VESICULAR TRAFFIC, SECRETION, AND ENDOCYTOSISkbp-srmc.yolasite.com/resources/Chapter 17.pdf · 2010-09-24 · green fluorescent protein (GFP), a naturally fluorescent pro-tein (Chapter

proteins physically moves from the cis position (nearest theER) to the trans position (farthest from the ER), successivelybecoming first a medial-Golgi cisterna and then a trans-Golgicisterna. This process, known as cisternal progression, doesnot involve the budding off and fusion of anterograde trans-port vesicles. During cisternal progression, enzymes andother Golgi-resident proteins are constantly being retrievedfrom later to earlier Golgi cisternae by retrograde transportvesicles, thereby remaining localized to the cis-, medial-, ortrans-Golgi cisternae.

Proteins in the secretory pathway that are destined forcompartments other than the ER or Golgi eventually reacha complex network of membranes and vesicles termed thetrans-Golgi network (TGN). From this major branch pointin the secretory pathway, a protein can be loaded into one

17

Electron micrograph of clathrin cages, like those that surround

clathrin-coated transport vesicles, formed by the in vitro

polymerization of clathrin heavy and light chains. [John Heuser,Washington University School of Medicine.]

VESICULAR TRAFFIC,SECRETION, AND ENDOCYTOSIS

In the previous chapter we explored how proteins are tar-geted to and translocated across the membranes of dif-ferent intracellular organelles. In this chapter we turn our

attention to the mechanisms that allow soluble and mem-brane proteins synthesized on the rough endoplasmic reticu-lum (ER) to move to their final destinations via the secretorypathway. A single unifying principle governs all protein traf-ficking in the secretory pathway: transport of membrane andsoluble proteins from one membrane-bounded compartmentto another is mediated by transport vesicles that collect“cargo” proteins in buds arising from the membrane of onecompartment and then deliver these cargo proteins to thenext compartment by fusing with the membrane of that com-partment. Importantly, as transport vesicles bud from onemembrane and fuse with the next, the same face of the mem-brane remains oriented toward the cytosol. Therefore oncea protein has been inserted into the membrane or the lumenof the ER, the protein can be carried along the secretorypathway, moving from one organelle to the next withoutbeing translocated across another membrane or altering itsorientation within the membrane.

Figure 17-1 outlines the major routes for protein traf-ficking in the secretory pathway. Once newly synthesizedproteins are incorporated into the ER lumen or membrane asdiscussed in Chapter 16, they can be packaged into antero-grade (forward-moving) transport vesicles. These vesicles fusewith each other to form a flattened membrane-bounded com-partment known as the cis-Golgi cisterna. Certain proteins,mainly ER-localized proteins, are retrieved from the cis-Golgito the ER via a different set of retrograde (backward-moving)transport vesicles. A new cis-Golgi cisterna with its cargo of

701

O U T L I N E

17.1 Techniques for Studying the SecretoryPathway

17.2 Molecular Mechanisms of Vesicular Traffic

17.3 Vesicle Trafficking in the Early Stages of the Secretory Pathway

17.4 Protein Sorting and Processing in Later Stagesof the Secretory Pathway

17.5 Receptor-Mediated Endocytosis and the Sorting of Internalized Proteins

17.6 Synaptic Vesicle Function and Formation

702 CHAPTER 17 • Vesicular Traffic, Secretion, and Endocytosis

Constitutivesecretion

Plasma membrane

Medial-Golgi

Cytosol

Exterior

Trans-Golgi

Trans-Golgi

network

Cis-Golgi

Cis-Golginetwork

Rough ER

Regulated secretion

Transportvesicle

Secretoryvesicle

Sorting tolysosomes

Budding and fusion of ER-to-Golgi vesicles to form cis-Golgi

Retrograde Golgi-to-ERtransport

Retrogradetransport fromlater to earlierGolgi cisternae

Protein synthesis on bound ribosomes;cotranslational transport of proteinsinto or across ER membrane

Lysosome

ER lumen

Endocytosis

Endocytic vesicle

Late endosome

Cisternalprogression

5

6

7

8

9

1

2 3

4

� FIGURE 17-1 Overview of

the secretory and endocytic

pathways of protein sorting.

Secretory pathway: Synthesis ofproteins bearing an ER signalsequence is completed on therough ER ( ), and the newly madepolypeptide chains are inserted intothe ER membrane or cross it intothe lumen (Chapter 16). Someproteins (e.g., ER enzymes orstructural proteins) remain withinthe ER. The remainder arepackaged into transport vesicles ( )that bud from the ER and fusetogether to form new cis-Golgicisternae. Missorted ER-residentproteins and vesicle membraneproteins that need to be reused areretrieved to the ER by vesicles ( )that bud from the cis-Golgi andfuse with the ER. Each cis-Golgicisterna, with its protein content,physically moves from the cis tothe trans face of the Golgi complex ( ) by a nonvesicularprocess called cisternalprogression. Retrograde transportvesicles ( ) move Golgi-residentproteins to the proper Golgicompartment. In all cells, certainsoluble proteins move to the cellsurface in transport vesicles ( )and are secreted continuously(constitutive secretion). In certaincell types, some soluble proteinsare stored in secretory vesicles ( )and are released only after the cellreceives an appropriate neural orhormonal signal (regulatedsecretion). Lysosome-destinedmembrane and soluble proteins,which are transported in vesiclesthat bud from the trans-Golgi ( ),first move to the late endosomeand then to the lysosome.Endocytic pathway: Membrane andsoluble extracellular proteins takenup in vesicles that bud from theplasma membrane ( ) also canmove to the lysosome via theendosome.

9

8

7

6

5

4

3

2

1

of at least three different kinds of vesicles. After buddingfrom the trans-Golgi network, the first type of vesicle imme-diately moves to and fuses with the plasma membrane, re-leasing its contents by exocytosis. In all cell types, at leastsome proteins are loaded into such vesicles and secreted con-tinuously in this manner. Examples of proteins released bysuch constitutive (or continuous) secretion include collagenby fibroblasts, serum proteins by hepatocytes, and antibod-ies by activated B lymphocytes. The second type of vesicleto bud from the trans-Golgi network, known as secretoryvesicles, are stored inside the cell until a signal for exocyto-sis causes release of their contents at the plasma membrane.Among the proteins released by such regulated secretion arepeptide hormones (e.g., insulin, glucagon, ACTH) from var-ious endocrine cells, precursors of digestive enzymes frompancreatic acinar cells, milk proteins from the mammarygland, and neurotransmitters from neurons.

The third type of vesicle that buds from the trans-Golginetwork is directed to the lysosome, an organelle responsiblefor the intracellular degradation of macromolecules, and tolysosome-like storage organelles in certain cells. Secretoryproteins destined for lysosomes first are transported by vesi-cles from the trans-Golgi network to a compartment usuallycalled the late endosome; proteins then are transferred to thelysosome by a mechanism that is not well understood butmay involve direct fusion of the endosome with the lysoso-mal membrane. Soluble proteins delivered by this pathwayinclude lysosomal digestive enzymes (e.g., proteases, glycosi-dases, and phosphatases) and membrane proteins (e.g., V-class proton pump) that pump H� from the cytosol intothe acidic lumen of the endosome and lysosome. As we willsee, some of the specific protein-processing and -sortingevents that take place within these organelles depend on theirlow luminal pH.

The endosome also functions in the endocytic pathway inwhich vesicles bud from the plasma membrane bringingmembrane proteins and their bound ligands into the cell (seeFigure 17-1). After being internalized by endocytosis, someproteins are transported to lysosomes, while others are re-cycled back to the cell surface. Endocytosis is a way for cellsto take up nutrients that are in macromolecular form—forexample, cholesterol in the form of lipoprotein particles andiron complexed with the serum protein transferrin. Endocy-tosis also can function as a regulatory mechanism to decreasesignaling activity by withdrawing receptors for a particularsignaling molecule from the cell surface.

Techniques for Studying the Secretory PathwayThe key to understanding how proteins are transportedthrough the organelles of the secretory pathway has been todevelop a basic description of the function of transport vesi-cles. Many components required for the formation and fu-

17.1

sion of transport vesicles have been identified in the pastdecade by a remarkable convergence of the genetic and bio-chemical approaches described in this section. All studies ofintracellular protein trafficking employ some method for as-saying the transport of a given protein from one compart-ment to another. We begin by describing how intracellularprotein transport can be followed in living cells and thenconsider genetic and in vitro systems that have proved use-ful in elucidating the secretory pathway.

Transport of a Protein Through the SecretoryPathway Can Be Assayed in Living CellsThe classic studies of G. Palade and his colleagues in the 1960sfirst established the order in which proteins move from or-ganelle to organelle in the secretory pathway. These early stud-ies also showed that secretory proteins were never releasedinto the cytosol, the first indication that transported proteinsare associated with some type of membrane-bounded inter-mediate. In these experiments, which combined pulse-chaselabeling (see Figure 3-36) and autoradiography, radioactivelylabeled amino acids were injected into the pancreas of a ham-ster. At different times after injection, the animal was sacrificedand the pancreatic cells were chemically fixed, sectioned, andsubjected to autoradiography to visualize the location of theradiolabeled proteins. Because the radioactive amino acidswere administered in a short pulse, only those proteins syn-thesized immediately after injection were labeled, forming adistinct group, or cohort, of labeled proteins whose transportcould be followed. In addition, because pancreatic acinar cellsare dedicated secretory cells, almost all of the labeled aminoacids in these cells are incorporated into secretory proteins,facilitating the observation of transported proteins.

Although autoradiography is rarely used today to local-ize proteins within cells, these early experiments illustrate thetwo basic requirements for any assay of intercompartmentaltransport. First, it is necessary to label a cohort of proteins inan early compartment so that their subsequent transfer tolater compartments can be followed with time. Second, it isnecessary to have a way to identify the compartment inwhich a labeled protein resides. Here we describe two mod-ern experimental procedures for observing the intracellulartrafficking of a secretory protein in almost any type of cell.

In both procedures, a gene encoding an abundant mem-brane glycoprotein (G protein) from vesicular stomatitisvirus (VSV) is introduced into cultured mammalian cells ei-ther by transfection or simply by infecting the cells with thevirus. The treated cells, even those that are not specialized forsecretion, rapidly synthesize the VSV G protein on the ERlike normal cellular secretory proteins. Use of a mutant encoding a temperature-sensitive VSV G protein allows re-searchers to turn subsequent protein transport on and off. Atthe restrictive temperature of 40 ̊ C, newly made VSV G pro-tein is misfolded and therefore retained within the ER byquality control mechanisms discussed in Chapter 16, whereasat the permissive temperature of 32 ˚C, the accumulated

17.1 • Techniques for Studying the Secretory Pathway 703

protein is correctly folded and is transported through the se-cretory pathway to the cell surface. This clever use of a tem-perature-sensitive mutation in effect defines a protein cohortwhose subsequent transport can be followed.

In two variations of this basic procedure, transport ofVSV G protein is monitored by different techniques. Studiesusing both of these modern trafficking assays and Palade’searly experiments all came to the same conclusion: in mam-malian cells vesicle-mediated transport of a protein moleculefrom its site of synthesis on the rough ER to its arrival at theplasma membrane takes from 30 to 60 minutes.

Microscopy of GFP-Labeled VSV G Protein One approachfor observing transport of VSV G protein employs a hybridgene in which the viral gene is fused to the gene encodinggreen fluorescent protein (GFP), a naturally fluorescent pro-tein (Chapter 5). The hybrid gene is transfected into culturedcells by techniques described in Chapter 9. When cells ex-pressing the temperature-sensitive form of the hybrid protein(VSVG-GFP) are grown at the restrictive temperature,VSVG-GFP accumulates in the ER, which appears as a lacynetwork of membranes when cells are observed in a fluores-cent microscope. When the cells are subsequently shifted to apermissive temperature, the VSVG-GFP can be seen to movefirst to the membranes of the Golgi apparatus, which aredensely concentrated at the edge of the nucleus, and then tothe cell surface (Figure 17-2a). By analyzing the distributionof VSVG-GFP at different times after shifting cells to the per-missive temperature, researchers have determined how longVSVG-GFP resides in each organelle of the secretory path-way (Figure 17-2b).

Detection of Compartment-Specific Oligosaccharide Modifi-cations A second way to follow the transport of secretoryproteins takes advantage of modifications to their carbohy-drate side chains that occur at different stages of the secretorypathway. To understand this approach, recall that many se-cretory proteins leaving the ER contain one or more copies ofthe N-linked oligosaccharide Man8(GlcNAc)2, which are syn-thesized and attached to secretory proteins in the ER (see Fig-ure 16-18). As a protein moves through the Golgi complex,different enzymes localized to the cis-, medial-, and trans-Golgi cisternae catalyze an ordered series of reactions to thesecore Man8(GlcNAc)2 chains. For instance, glycosidases thatreside specifically in the cis-Golgi compartment sequentiallytrim mannose residues off of the core oligosaccharide to yielda “trimmed” form Man5(GlcNAc)2 (Figure 17-3, reaction 1 ).Scientists can use a specialized carbohydrate-cleaving enzymeknown as endoglycosidase D to distinguish glycosylated pro-teins that remain in the ER from those that have entered thecis-Golgi: trimmed cis-Golgi–specific oligosaccharides arecleaved from proteins by endoglycosidase D, whereas the core(untrimmed) oligosaccharide chains on secretory proteinswithin the ER are resistant to cleavage by this enzyme. Becausea deglycosylated protein produced by endoglycosidase D digestion moves faster on an SDS gel than the correspondingglycosylated protein, they can be readily distinguished.

This type of assay can be used to track movement of VSVG protein in virus-infected cells pulse-labeled with radioac-tive amino acids. Immediately after labeling, all the extractedlabeled VSV G protein is still in the ER and is resistant to di-gestion by endoglycosidase D, but with time an increasingfraction of the glycoprotein becomes sensitive to digestion


600500400300Time (min)

2001000

20

15

10

5

0

VS

VG

–GFP

(×

106 )

TotalER

PM

Golgi

ER Plasmamembrane

Golgi

(a) (b)0 min 40 min 180 min

▲ EXPERIMENTAL FIGURE 17-2 Protein

transport through the secretory pathway can be

visualized by fluorescence microscopy of cells

producing a GFP-tagged membrane protein. Culturedcells were transfected with a hybrid gene encoding theviral membrane glycoprotein VSV G protein linked to thegene for green fluorescent protein (GFP). A mutantversion of the viral gene was used so that newly madehybrid protein (VSVG-GFP) is retained in the ER at 40 �Cbut is released for transport at 32 �C. (a) Fluorescencemicrographs of cells just before and at two times after

they were shifted to the lower temperature. Movement of VSVG-GFP from the ER to the Golgi and finally to the cell surfaceoccurred within 180 minutes. (b) Plot of the levels of VSVG-GFPin the endoplasmic reticulum (ER), Golgi, and plasma membrane(PM) at different times after shift to lower temperature. Thekinetics of transport from one organelle to another can bereconstructed from computer analysis of these data. Thedecrease in total fluorescence that occurs at later times probablyresults from slow inactivation of GFP fluorescence. [From JenniferLippincott-Schwartz and Koret Hirschberg, Metabolism Branch, NationalInstitute of Child Health and Human Development.]

ME

DIA

C

ON

NE

CT

IO

NS

Vid

eo:T

rans

port

of

VSV

G-G

FP T

hrou

gh t

he

Secr

etor

y Pa

thw

ay

(Figure 17-4). This conversion of VSV G protein from an en-doglycosidase D–resistant form to an endoglycosidaseD–sensitive form corresponds to vesicular transport of theprotein from the ER to the cis-Golgi. Note that transport ofVSV G protein from the ER to the Golgi takes about 30 min-utes as measured by either the assay based on oligosaccha-ride processing or fluorescence microscopy of VSVG-GFP.

Yeast Mutants Define Major Stages and ManyComponents in Vesicular TransportThe general organization of the secretory pathway and manyof the molecular components required for vesicle traffickingare similar in all eukaryotic cells. Because of this conserva-tion, genetic studies with yeast have been useful in confirmingthe sequence of steps in the secretory pathway and in identi-fying many of the proteins that participate in vesicular traffic.Although yeasts secrete few proteins into the growth medium,they continuously secrete a number of enzymes that remain

17.1 • Techniques for Studying the Secretory Pathway 705

Cis

Golgi

(Man)8(GlcNAc)2(Man)5(GlcNAc)2

Medial

(Man)5(GlcNAc)2

(GlcNAc)(Man)5(GlcNAc)2

UDPUDP

GDP

Transport vesiclefrom ER

Trans UDP

Exit

= N-Acetylglucosamine= Mannose = Galactose

= Fucose = N-Acetylneuraminic acid

CMP

6 7

2 3 4 5

1

▲ FIGURE 17-3 Processing of N-linked oligosaccharide

chains on glycoproteins within cis-, medial-, and trans-Golgi

cisternae in vertebrate cells. The enzymes catalyzing each stepare localized to the indicated compartments. After removal ofthree mannose residues in the cis-Golgi (step ), the proteinmoves by cisternal progression to the medial-Golgi. Here, threeGlcNAc residues are added (steps and ), two more mannoseresidues are removed (step ), and a single fucose is added(step ). Processing is completed in the trans-Golgi by additionof three galactose residues (step ) and finally by linkage of anN-acetylneuraminic acid residue to each of the galactose residues(step ). Specific transferase enzymes add sugars to theoligosaccharide, one at a time, from sugar nucleotide precursorsimported from the cytosol. This pathway represents the Golgiprocessing events for a typical mammalian glycoprotein. Variationsin the structure of N-linked oligosaccharides can result fromdifferences in processing steps in the Golgi. [See R. Kornfeld and S. Kornfeld, 1985, Ann. Rev. Biochem. 45:631.]

7

65

342

1

60504030

Time (min)

200

0.2

0.4

0.6

Frac

tio

n o

f to

tal G

pro

tein

sen

siti

ve t

o e

nd

og

lyco

sid

ase

D

0.8

1.0

10

40 °C

32 °C

Resistant

Sensitive

Time at 32 °C (min) 0 5 10 15 20 30 45 60

(b)

(a)

▲ EXPERIMENTAL FIGURE 17-4 Transport of a membrane

glycoprotein from the ER to the Golgi can be assayed based

on sensititivity to cleavage by endoglycosidase D. Cellsexpressing a temperature-sensitive VSV G protein (VSVG) werelabeled with a pulse of radioactive amino acids at thenonpermissive temperature so that labeled protein was retainedin the ER. At periodic times after a return to the permissivetemperature of 32 �C, VSVG was extracted from cells anddigested with endoglycosidase D, which cleaves theoligosaccharide chains from proteins processed in the cis-Golgibut not from proteins in the ER. (a) SDS gel electrophoresis ofthe digestion mixtures resolves the resistant, uncleaved (slowermigrating) and sensitive, cleaved (faster migrating) forms oflabeled VSVG. As this electrophoretogram shows, initially all ofthe VSVG was resistant to digestion, but with time an increasingfraction is sensitive to digestion, reflecting protein transportedfrom the ER to the Golgi and processed there. In control cellskept at 40 �C, only slow-moving, digestion-resistant VSVG wasdetected after 60 minutes (not shown). (b) Plot of the proportionof VSVG that is sensitive to digestion, derived fromelectrophoretic data, reveals the time course of ER → Golgitransport. [From C. J. Beckers et al., 1987, Cell 50:523.]

localized in the narrow space between the plasma membraneand the cell wall. The best-studied of these, invertase, hydro-lyzes the disaccharide sucrose to glucose and fructose.

A large number of yeast mutants initially were identifiedbased on their ability to secrete proteins at one temperatureand inability to do so at a higher, nonpermissive temperature.When these temperature-sensitive secretion (sec) mutants aretransferred from the lower to the higher temperature, theyaccumulate secreted proteins at the point in the pathwayblocked by the mutation. Analysis of such mutants identifiedfive classes (A–E) characterized by protein accumulation inthe cytosol, rough ER, small vesicles taking proteins from theER to the Golgi complex, Golgi cisternae, or constitutive se-cretory vesicles (Figure 17-5). Subsequent characterization ofsec mutants in the various classes has helped elucidate thefundamental components and molecular mechanisms of vesi-cle trafficking that we discuss in later sections.

To determine the order of the steps in the pathway, re-searchers analyzed double sec mutants. For instance, whenyeast cells contain mutations in both class B and class D func-tions, proteins accumulate in the rough ER, not in the Golgicisternae. Since proteins accumulate at the earliest blockedstep, this finding shows that class B mutations must act at anearlier point in the secretory pathway than class D mutationsdo. These studies confirmed that as a secreted protein is syn-thesized and processed it moves sequentially from the cytosol→ rough ER → ER-to-Golgi transport vesicles → Golgi cis-ternae → secretory vesicles and finally is exocytosed.

Cell-free Transport Assays Allow Dissection of Individual Steps in Vesicular TransportIn vitro assays for intercompartmental transport are power-ful complementary approaches to studies with yeast sec mu-

tants for identifying and analyzing the cellular componentsresponsible for vesicular trafficking. In one application ofthis approach, cultured mutant cells lacking one of the en-zymes that modify N-linked oligosaccharide chains in theGolgi are infected with vesicular stomatitis virus (VSV). Forexample, if infected cells lack N-acetylglucosamine trans-ferase I, they produce abundant amounts of VSV G proteinbut cannot add N-acetylglucosamine residues to theoligosaccharide chains in the medial-Golgi as wild-type cellsdo (Figure 17-6a). When Golgi membranes isolated fromsuch mutant cells are mixed with Golgi membranes fromwild-type, uninfected cells, the addition of N-acetylglu-cosamine to VSV G protein is restored (Figure 17-6b). Thismodification is the consequence of the retrograde vesiculartransport of N-acetylglucosamine transferase I from thewild-type medial-Golgi to the cis-Golgi compartment fromvirally infected mutant cells. Successful intercompartmentaltransport in this cell-free system depends on requirementsthat are typical of a normal physiological process including acytosolic extract, a source of chemical energy in the form ofATP and GTP, and incubation at physiological temperatures.

In addition, under appropriate conditions a uniform pop-ulation of the retrograde transport vesicles that move N-acetylglucosamine transferase I from the medial- to cis-Golgican be purified away from the donor wild-type Golgi mem-branes by centrifugation. By examining the proteins that areenriched in these vesicles, scientists have been able to identifymany of the integral membrane proteins and peripheral vesi-cle coat proteins that are the structural components of thistype of vesicle. Moreover, fractionation of the cytosolic ex-tract required for transport in cell-free reaction mixtures haspermitted isolation of the various proteins required for for-mation of transport vesicles and of proteins required for thetargeting and fusion of vesicles with appropriate acceptor


ER

Golgi

Class A Class B Class C Class D Class E

Fate ofsecretedproteins

Defectivefunction

Normalsecretion

Accumulationin rough ER

Budding ofvesicles fromthe rough ER

Accumulationin ER-to-Golgitransport vesicles

Fusion oftransport vesicleswith Golgi

Accumulationin secretoryvesicles

Transport fromsecretory vesicles to cell surface

Accumulationin the cytosol

Transportinto the ER

Accumulationin Golgi

Transport from Golgi to secretoryvesicles

▲ EXPERIMENTAL FIGURE 17-5 Phenotypes of yeast sec

mutants identified stages in the secretory pathway. Thesetemperature-sensitive mutants can be grouped into five classesbased on the site where newly made secreted proteins (reddots) accumulate when cells are shifted from the permissive

temperature to the higher nonpermissive one. Analysis of doublemutants permitted the sequential order of the steps to bedetermined. [See P. Novick et al., 1981, Cell 25:461, and C. A. Kaiserand R. Schekman, 1990, Cell 61:723.]

membranes. In vitro assays similar in general design to theone shown in Figure 17-6 have been used to study varioustransport steps in the secretory pathway.

KEY CONCEPTS OF SECTION 17.1

Techniques for Studying the Secretory Pathway

■ All assays for following the trafficking of proteinsthrough the secretory pathway in living cells require a wayto label a cohort of secretory proteins and a way to iden-tify the compartments where labeled proteins subsequentlyare located.

■ Pulse-labeling with radioactive amino acids can specifi-cally label a cohort of newly made proteins in the ER. Al-ternatively, a temperature-sensitive mutant protein that isretained in the ER at the nonpermissive temperature willbe released as a cohort for transport when cells are shiftedto the permissive temperature.

■ Transport of a fluorescently labeled protein along the se-cretory pathway can be observed by microscopy (see Fig-ure 17-2). Transport of a radiolabeled protein commonlyis tracked by following compartment-specific covalentmodifications to the protein.

■ Many of the components required for intracellular pro-tein trafficking have been identified in yeast by analysis

of temperature-sensitive sec mutants defective for the se-cretion of proteins at the nonpermissive temperature (seeFigure 17-5).

■ Cell-free assays for intercompartmental protein trans-port have allowed the biochemical dissection of individualsteps of the secretory pathway. Such in vitro reactions canbe used to produce pure transport vesicles and to test thebiochemical function of individual transport proteins.

Molecular Mechanisms of Vesicular TrafficSmall membrane-bounded vesicles that transport proteinsfrom one organelle to another are common elements in thesecretory and endocytic pathways (see Figure 17-1). Thesevesicles bud from the membrane of a particular “parent”(donor) organelle and fuse with the membrane of a particu-lar “target” (destination) organelle. Although each step inthe secretory and endocytic pathways employs a differenttype of vesicle, studies employing genetic and biochemicaltechniques described in the previous section have revealedthat each of the different vesicular transport steps is simplya variation on a common theme. In this section we explorethat common theme, the basic mechanisms underlying vesi-cle budding and fusion.

17.2

17.2 • Molecular Mechanisms of Vesicular Traffic 707

Addition ofN-acetyl-glucosamineto G protein

Incubation

(a) (b)

Golgi isolated fromuninfected wild-type cells

G protein in Golgi from infected mutant cells

VSV-infected wild-type cells

VSV-infected mutant cells

(no N-acetylglucosaminetransferase I)

N-Acetylglucosaminetransferase I reaction

G protein

Cis-Golgi Medial-Golgi Trans-Golgi

= N-Acetylglucosamine= Mannose

= Galactose

= N-Acetylneuraminic acid

▲ EXPERIMENTAL FIGURE 17-6 Protein transport from

one Golgi cisternae to another can be assayed in a cell-free

system. (a) A mutant line of cultured fibroblasts is essential inthis type of assay. In this example, the cells lack the enzyme N-acetylglucosamine transferase I (step in Figure 17-3). Inwild-type cells, this enzyme is localized to the medial-Golgi andmodifies N-linked oligosaccharides by the addition of one N-acetylglucosamine. In VSV-infected wild-type cells, theoligosaccharide on the viral G protein is modified to a typicalcomplex oligosaccharide, as shown in the trans-Golgi panel. Ininfected mutant cells, however, the G protein reaches the cell

2

surface with a simpler high-mannose oligosaccharide containingonly two N-acetylglucosamine and five mannose residues. (b)When Golgi cisternae isolated from infected mutant cells areincubated with Golgi cisternae from normal, uninfected cells, theVSV G protein produced in vitro contains the additional N-acetylglucosamine. This modification is carried out by transferaseenzyme that is moved by retrograde transport vesicles from thewild-type medial-Golgi cisternae to the mutant cis-Golgi cisternaein the reaction mixture. [See W. E. Balch et al., 1984, Cell 39:405 and525; W. A. Braell et al., 1984, Cell 39:511; and J. E. Rothman and T. Söllner, 1997, Science 276:1212.]

The budding of vesicles from their parent membrane isdriven by the polymerization of soluble protein complexesonto the membrane to form a proteinaceous vesicle coat (Fig-ure 17-7a). Interactions between the cytosolic portions of in-tegral membrane proteins and the vesicle coat gather theappropriate cargo proteins into the forming vesicle. Thus thecoat not only adds curvature to the membrane to form a vesi-cle but also acts as the filter to determine which proteins areadmitted into the vesicle.

The integral membrane proteins in a budding vesicle in-clude v-SNAREs, which are crucial to eventual fusion of thevesicle with the correct target membrane. Shortly after for-mation of a vesicle is completed, the coat is shed exposingits v-SNARE proteins. The specific joining of v-SNAREs in

the vesicle membrane with cognate t-SNAREs in the targetmembrane brings the membranes into close apposition, al-lowing the two bilayers to fuse (Figure 17-7b).

Assembly of a Protein Coat Drives VesicleFormation and Selection of Cargo MoleculesThree types of coated vesicles have been characterized, eachwith a different type of protein coat and each formed by re-versible polymerization of a distinct set of protein subunits(Table 17-1). Each type of vesicle, named for its primary coatproteins, transports cargo proteins from particular parent or-ganelles to particular destination organelles:

■ COPII vesicles transport proteins from the rough ER tothe Golgi.

■ COPI vesicles mainly transport proteins in the retro-grade direction between Golgi cisternae and from the cis-Golgi back to the rough ER.

■ Clathrin vesicles transport proteins from the plasmamembrane (cell surface) and the trans-Golgi network tolate endosomes.

Researchers have not yet identified the coat proteins sur-rounding the vesicles that move proteins from the trans-Golgi to the plasma membrane during either constitutive orregulated secretion.

The general scheme of vesicle budding shown in Figure17-7a applies to all three known types of coated vesicles.Experiments with isolated or artificial membranes and puri-fied coat proteins have shown that polymerization of thecoat proteins onto the cytosolic face of the parent mem-brane is necessary to produce the high curvature of the


Coat proteinsDonormembrane

GTP-binding protein

Membrane cargo-receptor protein

Membranecargo protein

Solublecargoprotein

(b) Uncoated vesicle fusion

t-SNAREproteins

(a) Coated vesicle budding

Targetmembrane

t-SNAREcomplex

v-SNAREprotein

Cytosol

Cytosol

▲ FIGURE 17-7 Overview of vesicle budding and fusion

with a target membrane. (a) Budding is initiated by recruitmentof a small GTP-binding protein to a patch of donor membrane.Complexes of coat proteins in the cytosol then bind to thecytosolic domain of membrane cargo proteins, some of whichalso act as receptors that bind soluble proteins in the lumen,thereby recruiting luminal cargo proteins into the budding vesicle.(b) After being released and shedding its coat, a vesicle fuseswith its target membrane in a process that involves interaction of cognate SNARE proteins.

▲ EXPERIMENTAL FIGURE 17-8 Vesicle buds can be

visualized during in vitro budding reactions. When purifiedCOPII coat components are incubated with isolated ER vesiclesor artificial phospholipid vesicles (liposomes), polymerization ofthe coat proteins on the vesicle surface induces emergence ofhighly curved buds. In this electron micrograph of an in vitrobudding reaction, note the distinct membrane coat, visible as adark protein layer, present on the vesicle buds. [From K. Matsuokaet al., 1988, Cell 93(2):263.]

100 nm

membrane that is typical of a transport vesicle about 50 nmin diameter. Electron micrographs of in vitro budding reac-tions often reveal structures that exhibit discrete regions ofthe parent membrane bearing a dense coat accompanied bythe curvature characteristic of a completed vesicle (Figure17-8). Such structures, usually called vesicle buds, appear tobe intermediates that are visible after the coat has begun topolymerize but before the completed vesicle pinches offfrom the parent membrane. The polymerized coat proteinsare thought to form some type of curved lattice that drivesthe formation of a vesicle bud by adhering to the cytosolicface of the membrane.

A Conserved Set of GTPase Switch ProteinsControls Assembly of Different Vesicle CoatsBased on in vitro vesicle-budding reactions with isolatedmembranes and purified coat proteins, scientists have deter-mined the minimum set of coat components required to formeach of the three major types of vesicles. Although most ofthe coat proteins differ considerably from one type of vesi-cle to another, the coats of all three vesicles contain a smallGTP-binding protein that acts as a regulatory subunit to con-trol coat assembly (see Figure 17-7a). For both COPI andclathrin vesicles, this GTP-binding protein is known as ARF.A different but related GTP-binding protein known as Sar1is present in the coat of COPII vesicles. Both ARF and Sar1are monomeric proteins with an overall structure similar tothat of Ras, a key intracellular signal-transducing protein(see Figure 14-20). ARF and Sar1 proteins, like Ras, belongto the GTPase superfamily of switch proteins that cycle be-tween inactive GDP-bound and active GTP-bound forms (seeFigure 3-29).

The cycle of GTP binding and hydrolysis by ARF andSar1 are thought to control the initiation of coat assembly

as schematically depicted for the assembly of COPII vesiclesin Figure 17-9. First, an ER membrane protein known asSec12 catalyzes release of GDP from cytosolic Sar1 � GDPand binding of GTP. The Sec12 guanine nucleotide–exchangefactor apparently receives and integrates multiple, as yet un-known signals, probably including the presence of cargo pro-teins in the ER membrane that are ready to be transported.Binding of GTP causes a conformational change in Sar1 thatexposes its hydrophobic N-terminus, which then becomesembedded in the phospholipid bilayer and tethers Sar1 � GTPto the ER membrane. The membrane-attached Sar1 � GTPdrives polymerization of cytosolic complexes of COPII sub-units on the membrane, eventually leading to formation ofvesicle buds. Once COPII vesicles are released from thedonor membrane, the Sar1 GTPase activity hydrolyzes Sar1 � GTP in the vesicle membrane to Sar1 � GDP with theassistance of one of the coat subunits. This hydrolysis trig-gers disassembly of the COPII coat. Thus Sar1 couples acycle of GTP binding and hydrolysis to the formation andthen dissociation of the COPII coat.

ARF protein undergoes a similar cycle of nucleotide ex-change and hydrolysis coupled to the assembly of vesiclecoats composed either of COPI or of clathrin and other coatproteins (AP complexes) discussed later. A myristate anchorcovalently attached to the N-terminus of ARF proteinweakly tethers ARF � GDP to the Golgi membrane. WhenGTP is exchanged for the bound GDP by a nucleotide-exchange factor attached to the Golgi membrane, the result-ing conformational change in ARF allows hydrophobicresidues in its N-terminal segment to insert into the mem-brane bilayer. The resulting tight association of ARF � GTPwith the membrane serves as the foundation for further coatassembly.

Drawing on the structural similarities of Sar1 and ARF to other small GTPase switch proteins, researchers have


TABLE 17-1 Coated Vesicles Involved in Protein Trafficking

Vesicle Type Coat Proteins Associated GTPase Transport Step Mediated

COPII Sec23/Sec24 and Sec13/Sec31 Sar1 ER to cis-Golgicomplexes, Sec16

COPI Coatomers containing seven ARF cis-Golgi to ERdifferent COP subunits Later to earlier Golgi cisternae

Clathrin and Clathrin � AP1 complexes ARF trans-Golgi to endosome adapter proteins*

Clathrin � GGA ARF trans-Golgi to endosome

Clathrin � AP2 complexes ARF Plasma membrane to endosome

AP3 complexes ARF Golgi to lysosome, melanosome, or platelet vesicles

*Each type of AP complex consists of four different subunits. It is not known whether the coat of AP3 vesicles contains clathrin.

constructed genes encoding mutant versions of the two pro-teins that have predictable effects on vesicular traffic whentransfected into cultured cells. For example, in cells express-ing mutant versions of Sar1 or ARF that cannot hydrolyzeGTP, vesicle coats form and vesicle buds pinch off. However,because the mutant proteins cannot trigger disassembly ofthe coat, all available coat subunits eventually become per-manently assembled into coated vesicles that are unable tofuse with target membranes. Addition of a nonhydrolyzableGTP analog to in vitro vesicle-budding reactions causes asimilar blocking of coat disassembly. The vesicles that formin such reactions have coats that never dissociate, allowingtheir composition and structure to be more readily analyzed.The purified COPI vesicles shown in Figure 17-10 were pro-duced in such a budding reaction.

Targeting Sequences on Cargo Proteins MakeSpecific Molecular Contacts with Coat ProteinsIn order for transport vesicles to move specific proteins fromone compartment to the next, vesicle buds must be able todiscriminate among potential membrane and soluble cargoproteins, accepting only those cargo proteins that should ad-vance to the next compartment and excluding those thatshould remain as residents in the donor compartment. In ad-dition to sculpting the curvature of a donor membrane, thevesicle coat also functions in selecting specific proteins ascargo. The primary mechanism by which the vesicle coat selects cargo molecules is by directly binding to specific


Sec12

GDPGTP

Sar1

ER lumen

2COPII coat

assembly

Uncoated vesicle

Pi

Pi

Pi

1 Sar1 membrane binding,

GTP exchange

Sec23/Sec24

GDP

GTP

Hydrophobic N-terminus

Cytosol

GTP hydrolysis3

Coat disassembly4

▲ FIGURE 17-9 Model for the role of Sar1 in the assembly

and disassembly of COPII coats. Step : Interaction of solubleGDP-bound Sar1 with the exchange factor Sec12, an ER integralmembrane protein, catalyzes exchange of GTP for GDP on Sar1.In the GTP-bound form of Sar1, its hydrophobic N-terminusextends outward from the protein’s surface and anchors Sar1 tothe ER membrane. Step : Sar1 attached to the membraneserves as a binding site for the the Sec23/Sec24 coat proteincomplex. Cargo proteins are recruited to the forming vesicle budby binding of specific short sequences (sorting signals) in theircytosolic regions to sites on the Sec23/Sec24 complex. The coatis completed by assembly of a second type of coat complexcomposed of Sec13/and Sec31 (not shown). Step : After thevesicle coat is complete, the Sec23 coat subunit promotes GTPhydrolysis by Sar1. Step : Release of Sar1 · GDP from thevesicle membrane causes disassembly of the coat. [See S. Springer et al., 1999, Cell 97:145.]

4

3

2

1

▲ EXPERIMENTAL FIGURE 17-10 Coated vesicles

accumulate during in vitro budding reactions in the presence

of a nonhydrolyzable analog of GTP. When isolated Golgimembranes are incubated with a cytosolic extract containingCOPI coat proteins and ATP, vesicles form and bud off from themembranes. Inclusion of a nonhydrolyzable analog of GTP in thebudding reaction prevents disassembly of the coat after vesiclerelease. This micrograph shows COPI vesicles generated in sucha reaction and separated from membranes by centrifugation.Coated vesicles prepared in this way can be analyzed todetermine their components and properties. [Courtesy of L. Orci.]

sequences, or sorting signals, in the cytosolic portion ofmembrane cargo proteins (see Figure 17-7a). The polymer-ized coat thus acts as an affinity matrix to cluster selectedmembrane cargo proteins into forming vesicle buds. Solubleproteins within the lumen of parent organelles can in turnbe selected by binding to the luminal domains of certainmembrane cargo proteins, which act as receptors for luminalcargo proteins. The properties of several known sorting sig-nals in membrane and soluble proteins are summarized inTable 17-2. We describe the role of these signals in more de-tail in later sections.

Rab GTPases Control Docking of Vesicles on Target MembranesA second set of small GTP-binding proteins, known as Rabproteins, participate in the targeting of vesicles to the appro-priate target membrane. Like Sar1 and ARF, Rab proteinsbelong to the GTPase superfamily of switch proteins. Con-version of cytosolic Rab � GDP to Rab � GTP, catalyzed bya specific guanine nucleotide–exchange factor, induces aconformational change in Rab that enables it to interact witha surface protein on a particular transport vesicle and insertits isoprenoid anchor into the vesicle membrane. Once

Rab � GTP is tethered to the vesicle surface, it is thought tointeract with one of a number of different large proteins,known as Rab effectors, attached to the target membrane.Binding of Rab � GTP to a Rab effector docks the vesicle onan appropriate target membrane (Figure 17-11, step 1 ).After vesicle fusion occurs, the GTP bound to the Rab pro-tein is hydrolyzed to GDP, triggering the release of Rab � GDP,which then can undergo another cycle of GDP-GTP ex-change, binding, and hydrolysis.

Several lines of evidence support the involvement of spe-cific Rab proteins in vesicle-fusion events. For instance, theyeast SEC4 gene encodes a Rab protein, and yeast cells ex-pressing mutant Sec4 proteins accumulate secretory vesiclesthat are unable to fuse with the plasma membrane (class Emutants in Figure 17-5). In mammalian cells, Rab5 protein islocalized to endocytic vesicles, also known as early endo-somes. These uncoated vesicles form from clathrin-coatedvesicles just after they bud from the plasma membrane dur-ing endocytosis (see Figure 17-1, step 9 ). The fusion of earlyendosomes with each other in cell-free systems requires thepresence of Rab5, and addition of Rab5 and GTP to cell-freeextracts accelerates the rate at which these vesicles fuse witheach other. A long coiled protein known as EEA1 (early endosome antigen 1), which resides on the membrane of the


TABLE 17-2 Known Sorting Signals That Direct Proteins to Specific Transport Vesicles

Vesicles That Incorporate Signal Sequence* Proteins with Signal Signal Receptor Signal-bearing Protein

Lys-Asp-Glu-Leu ER-resident luminal proteins KDEL receptor in COPI(KDEL) cis-Golgi membrane

Lys-Lys-X-X ER-resident membrane COPI � and � subunits COPI(KKXX) proteins (cytosolic domain)

Di-acidic Cargo membrane proteins in COPII Sec24 subunit COPII(e.g., Asp-X-Glu) ER (cytosolic domain)

Mannose 6-phosphate Soluble lysosomal enzymes M6P receptor in trans- Clathrin/AP1(M6P) after processing in cis-Golgi Golgi membrane

Secreted lysosomal enzymes M6P receptor in plasma Clathrin/AP2membrane

Asn-Pro-X-Tyr LDL receptor in the plasma AP2 complex Clathrin/AP2(NPXY) membrane (cytosolic domain)

Tyr-X-X-� Membrane proteins in trans- AP1 (�1 subunit) Clathrin/AP1(YXX�) Golgi (cytosolic domain)

Plasma membrane proteins AP2 (�2 subunit) Clathrin/AP2(cytosolic domain)

Leu-Leu Plasma membrane proteins AP2 complexes Clathrin/AP2 (LL) (cytosolic domain)

*X � any amino acid; � � hydrophobic amino acid. Single-letter amino acid abbreviations are in parentheses.

early endosome, functions as the effector for Rab5. In thiscase, Rab5 � GTP on one endocytic vesicle is thought tospecifically bind to EEA1 on the membrane of another en-docytic vesicle, setting the stage for fusion of the two vesicles.

Similarly, Rab1 is essential for ER-to-Golgi transport re-actions to occur in cell-free extracts. Rab1 � GTP binds to a

long coiled-coil protein known as p115, which specificallytethers COPII vesicles carrying Rab1� GTP to the targetGolgi membrane. A different type of Rab effector appearsto function for each vesicle type and at each step of the se-cretory pathway. Many questions remain about how Rabproteins are targeted to the correct membrane and how spe-cific complexes form between the different Rab proteins andtheir corresponding effector proteins.

Paired Sets of SNARE Proteins Mediate Fusion of Vesicles with Target MembranesAs noted previously, shortly after a vesicle buds off from thedonor membrane, the vesicle coat disassembles to uncover avesicle-specific membrane protein, a v-SNARE (see Figure17-7b). Likewise, each type of target membrane in a cell con-tains t-SNARE membrane proteins. After Rab-mediateddocking of a vesicle on its target (destination) membrane, theinteraction of cognate SNAREs brings the two membranesclose enough together that they can fuse.

One of the best-understood examples of SNARE-mediatedfusion occurs during exocytosis of secreted proteins (Figure17-11, steps 2 and 3 ). In this case, the v-SNARE, known asVAMP (vesicle-associated membrane protein), is incorporatedinto secretory vesicles as they bud from the trans-Golgi network. The t-SNAREs are syntaxin, an integral membraneprotein in the plasma membrane, and SNAP-25, which is attached to the plasma membrane by a hydrophobic lipid anchor in the middle of the protein. The cytosolic region ineach of these three SNARE proteins contains a repeating hep-tad sequence that allows four � helices—one from VAMP, one


ADP + Pi

ATP

VAMP

Transportvesicle

Targetmembrane

Syntaxin

SNAP-25

α -SNAP

NSF

Rab • GTP

Rab effector

4

Vesicle docking

Assembly of SNARE complexes

Membrane fusion

Disassembly of SNARE complexes

SNAREcomplex

cis-SNAREcomplex

1

2

3

VAMP

SyntaxinSNAP-25

� FIGURE 17-11 Model for docking and fusion of transport

vesicles with their target membranes. The proteins shown inthis example participate in fusion of secretory vesicles with theplasma membrane, but similar proteins mediate all vesicle-fusionevents. Step : A Rab protein tethered via a lipid anchor to asecretory vesicle binds to an effector protein complex on theplasma membrane, thereby docking the transport vesicle on theappropriate target membrane. Step : A v-SNARE protein (in thiscase, VAMP) interacts with the cytosolic domains of the cognatet-SNAREs (in this case, syntaxin and SNAP-25). The very stablecoiled-coil SNARE complexes that are formed hold the vesicleclose to the target membrane. Inset: Numerous noncovalentinteractions between four long � helices, two from SNAP-25 and one each from syntaxin and VAMP, stabilize the coiled-coilstructure. Step : Fusion of the two membranes immediatelyfollows formation of SNARE complexes, but precisely how thisoccurs is not known. Step : Following membrane fusion, NSF in conjunction with �-SNAP protein binds to the SNAREcomplexes. The NSF-catalyzed hydrolysis of ATP then drivesdissociation of the SNARE complexes, freeing the SNAREproteins for another round of vesicle fusion. [See J. E. Rothman and T. Söllner, 1997, Science 276:1212, and W. Weis and R. Scheller, 1998, Nature 395:328. Inset from Y. A. Chen and R. H. Scheller, 2001, Nat. Rev. Mol. Cell Biol. 2(2):98.]

4

3

2

1

from syntaxin, and two from SNAP-25—to coil around oneanother to form a four-helix bundle. The unusual stability ofthis bundled SNARE complex is conferred by the arrangementof hydrophobic and charged amino residues in the heptad repeats. The hydrophobic amino acids are buried in the centralcore of the bundle, and amino acids of opposite charge arealigned to form favorable electrostatic interactions betweenhelices. As the four-helix bundles form, the vesicle and targetmembranes are drawn into close apposition by the embeddedtransmembrane domains of VAMP and syntaxin.

In vitro experiments have shown that when liposomescontaining purified VAMP are incubated with other lipo-somes containing syntaxin and SNAP-25, the two classes ofmembranes fuse, albeit slowly. This finding is strong evidencethat the close apposition of membranes resulting from for-mation of SNARE complexes is sufficient to bring aboutmembrane fusion. Fusion of a vesicle and target membraneoccurs much more rapidly and efficiently in the cell than itdoes in liposome experiments in which fusion is catalyzedonly by SNARE proteins. The likely explanation for this dif-ference is that in the cell the interactions between specific Rabproteins and their effectors promote the formation of specificSNARE bundles by tethering a vesicle to its target membrane.

Yeast cells, like all eukaryotic cells, express more than20 different related v-SNARE and t-SNARE proteins. Analy-ses of yeast sec mutants defective in each of the SNAREgenes have identified the specific membrane-fusion event inwhich each SNARE protein participates. For all fusion eventsthat have been examined, the SNAREs form four-helix bun-dled complexes, similar to the VAMP/syntaxin/SNAP-25complexes that mediate fusion of secretory vesicles with theplasma membrane. However, in other fusion events (e.g., fusion of COPII vesicles with the cis-Golgi network), eachparticipating SNARE protein contributes only one � helixto the bundle (unlike SNAP-25, which contributes two helices); in these cases the SNARE complexes comprise onev-SNARE and three t-SNARE molecules.

Using the in vitro liposome fusion assay, researchers havetested the ability of various combinations of individual v-SNARE and t-SNARE proteins to mediate fusion of donorand target membranes. Of the very large number of differentcombinations tested, only a small number mediated mem-brane fusion. To a remarkable degree the functional combi-nations of v-SNAREs and t-SNAREs revealed in these in vitroexperiments correspond to the actual SNARE protein inter-actions that mediate known membrane-fusion events in theyeast cell. Thus the specificity of the interaction betweenSNARE proteins can account for the specificity of fusion be-tween a particular vesicle and its target membranes.

Dissociation of SNARE Complexes AfterMembrane Fusion Is Driven by ATP HydrolysisAfter a vesicle and its target membrane have fused, theSNARE complexes must dissociate to make the individualSNARE proteins available for additional fusion events. Be-

cause of the stability of SNARE complexes, which are heldtogether by numerous noncovalent intermolecular interac-tions, their dissociation depends on additional proteins andthe input of energy.

The first clue that dissociation of SNARE complexes re-quired the assistance of other proteins came from in vitrotransport reactions depleted of certain cytosolic proteins.The observed accumulation of vesicles in these reactions in-dicated that vesicles could form but were unable to fuse witha target membrane. Eventually two proteins, designated NSFand �-SNAP, were found to be required for ongoing vesiclefusion in the in vitro transport reaction. The function of NSFin vivo can be blocked selectively by N-ethylmaleimide(NEM), a chemical that reacts with an essential –SH groupon NSF (hence the name, NEM-sensitive factor).

Among the class C yeast sec mutants are strains that lackfunctional Sec18 or Sec17, the yeast counterparts of mam-malian NSF and �-SNAP, respectively. When these class Cmutants are placed at the nonpermissive temperature, theyaccumulate ER-to-Golgi transport vesicles; when the cells areshifted to the lower, permissive temperature, the accumu-lated vesicles are able to fuse with the cis-Golgi.

Subsequent to the initial biochemical and genetic studiesidentifying NSF and �-SNAP, more sophisticated in vitrotransport assays were developed. Using these newer assays,researchers have shown that NSF and �-SNAP proteins arenot necessary for actual membrane fusion, but rather are re-quired for regeneration of free SNARE proteins. NSF, ahexamer of identical subunits, associates with a SNAREcomplex with the aid �-SNAP (soluble NSF attachment pro-tein). The bound NSF then hydrolyzes ATP, releasing suffi-cient energy to dissociate the SNARE complex (Figure 17-11,step 4 ). Evidently, the defects in vesicle fusion observed inthe earlier in vitro fusion assays and in the yeast mutantsafter a loss of Sec17 or Sec18 were a consequence of freeSNARE proteins rapidly becoming sequestered in undissoci-ated SNARE complexes and thus unavailable to mediatemembrane fusion.

Conformational Changes in Viral EnvelopeProteins Trigger Membrane FusionSome animal viruses, including influenza virus, rabies virus,and human immunodeficiency virus (HIV), have an outerphospholipid bilayer membrane, or envelope, surroundingthe core of the virus particle composed of viral proteins andgenetic material. The viral envelope is derived by buddingfrom the host-cell plasma membrane, which contains virus-encoded glycoproteins. Enveloped viruses enter a host cell byendocytosis following binding of one or more viral envelopeglycoproteins with a host’s cell-surface molecules. Subsequentfusion of the viral envelope with the endosomal membrane re-leases the viral genome into the cytosol of the host cell, initi-ating replication of the virus (see Figure 4-41, step 3 ). Themolecular events of this fusion process have been elucidatedin considerable detail in the case of influenza virus.


The predominant glycoprotein of the influenza virus ishemagglutinin (HA), which forms the larger spikes on the sur-face of the virus. There is considerable evidence that followingendocytosis of an influenza virion, the low pH within the en-closing late endosome triggers fusion of its membrane with theviral envelope. For instance, viral infection is inhibited by theaddition of lipid-soluble bases, such as ammonia or trimethyl-amine, which raise the normally acidic pH of late endosomes.Also, a conformational change in the HA protein that is criticalfor infectivity occurs over a very narrow range in pH (5.0–5.5).

Each HA spike on an influenza virion consists of threeHA1 and three HA2 subunits. At the N-terminus of HA2 is astrongly hydrophobic 11-residue sequence, called the fusion

peptide. Structural studies have shown that at pH 7.0, the N-terminus of each HA2 subunit is tucked into a crevice in thespike (Figure 17-12a). This is the normal HA conformationwhen a viral particle encounters the surface of a host cell. Atthe acidic pH characteristic of late endosomes, HA undergoesseveral conformational changes that cause a major rearrange-ment of the subunits. As a result, the three HA2 subunits twisttogether into a three-stranded coiled-coil rod that protrudesmore than 13 nm outward from the viral envelope with thefusion peptides at the tip of the rod (Figure 17-12b). In thisconformation, the highly hydrophobic fusion peptides are ex-posed and can insert into the lipid bilayer of the endosomalmembrane, triggering fusion of the viral envelope and themembrane. Thus at pH 7 HA can be said to be trapped in ametastable, “spring-loaded” state, which is converted to thelower-energy fusogenic state by shifting the pH to 5–5.5.

Multiple low pH–activated HA spikes are essential for mem-brane fusion to occur. Figure 17-13 suggests one way by which the protein scaffold formed by many HA spikes, possibly with the assistance of other cellular proteins, could link together the


(a) pH ∼ 7.0

Cell-surface membrane

Sialicacid

Fusionpeptide

(b) pH 5.0–5.5

Endosomalmembrane

Viral envelope

Disulfidebond

▲ FIGURE 17-12 Schematic models of the structure

of influenza hemagglutinin (HA) at pH 7 and 5. Three HA1 andthree HA2 subunits compose a hemagglutinin molecule, whichprotrudes from the viral envelope like a spike. (a) At pH ≈7, partof each HA1 subunit forms a globular domain (green) at the tip ofthe native spike. These domains bind to sialic acid residues onthe host-cell plasma membrane, initiating viral entry. Each HA1

subunit is linked to one HA2 subunit by a disulfide bond at thebase of the molecule near the viral envelope. Each HA2 subunitcontains a fusion peptide (red) at its N-terminus (only two arevisible), followed by a short � helix (orange cylinder), a nonhelicalloop (brown), and a longer � helix (light purple). The longer �helices from the three HA2 subunits form a three-strandedcoiled-coil structure (see Figure 3-7). In this conformation, thefusion peptides are buried within the molecule. (b) At the acidicpH within a late endosome, the binding of the fusion peptide toother segments of HA2 is disrupted, inducing major structuralrearrangements in the protein. First, the three HA1 globulardomains separate from each other but remain tethered to theHA2 subunits by the disulfide bonds at the base of the molecule.Second, the loop segment of each HA2 rearranges into an � helix(brown) and combines with the short and long �-helicalsegments to form a continuous 88-residue � helix. The threelong � helices thus form a 13.5-nm-long three-stranded coiledcoil that protrudes outward from the viral envelope. In thisconformation, the fusion peptides are at the tip of the coiled coiland can insert into the endosomal membrane. [Adapted from C. M.Carr et al., 1997, Proc. Nat’l. Acad. Sci. 94:14306; courtesy of Peter Kim.]

Viral envelopeActivatedHA proteins

Exoplasmicleaflet

Cystolicleaflet

Fused membranes

Endosomal membrane

▲ FIGURE 17-13 Model for membrane fusion directed by

hemagglutinin (HA). A number of low pH–activated HA spikes,possibly in concert with host-cell membrane proteins, form ascaffold that connects a small region of the viral envelope andthe endosomal membrane. By unknown mechanisms, theexoplasmic leaflets of the two membranes fuse and then thecytosolic leaflets fuse, forming a pore that widens until the twomembranes are completely joined. A similar interaction betweenmembrane bilayers may be brought about during SNARE-mediatedvesicle fusion. [Adapted from J. R. Monck and J. M. Fernandez, 1992, J. Cell Biol. 119:1395.]

viral envelope and endosomal membrane and induce their fu-sion. This figure also illustrates how cellular membranes brought into close apposition by SNARE complexes might fuse. Note that each HA molecule participates in only one fusion event, whereas the cellular fusion proteins, such as SNAREs, are recycled and catalyze multiple cycles of membrane fusion.


Molecular Mechanisms of Vesicular Traffic

■ The three well-characterized transport vesicles—COPI,COPII, and clathrin vesicles—are distinguished by the pro-teins that form their coats and the transport routes theymediate (see Table 17-1).

■ All types of coated vesicles are formed by polymeriza-tion of cytosolic coat proteins onto a donor (parent) mem-brane to form vesicle buds that eventually pinch off fromthe membrane to release a complete vesicle. Shortly aftervesicle release, the coat is shed exposing proteins requiredfor fusion with the target membrane (see Figure 17-7).

■ Small GTP-binding proteins (ARF or Sar1) belonging tothe GTPase superfamily control polymerization of coat pro-teins, the initial step in vesicle budding (see Figure 17-9).After vesicles are released from the donor membrane, hy-drolysis of GTP bound to ARF or Sar1 triggers disassem-bly of the vesicle coats.

■ Specific sorting signals in membrane and luminal pro-teins of donor organelles interact with coat proteins dur-ing vesicle budding, thereby recruiting cargo proteins tovesicles (see Table 17-2).

■ A second set of GTP-binding proteins, the Rab proteins,regulate docking of vesicles with the correct target mem-brane. Each Rab appears to bind to a specific Rab effec-tor, a typically long coiled-coil protein, associated with thetarget membrane.

■ Each v-SNARE in a vesicular membrane specificallybinds to a complex of cognate t-SNARE proteins in thetarget membrane, inducing fusion of the two membranes.After fusion is completed, the SNARE complex is disas-sembled in an ATP-dependent reaction mediated by othercytosolic proteins (see Figure 17-11).

■ After an enveloped animal virus is endocytosed, the viralenvelope fuses with the surrounding endosomal membrane.In the case of influenza virus, the acidic pH within late en-dosomes causes a conformational change in the HA pro-tein in the viral envelope that permits insertion of HA intothe endosomal membrane.

Early Stages of the Secretory PathwayIn this section we take a closer look at vesicular trafficthrough the ER and Golgi stages of the secretory pathway

17.3

and some of the evidence supporting the general mechanismsdiscussed in the previous section. Recall that anterogradetransport from the ER to Golgi, the first step in the secre-tory pathway, is mediated by COPII vesicles, whereas the re-verse retrograde transport from the cis-Golgi to the ER ismediated by COPI vesicles (Figure 17-14). This retrograde

17.3 • Early Stages of the Secretory Pathway 715

RoughER

Cis-Golginetwork

Soluble cargo

COPIvesicle

COPII vesicle

SNARE pair

Membrane receptor

Coat protein

Membranecargo

SNAREprotein

SNARE pair

3

2

1

4

5

6

▲ FIGURE 17-14 Vesicle-mediated protein trafficking

between the ER and cis-Golgi. Steps – : Forward(anterograde) transport is mediated by COPII vesicles, which areformed by polymerization of soluble COPII coat proteincomplexes (blue) on the ER membrane. v-SNAREs (red) andother cargo proteins (green) in the ER membrane areincorporated into the vesicle by interacting with coat proteins.Soluble cargo proteins (purple) are recruited by binding toappropriate receptors in the membrane of budding vesicles.Dissociation of the coat recycles free coat complexes andexposes v-SNARE proteins on the vesicle surface. After theuncoated vesicle becomes tethered to the cis-Golgi membrane ina Rab-mediated process, pairing between the exposed v-SNAREsand cognate t-SNAREs in the Golgi membrane allow vesiclefusion, releasing the contents into the cis-Golgi compartment(see Figure 17-11). Steps – : Reverse (retrograde) transport,mediated by vesicles coated with COPI proteins (green), recyclesthe membrane bilayer and certain proteins, such as v-SNAREsand missorted ER-resident proteins (not shown), from the cis-Golgi to the ER. All SNARE proteins are shown in red althougheach v-SNARE and t-SNARE are distinct proteins.

64

31

vesicle transport serves to retrieve v-SNARE proteins and themembrane itself back to the ER to provide the necessary ma-terial for additional rounds of vesicle budding from the ER.COPI-mediated retrograde transport also retrieves missortedER-resident proteins from the cis-Golgi to correct sortingmistakes. Proteins that have been correctly delivered to theGolgi advance through successive compartments of the Golgiby cisternal progression.

COPII Vesicles Mediate Transport from the ER to the GolgiCOPII vesicles were first recognized when cell-free extractsof yeast rough ER membranes were incubated with cytosol,ATP, and a nonhydrolyzable analog of GTP. The vesicles thatformed from the ER membranes had a distinct coat, similarto that on COPI vesicles but composed of different proteins,designated COPII proteins. Yeast cells with mutations in thegenes for COPII proteins are class B sec mutants and accu-mulate proteins in the rough ER (see Figure 17-5). Analysisof such mutants has revealed several proteins required forformation of COPII vesicles.

As described previously, formation of COPII vesicles istriggered when Sec12, a guanine nucleotide–exchange fac-tor, catalyzes the exchange of bound GDP for GTP on Sar1.This exchange induces binding of Sar1 to the ER membranefollowed by binding of a complex of Sec23 and Sec24 pro-teins (see Figure 17-9). The resulting ternary complex formedbetween Sar1� GTP, Sec23, and Sec24 is shown in Figure 17-15. After this complex forms on the ER membrane, a sec-ond complex comprising Sec13 and Sec31 proteins thenbinds to complete the coat structure. A large fibrous protein,called Sec16, which is bound to the cytosolic surface of theER, interacts with the Sec13/31 and Sec23/24 complexes,and acts to organize the other coat proteins, increasing theefficiency of coat polymerization.

Certain integral ER membrane proteins are specificallyrecruited into COPII vesicles for transport to the Golgi. Thecytosolic segments of many of these proteins contain a di-acidic sorting signal (Asp-X-Glu, or DXE in the one-letter code). This sorting signal binds to the Sec24 subunitof the COPII coat and is essential for the selective export ofcertain membrane proteins from the ER (see Figure 17-15).Biochemical and genetic studies currently are under way toidentify additional signals that help direct membrane cargoproteins into COPII vesicles. Other ongoing studies seek todetermine how soluble cargo proteins are selectively loadedinto COPII vesicles. Although purified COPII vesicles fromyeast cells have been found to contain a membrane proteinthat binds the soluble � mating factor, the receptors for othersoluble cargo proteins such as invertase are not yet known.

The experiments described previously in which the tran-sit of VSVG-GFP in cultured mammalian cells is followedby fluorescence microscopy (see Figure 17-2) provided in-sight into the intermediates in ER-to-Golgi transport. Insome cells, small fluorescent vesicles containing VSVG-GFPcould be seen to form from the ER, move less than 1 �m, and

then fuse directly with the cis-Golgi. In other cells, in whichthe ER was located several micrometers from the Golgi com-plex, several ER-derived vesicles were seen to fuse with eachother shortly after their formation, forming what is termedthe “ER-to-Golgi intermediate compartment.” These largerstructures then were transported along microtubules to thecis-Golgi, much in the way vesicles in nerve cells are trans-ported from the cell body, where they are formed, down thelong axon to the axon terminus (Chapter 20). Microtubulesfunction much as “railroad tracks” enabling these large ag-gregates of transport vesicles to move long distances to theircis-Golgi destination. At the time the ER-to-Golgi intermedi-ate compartment is formed, some COPI vesicles bud off fromit, recycling some proteins back to the ER.

COPI Vesicles Mediate Retrograde Transportwithin the Golgi and from the Golgi to the ERCOPI vesicles were first discovered when isolated Golgi frac-tions were incubated in a solution containing ATP, cytosol, anda nonhydrolyzable analog of GTP (see Figure 17-10). Subse-quent analysis of these vesicles showed that the coat is formedfrom large cytosolic complexes, called coatomers, composed ofseven polypeptide subunits. Yeast cells containing temperature-sensitive mutations in COPI proteins accumulate proteins in the


Sec23

Sec24

Sar1Vesiclemembrane

Transmembranesegmentof cargo protein

COO−

▲ FIGURE 17-15 Three-dimensional structure of ternary

complex comprising the COPII coat proteins Sec23 and

Sec24 and Sar1 · GTP. Early in the formation of the COPII coat,Sec23 (orange)/Sec24 (green) complexes are recruited to the ERmembrane by Sar1 (red) in its GTP-bound state. In order to forma stable ternary complex in solution for structural studies, thenonhydrolyzable GTP analog GppNHp was used. A cargo proteinin the ER membrane can be recruited to COPII vesicles byinteraction of a tripeptide di-acidic signal (purple) in the cargo’scytosolic domain with Sec24. The likely position of the COPIIvesicle membrane and the transmembrane segment of the cargoprotein are indicated. The N-terminal segment of Sar1 thattethers it to the membrane is not shown. [See X. Bi et al., 2002,Nature 419:271; interaction with peptide courtesy of J. Goldberg.]

rough ER at the nonpermissive temperature and thus are cate-gorized as class B sec mutants (see Figure 17-5). Although dis-covery of these mutants initially suggested that COPI vesiclesmediate ER-to-Golgi transport, subsequent experimentsshowed that their main function is retrograde transport, bothbetween Golgi cisternae and from the cis-Golgi to the rough ER(see Figure 17-14, right). Because COPI mutants cannot recyclekey membrane proteins back to the rough ER, the ER graduallybecomes depleted of ER proteins such as v-SNAREs necessaryfor COPII vesicle function. Eventually vesicle formation fromthe rough ER grinds to a halt; secretory proteins continue tobe synthesized but accumulate in the ER, the defining charac-teristic of class B sec mutants.

As discussed in Chapter 16, the ER contains several solu-ble proteins dedicated to the folding and modification ofnewly synthesized secretory proteins. These include the chap-erone BiP and the enzyme protein disulfide isomerase, whichare necessary for the ER to carry out its functions. Althoughsuch ER-resident luminal proteins are not specifically selectedby COPII vesicles, their sheer abundance causes them to becontinuously loaded passively into vesicles destined for the cis-Golgi. The transport of these soluble proteins back to the ER,mediated by COPI vesicles, prevents their eventual depletion

Most soluble ER-resident proteins carry a Lys-Asp-Glu-Leu (KDEL in the one-letter code) sequence at their C-terminus (see Table 17-2). Several experiments demonstratedthat this KDEL sorting signal is both necessary and sufficientfor retention in the ER. For instance, when a mutant proteindisulfide isomerase lacking these four residues is synthesizedin cultured fibroblasts, the protein is secreted. Moreover, ifa protein that normally is secreted is altered so that it con-tains the KDEL signal at its C-terminus, the protein is re-tained in the ER. The KDEL sorting signal is recognized andbound by the KDEL receptor, a transmembrane proteinfound primarily on small transport vesicles shuttling betweenthe ER and the cis-Golgi and on the cis-Golgi reticulum. Inaddition, soluble ER-resident proteins that carry the KDELsignal have oligosaccharide chains with modifications thatare catalyzed by enzymes found only in the cis-Golgi or cis-Golgi reticulum; thus at some time these proteins must haveleft the ER and been transported at least as far as the cis-Golgi network. These findings indicate that the KDEL re-ceptor acts mainly to retrieve soluble proteins containing theKDEL sorting signal that have escaped to the cis-Golgi net-work and return them to the ER (Figure 17-16).

The KDEL receptor and other membrane proteins thatare transported back to the ER from the Golgi contain a Lys-Lys-X-X sequence at the very end of their C-terminal seg-ment, which faces the cytosol (see Table 17-2). This KKXXsorting signal which binds to a complex of the COPI � and� subunits, is both necessary and sufficient to incorporatemembrane proteins into COPI vesicles for retrograde trans-port to the ER. Temperature-sensitive yeast mutants lackingCOPI� or COPI� not only are unable to bind the KKXX sig-nal but also are unable to retrieve proteins bearing this signalback to the ER, indicating that COPI vesicles mediate retro-grade Golgi-to-ER transport.

Clearly, the partitioning of proteins between the ER andGolgi complex is a highly selective and regulated process ul-timately controlled by the specificity of cargo loading intoboth COPII (anterograde) and COPI (retrograde) vesicles.The selective entry of proteins into membrane-boundedtransport vesicles, the recycling of membrane phospholipidsand proteins, and the recycling of soluble luminal proteinsbetween the two compartments are fundamental features ofvesicular protein trafficking that also occur in later stages ofthe secretory pathway.

17.3 • Early Stages of the Secretory Pathway 717

MissortedER-resident

protein

RoughER

KDELreceptor

ER-to-Golgitransport vesicle Retrieval

of KDEL-bearingproteinsto ER

Cis-Golginetwork

COPI coat

COPII coat

2

1

3

4

KDEL peptide

▲ FIGURE 17-16 Role of the KDEL receptor in

retrieval of ER-resident luminal proteins from the Golgi.

ER luminal proteins, especially those present at high levels,can be passively incorporated into COPII vesicles andtransported to the Golgi (steps and ). Many suchproteins bear a C-terminal KDEL (Lys-Asp-Glu-Leu)sequence (red) that allows them to be retrieved. The KDELreceptor, located mainly in the cis-Golgi network and inboth COPII and COPI vesicles, binds proteins bearing theKDEL sorting signal and returns them to the ER (steps and ). This retrieval system prevents depletion of ERluminal proteins such as those needed for proper folding of newly made secretory proteins. The binding affinity ofthe KDEL receptor is very sensitive to pH. The smalldifference in the pH of the ER and Golgi favors binding ofKDEL-bearing proteins to the receptor in Golgi-derivedvesicles and their release in the ER. [Adapted from J.Semenza et al., 1990, Cell 61:1349.]

43

21

ME

DIA

C

ON

NE

CT

IO

NS

Video: K

DEL

Receptor Trafficking

Anterograde Transport Through the Golgi Occursby Cisternal Progression

At one time it was thought that small transport vesicles carrysecretory proteins from the cis- to the medial-Golgi and fromthe medial- to the trans-Golgi. Indeed, electron microscopyreveals many small vesicles associated with the Golgi com-plex that move proteins from one Golgi compartment to an-other (Figure 17-17). However, these vesicles most likelymediate retrograde transport, retrieving ER or Golgi en-zymes from a later compartment and transporting them toan earlier compartment in the secretory pathway. In this wayenzymes that modify secretory proteins come to be localizedin the correct compartment.

The first evidence that the forward transport of cargoproteins from the cis- to the trans-Golgi occurs by a non-vesicular mechanism, called cisternal progression, came fromcareful microscopic analysis of the synthesis of algal scales.These cell-wall glycoproteins are assembled in the cis-Golgiinto large complexes visible in the electron microscope. Likeother secretory proteins, newly made scales move from thecis- to the trans-Golgi, but they can be 20 times larger thanthe usual transport vesicles that bud from Golgi cisternae.Similarly, in the synthesis of collagen by fibroblasts, large ag-gregates of the procollagen precursor often form in the

lumen of the cis-Golgi (see Figure 6-20). The procollagen ag-gregates are too large to be incorporated into small transportvesicles, and investigators could never find such aggregates intransport vesicles. These observations suggested that the for-ward movement of these and perhaps all secretory proteinsfrom one Golgi compartment to another does not occur viasmall vesicles.

In one test of the cisternal progression model, collagenfolding was blocked by an inhibitor of proline hydroxy-lation, and soon all pre-made, folded procollagen aggre-gates were secreted from the cell. When the inhibitor wasremoved, newly made procollagen peptides folded andthen formed aggregates in the cis-Golgi that subsequentlycould be seen to move as a “wave” from the cis- throughthe medial-Golgi cisternae to the trans-Golgi, followed bysecretion and incorporation into the extracellular matrix.In these experiments procollagen aggregates were nevervisible in small transport vesicles. Numerous controver-sial questions concerning membrane flow within the Golgistack remain unresolved. Nonetheless, the observedmovement of very large macromolecular assembliesthrough the Golgi stack and the evidence described pre-viously that COPI vesicles mediate retrograde transporthave led most researchers in the field to favor the cisternalprogression model.


0.5 �m

Transitional elements

Smooth protrusion

ER-to-Golgitransport vesicles

Cis-Golgi reticulum

Cis

Medial

TransGolgicisternae

Trans-Golgi network

Forming secretoryvesicle

▲ EXPERIMENTAL FIGURE 17-17 Electron micrograph of

the Golgi complex in an exocrine pancreatic cell reveals both

anterograde and retrograde transport vesicles. A largesecretory vesicle can be seen forming from the trans-Golginetwork. Elements of the rough ER are on the left in thismicrograph. Adjacent to the rough ER are transitional elements

from which smooth protrusions appear to be budding. These buds form the small vesicles that transport secretory proteinsfrom the rough ER to the Golgi complex. Interspersed among theGolgi cisternae are other small vesicles now known to function inretrograde, not anterograde, transport. [Courtesy G. Palade.]


Vesicle Traffic in the Early Stages of the SecretoryPathway

■ COPII vesicles transport proteins from the rough ER tothe cis-Golgi; COPI vesicles transport proteins in the re-verse direction (see Figure 17-14).

■ COPII coats comprise three components: the small GTP-binding protein Sar1, a Sec23/Sec24 complex, and aSec13/Sec31 complex.

■ Components of the COPII coat bind to membrane cargoproteins containing a di-acidic or other sorting signal intheir cytosolic regions (see Figure 17-15). Soluble cargoproteins probably are targeted to COPII vesicles by bind-ing to a membrane protein receptor.

■ Membrane proteins needed to form COPII vesicles canbe retrieved from the cis-Golgi by COPI vesicles. One ofthe sorting signals that directs membrane proteins intoCOPI vesicles is a KKXX sequence, which binds to sub-units of the COPI coat.

■ Many soluble ER-resident proteins contain a KDEL sort-ing signal. Binding of this retrieval sequence to a specificreceptor protein in the cis-Golgi membrane recruits missorted ER proteins into retrograde COPI vesicles (seeFigure 17-16).

■ COPI vesicles also carry Golgi-resident proteins fromlater to earlier compartments in the Golgi stack.

■ Soluble and membrane proteins advance through theGolgi complex by cisternal progression, a nonvesicularprocess of anterograde transport.

Later Stages of the Secretory PathwayAs cargo proteins move from the cis face to the trans face ofthe Golgi complex by cisternal progression, modificationsto their oligosaccharide chains are carried out by Golgi-resident enzymes. The retrograde trafficking of COPI vesi-cles from later to earlier Golgi compartments maintainssufficient levels of these carbohydrate-modifying enzymesin their functional compartments. Eventually, properlyprocessed cargo proteins reach the trans-Golgi network, themost-distal Golgi compartment. Here they are sorted intovesicles for delivery to their final destination. In this sectionwe discuss the different kinds of vesicles that bud from thetrans-Golgi network, the mechanisms that segregate cargoproteins among them, and key processing events that occurlate in the secretory pathway. The transport steps mediatedby the major types of coated vesicles are summarized inFigure 17-18.

17.4

17.4 • Later Stages of the Secretory Pathway 719

Clathrin

COPI

COPICis-Golgi network

COPII

Rough ER

Late endosome

COPI

?

COPIIsubunits

Plasmamembrane

Trans-Golgi

Medial-Golgi

Cis-Golgi

5

3

4

2

1

� FIGURE 17-18 Involvement of the

three major types of coat proteins in

vesicular traffic in the secretory and

endocytic pathways. After formation ofvesicles by budding from a donor membrane,the coats depolymerize into their subunits,which are re-used to form additional transportvesicles. COPII vesicles ( ) mediate antero-grade transport from the rough ER to the cis-Golgi/cis-Golgi network. COPI vesicles ( )mediate retrograde transport within the Golgiand from the cis-Golgi/cis-Golgi network tothe rough ER. The coat proteins surroundingsecretory vesicles ( ) are not yetcharacterized; these vesicles carry secretedproteins and plasma-membrane proteins fromthe trans-Golgi network to the cell surface.Vesicles coated with clathrin (red) bud fromthe trans-Golgi network ( ) and from theplasma membrane ( ); after uncoating, these vesicles fuse with late endosomes. The coat on most clathrin vesicles containsadditional proteins not indicated here. Notethat secretory proteins move from the cis- totrans-Golgi by cisternal progression, which isnot mediated by vesicles. [See H. Pelham, 1997,Nature 389:17, and J. F. Presley et al., 1997, Nature389:81.]

54

3

2

1

Vesicles Coated with Clathrin and/or AdapterProteins Mediate Several Transport StepsThe best-characterized vesicles that bud from the trans-Golgi network (TGN) have a two-layered coat: an outer layer composed of the fibrous protein clathrin and an inner layer composed of adapter protein (AP) complexes.Purified clathrin molecules, which have a three-limbed shape, are called triskelions from the Greek for three-legged (Figure 17-19a). Each limb contains one clathrin heavy chain (180,000 MW) and one clathrin light chain

(≈35,000– 40,000 MW). Triskelions polymerize to form a polygonal lattice with an intrinsic curvature (Figure 17-19b). When clathrin polymerizes on a donor membrane,it does so in association with AP complexes, which assem-ble between the clathrin lattice and the membrane. Each APcomplex (340,000 MW) contains one copy each of four dif-ferent adapter subunit proteins. A specific association be-tween the globular domain at the end of each clathrin heavychain in a triskelion and one subunit of the AP complex bothpromotes the co-assembly of clathrin triskelions with APcomplexes and adds to the stability of the completed vesiclecoat (Figure 17-19c).

By binding to the cytosolic face of membrane proteins,adapter proteins determine which cargo proteins are specifi-cally included in (or excluded from) a budding transportvesicle. Each type of AP complex (e.g., AP1, AP2, AP3) andthe recently identified GGAs are composed of different,though related, proteins. Vesicles containing each complexhave been found to mediate specific transport steps (seeTable 17-1). All vesicles whose coats contain one of thesecomplexes utilize ARF to initiate coat assembly onto thedonor membrane. As discussed previously, ARF also initiatesassembly of COPI coats. The additional features of the mem-brane or protein factors that determine which type of coatwill assemble after ARF attachment are not well understoodat this time.

Vesicles that bud from the trans-Golgi network en routeto the lysosome by way of the late endosome have clathrincoats associated with either AP1 or GGA. Both AP1 andGGA bind to the cytosolic domain of cargo proteins in thedonor membrane, but the functional differences betweenvesicles that contain AP1 or GGA are unclear. Recent stud-ies have shown that membrane proteins containing a Tyr-X-X-� sequence, where X is any amino acid and � is a bulkyhydrophobic amino acid, are recruited into clathrin/AP1vesicles budding from the trans-Golgi network. This YXX�sorting signal interacts with one of the AP1 subunits in thevesicle coat. As we discuss in the next section, vesicles withclathrin/AP2 coats, which bud from the plasma membraneduring endocytosis, also can recognize the YXX� sortingsignal.

Some vesicles that bud from the trans-Golgi networkhave coats composed of the AP3 complex. These vesicles me-diate trafficking to the lysosome, but they appear to bypassthe late endosome and fuse directly with the lysosomal mem-brane. In certain types of cells, such AP3 vesicles mediateprotein transport to specialized storage compartments re-lated to the lysosome. For example, AP3 is required for de-livery of proteins to melanosomes, which contain the blackpigment melanin in skin cells, and to platelet storage vesi-cles in megakaryocytes, a large cell that fragments intodozens of platelets. Mice with mutations in either of two dif-ferent subunits of AP3 not only have abnormal skin pigmen-tation but also exhibit bleeding disorders. The latter occurbecause tears in blood vessels cannot be repaired withoutplatelets that contain normal storage vesicles.


(a) Triskelion structure (b) Assembly intermediate

Heavychain

Light chain

Binding site for assemblyparticles

(c)

▲ FIGURE 17-19 Structure of clathrin coats. (a) Aclathrin molecule, called a triskelion, is composed of threeheavy and three light chains. It has an intrinsic curvaturedue to the bend in the heavy chains. (b) The fibrousclathrin coat around vesicles is constructed of 36 clathrintriskelions. Depicted here is an intermediate in assemblyof a clathrin coat, containing 10 of the final 36 triskelions,which illustrates the intrinsic curvature and the packing ofclathrin triskelions. (c) Clathrin coats were formed in vitroby mixing purified clathrin heavy and light chains with AP2complexes in the absence of membranes. Cryoelectronmicrographs of more than 1000 assembled particles wereanalyzed by digital image processing to generate anaverage structural representation. The left image showsthe reconstructed structure of a complete particle withAP2 complexes packed into the interior of the clathrincage. In the right image, the AP2 complexes have beensubtracted to show only the assembled clathrin heavy andlight chains. [See B. Pishvaee and G. Payne, 1998, Cell 95:443.Part (c) from Corinne J. Smith, Department of Biological Sciences,University of Warwick.]

ME

DIA

C

ON

NE

CT

IO

NS

Vid

eo:B

irth

of a

Cla

thri

n C

oat

Dynamin Is Required for Pinching Off of Clathrin VesiclesA fundamental step in the formation of a transport vesiclethat we have not yet considered is how a vesicle bud ispinched off from the donor membrane. In the case ofclathrin/AP-coated vesicles, a cytosolic protein called dy-namin is essential for release of complete vesicles. At the laterstages of bud formation, dynamin polymerizes around theneck portion and then hydrolyzes GTP. The energy derivedfrom GTP hydrolysis is thought to drive “contraction” of dynamin around the vesicle neck until the vesicle pinches off(Figure 17-20). Interestingly, COPI and COPII vesicles ap-pear to pinch off from donor membranes without the aid ofa GTPase such as dynamin. At present this fundamental dif-ference in the process of pinching off among the differenttypes of vesicles is not understood.

Incubation of cell extracts with a nonhydrolyzable deriv-ative of GTP provides dramatic evidence for the importanceof dynamin in pinching off of clathrin/AP vesicles during en-docytosis. Such treatment leads to accumulation of clathrin-coated vesicle buds with excessively long necks that are

surrounded by polymeric dynamin but do not pinch off (Fig-ure 17-21). Likewise, cells expressing mutant forms of dy-namin that cannot bind GTP do not form clathrin-coatedvesicles, and instead accumulate similar long-necked vesiclebuds encased with polymerized dynamin.

As with COPI and COPII vesicles, clathrin/AP vesicles nor-mally lose their coat soon after their formation. CytosolicHsc70, a constitutive chaperone protein found in all eukaryoticcells, is thought to use energy derived from the hydrolysis ofATP to drive depolymerization of the clathrin coat into triske-lions. Uncoating not only releases triskelions for reuse in theformation of additional vesicles, but also exposes v-SNAREsfor use in fusion with target membranes. Conformationalchanges that occur when ARF switches from the GTP-bound toGDP-bound state are thought to regulate the timing of clathrincoat depolymerization. How the action of Hsc70 might be cou-pled to ARF switching is not well understood.

Mannose 6-Phosphate Residues Target SolubleProteins to LysosomesMost of the sorting signals that function in vesicular traf-ficking are short amino acid sequences in the targeted pro-tein. In contrast, the sorting signal that directs soluble


Clathrin-coated vesicle

Dynamin

GTP

GDP + Pi

APcomplex

Cytosolic face

Exoplasmic face

Fibrousclathrincoat

Solublecargoprotein

Integralcargo protein

Integralreceptorprotein

▲ FIGURE 17-20 Model for dynamin-mediated pinching off

of clathrin/AP-coated vesicles. After a vesicle bud forms,dynamin polymerizes over the neck. By a mechanism that is notwell understood, dynamin-catalyzed hydrolysis of GTP leads torelease of the vesicle from the donor membrane. Note thatmembrane proteins in the donor membrane are incorporated intovesicles by interacting with AP complexes in the coat. [Adaptedfrom K. Takel et al., 1995, Nature 374:186.]

▲ EXPERIMENTAL FIGURE 17-21 GTP hydrolysis by

dynamin is required for pinching off of clathrin-coated

vesicles in cell-free extracts. A preparation of nerve terminals,which undergo extensive endocytosis, was lysed by treatmentwith distilled water and incubated with GTP--S, anonhydrolyzable derivative of GTP. After sectioning, thepreparation was treated with gold-tagged anti-dynamin antibodyand viewed in the electron microscope. This image, which showsa long-necked clathrin/AP-coated bud with polymerized dynaminlining the neck, reveals that buds can form in the absence ofGTP hydrolysis, but vesicles cannot pinch off. The extensivepolymerization of dynamin that occurs in the presence of withGTP--S probably does not occur during the normal buddingprocess. [From K. Takel et al., 1995, Nature 374:186; courtesy of Pietro De Camilli.]


U

UDP-GlcNAc

GlcNAc phosphotransferase Catalytic site Recognition site

Phosphodiesterase

UMP

Lysosomal enzymeU

P P

P P

P

P

Recognition sequences

1 2

▲ FIGURE 17-22 Formation of mannose 6-phosphate (M6P)

residues that target soluble enzymes to lysosomes. The M6Presidues that direct proteins to lysosomes are generated in thecis-Golgi by two Golgi-resident enzymes. Step : An N-acetylglucosamine (GlcNAc) phosphotransferase transfers aphosphorylated GlcNAc group to carbon atom 6 of one or moremannose residues. Because only lysosomal enzymes contain

1

sequences (red) that are recognized and bound by this enzyme,phosphorylated GlcNAc groups are added specifically tolysosomal enzymes. Step : After release of a modified proteinfrom the phosphotransferase, a phosphodiesterase removes theGlcNAc group, leaving a phosphorylated mannose residue on thelysosomal enzyme. [See A. B. Cantor et al., 1992, J. Biol. Chem.267:23349, and S. Kornfeld, 1987, FASEB J. 1:462.]

2

Plasma membrane

Cytosol

Trans- Golgi

network

Exterior

Lysosome

Clathrin-coatedvesicle

Clathrin triskelions

APcomplex

receptor

Uncoated transportvesicle

Late

endosome

(low pH)

Clathrin-coatedbud

Clathrin-coated pit

Receptor-

mediated

endocytosis

Uncoatedendocyticvesicle

Clathrin-coated vesicle

1

2

3

4

8

7

6

5Recycling of

M6P receptor

Constitutive

secretion

2a

4a

P

P

P

P

P

P

P

lysosomal enzymes from the trans-Golgi network to the lateendosome is a carbohydrate residue, mannose 6-phosphate(M6P), which is formed in the cis-Golgi. The addition andinitial processing of one or more preformed N-linkedoligosaccharide precursors in the rough ER is the same forlysosomal enzymes as for membrane and secreted proteins,yielding core Man8(GlcNAc)2 chains (see Figure 16-18). Inthe cis-Golgi, the N-linked oligosaccharides present on mostlysosomal enzymes undergo a two-step reaction sequencethat generates M6P residues (Figure 17-22). The addition ofM6P residues to the oligosaccharide chains of soluble lyso-somal enzymes prevents these proteins from undergoing thefurther processing reactions characteristic of secreted andmembrane proteins (see Figure 17-3).

As shown in Figure 17-23, the segregation of M6P-bearinglysosomal enzymes from secreted and membrane proteins occurs in the trans-Golgi network. Here transmembranemannose 6-phosphate receptors bind the M6P residues onlysosome-destined proteins very tightly and specifically.Clathrin/AP1 vesicles containing the M6P receptor andbound lysosomal enzymes then bud from the trans-Golgi net-work, lose their coats, and subsequently fuse with the lateendosome by mechanisms described previously. BecauseM6P receptors can bind M6P at the slightly acidic pH (≈6.5)of the trans-Golgi network but not at a pH less than 6, thebound lysosomal enzymes are released within late endo-somes, which have an internal pH of 5.0–5.5. Furthermore,a phosphatase within late endosomes usually removes thephosphate from M6P residues on lysosomal enzymes, pre-venting any rebinding to the M6P receptor that might occurin spite of the low pH in endosomes. Vesicles budding fromlate endosomes recycle the M6P receptor back to the trans-

Golgi network or, on occasion, to the cell surface. Eventually,mature late endosomes fuse with lysosomes, delivering thelysosomal enzymes to their final destination.

The sorting of soluble lysosomal enzymes in the trans-Golgi network (see Figure 17-23, steps 1 – 4 ) shares many ofthe features of trafficking between the ER and cis-Golgi com-partments mediated by COPII and COPI vesicles. First, man-nose 6-phosphate acts as a sorting signal by interacting withthe luminal domain of a receptor protein in the donor mem-brane. Second, the membrane-embedded receptors with theirbound ligands are incorporated into the appropriate vesi-cles—in this case, AP1-containing clathrin vesicles—by interacting with the vesicle coat. Third, these transport vesiclesfuse only with one specific organelle, here the late endosome,as the result of interactions between specific v-SNAREs andt-SNAREs. And finally, intracellular transport receptors arerecycled after dissociating from their bound ligand.

Study of Lysosomal Storage Diseases RevealedKey Components of the Lysosomal Sorting Pathway

A group of genetic disorders, termed lysosomalstorage diseases, are caused by the absence of oneor more lysosomal enzymes. As a result, undi-

gested glycolipids and extracellular components that wouldnormally be degraded by lysosomal enzymes accumulate inlysosomes as large inclusions. I-cell disease is a particularlysevere type of lysosomal storage disease in which multipleenzymes are missing from the lysosomes. Cells from affectedindividuals lack the N-acetylglucosamine phosphotrans-ferase that is required for formation of M6P residues on lysosomal enzymes in the cis-Golgi (see Figure 17-22). Bio-chemical comparison of lysosomal enzymes from normal individuals with those from patients with I-cell disease ledto the initial discovery of mannose 6-phosphate as the lyso-somal sorting signal. Lacking the M6P sorting signal, thelysosomal enzymes in I-cell patients are secreted rather thanbeing sorted to and sequestered in lysosomes.

When fibroblasts from patients with I-cell disease aregrown in a medium containing lysosomal enzymes bearingM6P residues, the diseased cells acquire a nearly normal in-tracellular content of lysosomal enzymes. This finding indi-cates that the plasma membrane of these cells contain M6Preceptors, which can internalize extracellular phosphorylatedlysosomal enzymes by receptor-mediated endocytosis. Thisprocess, used by many cell-surface receptors to bring boundproteins or particles into the cell, is discussed in detail in thenext section. It is now known that even in normal cells, someM6P receptors are transported to the plasma membrane andsome phosphorylated lysosomal enzymes are secreted (seeFigure 17-23). The secreted enzymes can be retrieved by re-ceptor-mediated endocytosis and directed to lysosomes. Thispathway thus scavenges any lysosomal enzymes that escapethe usual M6P sorting pathway.


� FIGURE 17-23 Trafficking of soluble lysosomal enzymes

from the trans-Golgi network and cell surface to lysosomes.

Newly synthesized lysosomal enzymes, produced in the ER,acquire mannose 6-phosphate (M6P) residues in the cis-Golgi(see Figure 17-22). For simplicity, only one phosphorylatedoligosaccharide chain is depicted, although lysosomal enzymestypically have many such chains. In the trans-Golgi network,proteins that bear the M6P sorting signal interact with M6Preceptors in the membrane and thereby are directed intoclathrin/AP1 vesicles (step ). The coat surrounding releasedvesicles is rapidly depolymerized (step ), and the uncoatedtransport vesicles fuse with late endosomes (step ). After thephosphorylated enzymes dissociate from the M6P receptors andare dephosphorylated, late endosomes subsequently fuse with alysosome (step ). Note that coat proteins and M6P receptorsare recycled (steps and ), and some receptors aredelivered to the cell surface (step ). Phosphorylated lysosomalenzymes occasionally are sorted from the trans-Golgi to the cellsurface and secreted. These secreted enzymes can be retrievedby receptor-mediated endocytosis (steps – ), a process thatclosely parallels trafficking of lysosomal enzymes from the trans-Golgi network to lysosomes. [See G. Griffiths et al., 1988, Cell 52:329;S. Kornfeld, 1992, Ann. Rev. Biochem. 61:307; and G. Griffiths and J. Gruenberg, 1991, Trends Cell Biol. 1:5.]

86

54a2a

4

32

1

Hepatocytes from patients with I-cell disease contain anormal complement of lysosomal enzymes and no inclusions,even though these cells are defective in mannose phosphory-lation. This finding implies that hepatocytes (the most abun-dant type of liver cell) employ a M6P-independent pathwayfor sorting lysosomal enzymes. The nature of this pathway,which also may operate in other cells types, is unknown. ❚

Protein Aggregation in the Trans-Golgi MayFunction in Sorting Proteins to RegulatedSecretory VesiclesAs noted in the chapter introduction, all eukaryotic cells con-tinuously secrete certain proteins, a process commonly calledconstitutive secretion. Specialized secretory cells also storeother proteins in vesicles and secrete them only when trig-gered by a specific stimulus. One example of such regulatedsecretion occurs in pancreatic � cells, which store newlymade insulin in special secretory vesicles and secrete insulinin response to an elevation in blood glucose (see Figure 15-7).These and other secretory cells simultaneously utilize twodifferent types of vesicles to move proteins from the trans-Golgi network to the cell surface: regulated transport vesi-cles, often simply called secretory vesicles, and unregulatedtransport vesicles, also called constitutive secretory vesicles.

A common mechanism appears to sort regulated proteinsas diverse as ACTH (adrenocorticotropic hormone), insulin,and trypsinogen into regulated secretory vesicles. Evidence fora common mechanism comes from experiments in which re-combinant DNA techniques are used to induce the synthesisof insulin and trypsinogen in pituitary tumor cells already syn-thesizing ACTH. In these cells all three proteins segregate intothe same regulated secretory vesicles and are secreted togetherwhen a hormone binds to a receptor on the pituitary cells andcauses a rise in cytosolic Ca2�. Although these three proteinsshare no identical amino acid sequences that might serve as asorting sequence, they obviously have some common featurethat signals their incorporation into regulated secretory vesicles.

Morphologic evidence suggests that sorting into the reg-ulated pathway is controlled by selective protein aggregation.For instance, immature vesicles in this pathway—those thathave just budded from the trans-Golgi network—contain dif-fuse aggregates of secreted protein that are visible in the elec-tron microscope. These aggregates also are found in vesiclesthat are in the process of budding, indicating that proteinsdestined for regulated secretory vesicles selectively aggregatetogether before their incorporation into the vesicles.

Other studies have shown that regulated secretory vesiclesfrom mammalian secretory cells contain three proteins, chro-mogranin A, chromogranin B, and secretogranin II, that to-gether form aggregates when incubated at the ionic conditions(pH ≈6.5 and 1 mM Ca2�) thought to occur in the trans-Golginetwork; such aggregates do not form at the neutral pH of theER. The selective aggregation of regulated secreted proteins to-gether with chromogranin A, chromogranin B, or secre-togranin II could be the basis for sorting of these proteins into

regulated secretory vesicles. Secreted proteins that do not as-sociate with these proteins, and thus do not form aggregates,would be sorted into unregulated transport vesicles by default.

Some Proteins Undergo Proteolytic ProcessingAfter Leaving the Trans-GolgiFor some secretory proteins (e.g., growth hormone) and cer-tain viral membrane proteins (e.g., the VSV glycoprotein), re-moval of the N-terminal ER signal sequence from the nascentchain is the only known proteolytic cleavage required to con-vert the polypeptide to the mature, active species (see Figure16-6). However, some membrane and many soluble secretoryproteins initially are synthesized as relatively long-lived, in-active precursors, termed proproteins, that require furtherproteolytic processing to generate the mature, active pro-teins. Examples of proteins that undergo such processing aresoluble lysosomal enzymes, many membrane proteins suchas influenza hemagglutinin (HA), and secreted proteins suchas serum albumin, insulin, glucagon, and the yeast � matingfactor. In general, the proteolytic conversion of a proproteinto the corresponding mature protein occurs after the pro-protein has been sorted in the trans-Golgi network to appro-priate vesicles.

In the case of soluble lysosomal enzymes, the proproteinsare called proenzymes, which are sorted by the M6P receptoras catalytically inactive enzymes. In the late endosome orlysosome a proenzyme undergoes a proteolytic cleavage thatgenerates a smaller but enzymatically active polypeptide. Delaying the activation of lysosomal proenzymes until theyreach the lysosome prevents them from digesting macromol-ecules in earlier compartments of the secretory pathway.

Normally, mature vesicles carrying secreted proteins tothe cell surface are formed by fusion of several immature


� EXPERIMENTAL FIGURE 17-24 Proteolytic cleavage of

proinsulin occurs in secretory vesicles after they have

budded from the trans-Golgi network. Serial sections of theGolgi region of an insulin-secreting cell were stained with (a) amonoclonal antibody that recognizes proinsulin but not insulin or(b) a different antibody that recognizes insulin but not proinsulin.The antibodies, which were bound to electron-opaque goldparticles, appear as dark dots in these electron micrographs (seeFigure 5-51). Immature secretory vesicles (closed arrowheads)and vesicles budding from the trans-Golgi (arrows) stain with theproinsulin antibody but not with insulin antibody. These vesiclescontain diffuse protein aggregates that include proinsulin andother regulated secreted proteins. Mature vesicles (openarrowheads) stain with insulin antibody but not with proinsulinantibody and have a dense core of almost crystalline insulin.Since budding and immature secretory vesicles contain proinsulin(not insulin), the proteolytic conversion of proinsulin to insulinmust take place in these vesicles after they bud from the trans-Golgi network. The inset in (a) shows a proinsulin-rich secretoryvesicle surrounded by a protein coat (dashed line). [From L. Orci etal., 1987, Cell 49:865; courtesy of L. Orci.]

ones containing proprotein. Proteolytic cleavage of propro-teins, such as proinsulin, occurs in vesicles after they moveaway from the trans-Golgi network (Figure 17-24). The pro-proteins of most constitutively secreted proteins (e.g., albu-min) are cleaved only once at a site C-terminal to a dibasic

recognition sequence such as Arg-Arg or Lys-Arg (Figure 17-25a). Proteolytic processing of proteins whose secretion isregulated generally entails additional cleavages. In the case ofproinsulin, multiple cleavages of the single polypeptide chainyields the N-terminal B chain and the C-terminal A chain ofmature insulin, which are linked by disulfide bonds, and thecentral C peptide, which is lost and subsequently degraded(Figure 17-25b).

The breakthrough in identifying the proteases responsiblefor such processing of secreted proteins came from analysis


Arg Arg

(a) Constitutive secreted proteins

NH3+

NH3+

NH3+

NH3+ COO−

COO−

COO−

COO−

Proalbumin

(b) Regulated secreted proteins

Proinsulin

Albumin

Insulin

Arg Arg

Arg Arg

Arg Arg

Lys Arg

Lys Arg

AB

Furin endoprotease

PC3 endoprotease

Carboxypeptidase

PC2 endoprotease

Arg Arg

S

S S

SS S

AB

S

S S

SS S

AB C

C

S

S S

SS S

▲ FIGURE 17-25 Proteolytic processing of proproteins in

the constitutive and regulated secretion pathways. Theprocessing of proalbumin and proinsulin is typical of theconstitutive and regulated pathways, respectively. Theendoproteases that function in such processing cleave C-terminalto sequences of two consecutive basic amino acids. (a) Theendoprotease furin acts on the precursors of constitutivesecreted proteins. (b) Two endoproteases, PC2 and PC3, act onthe precursors of regulated secreted proteins. The finalprocessing of many such proteins is catalyzed by acarboxypeptidase that sequentially removes two basic amino acidresidues at the C-terminus of a polypeptide. [See D. Steiner et al.,1992, J. Biol. Chem. 267:23435.]

0.2 �m

G

(a) Proinsulin antibody

0.5 �m

G

(b) Insulin antibody

of yeast with a mutation in the KEX2 gene. These mutantcells synthesized the precursor of the � mating factor butcould not proteolytically process it to the functional form,and thus were unable to mate with cells of the opposite mat-ing type (see Figure 22-13). The wild-type KEX2 gene en-codes an endoprotease that cleaves the �-factor precursor ata site C-terminal to Arg-Arg and Lys-Arg residues. Using theKEX2 gene as a DNA probe, researchers were able to clonea family of mammalian endoproteases, all of which cleave aprotein chain on the C-terminal side of an Arg-Arg or Lys-Arg sequence. One, called furin, is found in all mammaliancells; it processes proteins such as albumin that are secretedby the continuous pathway. In contrast, the PC2 and PC3endoproteases are found only in cells that exhibit regulatedsecretion; these enzymes are localized to regulated secretoryvesicles and proteolytically cleave the precursors of manyhormones at specific sites.

Several Pathways Sort Membrane Proteins to theApical or Basolateral Region of Polarized CellsThe plasma membrane of polarized epithelial cells is di-vided into two domains, apical and basolateral; tight junc-tions located between the two domains prevent themovement of plasma-membrane proteins between the do-mains (see Figure 6-5). Several sorting mechanisms directnewly synthesized membrane proteins to either the apicalor basolateral domain of epithelial cells, and any one pro-tein may be sorted by more than one mechanism. Althoughthese sorting mechanisms are understood in general terms,the molecular signals underlying the vesicle-mediated trans-port of membrane proteins in polarized cells are not yet

known. As a result of this sorting and the restriction onprotein movement within the plasma membrane due totight junctions, distinct sets of proteins are found in the api-cal or basolateral domain. This preferential localization ofcertain transport proteins is critical to a variety of impor-tant physiological functions, such as absorption of nutri-ents from the intestinal lumen and acidification of thestomach lumen (see Figures 7-27 and 7-28).

Microscopic and cell-fractionation studies indicate thatproteins destined for either the apical or the basolateralmembranes are initially located together within the mem-branes of the trans-Golgi network. In some cases, proteinsdestined for the apical membrane are sorted into their owntransport vesicles that bud from the trans-Golgi network andthen move to the apical region, whereas proteins destined forthe basolateral membrane are sorted into other vesicles thatmove to the basolateral region. The different vesicle typescan be distinguished by their protein constituents, includingdistinct Rab and v-SNARE proteins, which apparently targetthem to the appropriate plasma-membrane domain. In thismechanism, segregation of proteins destined for either theapical or basolateral membranes occurs as cargo proteins areincorporated into particular types of vesicles budding fromthe trans-Golgi network.

Such direct basolateral-apical sorting has been investi-gated in cultured Madin-Darby canine kidney (MDCK)cells, a line of cultured polarized epithelial cells (see Figure6-6). In MDCK cells infected with the influenza virus, prog-eny viruses bud only from the apical membrane, whereas incells infected with vesicular stomatitis virus (VSV), progenyviruses bud only from the basolateral membrane. This dif-ference occurs because the HA glycoprotein of influenza


Influenza virus HA glycoprotein

GPI anchor

Apical protein

Tightjunction

Trans-Golginetwork

Recycling

Basolateralplasma membrane

Clathrin-coated pit

Basolateralsorting

Direct apicalsorting

Transcytosis

Apical plasmamembrane

VSV G glycoprotein

Endocytosis

� FIGURE 17-26 Sorting of proteins

destined for the apical and basolateral

plasma membranes of polarized cells.

When cultured MDCK cells are infectedsimultaneously with VSV and influenza virus,the VSV G glycoprotein (purple) is foundonly on the basolateral membrane, whereasthe influenza HA glycoprotein (green) isfound only on the apical membrane. Somecellular proteins (orange circle), especiallythose with a GPI anchor, are likewise sorteddirectly to the apical membrane and othersto the basolateral membrane (not shown)via specific transport vesicles that bud fromthe trans-Golgi network. In certain polarizedcells, some apical and basolateral proteinsare transported together to the basolateralsurface; the apical proteins (orange oval)then move selectively, by endocytosis andtranscytosis, to the apical membrane. [AfterK. Simons and A. Wandinger-Ness, 1990, Cell62:207, and K. Mostov et al., 1992, J. Cell Biol.116:577.]

ME

DIA

C

ON

NE

CT

IO

NS

Vid

eo:S

egre

gatio

n of

Api

cal a

nd B

asol

ater

al C

argo

in th

e G

olgi

of L

ive

Cel

ls

virus is transported from the Golgi complex exclusively tothe apical membrane, and the VSV G protein is transportedonly to the basolateral membrane (Figure 17-26). Further-more, when the gene encoding HA protein is introducedinto uninfected cells by recombinant DNA techniques, allthe expressed HA accumulates in the apical membrane, in-dicating that the sorting signal resides in the HA glycopro-tein itself and not in other viral proteins produced duringviral infection.

Among the cellular proteins that undergo similar apical-basolateral sorting in the Golgi are those with a glyco-sylphosphatidylinositol (GPI) membrane anchor. In MDCKcells and most other types of epithelial cells, GPI-anchoredproteins are targeted to the apical membrane. In membranesGPI-anchored proteins are clustered into lipid rafts, whichare rich in sphingolipids (see Figure 5-10). This finding sug-gests that lipid rafts are localized to the apical membranealong with proteins that preferentially partition into themin many cells. However, the GPI anchor is not an apical sort-ing signal in all polarized cells; in thyroid cells, for example,GPI-anchored proteins are targeted to the basolateral mem-brane. Other than GPI anchors no unique sequences havebeen identified that are both necessary and sufficient to tar-get proteins to either the apical or basolateral domain. In-stead, each membrane protein may contain multiple sortingsignals, any one of which can target it to the appropriateplasma-membrane domain. The identification of such com-plex signals and of the vesicle coat proteins that recognizethem is currently being pursued for a number of differentproteins that are sorted to specific plasma-membrane do-mains of polarized epithelial cells.

Another mechanism for sorting apical and basolateralproteins, also illustrated in Figure 17-26, operates in hepa-tocytes. The basolateral membranes of hepatocytes face theblood (as in intestinal epithelial cells), and the apical mem-branes line the small intercellular channels into which bile issecreted. In hepatocytes, newly made apical and basolateralproteins are first transported in vesicles from the trans-Golginetwork to the basolateral region and incorporated into theplasma membrane by exocytosis (i.e., fusion of the vesiclemembrane with the plasma membrane). From there, both ba-solateral and apical proteins are endocytosed in the samevesicles, but then their paths diverge. The endocytosed ba-solateral proteins are sorted into transport vesicles that re-cycle them to the basolateral membrane. In contrast, theapically destined endocytosed proteins are sorted into trans-port vesicles that move across the cell and fuse with the api-cal membrane, a process called transcytosis. As discussed inthe next section, transcytosis also is used to move extracel-lular materials from one side of an epithelium to another.Even in epithelial cells, such as MDCK cells, in which apical-basolateral protein sorting occurs in the Golgi, transcytosismay provide a “fail-safe” sorting mechanism. That is, an api-cal protein sorted incorrectly to the basolateral membranewould be subjected to endocytosis and then correctly deliv-ered to the apical membrane.


Protein Sorting and Processing in Later Stages of the Secretory Pathway

■ The trans-Golgi network (TGN) is a major branch pointin the secretory pathway where soluble secreted proteins,lysosomal proteins, and in some cells membrane proteinsdestined for the basolateral or apical plasma membrane aresegregated into different transport vesicles.

■ Many vesicles that bud from the trans-Golgi network aswell as endocytic vesicles bear a coat composed of AP(adapter protein) complexes and clathrin (see Figure 17-19).

■ Pinching off of clathrin-coated vesicles requires dy-namin, which forms a collar around the neck of the vesi-cle bud and hydrolyzes GTP (see Figure 17-20).

■ Soluble enzymes destined for lysosomes are modified inthe cis-Golgi yielding multiple mannose 6-phosphate(M6P) residues on their oligosaccharide chains.

■ M6P receptors in the membrane of the trans-Golgi net-work bind proteins bearing M6P residues and direct theirtransfer to late endosomes, where receptors and their lig-and proteins dissociate. The receptors then are recycled tothe Golgi or plasma membrane, and the lysosomal enzymesare delivered to lysosomes (see Figure 17-23).

■ Regulated secreted proteins are concentrated and storedin secretory vesicles to await a neural or hormonal signalfor exocytosis. Protein aggregation within the trans-Golginetwork may play a role in sorting secreted proteins to theregulated pathway.

■ Many proteins transported through the secretory path-way undergo post-Golgi proteolytic cleavages that yield themature, active proteins. Generally, proteolytic maturationcan occur in vesicles carrying proteins from the trans-Golginetwork to the cell surface, in the late endosome, or in thelysosomal.

■ In polarized epithelial cells, membrane proteins destinedfor the apical or basolateral domains of the plasma mem-brane are sorted in the trans-Golgi network into differenttransport vesicles (see Figure 17-26). The GPI anchor isthe only apical-basolateral sorting signal identified so far.

■ In hepatocytes and some other polarized cells, allplasma-membrane proteins are directed first to the baso-lateral membrane. Apically destined proteins then are en-docytosed and moved across the cell to the apical mem-brane (transcytosis).

Receptor-Mediated Endocytosisand the Sorting of Internalized ProteinsIn previous sections we have explored the main pathwayswhereby secretory and membrane proteins synthesized on

17.5

17.5 • Receptor-Mediated Endocytosis and the Sorting of Internalized Proteins 727

the rough ER are delivered to the cell surface or other desti-nations. Cells also can internalize materials from their sur-roundings and sort these to particular destinations. A fewcell types (e.g., macrophages) can take up whole bacteria andother large particles by phagocytosis, a nonselective actin-mediated process in which extensions of the plasma mem-brane envelop the ingested material, forming large vesiclescalled phagosomes (see Figure 5-20). In contrast, all eukary-otic cells continually engage in endocytosis, a process inwhich a small region of the plasma membrane invaginatesto form a membrane-limited vesicle about 0.05–0.1 �m indiameter. In one form of endocytosis, called pinocytosis,small droplets of extracellular fluid and any material dis-solved in it are nonspecifically taken up. Our focus in thissection, however, is on receptor-mediated endocytosis inwhich a specific receptor on the cell surface binds tightly toan extracellular macromolecular ligand that it recognizes; theplasma-membrane region containing the receptor-ligandcomplex then buds inward and pinches off, becoming atransport vesicle.

Among the common macromolecules that vertebrate cellsinternalize by receptor-mediated endocytosis are cholesterol-containing particles called low-density lipoprotein (LDL); theiron-binding protein transferrin; many protein hormones(e.g., insulin); and certain glycoproteins. Receptor-mediatedendocytosis of such ligands generally occurs via clathrin/AP2-

coated pits and vesicles in a process similar to the packag-ing of lysosomal enzymes by mannose 6-phosphate (M6P)in the trans-Golgi network (see Figure 17-23). As noted ear-lier, some M6P receptors are found on the cell surface, andthese participate in the receptor-mediated endocytosis oflysosomal enzymes that are secreted. In general, transmem-brane receptor proteins that function in the uptake of extra-cellular ligands are internalized from the cell surface duringendocytosis and are then sorted and recycled back to the cellsurface, much like the recycling of M6P receptors to theplasma membrane and trans-Golgi. The rate at which a lig-and is internalized is limited by the amount of its correspon-ding receptor on the cell surface.

Clathrin/AP2 pits make up about 2 percent of the surfaceof cells such as hepatocytes and fibroblasts. Many internal-ized ligands have been observed in these pits and vesicles,which are thought to function as intermediates in the endo-cytosis of most (though not all) ligands bound to cell-surfacereceptors (Figure 17-27). Some receptors are clustered overclathrin-coated pits even in the absence of ligand. Other re-ceptors diffuse freely in the plane of the plasma membranebut undergo a conformational change when binding to lig-and, so that when the receptor-ligand complex diffuses intoa clathrin-coated pit, it is retained there. Two or more typesof receptor-bound ligands, such as LDL and transferrin, canbe seen in the same coated pit or vesicle.


LDL-ferritin

Clathrin-coated pitLDL-ferritin

(a)

(c) (d)

(b)

0.2 �m

� EXPERIMENTAL FIGURE 17-27The initial stages of receptor-mediated

endocytosis of low-density lipoprotein

(LDL) particles are revealed by electron

microscopy. Cultured human fibroblastswere incubated in a medium containingLDL particles covalently linked to theelectron-dense, iron-containing proteinferritin; each small iron particle in ferritin is visible as a small dot under the electronmicroscope. Cells initially were incubated at4 °C; at this temperature LDL can bind to itsreceptor but internalization does not occur.After excess LDL not bound to the cellswas washed away, the cells were warmedto 37 °C and then prepared for microscopyat periodic intervals. (a) A coated pit,showing the clathrin coat on the inner(cytosolic) surface of the pit, soon after thetemperature was raised. (b) A pit containingLDL apparently closing on itself to form acoated vesicle. (c) A coated vesiclecontaining ferritin-tagged LDL particles. (d) Ferritin-tagged LDL particles in a smooth-surfaced early endosome 6 minutes afterinternalization began. [Photographs courtesy ofR. Anderson. Reprinted by permission from J. Goldstein et al., Nature 279:679. Copyright1979, Macmillan Journals Limited. See also M. S.Brown and J. Goldstein, 1986, Science 232:34.]

Receptors for Low-Density Lipoprotein and OtherLigands Contain Sorting Signals That TargetThem for Endocytosis

As will be discussed in detail in the next chapter, low-density lipoprotein (LDL) is one of several complexes thatcarry cholesterol through the bloodstream (see Figure 17-28).A LDL particle, a sphere 20–25 nm in diameter, has an outerphospholipid shell containing a single molecule of a large

protein known as apoB-100; the core of a particle is packedwith cholesterol in the form of cholesteryl esters (see Figure18-12). Most mammalian cells produce cell-surface receptorsthat specifically bind to apoB-100 and internalize LDL par-ticles by receptor-mediated endocytosis. After endocytosis,the LDL particles are transported to lysosomes via the en-docytic pathway and then are degraded by lysosomal hydro-lases. LDL receptors, which dissociate from their ligands inthe late endosome, recycle to the cell surface.


LDL receptor

LDL particle

pH 5.0

Plasma membrane

Coated

vesicle

Amino acids

Cholesterol

Fatty acids

Lysosome

At neutral pH, ligand-binding arm is free to bind another

LDL particle

Late endosome

AP2 complex

Clathrin

Earlyendosome

Coated

pit

Phospholipid monolayer

ApoB protein

2

5

4

3

1

▲ FIGURE 17-28 Endocytic pathway for internalizing low-

density lipoprotein (LDL). Step : Cell-surface LDL receptorsbind to an apoB protein embedded in the phospholipid outerlayer of LDL particles. Interaction between the NPXY sortingsignal in the cytosolic tail of the LDL receptor and the AP2complex incorporates the receptor-ligand complex into formingendocytic vesicles. Step : Clathrin-coated pits (or buds)containing receptor-LDL complexes are pinched off by the samedynamin-mediated mechanism used to form clathrin/AP1 vesicleson the trans-Golgi network (see Figure 17-20). Step : After thevesicle coat is shed, the uncoated endocytic vesicle (early

3

2

1endosome) fuses with the late endosome. The acidic pH in thiscompartment causes a conformational change in the LDLreceptor that leads to release of the bound LDL particle. Step :The late endosome fuses with the lysosome, and the proteinsand lipids of the free LDL particle are broken down to theirconstituent parts by enzymes in the lysosome. Step : The LDLreceptor recycles to the cell surface where at the neutral pH ofthe exterior medium the receptor undergoes a conformationalchange so that it can bind another LDL particle. [See M. S. Brownand J. L. Goldstein, 1986, Science 232:34, and G. Rudenko et al., 2002,Science 298:2353.]

5

4

Studies of the inherited disorder familial hypercholes-terolemia led to discovery of the LDL receptor and the initialunderstanding of the endocytic pathway. An individual withthis disorder produces one of several mutant forms of theLDL receptor, causing impaired endocytosis of LDL and highserum levels of cholesterol (Chapter 18). The major featuresof the LDL endocytic pathway as currently understood are depicted in Figure 17-28. The LDL receptor is an 839-residue glycoprotein with a single transmembrane segment; ithas a short C-terminal cytosolic segment and a long N-terminal exoplasmic segment that contains a �-propeller domain and a ligand-binding domain. Seven cysteine-richimperfect repeats form the ligand-binding domain, which in-teracts with the apoB-100 molecule in a LDL particle.

Mutant receptors from some individuals with fa-milial hypercholesterolemia bind LDL normally,but the LDL-receptor complex cannot be internal-

ized by the cell and is distributed evenly over the cell surfacerather than being confined to clathrin/AP2-coated pits. In in-dividuals with this type of defect, plasma-membrane recep-tors for other ligands are internalized normally, but themutant LDL receptor apparently is not recruited into coatedpits. Analysis of this mutant receptor and other mutant LDLreceptors generated experimentally and expressed in fibrob-lasts identified a four-residue motif in the cytosolic segmentof the receptor that is crucial for its internalization: Asn-Pro-X-Tyr where X can be any amino acid. This NPXY sortingsignal binds to the AP2 complex, linking the clathrin/AP2coat to the cytosolic segment of the LDL receptor in form-ing coated pits. A mutation in any of the conserved residuesof the NPXY signal will abolish the ability of the LDL re-ceptor to be incorporated into coated pits.

A small number of individuals who exhibit the usualsymptoms associated with familial hypercholesterolemiaproduce normal LDL receptors. In these individuals, the geneencoding the AP2 subunit protein that binds the NPXY sort-ing signal is defective. As a result, LDL receptors are not in-corporated into clathrin/AP2 vesicles and endocytosis ofLDL particles is compromised. Analysis of patients with thisgenetic disorder highlights the importance of adapter pro-teins in protein trafficking mediated by clathrin vesicles. ❚

Mutational studies have shown that other cell-surface re-ceptors can be directed into forming clathrin/AP2 pits by adifferent sorting signal: Tyr-X-X-�, where X can be anyamino acid and � is a bulky hydrophobic amino acid. ThisYXX� sorting signal in the cytosolic segment of a receptorprotein binds to a specific cleft in the �2 subunit of the AP2complex. Because the tyrosine and � residues mediate thisbinding, a mutation in either one reduces or abolishes theability of the receptor to be incorporated into clathrin/AP2-coated pits. Moreover, if influenza HA protein, which is notnormally endocytosed, is genetically engineered to containthis four-residue sequence in its cytosolic domain, the mutantHA is internalized. Recall from our earlier discussion that

this same sorting signal recruits membrane proteins intoclathrin/AP1 vesicles that bud from the trans-Golgi networkby binding to the µ1 subunit of AP1 (see Table 17-2). All theseobservations indicate that YXX� is a widely used signal forsorting membrane proteins to clathrin-coated vesicles.

In some cell-surface proteins, however, other sequences(e.g., Leu-Leu) or covalently linked ubiquitin molecules sig-nal endocytosis. Among the proteins associated with clathrin/AP2 vesicles, several contain domains that specifically bindto ubiquitin, and it has been hypothesized that these vesicle-associated proteins mediate the selective incorporation ofubiquitinated membrane proteins into endocytic vesicles. Asdescribed later, the ubiquitin tag on endocytosed membraneproteins is also recognized at a later stage in the endocyticpathway and plays a role in delivering these proteins into theinterior of the lysosome where they are degraded.

The Acidic pH of Late Endosomes Causes MostReceptor-Ligand Complexes to DissociateThe overall rate of endocytic internalization of the plasmamembrane is quite high; cultured fibroblasts regularly inter-nalize 50 percent of their cell-surface proteins and phospho-lipids each hour. Most cell-surface receptors that undergoendocytosis will repeatedly deposit their ligands within the


0.2 �m

Ligand inlumen

Receptors invesicleextensions

▲ EXPERIMENTAL FIGURE 17-29 Electron microscopy

demonstrates that endocytosed receptor-ligand complexes

dissociate in late endosomes. Liver cells were perfused with anasialoglycoprotein ligand and then were fixed and sectioned forelectron microscopy. The sections were stained with receptor-specific antibodies, tagged with gold particles 8 nm in diameter,to localize the receptor and with asialoglycoprotein-specificantibody, linked to gold particles 5 nm in diameter, to localize theligand (see Figure 5-51). As seen in this electron micrograph of alate endosome, the ligand (smaller dark grains) is localized in thevesicle lumen and the asialoglycoprotein receptor (larger darkgrains) is localized in the tubular extensions budding off from thevesicle. [Courtesy of H. J. Geuze. Copyright 1983, M.I.T. See H. J. Geuzeet al., 1983, Cell 32:277.]

cell and then recycle to the plasma membrane, once again tomediate internalization of ligand molecules. For instance, theLDL receptor makes one round trip into and out of the cellevery 10–20 minutes, for a total of several hundred trips inits 20-hour life span.

Internalized receptor-ligand complexes commonly followthe pathway depicted for the M6P receptor in Figure 17-23and the LDL receptor in Figure 17-28. Endocytosed cell-surface receptors typically dissociate from their ligands withinlate endosomes, which appear as spherical vesicles with tubu-lar branching membranes located a few micrometers from thecell surface. The original experiments that defined the late en-dosome sorting vesicle utilized the asialoglycoprotein receptor.This liver-specific protein mediates the binding and internal-ization of abnormal glycoproteins whose oligosaccharides ter-minate in galactose rather than the normal sialic acid, hencethe name asialoglycoprotein. Electron microscopy of liver cellsperfused with asialoglycoprotein reveal that 5–10 minutesafter internalization, ligand molecules are found in the lumenof late endosomes, while the tubular membrane extensions arerich in receptor and rarely contain ligand (Figure 17-29).These findings indicate that the late endosome is the organellein which receptors and ligands are uncoupled.

The dissociation of receptor-ligand complexes in late en-dosomes occurs not only in the endocytic pathway but alsoin the delivery of soluble lysosomal enzymes via the secretorypathway (see Figure 17-23). As discussed in Chapter 7, themembranes of late endosomes and lysosomes contain V-class

proton pumps that act in concert with Cl channels to acid-ify the vesicle lumen (see Figure 7-10). Most receptors, in-cluding the M6P receptor and cell-surface receptors for LDLparticles and asialoglycoprotein, bind their ligands tightly atneutral pH but release their ligands if the pH is lowered to6.0 or below. The late endosome is the first vesicle encoun-tered by receptor-ligand complexes whose luminal pH is suf-ficiently acidic to promote dissociation of most endocytosedreceptors from their tightly bound ligands.

The mechanism by which the LDL receptor releases boundLDL particles is now understood in detail (Figure 17-30). Atthe endosomal pH of 5.0–5.5, histidine residues in the �-propeller domain of the receptor become protonated, form-ing a site that can bind with high affinity to the negativelycharged repeats in the ligand-binding domain. This intramol-ecular interaction sequesters the repeats in a conformation thatcannot simultaneously bind to apoB-100, thus causing releaseof the bound LDL particle.

The Endocytic Pathway Delivers Iron to CellsWithout Dissociation of Receptor-TransferrinComplex in EndosomesAn exception to the general theme of pH-dependent recep-tor-ligand dissociation in the late endosome occurs in the en-docytic pathway that delivers transferrin-bound iron to cells.A major glycoprotein in the blood, transferrin transportsiron to all tissue cells from the liver (the main site of iron


LDL receptor

LDL particle

Ligand-bindingarm (R1–R7)

β-propellerdomain

NPXY sorting signal

Cholesterol esters

Phospholipidmonolayer

ApoB protein

Cell surface [pH ~7.0]

Ligand- binding arm

β-propellerdomain

R2

R3

R4

R5

R6

R7

Released LDL particleEndosome [pH ~5]

Surface ofβ-propellerdomain becomespositively charged, and then bindsto the ligand-binding arm

▲ FIGURE 17-30 Model for pH-dependent binding of LDL particles

by the LDL receptor. Schematic depiction of LDL receptor at neutral pHfound at the cell surface (left) and at acidic pH found in the interior of thelate endosome (right). At the cell surface, apoB-100 on the surface of a LDL particle binds tightly to the receptor. Of the seven repeats (R1–R7) inthe ligand-binding arm, R4 and R5 appear to be most critical for LDL binding.Within the endosome, histidine residues in the �-propeller domain of theLDL receptor become protonated. The positively charged propeller can bindwith high affinity to the ligand-binding arm, which contains negativelycharged residues, causing release of the LDL particle. Experimental electrondensity and C� trace model of the extracellular region of the LDL receptor at pH 5.3 based on X-ray crystallographic analysis. In this conformation,extensive hydrophobic and ionic interactions occur between the � propellerand the R4 and R5 repeats. Red spheres represent Ca2� ions. [Part (b) from G. Rudenko et al., 2002, Science 298:2353.]

storage in the body) and from the intestine (the site of ironabsorption). The iron-free form, apotransferrin, binds twoFe3� ions very tightly to form ferrotransferrin. All mam-malian cells contain cell-surface transferrin receptors thatavidly bind ferrotransferrin at neutral pH, after which the re-ceptor-bound ferrotransferrin is subjected to endocytosis.Like the components of a LDL particle, the two bound Fe3�

atoms remain in the cell, but the apotransferrin part of theligand does not dissociate from the receptor and is secretedfrom the cell within minutes after being endocytosed.

Although apotransferrin remains bound to the transfer-rin receptor at the low pH of late endosomes, changes in pHare critical to functioning of the transferrin endocytic path-way. At a pH below 6.0, the two bound Fe3� atoms dissoci-ate from ferrotransferrin, are reduced to Fe2� by anunknown mechanism, and then are exported into the cytosolby an endosomal transporter specific for divalent metal ions.The receptor-apotransferrin complex remaining after disso-ciation of the iron atoms is recycled back to the cell surface.Although apotransferrin binds tightly to its receptor at a pHof 5.0 or 6.0, it does not bind at neutral pH. Hence thebound apotransferrin dissociates from the transferrin recep-tor when the recycling vesicles fuse with the plasma mem-brane and the receptor-ligand complex encounters the

neutral pH of the extracellular interstitial fluid or growthmedium. The recycled receptor is then free to bind anothermolecule of ferrotransferrin, and the released apotransferrinis carried in the bloodstream to the liver or intestine to bereloaded with iron.

Specialized Vesicles Deliver Cell Components to the Lysosome for DegradationThe major function of lysosomes is to degrade extracellularmaterials taken up by the cell and intracellular componentsunder certain conditions. Materials to be degraded must bedelivered to the lumen of the lysosome where the variousdegradative enzymes reside. As just discussed, endocytosedligands (e.g., LDL particles) that dissociate from their recep-tors in the late endosome subsequently enter the lysosomallumen when the membrane of the late endosome fuses withthe membrane of the lysosome (see Figure 17-28). Likewise,phagosomes carrying bacteria or other particulate matter canfuse with lysosomes, releasing their contents into the lumenfor degradation. However, the delivery of endocytosed mem-brane proteins and of cytoplasmic materials to lysosomes fordegradation poses special problems and involves two un-usual types of vesicles.


Autophagicvesicle

Lysosome

Late endosome/multivesicular body

Earlyendosome

Peroxisome

Transportvesicle

Late endosome

Autophagic pathway

Multivesicular endosomalpathway

1

2

32

1

▲ FIGURE 17-31 Delivery of plasma-membrane proteins

and cytoplasmic components to the lysosomal interior for

degradation. Left: Early endosomes carrying endocytosedplasma-membrane proteins (blue) and vesicles carrying lysosomal membrane proteins (red) from the trans-Golgi networkfuse with the late endosome, transferring their membraneproteins to the endosomal membrane (step ). Proteins to bedegraded are incorporated into vesicles that bud into the interiorof the late endosome, eventually forming a multivesicularendosome containing many such internal vesicles (step ).Fusion of a multivesicular endosome directly with a lysosomereleases the internal vesicles into the lumen of the lysosomewhere they can be degraded (step ). Because proton pumps3

2

1

and other lysomal membrane proteins normally are notincorporated into internal endosomal vesicles, they are deliveredto the lysosomal membrane and are protected from degradation.Right: In the autophagic pathway, a cup-shaped structure forms around portions of the cytosol or an organelle such as aperoxisome, as shown here. Continued addition of membraneeventually leads to the formation of an autophagic vesicle that envelopes its contents by two complete membranes (step ). Fusion of the outer membrane with the membrane ofa lysosome releases a single-layer vesicle and its contents intothe lysosome interior (step ). [See F. Reggiori and D. J. Klionsky,2002, Eukaryot. Cell 1:11, and D. J. Katzmann et al., 2002, Nature Rev.Mol. Cell Biol. 3:893.]

2

1

Multivesicular Endosomes Resident lysosomal proteins,such as V-class proton pumps and other lysosomal mem-brane proteins, can carry out their functions and remain inthe lysosomal membrane where they are protected fromdegradation by the soluble hydrolytic enzymes in the lumen.Such proteins are delivered to the lysosomal membrane bytransport vesicles that bud from the trans-Golgi network bythe same basic mechanisms described in earlier sections. Incontrast, endocytosed membrane proteins to be degraded aretransferred in their entirety to the interior of the lysosome bya specialized delivery mechanism. Lysosomal degradation ofcell-surface receptors for extracellular signaling molecules isa common mechanism for controlling the sensitivity of cellsto such signals (Chapter 13). Receptors that become dam-aged also are targeted for lysosomal degradation.

Early evidence that membranes can be delivered to thelumen of compartments came from electron micrographsshowing membrane vesicles and fragments of membraneswithin endosomes and lysosomes (see Figure 5-20c). Paral-lel experiments in yeast revealed that endocytosed receptorproteins targeted to the vacuole (the yeast organelle equiva-lent to the lysosome) were primarily associated with mem-brane fragments and small vesicles within the interior of thevacuole rather than with the vacuole surface membrane.

These observations suggest that endocytosed membraneproteins can be incorporated into specialized vesicles thatform at the endosomal membrane (Figure 17-31, left). Al-though these vesicles are similar in size and appearance totransport vesicles, they differ topologically. Transport vesiclesbud outward from the surface of a donor organelle into thecytosol, whereas vesicles within the endosome bud inwardfrom the surface into the lumen (away from the cytosol). Mature endosomes containing numerous vesicles in their in-terior are usually called multivesicular endosomes (or bodies).Eventually the surface membrane of a multivesicular endo-some fuses with the membrane of a lysosome, thereby deliv-ering its internal vesicles and the membrane proteins theycontain into the lysosome interior for degradation. Thus thesorting of proteins in the endosomal membrane determineswhich ones will remain on the lysosome surface (e.g., pumpsand transporters) and which ones will be incorporated intointernal vesicles and ultimately degraded in lysosomes.

Autophagic Vesicles The delivery of bulk amounts of cy-tosol or entire organelles to lysosomes and their subsequentdegradation is known as autophagy (“eating oneself”). Au-tophagy is often a regulated process and is typically inducedin cells placed under conditions of starvation or other typesof stress, allowing the cell to recycle macromolecules for useas nutrients.

The autophagic pathway begins with the formation of aflattened double-membraned cup-shaped structure (Figure17-31, right). This structure can grow by vesicle fusion andeventually seals to form an autophagic vesicle that envelopsa region of the cytosol or an entire organelle (e.g., peroxi-some, mitochondrion). Unknown at this time is the origin

of the membranes that form the initial cup-shaped organelleand the vesicles that are added to it, but the endosome itselfis a likely candidate. The outer membrane of an autophagicvesicle can fuse with the lysosome delivering a large vesicle,bounded by a single membrane bilayer, to the interior of thelysosome. The lipases and proteases within the lysosomeeventually will degrade this vesicle and its contents into theirmolecular components.

Retroviruses Bud from the Plasma Membrane by a Process Similar to Formation of Multivesicular EndosomesThe vesicles that bud into the interior of endosomes have atopology similar to that of enveloped virus particles that budfrom the plasma membrane of virus-infected cells. Moreover,recent experiments demonstrate that a common set of proteinsare required for both types of membrane-budding events. Infact, the two processes so closely parallel one another in mech-anistic detail as to suggest that enveloped viruses have evolvedmechanisms to recruit the cellular proteins used in inward en-dosomal budding for their own purposes.

Many of the proteins required for inward budding of theendosomal membrane were first identified by mutations inyeast that blocked delivery of membrane proteins to the in-terior of the vacuole. More than 10 such “budding” proteinshave been identified in yeast, most with significant similari-ties to mammalian proteins that evidently perform the samefunction in mammalian cells. The current model of endoso-mal budding to form multivesicular endosomes in mam-malian cells is based primarily on studies in yeast (Figure17-32). A ubiquitin-tagged peripheral membrane protein ofthe endosome, known as Hrs, facilitates loading of specificubiquitinated membrane cargo proteins into vesicle buds di-rected into the interior of the endosome. The ubiquitinatedHrs protein then recruits a set of three different protein com-plexes to the membrane. These ESCRT (endosomal sortingcomplexes required for transport) complexes include theubiquitin-binding protein Tsg101. The membrane-associatedESCRT complexes act to complete vesicle budding, leadingto release of a vesicle carrying specific membrane cargo intothe interior of the endosome. Finally, an ATPase, known asVps4, uses the energy from ATP hydrolysis to disassemblethe ESCRT complexes, releasing them into the cytosol foranother round of budding. In the fusion event that pinchesoff a completed endosomal vesicle, the ESCRT proteins andVps4 may function like SNAREs and NSF, respectively, inthe typical membrane-fusion process discussed previously(see Figure 17-11).

The human immunodeficiency virus (HIV) is an en-veloped retrovirus that buds from the plasma membraneof infected cells in a process driven by viral Gag protein, themajor structural component of completed virus particles.Gag protein binds to the plasma membrane of an infectedcell and ≈4000 Gag molecules polymerize into a spherical


shell, producing a structure that looks like a vesicle budprotruding outward from the plasma membrane. Muta-tional studies with HIV have revealed that the N-terminalsegment of Gag protein is required for association with theplasma membrane, whereas the C-terminal segment is re-quired for pinching off of complete HIV particles. For in-stance, if the portion of the viral genome encoding theC-terminus of Gag is removed, HIV buds will form in in-fected cells, but pinching off does not occur and thus nofree virus particles are released.

The first indication that HIV budding employs the samemolecular machinery as vesicle budding into endosomes

came from the observation that Tsg101, a component of theESCRT complex, binds to the C-terminus of Gag protein.Subsequent findings have clearly established the mechanis-tic parallels between the two processes (Figure 17-32). Forexample, Gag is ubiquitinated as part of the process of virusbudding, and in cells with mutations in Tsg101 or Vps4, HIVvirus buds accumulate but cannot pinch off from the mem-brane (Figure 17-33). Moreover, when a segment from thecellular Hrs protein is added to a truncated Gag protein,proper budding and release of virus particles is restored.Taken together, these results indicate that Gag protein mim-ics the function of Hrs, redirecting ESCRT complexes to the


Plasma membrane

Cytosol

Lumen of endosome

Extracellular space

HIV Gag protein

Ubiquitin

ESCRT complexassembly

Core particle

HIV virus

Hrs proteinUbiquitin

Endosomalvesicle

ATP

ADP + Pi

Vps4ATP

ADP + Pi

HIV Env

Vps4

ESCRT complexdisassembly

Cargo proteins

3

6

2

4 5

1

� FIGURE 17-32 Model of the

common mechanism for formation of

multivesicular endosomes and budding

of HIV from the plasma membrane.

Bottom: In endosomal budding, ubiquitinatedHrs on the endosomal membrane directsloading of specific membrane cargo proteins(blue) into vesicle buds and then recruitscytosolic ESCRT complexes to themembrane (step ). Note that both Hrs andthe recruited cargo proteins are tagged withubiquitin. After the set of bound ESCRTcomplexes mediate membrane fusion andpinching off of the completed vesicle (step ), they are disasssembled by theATPase Vps4 and returned to the cytosol(step ). Top: Budding of HIV particles fromHIV-infected cells occurs by a similarmechanism using the virally encoded Gagprotein and cellular ESCRT complexes andVps4 (steps – ). Ubiquitinated Gag neara budding particle functions like Hrs. Seetext for discussion. [Adapted from O. Pornilloset al., 2002, Trends Cell Biol. 12:569.]

64

3

2

1

(a) (b)

� FIGURE 17-33 Electron micrographs of virus budding

from wild-type and ESCRT-deficient HIV-infected cells. (a) Inwild-type cells infected with HIV, virus particles bud from theplasma membrane and are rapidly released into the extracellularspace. (b) In cells that lack the functional ESCRT protein Tsg101,the viral Gag protein forms dense virus-like structures, butbudding of these structures from the plasma membrane cannotbe completed and chains of incomplete viral buds still attached tothe plasma membrane accumulate. [Wes Sundquist, University ofUtah.]

plasma membrane where they can function in the budding ofvirus particles.

Other enveloped retroviruses such as murine leukemiavirus and Rous sarcoma virus also have been shown to re-quire ESCRT complexes for their budding, although eachvirus appears to have evolved a somewhat different mecha-nism to recruit ESCRT complexes to the site of virus budding.

Transcytosis Moves Some Endocytosed LigandsAcross an Epithelial Cell LayerAs noted previously, transcytosis is used by some cells in theapical-basolateral sorting of certain membrane proteins (seeFigure 17-26). This process of transcellular transport, whichcombines endocytosis and exocytosis, also can be employedto import an extracellular ligand from one side of a cell,transport it across the cytoplasm, and secrete it from theplasma membrane at the opposite side. Transcytosis occursmainly in sheets of polarized epithelial cells.

Maternal immunoglobulins (antibodies) contained in in-gested breast milk are transported across the intestinal ep-ithelial cells of the newborn mouse and human by transcytosis(Figure 17-34). The Fc receptor that mediates this movementbinds antibodies at the acidic pH of 6 found in the intestinallumen but not at the neutral pH of the extracellular fluid onthe basal side of the intestinal epithelium. This difference inthe pH of the extracellular media on the two sides of intes-

tinal epithelial cells allows maternal immunoglobulins tomove in one direction—from the lumen to the blood. Thesame process also moves circulating maternal immunoglobu-lins across mammalian yolk-sac cells into the fetus.


Receptor-Mediated Endocytosis and the Sorting of Internalized Proteins

■ Some extracellular ligands that bind to specific cell-sur-face receptors are internalized, along with their receptors,in clathrin-coated vesicles whose coats also contain AP2complexes.

■ Sorting signals in the cytosolic domain of cell-surface re-ceptors target them into clathrin/AP2-coated pits for in-ternalization. Known signals include the Asn-Pro-X-Tyr,Tyr-X-X-�, and Leu-Leu sequences (see Table 17-2).

■ The endocytic pathway delivers some ligands (e.g., LDLparticles) to lysosomes where they are degraded. Most re-ceptor-ligand complexes dissociate in the acidic milieu ofthe late endosome; the receptors are recycled to the plasmamembrane, while the ligands are sorted to lysosomes (seeFigure 17-28).

■ Iron is imported into cells by an endocytic pathway inwhich Fe3� ions are released from ferrotransferrin in thelate endosome. The receptor-apotransferrin complex is re-cycled to the cell surface where the complex dissociates,releasing both the receptor and apotransferrin for reuse.

■ Endocytosed membrane proteins destined for degrada-tion in the lysosome are incorporated into vesicles that budinto the interior of the endosome. Multivesicular endo-somes, which contain many of these internal vesicles, canfuse with the lysosome to deliver the vesicles to the inte-rior of the lysosome (see Figure 17-31).

■ A portion of the cytoplasm or an entire organelle (e.g.,peroxisome) can be enveloped in a flattened membrane andeventually incorporated into a double-membraned au-tophagic vesicle. Fusion of the outer vesicle membrane withthe lysosome delivers the enveloped contents to the inte-rior of the lysosome for degradation.

■ Some of the cellular components (e.g., ESCRT com-plexes) that mediate inward budding of endosomal mem-branes are used in the budding and pinching off of en-veloped viruses such as HIV from the plasma membraneof virus-infected cells (see Figure 17-32).

Synaptic Vesicle Function and FormationIn this final section we consider the regulated secretion ofneurotransmitters that is the basis for signaling by manynerve cells. These small, water-soluble molecules (e.g., acetyl-

17.6

17.6 • Synaptic Vesicle Function and Formation 735

Basalmembrane

Tight junction

Luminalmembrane

Intestinallumen(pH ~6)

Blood andinterstitialfluid(pH ~7)

Epithelial cells

Endosome

Fc receptor

Fc regionIgG

Membrane

▲ FIGURE 17-34 Transcytosis of maternal IgG

immunoglobulins across the intestinal epithelial cells of

newborn mice. This transcellular movement of a ligand involvesboth endocytosis and exocytosis. The one-way movement ofligand from the intestinal lumen to the blood depends on thedifferential affinity of the Fc receptor for antibody at pH 6 (strongbinding) and at pH 7 (weak binding). Transcytosis in the oppositedirection returns the empty Fc receptor to the luminalmembrane. See text for discussion.

choline, dopamine) are released at chemical synapses, spe-cialized sites of contact between a signaling neuron and a re-ceiving cell. Generally signals are transmitted in only onedirection: an axon terminal from a presynaptic cell releasesneurotransmitter molecules that diffuse through a narrowextracellular space (the synaptic cleft) and bind to receptorson a postsynaptic cell (see Figure 7-31). The membrane ofthe postsynaptic cell, which can be another neuron, a mus-cle cell, or a gland cell, is located within approximately 50nm of the presynaptic membrane.

Neurotransmitters are stored in specialized regulated secretory vesicles, known as synaptic vesicles, which are40–50 nm in diameter. Exocytosis of these vesicles and re-lease of neurotransmitters is initiated when a stimulatoryelectrical impulse (action potential) travels down the axon ofa presynaptic cell to the axon terminal where it triggers open-ing of voltage-gated Ca2� channels. The subsequent local-ized rise in the cytosolic Ca2� concentration induces somesynaptic vesicles to fuse with the plasma membrane, releas-ing their contents into the synaptic cleft. We described themajor events in signal transmission at chemical synapses andthe effects of neurotransmitter binding on postsynaptic cellsin Chapter 7. Here we focus on the regulated secretion ofneurotransmitters and the formation of synaptic vesicles inthe context of the basic principles of vesicular trafficking al-ready outlined in this chapter.

Synaptic Vesicles Loaded with NeurotransmitterAre Localized Near the Plasma MembraneThe exocytosis of neurotransmitters from synaptic vesicles in-volves targeting and fusion events similar to those that leadto release of secreted proteins in the secretory pathway. How-ever, several unique features permit the very rapid release ofneurotransmitters in response to arrival of an action potentialat the presynaptic axon terminal. For example, in resting neu-rons some neurotransmitter-filled synaptic vesicles are“docked” at the plasma membrane; others are in reserve in theactive zone near the plasma membrane at the synaptic cleft.In addition, the membrane of synaptic vesicles contains a spe-cialized Ca2�-binding protein that senses the rise in cytosolicCa2� after arrival of an action potential, triggering rapid fu-sion of docked vesicles with the presynaptic membrane.

A highly organized arrangement of cytoskeletal fibers inthe axon terminal helps localize synaptic vesicles in the activezone (Figure 17-35). The vesicles themselves are linked to-gether by synapsin, a fibrous phosphoprotein associated withthe cytosolic surface of all synaptic-vesicle membranes. Fila-ments of synapsin also radiate from the plasma membraneand bind to vesicle-associated synapsin. These interactionsprobably keep synaptic vesicles close to the part of theplasma membrane facing the synapse. Indeed, synapsinknockout mice, although viable, are prone to seizures; duringrepetitive stimulation of many neurons in such mice, thenumber of synaptic vesicles that fuse with the plasma mem-

brane is greatly reduced. Thus synapsins are thought to re-cruit synaptic vesicles to the active zone.

Rab3A, a GTP-binding protein located in the membraneof synaptic vesicles, also is required for targeting of neuro-transmitter-filled vesicles to the active zone of presynapticcells facing the synaptic cleft. Rab3A knockout mice, likesynapsin-deficient mice, exhibit a reduced number of synap-tic vesicles able to fuse with the plasma membrane afterrepetitive stimulation. The neuron-specific Rab3 is similarin sequence and function to other Rab proteins that partici-pate in docking vesicles on particular target membranes inthe secretory pathway.

A Calcium-Binding Protein Regulates Fusion of Synaptic Vesicles with the Plasma Membrane

Fusion of synaptic vesicles with the plasma membrane ofaxon terminals depends on the same proteins that mediatemembrane fusion of other regulated secretory vesicles. Theprincipal v-SNARE in synaptic vesicles (VAMP) tightly bindssyntaxin and SNAP-25, the principal t-SNAREs in theplasma membrane of axon terminals, to form four-helixSNARE complexes. After fusion, SNAP proteins and NSF


Axon terminal Synapsin-containingfibers

Postsynaptic cell

Docked synapticvesicle

Activezone

0.1 �m

▲ EXPERIMENTAL FIGURE 17-35 Fibrous proteins

help localize synaptic vesicles to the active zone of axon

terminals. In this micrograph of an axon terminal obtained by therapid-freezing deep-etch technique, synapsin fibers can be seento interconnect the vesicles and to connect some to the activezone of the plasma membrane. Docked vesicles are ready to beexocytosed. Those toward the center of the terminal are in theprocess of being filled with neurotransmitter. [From D. M. D. Landiset al., 1988, Neuron 1:201.]

within the axon terminal promote disassociation of VAMPfrom t-SNAREs, as in the fusion of secretory vesicles de-picted previously (see Figure 17-11).

Strong evidence for the role of VAMP in neurotrans-mitter exocytosis is provided by the mechanism ofaction of botulinum toxin, a bacterial protein that

can cause the paralysis and death characteristic of botulism, atype of food poisoning. The toxin is composed of two polypep-tides: One binds to motor neurons that release acetylcholine atsynapses with muscle cells, facilitating entry of the otherpolypeptide, a protease, into the cytosol of the axon terminal.The only protein this protease cleaves is VAMP. After the bot-ulinum protease enters an axon terminal, synaptic vesicles thatare not already docked rapidly lose their ability to fuse with theplasma membrane because cleavage of VAMP prevents assem-bly of SNARE complexes. The resulting block in acetylcholinerelease at neuromuscular synapses causes paralysis. However,vesicles that are already docked exhibit remarkable resistanceto the toxin indicating that SNARE complexes may already bein a partially assembled, protease-resistant state when vesiclesare docked on the presynaptic membrane. ❚

The signal that triggers exocytosis of docked synapticvesicles is a rise in the Ca2� concentration in the cytosol nearvesicles from �0.1 �M, characteristic of resting cells, to1–100 �M following arrival of an action potential in stimu-lated cells. The speed with which synaptic vesicles fuse withthe presynaptic membrane after a rise in cytosolic Ca2� (lessthan 1 msec) indicates that the fusion machinery is entirelyassembled in the resting state and can rapidly undergo a con-formational change leading to exocytosis of neurotransmit-ter. A Ca2�-binding protein called synaptotagmin, locatedin the membrane of synaptic vesicles, is thought to be a keycomponent of the vesicle fusion machinery that triggers exo-cytosis in response to Ca2� (Figure 17-36).

Several lines of evidence support a role for synaptotag-min as the Ca2� sensor for exocytosis of neurotransmitters.For instance, mutant embryos of Drosophila and C. elegansthat completely lack synaptotagmin fail to hatch and exhibitvery reduced, uncoordinated muscle contractions. Larvaewith partial loss-of-function mutations of synaptotagminsurvive, but their neurons are defective in Ca2�-stimulatedvesicle exocytosis. Moreover, in mice, mutations in synapto-tagmin that decrease its affinity for Ca2� cause a corresponding

17.6 • Synaptic Vesicle Function and Formation 737

Synaptic

cleft

Neurotransmitterreceptors

Presynaptic cell

Postsynaptic cell

Neurotransmitter

Shibiremutation

SNAREcomplex

Botulinum toxin

Synaptic vesicle

+ Ca2+

Neurotransmitter/transporter

Synoptotagmin

Clathrin

4

3

2

1

Dynamin

▲ FIGURE 17-36 Release of neurotransmitters and the

recycling of synaptic vesicles. Step : Synaptic vesicles loadedwith neurotransmitter (red circles) move to the active zone andthen dock at defined sites on the plasma membrane of apresynaptic cell. Synpatotagmin prevents membrane fusion andrelease of neurotransmitter. Botulinum toxin prevents exocytosisby proteolytically cleaving VAMP, the v-SNARE on vesicles. Step

: In response to a nerve impulse (action potential), voltage-gated Ca2� channels in the plasma membrane open, allowing aninflux of Ca2� from the extracellular medium. The resulting Ca2�-induced conformational change in synaptotagmin leads to fusionof docked vesicles with the plasma membrane and release ofneurotransmitters into the synaptic cleft. Step : After3

2

1clathrin/AP vesicles containing v-SNARE and neurotransmittertransporter proteins bud inward and are pinched off in a dynamin-mediated process, they lose their coat proteins. Dynaminmutations such as shibire in Drosophila block the re-formation ofsynaptic vesicles, leading to paralysis. Step : The uncoatedvesicles import neurotransmitters from the cytosol, generatingfully reconstituted synaptic vesicles and completing the cycle.Most synaptic vesicles are formed by endocytic recycling asdepicted here. However, endocytic vesicles containing membranefrom the axon terminus can fuse with the endosome; buddingfrom this compartment can then form “new” synaptic vesicles.[See K. Takei et al., 1996, J. Cell. Biol. 133:1237; V. Murthy and C. Stevens, 1998, Nature 392:497; and R. Jahn et al., 2003, Cell 112:519.]

4

increase in the amount of cytosolic Ca2� needed to triggerrapid exocytosis.

Several hypotheses concerning how synaptotagmin pro-motes neurotransmitter exocytosis have been proposed, butthe precise mechanism of its function is still unresolved.Synaptotagmin is known to bind phospholipids and after un-dergoing a Ca2�-induced conformational change, it maypromote association of the phospholipids in the vesicle andplasma membranes. Synaptotagmin also binds to SNAREproteins and may catalyze a late stage in assembly of SNAREcomplexes when bound to Ca2�. Finally, synaptotagmin mayalso act to inhibit inappropriate exocytosis in resting cells. Atthe low cytosolic Ca2� levels found in resting nerve cells,synaptotagmin apparently binds to a complex of the plasma-membrane proteins neurexin and syntaxin. The presence ofsynaptotagmin blocks binding of other essential fusion pro-teins to the neurexin-syntaxin complex, thereby preventingvesicle fusion. When synaptotagmin binds Ca2�, it is dis-placed from the complex, allowing other proteins to bindand thus initiating membrane docking or fusion. Thus synap-totagmin may operate as a “clamp” to prevent fusion fromproceeding in the absence of a Ca2� signal.

Fly Mutants Lacking Dynamin Cannot RecycleSynaptic VesiclesSynaptic vesicles are formed primarily by endocytic buddingfrom the plasma membrane of axon terminals. Endocytosisusually involves clathrin-coated pits and is quite specific, in thatseveral membrane proteins unique to the synaptic vesicles (e.g.,neurotransmitter transporters) are specifically incorporatedinto the endocytosed vesicles. In this way, synaptic-vesiclemembrane proteins can be reused and the recycled vesicles refilled with neurotransmitter (see Figure 17-36).

As in the formation of other clathrin/AP-coated vesicles,pinching off of endocytosed synaptic vesicles requires theGTP-binding protein dynamin (see Figure 17-20). Indeed,analysis of a temperature-sensitive Drosophila mutant calledshibire (shi), which encodes the fly dynamin protein, pro-vided early evidence for the role of dynamin in endocytosis.At the permissive temperature of 20�C, the mutant flies arenormal, but at the nonpermissive temperature of 30�C, theyare paralyzed (shibire, paralyzed in Japanese) because pinch-ing off of clathrin-coated pits in neurons and other cells isblocked. When viewed in the electron microscope, the shineurons at 30�C show abundant clathrin-coated pits withlong necks but few clathrin-coated vesicles. The appearanceof nerve terminals in shi mutants at the nonpermissive tem-perature is similar to that of terminals from normal neuronsincubated in the presence of a nonhydrolyzable analog ofGTP (see Figure 17-21). Because of their inability to pinchoff new synaptic vesicles, the neurons in shi mutants even-tually become depleted of synaptic vesicles when flies areshifted to the nonpermissive temperature, leading to a ces-sation of synaptic signaling and paralysis.


Synaptic Vesicle Formation and Function

■ Transmission of nerve impulses at chemical synapses de-pends on the exocytosis of neurotransmitter-filled synap-tic vesicles and the regeneration of empty vesicles by en-docytosis.

■ Efficient recruitment of vesicles to the presynaptic mem-brane adjacent to the synaptic cleft requires cytosolic pro-teins, such as synapsin, and Rab3a, a GTP-binding proteinthat is tethered to the vesicle membrane.

■ In resting neurons, synaptotagmin in the synaptic-vesi-cle membrane prevents fusion of docked vesicles with themembrane. The influx of Ca2� following arrival of an ac-tion potential at the axon terminus leads to Ca2� bindingby synaptotagmin, causing a change in its conformationthat permits vesicle fusion to proceed (see Figure 17-36).

■ Synaptic vesicles are rapidly regenerated by endocyticbudding of clathrin-coated vesicles from the plasma mem-brane, a process that requires dynamin. After the clathrincoat is shed, vesicles are refilled with neurotransmitter andmove to the active zone for another round of docking andfusion.

P E R S P E C T I V E S F O R T H E F U T U R E

The biochemical, genetic, and structural information pre-sented in this chapter shows that we now have a basic un-derstanding of how protein traffic flows from onemembrane-bounded compartment to another. Our under-standing of these processes has come largely from experi-ments on the function of various types of transport vesicles.These studies have led to the identification of many vesiclecomponents and the discovery of how these componentswork together to drive vesicle budding, to incorporate thecorrect set of cargo molecules from the donor organelle, andthen to mediate fusion of a completed vesicle with the mem-brane of a target organelle.

Despite these advances, there remain important stagesof the secretory and endocytic pathways about which weknow relatively little. For example, we do not yet knowwhat types of proteins form the coats of either the regu-lated or constitutive secretory vesicles that bud from thetrans-Golgi network. Indeed, it is not clear whether assem-bly of a cytosolic coat drives their budding at all. Moreover,the types of signals on cargo proteins that might targetthem for packaging into secretory vesicles have not yet beendefined. Another baffling process is the formation of vesi-cles that bud away from the cytosol, such as the vesiclesthat enter multivesicular endosomes. Although some of theproteins that participate in formation of these “internal”endosome vesicles are known, we do not know what de-


termines their shape or what type of process causes them topinch off from the donor membrane. In the future, it shouldbe possible for these and other poorly understood vesicle-trafficking steps to be dissected through the use of the samepowerful combination of biochemical and genetic methodsthat have delineated the working parts of COPI, COPII,and clathrin/AP vesicles.

Questions still remain about vesicle trafficking betweenthe ER and cis-Golgi, between Golgi stacks, and betweenthe trans-Golgi and endosome, the best-characterized trans-port steps. In particular, our understanding of how proteinsare actually sorted between these organelles is incompletelargely because of the highly dynamic nature of all the or-ganelles along the secretory pathway. Although we knowmany of the details of how particular vesicle componentsfunction, we cannot account for why their functions are re-stricted to specific stages in the overall flow of anterogradeand retrograde transport steps. For example, we cannot ex-plain why COPII vesicles fuse with one another to form anew cis-Golgi stack, whereas COPI vesicles fuse with themembrane of the ER, since both vesicle types appear to con-tain similar sets of v-SNARE proteins. In the same vein, wedo not know what feature of the Golgi membrane actuallydistinguishes a COPI-coated vesicle bud from a clathrin/AP-coated bud. In both cases binding of ARF protein to theGolgi membrane appears to initiate vesicle budding. The so-lution to these problems will require a more integrated un-derstanding of the flow of vesicular traffic in the context ofthe entire secretory pathway. Recent improvements in ourability to image vesicular transport of cargo proteins in livecells gives hope that some of these more subtle aspects ofvesicle function may be clarified in the near future.

KEY TERMS

AP (adapter protein) complexes 720

anterograde transport 715ARF protein 709autophagy 733cisternal progression 715clathrin 708constitutive secretion 724COPI 708COPII 708dynamin 721ESCRT complexes 733late endosome 702mannose 6-phosphate

(M6P) 723multivesicular

endosomes 733

REVIEW THE CONCEPTS

1. The studies of Palade and colleagues using pulse-chaselabeling with radioactively labeled amino acids and autora-diography to visualize the location of the radiolabeled pro-teins is a classic case of the experimentalist realizing whatworks well in his or her biological system. These experimentswere done with pancreatic acinar cells. Alternatively, HeLacells can be used. HeLa cells are a classic human cell lineoriginating from a cervical carcinoma. When applied toHeLa cells, the same experimental protocols are a dismalfailure with respect to tracking secretion. What would youexpect autoradiography of a HeLa cell to look like afterpulse-labeling with radioactive amino acids?

2. Sec18 is a yeast gene that encodes NSF. It is a class Cmutant in the yeast secretory pathway. What is the mecha-nistic role of NSF in membrane trafficking. As indicated byits class C phenotype, why does an NSF mutation produceaccumulation of vesicles at what appears to be only one stageof the secretory pathway?

3. Vesicle budding is associated with coat proteins. What isthe role of coat proteins in vesicle budding? How are coatproteins recruited to membranes? What kinds of moleculesare likely to be included or excluded from newly formed vesi-cles? What is the best-known example of a protein likely tobe involved in vesicle pinching off?

4. Treatment of cells with the drug brefeldin A (BFA) hasthe effect of decoating Golgi apparatus membranes, resultingin a cell in which the vast majority of Golgi proteins arefound in the ER. What inferences can be made from this ob-servation regarding roles of coat proteins other than pro-moting vesicle formation? Predict what type of mutation inArf1 might have the same effect as treating cells with BFA.

5. An antibody to an exposed “hinge” region of �COPIknown as EAGE blocks the function of �COPI when mi-croinjected into HeLa cells. Predict what the consequences ofthis functional block might be for anterograde transportfrom the ER to the plasma membrane. Propose an experi-ment to test whether the effect of EAGE microinjection is ini-tially on anterograde or retrograde transport.

6. Specificity in fusion between vesicles involves two discreteand sequential processes. Describe the first of the two processesand its regulation by GTPase switch proteins. What effect onthe size of early endosomes might result from overexpression ofa mutant form of Rab5 that is stuck in the GTP-bound state?

7. Two different protein-mediated membrane fusionprocesses are described in this chapter, SNARE- and viralHA-mediated fusion. Compare and contrast the two. In eachexample give particular attention to what the direct effect ofpolypeptide sequences in membrane fusion is and to whatcontrols the specificity of membrane fusion in each.

Review the Concepts 739

Rab proteins 711receptor-mediated

endocytosis 728regulated secretion 724retrograde transport 715sec mutants 706secretory pathway 701sorting signals 711synaptotagmin 737transcytosis 727trans-Golgi network

(TGN) 701transport vesicles 701t-SNAREs 708v-SNAREs 708

8. Sorting signals that cause retrograde transport of a pro-tein in the secretory pathway are sometimes known as re-trieval sequences. List the two known examples of retrievalsequences for soluble and membrane proteins of the ER?How does the presence of a retrieval sequence on a solubleER protein result in its retrieval from the cis-Golgi complex?Describe how the concept of a retrieval sequence is essentialto the cisternal-progression model.

9. Clathrin adapter protein (AP) complexes bind directly tothe cytosolic face of membrane proteins and also interactwith clathrin. What are the four known adapter proteincomplexes? Why may clathrin be considered to be an acces-sory protein to a core coat composed of adapter proteins?

10. I-cell disease is a classic example of an inherited humandefect in protein targeting that affects an entire class of pro-teins, soluble enzymes of the lysosome. What is the molecu-lar defect in I-cell disease? Why does it affect the targetingof an entire class of proteins? What other types of mutationsmight produce the same phenotype?

11. The TGN, trans-Golgi network, is the site of multiplesorting processes as proteins and lipids exit the Golgi com-plex. Compare and contrast the sorting of proteins to lyso-somes versus the packaging of proteins into regulatedsecretory granules such as those containing insulin. Compareand contrast the sorting of proteins to the basolateral versusapical cell surfaces in MDCK cells versus hepatocytes.

12. The efficiency of bacterial phagocytosis by macrophagesis increased greatly by first binding antibody molecules to thebacterial surface. On the basis of prior descriptions of anti-body structure, to what portion of the immunoglobulin mol-ecule do you predict a macrophage receptor for antibodybound to bacteria might be directed? Design an experimentto test this prediction.

13. Describe how pH plays a key role in regulating the in-teraction between mannose 6-phosphate and the mannose 6-phosphate receptor. Why does elevating endosomal pH leadto the secretion of newly synthesized lysosomal enzymes intothe extracellular medium?

14. What mechanistic features are shared by (a) the for-mation of multivesicular endosomes by budding into the interior of the endosome and (b) the outward budding of HIV virus at the cell surface? You wish to design a pep-tide inhibitor/competitor of HIV budding and decide to mimic in a synthetic peptide a portion of the HIV Gag pro-tein. Which portion of the HIV Gag protein would be a logical choice? What normal cellular process might this in-hibitor block?

15. The exocytosis of neurotransmitter-filled synaptic vesicles is an example of regulated exocytosis. How is the influx of Ca2� following arrival of an action potential at the axon terminus sensed and linked to the exocytosis of

synaptic vesicles? Why do normal Drosophila neurons in-cubated in the presence of a nonhydrolyzable analog of GTP have the same appearance as nerve terminals in shimutants?

ANALYZE THE DATA

A variety of protein toxins, such as the bacterial toxinPseudomonas and Shiga toxin and the plant toxin ricin, areheteromeric proteins consisting of A and B subunits. TheA subunit is catalytic. For Shiga toxin, the proximal causeof food poisoning due to bacterially contaminated ham-burger, the A subunit is an N-glycosidase and specificallycleaves 28S ribosomal RNA, thereby intoxicating cells byinhibiting protein synthesis. Amazingly, only one moleculeof A subunit when introduced into the cytosol is sufficientto kill a cell. Interestingly the A subunit of Shiga toxin istransferred into the cytosol from the lumen of the ER bythe Sec61 protein translocon. The B subunit targets Shigatoxin to the ER by binding to a glycolipid GM3 on the cellsurface that acts as the Shiga toxin internalization recep-tor. Shiga toxin is internalized into endosomes, from en-dosomes is transferred to the Golgi complex, and from theGolgi complex goes to the ER where the A and B subunitsdissociate, permitting the A subunit to translocate into thecytosol.

In a series of experiments designed to characterize thecomparative mechanisms of Pseudomonas and Shiga toxintransfer from the Golgi complex to the ER, investigators firstsequenced the respective targeting subunits. The C-terminal24 amino acids of the B subunits of Pseudomonas toxin andShiga toxin are shown below:

C-terminal 24 amino acids of Pseudomonas toxin B subunitKEQAISALPD YASQPGKPPR KDEL

C-terminal 24 amino acids of Shiga toxin B subunitTGMTVTIKTN ACHNGGGFSE VIFR

From inspection of these sequences, what is the probable tar-geting receptor for transfer of Pseudomonas toxins from theGolgi apparatus to the ER?

To test this prediction directly, investigators experimen-tally characterized the role of COPI coat proteins andKDEL receptors in intoxication. Monkey cells were mi-croinjected with antibodies directed against either COPIcoat proteins or the cytosolic domain of KDEL receptors.Cells then were incubated with Pseudomonas or Shigatoxin for 4 h. Protein synthesis was determined followinga 30-minute pulse labeling with [35S]methionine. Resultsare shown in the accompanying figure, with controls show-ing the low level of protein synthesis caused by incubationwith either Pseudomonas or Shiga toxin without antibodyinjection.


How do these results support your sequence-based pre-dictions and the known role of COPI coat protein in retro-grade transport? Can you formulate a hypothesis for howShiga toxin is transported from the Golgi to the ER?

To explore further whether or not Shiga toxin transferfrom the Golgi apparatus to the ER depends on COPI coatproteins, investigators prepared two different fluorescentdye–conjugated Shiga toxin B subunits and then assessed byfluorescence microscopy transport of the B subunits from theGolgi complex to the ER. The first preparation was Cy3-conjugated wild-type B subunit. The second preparation wasCy3-conjugated B subunit in which the C terminus was ex-tended by the four amino acids KDEL (B-KDEL). Cells weremicroinjected with antibody directed against COPI coat pro-teins. Following microinjection, cells were incubated with flu-orescent B subunit for various periods of time and B subunitdistributions scored. The results are shown in the figure below.

What evidence do these results provide for or againsttransport of wild-type Shiga toxin B subunit from the Golgicomplex to the ER in COPI-coated vesicles? What is the im-portance of the results with B-KDEL in interpreting the over-all results of these experiments?

REFERENCES

Techniques for Studying the Secretory PathwayBeckers, C. J., et al. 1987. Semi-intact cells permeable to macromolecules:

use in reconstitution of protein transport from the endoplasmic reticulum tothe Golgi complex. Cell 50:523–534.

Kaiser, C. A., and R. Schekman. 1990. Distinct sets of SEC genesgovern transport vesicle formation and fusion early in the secretorypathway. Cell 61:723–733.

Novick, P., et al. 1981. Order of events in the yeast secretorypathway. Cell 25:461–469.

Lippincott-Schwartz J., et al. 2001. Studying protein dynamicsin living cells. Nature Rev. Mol. Cell Biol. 2:444–456.

Orci, L., et al. 1989. Dissection of a single round of vesiculartransport: sequential intermediates for intercisternal movement in theGolgi stack. Cell 56:357–368.

Palade, G. 1975. Intracellular aspects of the process of proteinsynthesis. Science 189:347–358.

Molecular Mechanisms of Vesicular TrafficJahn, R., et al. 2003. Membrane fusion. Cell 112:519–533.Kirchhausen, T. 2000. Three ways to make a vesicle. Nature Rev.

Mol. Cell Biol. 1:187–198.McNew, J. A., et al. 2000. Compartmental specificity of cellu-

lar membrane fusion encoded in SNARE proteins. Nature407:153–159.

Ostermann, J., et al. 1993. Stepwise assembly of functionally ac-tive transport vesicles. Cell 75:1015–1025.

Schimmöller, F., I. Simon, and S. Pfeffer. 1998. Rab GTPases, di-rectors of vesicle docking. J. Biol. Chem. 273:22161–22164.

Skehel, J. J., and D. C. Wiley. 2000. Receptor binding and mem-brane fusion in virus entry: the influenza hemagglutinin. Ann. Rev.Biochem. 69:531–569.

Söllner, T., et al. SNAP receptors implicated in vesicle targetingand fusion. Nature 362:318–324.

Springer, S., et al. 1999. A primer on vesicle budding. Cell97:145–148.

Zerial, M., and H. McBride. 2001. Rab proteins as membraneorganizers. Nature Rev. Mol. Cell Biol. 2:107–117.

Vesicle Trafficking in the Early Stages of the SecretoryPathway

Barlowe, C. 2002. COPII-dependent transport from the endo-plasmic reticulum. Curr. Opin. Cell Biol. 14:417–422.

Bi, X., et al. 2002. Structure of the Sec23/24-Sar1 pre-buddingcomplex of the COPII vesicle coat. Nature 419:271–277.

Glick, B. S., and V. Malhotra. 1998. The curious status of theGolgi apparatus. Cell 95:883–889.

Kuehn, M. J., and R. Schekman. 1997. COPII and secre-tory cargo capture into transport vesicles. Curr. Opin. Cell Biol.9:477–483.

References 741

0

100

200

300

Anti-COPI

Anti-KDEL receptor

Pseudomonastoxin

Shigatoxin

Control

400P

rote

in s

ynth

esis

(%

of

con

tro

l)

Effect of microinjected antibodies on cell intoxication byPseudomonas and Shiga toxin.

0

20

40

60

80

NoninjectedAnti-EAGE

Noninjected/B-KDELAnti-EAGE/B-KDEL

4.03.02.01.0

100

ER

loca

lizat

ion

(%

of

cells

)

Time (h)

Effect of microinjected antibodies directed against COPI proteinson the transport of wild type Shiga toxin B-subunit or Shiga toxinB-KDEL from Golgi complex to the ER.

Letourneur, F., et al. 1994. Coatomer is essential for retrieval of dilysine-tagged proteins to the endoplasmic reticulum. Cell79:1199–1207.

Pelham, H. R. 1995. Sorting and retrieval between the endo-plasmic reticulum and Golgi apparatus. Curr. Opin. Cell Biol.7:530–535.

Schekman, R., and I. Mellman. 1997. Does COPI go both ways?Cell 90:197–200.

Protein Sorting and Processing in Later Stages of the Secretory Pathway

Bonifacino, J. S., and E. C. Dell’Angelica. 1999. Molecular basesfor the recognition of tyrosine-based sorting signals. J. Cell Biol.145:923–926.

Ghosh, P., et al. 2003. Mannose 6-phosphate receptors: newtwists in the tale. Nature Rev. Mol. Cell Bio. 4:202–213.

Hinshaw, J. E. 2000. Dynamin and its role in membrane fission.Ann. Rev. Cell Dev. Biol. 16:483–519.

Kirchhausen, T. 2000. Clathrin. Ann. Rev. Biochem. 69:699–727. Mostov, K. E., et al. 1995. Regulation of protein traffic in

polarized epithelial cells: the polymeric immunoglobulin receptormodel. Cold Spring Harbor Symp. Quant. Biol. 60:775–781.

Robinson, M. S., and J. S. Bonifacino. 2001. Adaptor-relatedproteins. Curr. Opin. Cell Biol. 13:444–453.

Schmid, S. 1997. Clathrin-coated vesicle formation and proteinsorting: an integrated process. Ann. Rev. Biochem. 66:511–548.

Simons, K., and E. Ikonen. 1997. Functional rafts in cell mem-branes. Nature 387:569–572.

Steiner, D. F., et al. 1996. The role of prohormone convertasesin insulin biosynthesis: evidence for inherited defects in their action in man and experimental animals. Diabetes Metab.22:94–104.

Tooze, S. A., et al. 2001. Secretory granule biogenesis: rafting tothe SNARE. Trends Cell Biol. 11:116–122.

Receptor-Mediated Endocytosis and the Sorting of Internalized Proteins

Brown, M. S., and J. L. Goldstein. 1986. Receptor-mediatedpathway for cholesterol homeostasis. Nobel Prize Lecture. Science232:34–47.

Katzmann, D. J., et al. 2002. Receptor downregulation and mul-tivesicular-body sorting. Nature Rev. Mol. Cell Biol. 3:893–905.

Khalfan, W. A., and D. J. Klionsky. 2002. Molecular machinery required for autophagy and the cytoplasm to vacuole tar-geting (Cvt) pathway in S. cerevisiae. Curr. Opin. Cell Biol.14:468–475.

Lemmon, S. K., and L. M. Traub. 2000. Sorting in the endosomalsystem in yeast and animal cells. Curr. Opin. Cell Biol. 12:457–466.

Pornillos, O., et al. 2002. Mechanisms of enveloped RNA virusbudding. Trends Cell Biol. 12:569–579.

Riezman, H., P. Woodman, G. van Meer, and M. Marsh. 1997.Molecular mechanisms of endocytosis. Cell 91:731–738.

Robinson, M. S., C. Watts, and M. Zerial. 1996. Membrane dy-namics in endocytosis. Cell 84:13–21.

Rudenko, G., et al. 2002. Structure of the LDL receptor extra-cellular domain at endosomal pH. Science 298:2353–2358.

Synaptic Vesicle Function and Formation Betz, W., and J. Angleson. 1998. The synaptic vesicle cycle. Ann.

Rev. Physiol. 60:347–364.Chapman, E. R. 2002. Synaptotagmin: a Ca2� sensor that trig-

gers exocytosis? Nature Rev. Mol. Cell Biol. 3:498–508.De Camilli, P., et al. 1995. The function of dynamin in endocy-

tosis. Curr. Opin. Neurobiol. 5:559–565.Geppert, M., and T. Sudhof. 1998. Rab3 and synaptotagmin:

the yin and yang of synaptic membrane fusion. Ann. Rev. Neurosci.21:75–96.

Neimann, H., J. Blasi, and R. Jahn. 1994. Clostridial neurotox-ins: new tools for dissecting exocytosis. Trends Cell Biol. 4:179–185.


Documents

VESICULAR TRAFFIC, SECRETION, AND ENDOCYTOSISkbp-srmc.yolasite.com/resources/Chapter 17.pdf · 2010-09-24 · green fluorescent protein (GFP), a naturally fluorescent pro-tein (Chapter