppet D. 1992. Sequence and expression of caveolin, a protein component of caveolae plasma membrane domains phosphorylated on tyrosine in Rous sarcoma virus-transformed fibroblasts. Proc. Natl. Acad. Sci. USA 89:10517–21 Goldberg MW, Wiese C, Allen TD, Wilson KL. 1997. Dimples, pores, star-rings, and thin rings on growing nuclear envelopes: evidence for structural intermediates in nuclear pore complex assembly. J. Cell Sci. 110(Pt 4):409–20 Griparic L, Van Der Bliek AM. 2001. The many shapes of mitochondrial membranes. Traffic 2:235–44 Guo X, Macleod GT, Wellington A, Hu F, Panchumarthi S, et al. 2005. The GTPase dMiro is required for axonal transport of mitochondria to Drosophila synapses. Neuron 47:379–93 Helfrich W. 1973. Elastic properties of lipid bilayers: theory and possible experiments. Z. Naturforsch. 28c:693– 703 Helfrich W. 1990. Elasticity and thermal undulations of fluid films of amphiphiles. In Les Houches, 1988: Liquids and Interfaces, ed. J Charvolin, J-F Joanny, J Zinn-Justin, pp. 212–37. Amsterdam: North-Holland Henne WM, Kent HM, Ford MG, Hegde BG, Daumke O, et al. 2007. Structure and analysis of FCHo2 FBAR domain: a dimerizing and membrane recruitment module that effects membrane curvature. Structure 15:839–52 Herlan M, Vogel F, Bornhovd C, Neupert W, Reichert AS. 2003. Processing of Mgm1 by the rhomboid-type protease Pcp1 is required for maintenance of mitochondrial morphology and of mitochondrial DNA. J. Biol. Chem. 278:27781–88 Annu. Rev. Cell Dev. Biol. 2009.25. Downloaded from arjournals.annualreviews.org by TEL-AVIV UNIVERSITY on 08/24/09. For personal use only. V I E W R E 17:48 S 14.22 C E I N A D V A N Shibata et al. 17:48 Hetzer MW, Walther TC, Mattaj IW. 2005. Pushing the envelope: structure, function, and dynamics of the nuclear periphery. Annu. Rev. Cell Dev. Biol. 21:347–80 Higgins MK, McMahon HT. 2002. Snap-shots of clathrin-mediated endocytosis. Trends Biochem. Sci. 27:257– 63 Hill MM, Bastiani M, Luetterforst R, Kirkham M, Kirkham A, et al. 2008. PTRF-Cavin, a conserved cytoplasmic protein required for caveola formation and function. Cell 132:113–24 Hinshaw JE, Schmid SL. 1995. Dynamin self-assembles into rings suggesting a mechanism for coated vesicle budding. Nature 374:190–92 Hoppins S, Lackner L, Nunnari J. 2007. The machines that divide and fuse mitochondria. Annu. Rev. Biochem. 76:751–80 Hsia KC, Stavropoulos P, Blobel G, Hoelz A. 2007. Architecture of a coat for the nuclear pore membrane. Cell 131:1313–26 Hu J, Shibata Y, Voss C, Shemesh T, Li Z, et al. 2008. Membrane proteins of the endoplasmic reticulum induce high-curvature tubules. Science 319:1247–50 Hu J, Shibata Y, Zhu P-P, Voss C, Rismanchi N, et al. 2009. A class of dynamin-like GTPases involved in the generation of the tubular ER network. Cell In press Ingerman E, Perkins EM, Marino M, Mears JA, McCaffery JM, et al. 2005. Dnm1 forms spirals that are structurally tailored to fit mitochondria. J. Cell Biol. 170:1021–27 Ishihara N, Fujita Y, Oka T, Mihara K. 2006. Regulation of mitochondrial morphology through proteolytic cleavage of OPA1. EMBO J. 25:2966–77 Itoh T, Erdmann KS, Roux A, Habermann B, Werner H, De Camilli P. 2005. Dynamin and the actin cytoskeleton cooperatively regulate plasma membrane invagination by BAR and F-BAR proteins. Dev. Cell 9:791–804 Karbowski M, Jeong SY, Youle RJ. 2004. Endophilin B1 is required for the maintenance of mitochondrial morphology. J. Cell Biol. 166:1027–39 King MC, Drivas TG, Blobel G. 2008. A network of nuclear envelope membrane proteins linking centromeres to microtubules. Cell 134:427–38 Kirchhausen T. 2000. Clathrin. Annu. Rev. Biochem. 69:699–727 Klopfenstein DR, Klumperman J, Lustig A, Kammerer RA, Oorschot V, Hauri HP. 2001. Subdomain-specific localization of CLIMP-63 (p63) in the endoplasmic reticulum is mediated by its luminal alpha-helical segment. J. Cell Biol. 153:1287–300 Kosaka T, Ikeda K. 1983. Possible temperature-dependent blockage of synaptic vesicle recycling induced by a single gene mutation in Drosophila. J. Neurobiol. 14:207–25 Koshiba T, Detmer SA, Kaiser JT, Chen H, McCaffery JM, Chan DC. 2004. Structural basis of mitochondrial tethering by mitofusin complexes. Science 305:858–62 Kozlov MM, Chernomordik LV. 1998. A mechanism of protein-mediated fusion: coupling between refolding of the influenza hemagglutinin and lipid rearrangements. Biophys. J. 75:1384–96 Kozlov MM, Chernomordik LV. 2002. The protein coat in membrane fusion: lessons from fission. Traffic 3:256–67 Kozlovsky Y, Kozlov MM. 2003. Membrane fission: model for intermediate structures. Biophys. J. 85:85–96 Krauss M, Jia JY, Roux A, Beck R, Wieland FT, et al. 2008. Arf1-GTP-induced tubule formation suggests a function of Arf family proteins in curvature acquisition at sites of vesicle budding. J. Biol. Chem. 283:27717– 23 Le Lan C, Neumann JM, Jamin N. 2006. Role of the membrane interface on the conformation of the caveolin scaffolding domain: a CD and NMR study. FEBS Lett. 580:5301–5 Lee C, Chen LB. 1988. Dynamic behavior of endoplasmic reticulum in living cells. Cell 54:37–46 Lee MC, Orci L, Hamamoto S, Futai E, Ravazzola M, Schekman R. 2005. Sar1p N-terminal helix initiates membrane curvature and completes the fission of a COPII vesicle. Cell 122:605–17 Li J, Mao X, Dong LQ, Liu F, Tong L. 2007. Crystal structures of the BAR-PH and PTB domains of human APPL1. Structure 15:525–33 Liu L, Brown D, McKee M, Lebrasseur NK, Yang D, et al. 2008. Deletion of Cavin/PTRF causes global loss of caveolae, dyslipidemia, and glucose intolerance. Cell Metab. 8:310–17 E V I E W S www.annualreviews.org • Mechanisms Shaping the Membranes of Cellular Organelles 14.23 I N E 29 June 2009 A C ARI R Annu. Rev. Cell Dev. Biol. 2009.25. Downloaded from arjournals.annualreviews.org by TEL-AVIV UNIVERSITY on 08/24/09. For personal use only. ANRV389-CB25-14 D V A N ANRV389-CB25-14 ARI 29 June 2009 Liu L, Pilch PF. 2008. A critical role of cavin (polymerase I and transcript release factor) in caveolae formation and organization. J. Biol. Chem. 283:4314–22 Low HH, Lowe J. 2006. A bacterial dynamin-like protein. Nature 444:766–69 Lundmark R, Doherty GJ, Vallis Y, Peter BJ, McMahon HT. 2008. Arf family GTP loading is activated by, and generates, positive membrane curvature. Biochem. J. 414:189–94 Mannan AU, Boehm J, Sauter SM, Rauber A, Byrne PC, et al. 2006. Spastin, the most commonly mutated protein in hereditary spastic paraplegia interacts with Reticulon 1 an endoplasmic reticulum protein. Neurogenetics 7:93–103 Mannella CA. 2000. Introduction: our changing views of mitochondria. J. Bioenerg. Biomembr. 32:1–4 Mannella CA. 2006. Structure and dynamics of the mitochondrial inner membrane cristae. Biochim. Biophys. Acta 1763:542–48 Martens S, Kozlov MM, McMahon HT. 2007. How synaptotagmin promotes membrane fusion. Science 316:1205–8 Masuda M, Takeda S, Sone M, Ohki T, Mori H, et al. 2006. Endophilin BAR domain drives membrane curvature by two newly identified structure-based mechanisms. EMBO J. 25:2889–97 McMahon HT, Gallop JL. 2005. Membrane curvature and mechanisms of dynamic cell membrane remodelling. Nature 438:590–96 Meeusen S, DeVay R, Block J, Cassidy-Stone A, Wayson S, et al. 2006. Mitochondrial inner-membrane fusion and crista maintenance requires the dynamin-related GTPase Mgm1. Cell 127:383–95 Meisinger C, Pfannschmidt S, Rissler M, Milenkovic D, Becker T, et al. 2007. The morphology proteins Mdm12/Mmm1 function in the major beta-barrel assembly pathway of mitochondria. EMBO J. 26:2229– 39 Meisinger C, Rissler M, Chacinska A, Szklarz LK, Milenkovic D, et al. 2004. The mitochondrial morphology protein Mdm10 functions in assembly of the preprotein translocase of the outer membrane. Dev. Cell 7:61–71 Merkwirth C, Dargazanli S, Tatsuta T, Geimer S, Lower B, et al. 2008. Prohibitins control cell proliferation and apoptosis by regulating OPA1-dependent cristae morphogenesis in mitochondria. Genes Dev. 22:476– 88 Minauro-Sanmiguel F, Wilkens S, Garcia JJ. 2005. Structure of dimeric mitochondrial ATP synthase: novel F0 bridging features and the structural basis of mitochondrial cristae biogenesis. Proc. Natl. Acad. Sci. USA 102:12356–58 Oertle T, Schwab ME. 2003. Nogo and its paRTNers. Trends Cell Biol. 13:187–94 Okamoto K, Shaw JM. 2005. Mitochondrial morphology and dynamics in yeast and multicellular eukaryotes. Annu. Rev. Genet. 39:503–36 Ostermeyer AG, Paci JM, Zeng Y, Lublin DM, Munro S, Brown DA. 2001. Accumulation of caveolin in the endoplasmic reticulum redirects the protein to lipid storage droplets. J. Cell Biol. 152:1071–78 Padmakumar VC, Libotte T, Lu W, Zaim H, Abraham S, et al. 2005. The inner nuclear membrane protein Sun1 mediates the anchorage of Nesprin-2 to the nuclear envelope. J. Cell Sci. 118:3419–30 Parton RG, Hanzal-Bayer M, Hancock JF. 2006. Biogenesis of caveolae: a structural model for caveolininduced domain formation. J. Cell Sci. 119:787–96 Parton RG, Simons K. 2007. The multiple faces of caveolae. Nat. Rev. Mol. Cell Biol. 8:185–94 Paumard P, Vaillier J, Coulary B, Schaeffer J, Soubannier V, et al. 2002. The ATP synthase is involved in generating mitochondrial cristae morphology. EMBO J. 21:221–30 Pelkmans L, Zerial M. 2005. Kinase-regulated quantal assemblies and kiss-and-run recycling of caveolae. Nature 436:128–33 Peter BJ, Kent HM, Mills IG, Vallis Y, Butler PJ, et al. 2004. BAR domains as sensors of membrane curvature: the amphiphysin BAR structure. Science 303:495–99 Pol A, Luetterforst R, Lindsay M, Heino S, Ikonen E, Parton RG. 2001. A caveolin dominant negative mutant associates with lipid bodies and induces intracellular cholesterol imbalance. J. Cell Biol. 152:1057–70 Praefcke GJ, McMahon HT. 2004. The dynamin superfamily: universal membrane tubulation and fission molecules? Nat. Rev. Mol. Cell Biol. 5:133–47 Prinz WA, Grzyb L, Veenhuis M, Kahana JA, Silver PA, Rapoport TA. 2000. Mutants affecting the structure of the cortical endoplasmic reticulum in Saccharomyces cerevisiae. J. Cell Biol. 150:461–74 Annu. Rev. Cell Dev. Biol. 2009.25. Downloaded from arjournals.annualreviews.org by TEL-AVIV UNIVERSITY on 08/24/09. For personal use only. V I E W R E 17:48 S 14.24 C E I N A D V A N Shibata et al. 17:48 Pucadyil TJ, Schmid SL. 2008. Real-time visualization of dynamin-catalyzed membrane fission and vesicle release. Cell 135:1263–75 Pylypenko O, Lundmark R, Rasmuson E, Carlsson SR, Rak A. 2007. The PX-BAR membrane-remodeling unit of sorting nexin 9. EMBO J. 26:4788–800 Rabl R, Soubannier V, Scholz R, Vogel F, Mendl N, Vasiljev-Neumeyer A, et al. 2009. Formation of cristae and crista junctions in mitochondria depends on antagonism between Fcj1 and Su e/g. J. Cell Biol. 185:1047–63 Ramachandran R, Schmid SL. 2008. Real-time detection reveals that effectors couple dynamin’s GTPdependent conformational changes to the membrane. EMBO J. 27:27–37 Rothberg KG, Heuser JE, Donzell WC, Ying YS, Glenney JR, Anderson RG. 1992. Caveolin, a protein component of caveolae membrane coats. Cell 68:673–82 Saarikangas J, Zhao H, Pykalainen A, Laurinmaki P, Mattila PK, et al. 2009. Molecular mechanisms of membrane deformation by I-BAR domain proteins. Curr. Biol. 19:95–107 Salinas S, Proukakis C, Crosby A, Warner TT. 2008. Hereditary spastic paraplegia: clinical features and pathogenetic mechanisms. Lancet Neurol. 7:1127–38 Sanderson CM, Connell JW, Edwards TL, Bright NA, Duley S, et al. 2006. Spastin and atlastin, two proteins mutated in autosomal-dominant hereditary spastic paraplegia, are binding partners. Hum. Mol. Genet. 15:307–18 Sargiacomo M, Scherer PE, Tang Z, Kubler E, Song KS, et al. 1995. Oligomeric structure of caveolin: implications for caveolae membrane organization. Proc. Natl. Acad. Sci. USA 92:9407–11 Schlegel A, Lisanti MP. 2000. A molecular dissection of caveolin-1 membrane attachment and oligomerization. Two separate regions of the caveolin-1 C-terminal domain mediate membrane binding and oligomer/oligomer interactions in vivo. J. Biol. Chem. 275:21605–17 Senda T, Yoshinaga-Hirabayashi T. 1998. Intermembrane bridges within membrane organelles revealed by quick-freeze deep-etch electron microscopy. Anat. Rec. 251:339–45 Sesaki H, Jensen RE. 2004. Ugo1p links the Fzo1p and Mgm1p GTPases for mitochondrial fusion. J. Biol. Chem. 279:28298–303 Sheetz MP, Singer SJ. 1974. Biological membranes as bilayer couples. A molecular mechanism of drugerythrocyte interactions. Proc. Natl. Acad. Sci. USA 71:4457–61 Shibata Y, Voeltz GK, Rapoport TA. 2006. Rough sheets and smooth tubules. Cell 126:435–39 Shibata Y, Voss C, Rist JM, Hu J, Rapoport TA, et al. 2008. The reticulon and DP1/Yop1p proteins form immobile oligomers in the tubular endoplasmic reticulum. J. Biol. Chem. 283:18892–904 Shimada A, Niwa H, Tsujita K, Suetsugu S, Nitta K, et al. 2007. Curved EFC/F-BAR-domain dimers are joined end to end into a filament for membrane invagination in endocytosis. Cell 129:761–72 Short B, Haas A, Barr FA. 2005. Golgins and GTPases, giving identity and structure to the Golgi apparatus. Biochim. Biophys. Acta 1744:383–95 Simons K, Toomre D. 2000. Lipid rafts and signal transduction. Nat. Rev. Mol. Cell Biol. 1:31–39 Song KS, Tang Z, Li S, Lisanti MP. 1997. Mutational analysis of the properties of caveolin-1. A novel role for the C-terminal domain in mediating homo-typic caveolin-caveolin interactions. J. Biol. Chem. 272:4398– 403 Stagg SM, LaPointe P, Razvi A, Gurkan C, Potter CS, et al. 2008. Structural basis for cargo regulation of COPII coat assembly. Cell 134:474–84 Stan RV. 2002. Structure and function of endothelial caveolae. Microsc. Res. Tech. 57:350–64 Stan RV. 2005. Structure of caveolae. Biochim. Biophys. Acta 1746:334–48 Stowell MH, Marks B, Wigge P, McMahon HT. 1999. Nucleotide-dependent conformational changes in dynamin: evidence for a mechanochemical molecular spring. Nat. Cell Biol. 1:27–32 Strauss M, Hofhaus G, Schroder RR, Kuhlbrandt W. 2008. Dimer ribbons of ATP synthase shape the inner mitochondrial membrane. EMBO J. 27:1154–60 Sudhof TC, Rothman JE. 2009. Membrane fusion: grappling with SNARE and SM proteins. Science 323: 474–77 Takei K, McPherson PS, Schmid SL, De Camilli P. 1995. Tubular membrane invaginations coated by dynamin rings are induced by GTP-gamma S in nerve terminals. Nature 374:186–90 Tatsuta T, Model K, Langer T. 2005. Formation of membrane-bound ring complexes by prohibitins in mitochondria. Mol. Biol. Cell 16:248–59 E V I E W S www.annualreviews.org • Mechanisms Shaping the Membranes of Cellular Organelles 14.25 I N E 29 June 2009 A C ARI R Annu. Rev. Cell Dev. Biol. 2009.25. Downloaded from arjournals.annualreviews.org by TEL-AVIV UNIVERSITY on 08/24/09. For personal use only. ANRV389-CB25-14 D V A N ANRV389-CB25-14 ARI 29 June 2009 17:48 Annu. Rev. Cell Dev. Biol. 2009.25. Downloaded from arjournals.annualreviews.org by TEL-AVIV UNIVERSITY on 08/24/09. For personal use only. Terasaki M, Chen LB, Fujiwara K. 1986. Microtubules and the endoplasmic reticulum are highly interdependent structures. J. Cell Biol. 103:1557–68 Tsujita K, Suetsugu S, Sasaki N, Furutani M, Oikawa T, Takenawa T. 2006. Coordination between the actin cytoskeleton and membrane deformation by a novel membrane tubulation domain of PCH proteins is involved in endocytosis. J. Cell Biol. 172:269–79 van Meer G, Voelker DR, Feigenson GW. 2008. Membrane lipids: where they are and how they behave. Nat. Rev. Mol. Cell Biol. 9:112–24 Voeltz GK, Prinz WA, Shibata Y, Rist JM, Rapoport TA. 2006. A class of membrane proteins shaping the tubular endoplasmic reticulum. Cell 124:573–86 Voeltz GK, Rolls MM, Rapoport TA. 2002. Structural organization of the endoplasmic reticulum. EMBO Rep. 3:944–50 Wang X, Schwarz TL. 2009. The mechanism of Ca2+ -dependent regulation of kinesin-mediated mitochondrial motility. Cell 136:163–74 Waterman-Storer CM, Salmon ED. 1998. Endoplasmic reticulum membrane tubules are distributed by microtubules in living cells using three distinct mechanisms. Curr. Biol. 8:798–806 Weissenhorn W. 2005. Crystal structure of the endophilin-A1 BAR domain. J. Mol. Biol. 351:653–61 Worman HJ, Gundersen GG. 2006. Here come the SUNs: a nucleocytoskeletal missing link. Trends Cell Biol. 16:67–69 Zhao L, Helms JB, Brugger B, Harter C, Martoglio B, et al. 1997. Direct and GTP-dependent interaction of ADP ribosylation factor 1 with coatomer subunit beta. Proc. Natl. Acad. Sci. USA 94:4418–23 Zhu G, Chen J, Liu J, Brunzelle JS, Huang B, et al. 2007. Structure of the APPL1 BAR-PH domain and characterization of its interaction with Rab5. EMBO J. 26:3484–93 Zimmerberg J, Kozlov MM. 2006. How proteins produce cellular membrane curvature. Nat. Rev. Mol. Cell Biol. 7:9–19 V I E W R E S 14.26 C E I N A D V A N Shibata et al.