A role for glycosphingolipids in protein sorting - Thesis

UvA-DARE is a service provided by the library of the University of Amsterdam (https://dare.uva.nl)

UvA-DARE (Digital Academic Repository)

A role for glycosphingolipids in protein sorting

Sprong, H.

Publication date

2001

Document Version

Final published version

Link to publication

Citation for published version (APA):

Sprong, H. (2001). A role for glycosphingolipids in protein sorting.

General rights

It is not permitted to download or to forward/distribute the text or part of it without the consent of the author(s) and/or copyright holder(s), other than for strictly personal, individual use, unless the work is under an open content license (like Creative Commons).

Disclaimer/Complaints regulations

If you believe that digital publication of certain material infringes any of your rights or (privacy) interests, please let the Library know, stating your reasons. In case of a legitimate complaint, the Library will make the material inaccessible and/or remove it from the website. Please Ask the Library: https://uba.uva.nl/en/contact, or a letter to: Library of the University of Amsterdam, Secretariat, Singel 425, 1012 WP Amsterdam, The Netherlands. You will be contacted as soon as possible.

AA role for glycosphingolipids

inin protein sorting

K. .

i i

AA role for glycosphingolipids in protein sorting

Academischh proefschrift

terr verkrijging van de graad van doctor aan de Universiteit van Amsterdam

opp gezag van de Rector Magnificus Prof. Dr. JJ.M. Franse

tenn overstaan van een door het college voor promoties ingestelde commissie,

inn het openbaar te verdedigen in de Aula der universiteit

opp dinsdag 29 mei 2001, te 12:00 uur

door r

Heinn Sprong

Promotiecommissie e

promotores: : Prof.. Dr. G.F.B.P. van Meer

verbondenverbonden aan de faculteit geneeskunde, Universiteit van Amsterdam

Prof.. Dr. G.J.A.M. Strous

verbondenverbonden aan de faculteit geneeskunde, Universiteit Utrecht

co-promotor:: Dr. P. van der Sluijs

verbondenverbonden aan de faculteit geneeskunde, Universiteit Utrecht

promotiecommissie: : Prof.. Dr. J.M.F.G. Aerts Prof.. Dr. B. de Kruijff Prof.. Dr. C.J.F, van Noorden Prof.. Dr. PJ.M. Rottier Prof.. Dr. H.F. Tabak Prof.. Dr. R.J.A. Wanders Prof.. Dr. F. Wieland

Thee research described in this thesis was carried out at the Department of Cell Biology and Histology,, Medicine school, University of Amsterdam and the Department of Cell Biology, Medicinee school, University of Utrecht, The Netherlands. The research was financed by grants fromfrom the Netherlands Foundations for Chemical Research and Life Sciences (to G.v.M. and P.v.d.S.),, European Community and Mizutani Foundation for Glycoscience (to G.v.M.), and thee Jan Dekker and Ludgardine Bouwman Stichting (to P.v.d.S.). The confocal microscope and STORMM 860 image facilities are supported by grants from the Netherlands Organization for Medicall Research (to G.S. and P.v.d.S., respectively). This thesis was printed by Thela Thesis, Amsterdam. .

Contents s

Chapterr Page

11 Introduction 1 22 Analysis of galactolipids and UDP-galactose:ceramide

galactosyltransferasee 17 33 UDP-Galactose:ceramide galactosyltransferase is a class I integral

membranee of the endoplasmic reticulum 27 44 Association of the galactose transporter with

UDP-galactosexeramidee galactosyltransferase allows UDP-galactose

importt in the endoplasmic reticulum 43 55 Glycosphingolipids are required for sorting of melanosomal

proteinss in the Golgi complex 57 Summarizingg discussion 79 Glossaryy 81 Abbreviationss 82 Referencess 83 Nederlandsee samenvatting 95 Dankwoordd 98 Curriculumm vitae 100 Listt of publications 101

Chapterr 1

Introduction n

Heinn Sprong, Peter van der Sluijs, and Gerrit van Meer

Summary y

Cellularr membranes differ in protein and lipid composition. For membrane proteins, this is largelyy caused by specificity in vesicular transport. Most membrane proteins are targeted by addresss labels in their structure that interact with proteins on the cytosolic surface of transport vesicless and target organelles. For lipids, local metabolism and selective transport are the importantt determinants. Specificity in transport is more complex for lipids since they can also bee transported as monomers through the cytosol and across membranes. While lipid transport andd sorting is mediated by membrane proteins, interactions between lipids play also a crucial role.. Some protein sorting is mediated by lipids. This concerns aggregation into domains but alsoo coat recruitment via signaling lipids. Topologically and temporally restricted metabolism off lipids apparently regulates fluxes through vesicular transport pathways.

Howw proteins move lipids and lipids move proteins

Eukaryoticc cells use membranes to compartmentalize their metabolism. Every cellular membranee possesses unique proteins to perform its specialized functions. The basic building blockss of membranes are lipids. Lipids provide the mechanical stability and strong tendency to formm closed structures. At the same time, lipids bestow the flexibility needed for vesiculation andd fusion, and for the snug fit of membrane-spanning proteins. Bilayers of the prototype membranee phospholipid phosphatidylcholine (PC; Box 1), display all of these properties in the testt tube. Cellular membranes, however, contain many different types of lipids that also serve inn signaling via specific interactions with proteins (Box 2). These two modes of action play a cruciall role in membrane transport and protein sorting, the two processes that underly the uniquee protein compositions of the membranes along the exo- and endocytotic recycling pathway. .

Too understand the biological functions of lipids, we must have insight in their physical propertiess in the cellular context, i.e. in lipid mixtures (Box 3). In addition, the activity of a lipidd is determined by its local concentration, which means its density at a certain lateral positionn on one surface of a particular membrane, in time. As this parameter is determined by synthesiss and hydrolysis, and by transport, we need to learn about these processes, the proteins involvedd and how their activity is regulated. During evolution, the selective transport of proteinss to specific cellular locations and the protein-mediated transport of lipids have evolved coordinately.. It can therefore be expected that membrane proteins and lipids make up indispensablee parts of an intricate sorting machinery. How does this work at the molecular level?? In other words: How do lipids and proteins move eachother?

Lipidd composition of organelles

Alll cellular membranes contain essentially the same lipids, but in different relative amounts. In hepatocytes,, the endoplasmic reticulum (ER) is rich in the glycerophospholipids PC, phosphatidylethanolaminee (PE) and phosphatidylinositol (PI). In contrast, the plasma membranee contains some 50 mol% cholesterol and sphingolipids, mostly sphingomyelin (SM), andd a reduced concentration of PC (1, 2). An increase in phosphatidylserine (PS) has been observedd in most studies. In addition, at least one-third of the phospholipids in the plasma membranee are disaturated, whereas these species are virtually absent from the ER (1). Generally,, the plasma membrane contains more than 65% of the cellular SM and an even higherr fraction of the cellular cholesterol. Finally, the lipids in the ER seem to be distributed symmetricallyy over both leaflets of the membrane, whereas in the plasma membrane lipids are asymmetricallyy arranged with most sphingolipids in the outer bilayer leaflet and most of the PS andd PE in the cytosolic leaflet (Figure 1).

Thee unique phospholipids cardiolipin and lysobisphosphatidic acid (LBPA) are synthesized at thee site where they are localized, the inner mitochondrial membrane and the endosome, respectivelyy (Figure 1). In contrast, the synthesis of PC, PE, PI, and PS is limited to the cytosolicc surface of the ER membrane, by transfer of an activated headgroup to diacylglycerol (DAG):: PC and PE (3), of inositol onto activated DAG: PI (4), or by base-exchange of the cholinee or ethanolamine of PC and PE for serine: PS (5). In addition, PE is synthesized by decarboxylationn of PS at the mitochondrial inner membrane (6, 7). Glycerophospholipids can bee degraded by phospholipases in the lysosomal lumen. Moreover, they undergo extensive deacylationn by phospholipases and reacylation from acyl-CoA by acyltransferases in the cytosol,, at both the CI and C2 position of the glycerol (8). It is presently unclear whether reacylationn could be responsible for the high content of saturated acyl chains of plasma membranee phospholipids.

Thee major sphingolipid, SM, is synthesized in the Golgi lumen. It can be hydrolyzed by acid sphingomyelinasee in the lysosome. However, neutral sphingomyelinases have been identified inn the ER and on the cytosolic surface of the plasma membrane (9,10). They may contribute to thee low SM concentrations at those locations (Figure 1). Also, glycosphingolipid synthesis occurss on the lumenal surface of the Golgi apparatus, with two exceptions: (I) galactosylceramidee (GalCer) is synthesized on the lumenal surface of the ER membrane (chapterr 3), and (II) glucosylceramide (GlcCer) is synthesized on the cytosolic surface of the Golgi.. GlcCer is then transported to the lumenal side of the Golgi membrane (see below), wheree it is converted to lactosylceramide, that serves as precursor for different series of complexx glycosphingolipids (Box 1). Most cells synthesize one of these series plus certain gangliosides.. Synthesis of the complex glycosphingolipids seems concentrated in the trans-Golgii and trans-Golgi network (TGN; 11). The enzymes for glycosphingolipid hydrolysis are concentratedd in the lysosomal lumen. Their inactivation results in lysosomal storage of sphingolipidss with characteristic and often severe pathologies in man (12).

Cholesteroll is synthesized in the ER while parts of the synthetic route may lead through the peroxisomess (13). In addition, cholesterol is directly internalized at the cell surface. Finally, it cann be released from cholesterol esters, in late endosomes from endocytosed lipoproteins or in lipidd droplets. In the ER, surplus cholesterol is removed by acylation to cholesterol ester, that is storedd in lipid droplets or secreted in lipoproteins (14), and it is removed from the plasma membranee by transport to extracellular high density lipoproteins (15).

Selectivityy in lipid transport

Lipidss are transported between organelles via various mechanisms (Box 4). Each of these wouldd result in dissipation of the compositional differences. Are the various pathways of intracellularr lipid transport specific for certain lipids, and, if so, how is specificity achieved? Sphingolipids Sphingolipids

Thee organelles along the secretory and endocytic pathways are connected by bidirectional vesicularr traffic (Figure 2). The distinct membrane composition of each organelle shows that vesicularr transport is not a random process. For membrane proteins, this means that after insertionn into the ER membrane they are preferentially incorporated into those budding transportt vesicles at the ER and each intermediate compartment, that travel to and fuse with the nextt compartment, while they are excluded from budding vesicles with a different destination. Thus,, the proteins destined for the various organelles are laterally segregated. For lipids the problemm of selectivity in transport is more complicated because they can in principle also be transportedd as monomers through the cytosol (Box 4). However, the major difference between ERR and plasma membrane is an enrichment of sphingolipids in the exoplasmic leaflet of the latter.. SM and the complex glycosphingolipids, synthesized in the lumen of the Golgi, have no accesss to monomelic transport (16, 17). Thus, sphingolipids must laterally segregate from glycerophospholipidss in the lumenal leaflet of a Golgi cisterna. A solid physico-chemical basis forr this behavior has been uncovered (Box 3): sphingolipids experience stronger van der Waals interactionss and hydrogen bonding. After segregation, the sphingolipid-rich domain must be incorporatedd into anterograde transport vesicles or excluded from retrograde transport vesicles (18).. The latter is more likely for a liquid-ordered domain (Box 3), and indeed COPI-coated transportt vesicles derived from the Golgi contained reduced levels of SM (19). This mechanismm may explain the low concentration of SM and complex glycosphingolipids in the ER,, and provides a simple explanation for the lack of these sphingolipids in mitochondria and peroxisomes,, which are not connected to vesicular transport pathways (20).

Glycerolipids s

BoxBox 1: Structure of mammalian membrane lipids

Glycerolipids:Glycerolipids: The most abundant lipid in animal membranes, phosphatidylcholine (PC),

consistsconsists of a glycerol with two fatty chains on the sn-1 and sn-2 positions and a phosphate (phosphatidic(phosphatidic acid, PA, or phosphatidyl-) carrying the choline base at sn-3. Various C16-C20 fattyfatty acids are found esterified to the sn-2 position but, generally, a saturated fatty acid (CI 6:0

oror C18.0) is esterified at sn-1 (diacy[glycerol; DAG). Alternatively, a long chain alcohol is

ether-bondedether-bonded at sn-1 (alkyl-acylglycerol; AAG). Plasmalogens are ether-lipids based on an unsaturatedunsaturated alcohol (alkenyl-acylglycerol). The double bonds in plasmalogens and acyl chains havehave the cis-configuration, which creates a kink in the chain and increases the membrane area

occupiedoccupied by the lipid. Glycerophospholipids are denoted as "phosphatidyl" or "P" followed by thethe name of the headgroup: choline, PC; ethanolamine, PE; serine, PS; inositol, PL PI can carrycarry additional phosphates as in PI(3)P, PI(4)P, P1(3,4)P2, PI(3,5)P2, PJ(4,5)P2, and PI(3,4,5)P3.PI(3,4,5)P3. Sphingoltpids: They contain a C18-C20 sphingoid base, mostly sphingosine (SPH;(SPH; 87). Membrane sphingolipids carry a fatty acid, amide-linked to their nitrogen (ceramide).(ceramide). Dihydroceramide contains sphingosine lacking the trans-double bond at C4 (sphinganine/(sphinganine/ dihydrosphingosine). Phytoceramide contains phytosphingosine (C4-0H sphinganine,sphinganine, PHS). The fatty acid is often long and saturated, sometimes cis-unsaturated at CIS,CIS, and sometimes hydroxylated at C2. LacCer: lactosylceramide. Sphingolipids of yeast are basedbased on inositolphospho-phytoceramide and carry longer fatty acids, notably C26.0. Sterols: CholesterolCholesterol can be esterified to a fatty acid. Like triacylglycerol, cholesterol esters are storage lipids.lipids. In animals, most other sterols act as hormones. The yeast sterol is ergosterol.

Thee sphingolipid-glycerolipid segregation is a basic feature of lipid sorting in the Golgi. However,, sorting in the distal Golgi is more complex than that. Multiple vesicular pathways originatee in the trans-Golgi and TGN, and except for the fact that glycerolipids are transported inn the retrograde direction, little is known about how lipids partition into the anterograde pathwayss to, for example, apical and basolateral cell surfaces, secretory granules, endosomes, andd other specialized compartments like melanosomes. A similar complexity is observed in the endosomall system. Experiments with fluorescent probes suggest that lipid sorting in this pathwayy follows the same principles as in the Golgi. From early endosomes, more fluid probes enteredd vesicles towards the recycling endosomes whereas the less fluid ones proceeded to late endosomess (21). Nevertheless, recycling endosomes were enriched in domain markers like SM (22).. Specificity was also found for transport from the endosomes to the TGN (23, 24). Exactly howw the native lipids behave remains to be determined (25, 26).

GlcCerr displays unique transport properties. After synthesis on the cytosolic surface of the Golgi,, one part translocates towards the Golgi lumen by an energy-independent process and is convertedd into complex glycosphingolipids. However, all GlcCer that reached the fibroblast surfacee was translocated directly across the plasma membrane by the ATP-binding cassette (ABC)) transporter MDR1 P-glycoprotein (Raggers, R. et al., manuscript in preparation; 27). MDR11 is a ubiquitously expressed multidrug transporter responsible for resistance of cancer cellss against chemotherapy. GlcCer translocation may regulate a vesicular transport step (see below). .

Glycerophospholipids Glycerophospholipids

Whichh glycerophospholipids are separated from the sphingolipids in the Golgi lumen? Informationn can be obtained from the phospholipids of lipoproteins in the Golgi lumen of hepatocytes.. These are covered by a phospholipid monolayer obtained from the lumenal leaflet off the Golgi via monomelic transfer (Box 4). Lipoproteins contain mostly PC and SM (65 and 122 mol%), minimal amounts of PE (5 mol%) and essentially no PS (28). From this and the compositionn of the Golgi membrane (28) most PE and PS must be on the cytosolic surface. Possibly,, the phospholipid asymmetry is generated in the Golgi or already in the ER: originally,, it was proposed that of the newly synthesized glycerophospholipids only PC could traversee the ER membrane by a PC-specific translocator (29). This has become uncertain since short-chainn analogs of numerous phospholipids readily move across the ER membrane (30). PS andd PE are continuously removed from the cell surface by the ATP-requiring aminophospholipidd translocator (31), which may also be active in the Golgi, translocating PS andd PE to the cytosolic surface. The phospholipid asymmetry in the plasma membrane can be disruptedd by the activation of an energy-independent scramblase, that exposes PS on the cell

surface.. A liver-specific ABC transporter with 80% homology to MDR1 P-glycoprotein translocatess PC into bile (32).

BoxBox 2: Signaling lipids

AA dazzling array of membrane lipids and fragments derived from them have been found to act asas signaling molecules within and between cells. These interact with specific proteins, or they changechange the local physical properties of a membrane. Cells do not strictly discriminate between thesethese modes of action, as is illustrated by ceramide signaling. The conversion of SM to ceramideceramide by a sphingomyelinase may activate a specific protein kinase and a phosphatase but alsoalso grossly affects membrane properties (88, 89). Cells must be capable of separating metabolicmetabolic pools from signaling pools of one and the same lipid: sphingosine, derived from ceramide,ceramide, once phosphorylated to sphingosine-1-phosphate, interacts with specific receptors (90)(90) that are related to the receptors for its glycerolipid homolog LPA (91). Sphingosine-1-phosphatephosphate is also the final intermediate in the normal breakdown of sphingolipids. One

beautifulbeautiful example of how cells utilize lipids for signaling purposes is the phosphoinositide system.system. The intricacy of the inositide signaling system is illustrated by the known involvement ofof close to a hundred different (isoforms of) kinases, phosphatases, and phospholipases in the

productionproduction and inactivation of these signaling molecules, and also their specific recognition by receptorreceptor domains on at least a hundred proteins more (74). Cholesterol is a signaling molecule inin the sense that its concentration in the cell is tightly regulated by feedback systems at the

transcriptionaltranscriptional level (92). A sensor in the ER membrane registers the local cholesterol concentration,concentration, which reflects the chemical activity of cholesterol in the cell (93), i.e. its availabilityavailability for equilibration with the ER. Since sphingolipids act as a cholesterol buffer in the plasmaplasma membrane, this system also senses the concentration of sphingolipids, notably SM (36). PlateletPlatelet activating factor (Alkyl-acetyl PC) and PC-derived peroxidation products with platelet activatingactivating factor-activity signal between cells (94). The eicosanoids: prostaglandins, thromboxanesthromboxanes and leukotrienes, are synthesized from arachidonic acid (C20:4), produced from phospholipidsphospholipids by regulated phospholipases A2. Steroid hormones are produced in

mitochondriamitochondria from cholesterol.

PhospholipasesPhospholipases A2 cleave the fatty acid from the sn-2 position of the glycerol. Phospholipases CC remove the headgroup leaving DAG, AAG, or, in the case of sphingomyelinase, ceramide.

PhospholipasesPhospholipases D cleave off the headgroup without the phosphate leaving PA. PA is hydrolyzedhydrolyzed by PA phosphohydrolases to DAG (or AAG) or by phospholipase A2 to LPA.

Mitochondriaa obtain their glycerophospholipids by monomelic transfer. This process probably occurss with high efficiency at ER-mitochondrial contact sites, of which presently no molecular detailss are known (6). Monomelic transport, especially of PC, is rapid. The relevance of the PC-transferr protein for PC transport is not clear (33). Pi-binding proteins may serve regulatory functionss in vesicle flow and lipid metabolism (34).

Cholesterol Cholesterol

Cholesteroll is concentrated at the same locations as sphingolipids. Its distribution is essentially governedd by the high affinity for sphingolipids (35, 36) and disaturated glycerophospholipids,

especiallyy PS. This suggests that also the cytosolic leaflet of the plasma membrane, that is enrichedd in disaturated PS and PC (1) contains a high concentration of cholesterol. Cholesterol wass found enriched in the cytosolic leaflet of the erythrocyte membrane (37, 38). Cholesterol couldd be continuously depleted from the ER and Golgi by the anterograde sorting of the sphingolipidss (18). However, cholesterol readily equilibrates through the aqueous phase: the half-timee (tVi) is 1-2 h from a PC surface in a dilute liposome suspension, probably shorter for transferr between organelles. However, this tVi is 10-20 h from a SM surface (39)! Newly synthesizedd sphingolipids in the trans-Golgi and TGN may attract cholesterol from both the closelyy apposed ER (40) and from the cytosol. The cholesterol then contributes to (or induces) thee sphingolipid/glycerolipid phase separation (Box 3). A function in cholesterol transport betweenn organelles has been claimed for caveolin. Caveolin constitutes the cytosolic coat of caveolae,, plasma membrane specializations enriched in sphingolipids and cholesterol. Caveolin iss tightly associated with membranes through its hydrophobic hairpin domain and three palmitoyll (CI6:0) chains, but may also occur in a soluble complex with cholesterol and three chaperoness in the cytosol. How this complex would shuttle cholesterol is not clear (41).

Forr a long time, cholesterol was assumed to move across membranes rapidly. However, measurementss on erythrocytes suggested a tV2 of 1-2 h (37). In addition, efficient "reverse"

transportt of intracellular cholesterol to extracellular lipoproteins depends on the activity of ABCA1,, a plasma membrane ABC transporter missing in Tangier disease. ABCA1 may translocatee cholesterol across the plasma membrane, possibly indirectly by translocating PS (42).. In the Niemann-Pick diseases type C (NPC1 and NPC2), endocytosed lipoprotein cholesteroll is unable to leave late endosomes and lysosomes (43). Apparently, under these conditions,, spontaneous transport of cholesterol from the lipoprotein to and across the endosomall membrane is too slow to cope with the incoming cholesterol. From the above, this mayy have two reasons. Cholesterol could be kinetically trapped by cholesterol-binding lipids accumulatedd in the endosomes due to defects in hydrolysis, as in the sphingolipid storage diseasess (44), due to defective removal of cholesterol-binding breakdown products like sphingosine,, or due to a change in the physical state of LBPA (45). Alternatively, cholesterol mayy accumulate due to a defect in a protein that is involved in its transport out of the endosome.. The NPC2 protein, HE1, is a cholesterol-binding protein in the lysosomal lumen (46),, that could be involved in the transfer step. The role of the LBPA-rich internal vesicles in transferr is unclear (45). The NPC1 protein is a proton motive force-driven lipid transporter (47),, that has a sterol-sensing domain. However, it does not translocate cholesterol. Since sphingoidd bases accumulate in NPC1 (48), and exogenous sphinganine induces an NPC phenotype,, possibly by neutralizing the low pH (43) or by interfering with LBPA (45), NPC1 mightt as well serve as a sphingosine transporter. Another cholesterol-binding membrane proteinn localized to late endosomes (49) displays high homology to StAR, a protein involved in cholesteroll transport from the outer to the inner mitochondrial membrane (50), and may stimulatee release of cholesterol from the cytosolic surface of endosomes. Also ether lipids (Box 1)) are required for cholesterol release (51). Unexpectedly, accumulation of cholesterol and sphingolipidss inhibits vesicle budding from late endosomes whereas the opposite is also true (52,, 53). Apparently, endosomal lipid processing is a delicate process that is easily frustrated (44,, 54, 55).

B B

R R

BoxBox 3: The molecular shape of lipids determines the physical properties of membranes

A:A: The molecular shape of a membrane lipid depends on the relative size of polar headgroup andand apolar tails (95). When the headgroup and lipid backbone have similar surface areas, the moleculemolecule has a cylindrical shape (PC and PS). Lipids with a small headgroup like PE are cone-shaped.cone-shaped. The cone shape of PA can be enhanced by interaction with Ca +. In contrast, whenwhen the hydrophobic part occupies a relatively smaller surface area, the molecule has the shapeshape of an inverted cone. This is true for LPC and, to some extent, for SM. The inverted cone-shapeshape of phosphorylated Pis can be modulated by Ca2+. This lipid polymorphism may play a physiologicalphysiological role in the generation of curvature as during vesicle budding (96). The example

isis based on the lipid asymmetry in the plasma membrane and may apply to TGN and

endosomes:endosomes: the cytosolic surface contains 40% PE, 60% PS plus PC, and the lumenal leaflet 60%60% PC, 30% SM and 10% PE. Note that the neck of the budding vesicle is constricted by the

assemblingassembling coat (COPI and COPIJ vesicles) or by the GTP-ase dynamin (clathrin-coated vesicles;vesicles; 97). Budding in the opposite direction, towards the lumen of endosomes involves membranemembrane sorting (66), may require LBPA, an inverted cone, on the lumenal surface (45) and PI(3)P,PI(3)P, which is also enriched in these vesicles (75, 76). B: The surface area ofSM or PC plus cholesterolcholesterol is lower than the sum of the individual areas (condensation). Especially in the case ofof PC, this results in an increased length of the molecule. SMwith or without cholesterol forms

bilayersbilayers with a thickness, i.e. headgroup separation, of 46/47 A for C18.0-SM (98) to 52-56 A forfor C24.0-SM (99). In contrast, the thickness of a C16:0/C18:1-PC bilayer is 35 A, and was

expandedexpanded to 40 A by cholesterol. The thickness of the hydrophobic core of the bilayer increasedincreased from 26 to 30 A (100). In a mixture of sphingolipids and glycerophospholipids, cholesterolcholesterol can induce fluid-fluid immiscibility resulting in a lateral segregation into two or moremore fluid phases, whereby cholesterol is enriched in the sphingolipid domain. Along the same line,line, cholesterol can interact with disaturated PS (101) on the cytoplasmic surface. The

cholesterol-richcholesterol-rich phases have a more ordered structure and have been termed liquid-ordered (vs.(vs. liquid-disordered). Depending on whether the domains on both surfaces colocalize, four

discretediscrete bilayer thicknesses could be present in these membranes. In the hypothetical example, thethe headgroup separations would be 35 A (a), 41 A (b), 47 A (c) and 43.5 A (d). The

hydrophobichydrophobic thickness would be three-quarters of these numbers. PE is typically excluded from rafts,rafts, and has low affinity for cholesterol. Being enriched on the cytosolic surface it will reduce thethe thickness of the cytosolic leaflet.

Physicall role of lipids in protein transport

SphingolipidSphingolipid domains sort proteins

Mixturess of cholesterol and SM constitute a thicker membrane leaflet than PC/cholesterol, whichh is in turn thicker than a monolayer of PC or PS (Box 3). Cells utilize this feature to sort membranee proteins destined for the plasma membrane from Golgi proteins according to the lengthh of their transmembrane domains (Figure 3). This has been demonstrated most convincinglyy for yeast (56). The domains may be excluded from retrograde COPI-coated vesicless because the bending energy of these rigid domains is higher than that of unsaturated lipids.. They will end up in TGN remnants after removal of all retrograde material and move to thee plasma membrane without going through a budding step. In order to dock and fuse, the vesicless need the proper SNARES, and sorting of SNARES by thickness has been demonstratedd (57). Although discrete increases in membrane thickness would allow the sorting off various populations of membrane proteins (Box 3), an attractive possibility is that the membranee gradually thickens along the cis-trans axis of the Golgi (58) and that membrane proteinss partition to the Golgi cistema that fits the length of their transmembrane domain. Recently,, we have uncovered a requirement for the glycosphingolipid GlcCer in transport of membrane-spanningg proteins from the Golgi to the melanosome (chapter 5). The molecular mechanismm is unclear.

Inn addition, sphingolipid/cholesterol domains are recognized by proteins that are anchored to thee non-cytoplasmic leaflet of the membrane by a glycosylphosphatidylinositol anchor and by transmembranee proteins that are palmitoylated. By the same methodology, domains are found onn the cytosolic side that are populated by peripheral proteins carrying myristoyl- and palmitoyll chains, whereas prenylated proteins, peripheral proteins with famesyl- or geranylgeranyl-tails,, are excluded (Figure 3; 59). The cholesterol-binding protein caveolin on thee cytosolic surface colocalized with the ganglioside GM1 in the opposite leaflet (60). The opposedd domains may colocalize by lipid-lipid interactions. Alternatively, caveolin may recognizee the domain in the opposite leaflet via its hydrophobic hairpin, and may be responsiblee for the transbilayer alignment of the domains. Although the physical details of lipid

domainss are not fully understood, it has become clear that besides their role in transport, they fulfill major functions in cell signaling (61, 62).

BB A

ganglioside e

BoxBox 4: Mechanisms of lipid transport between cellular membranes

A:A: Membrane lipids diffuse laterally in the membrane (I), they can translocate between the two

leafletsleaflets of the bilayer (II), diffuse into the cytosol as monomers and equilibrate with the cytosoliccytosolic surface of another organelle (III), and be included in carrier vesicles, whereby the transbilayertransbilayer orientation of the membrane lipids is maintained during fission and fusion (IV). B:B: The rates of spontaneous movement across the bilayer and diffusion into the aqueous phase

areare inversely related. As the polar headgroup becomes larger or more polar, it traverses the hydrophobichydrophobic phase less readily but leaves the bilayer more easily. An extreme example is the differencedifference between fluorescent analogs of DAG and PC (102). The replacement of the sole hydroxyhydroxy group by phosphocholine reduced transbilayer movement (t'A = 70 ms vs. 7.5 h) but increasedincreased desorption into the aqueous phase (t'A = 100 h vs. 6 min). The same is true when the hydrophobichydrophobic moiety becomes smaller, LPC vs. PC. On the contrary, when the hydrophobic tail becomesbecomes more hydrophobic, transbilayer translocation becomes easier whereas the tendency toto leave the membrane is reduced. For example, (fluorescent) DAG is more hydrophobic than (fluorescent)(fluorescent) ceramide, with a t'A of 70 ms vs. 22 min and a t'A of desorption of 100 h vs. 7 min (102).(102). When the size and/or hydrophobicity of the hydrophobic moiety increase concomitantly thesethese neutralize the effect of the larger head group on the off-rate. Complex glycosphingolipids havehave extremely low off-rates from membranes (t'A >1000 h; 39). Both rates are probably reducedreduced when lipids are closely packed in liquid-ordered domains.

Apical l

A A

* Golgi j . ;

c c

endoplasmicc reticulum

nucleus s

Basolateral l

FigureFigure 1: Topology of membrane lipids in intestinal epithelial cells

TheThe apical plasma membrane contains 30 mol% sphingolipids, that are concentrated in the outerouter bilayer leaflet and cover the apical surface. The basolateral membrane represents the typicaltypical plasma membrane of non-polarized cells with four times less sphingolipids, the remainderremainder being PC. The difference is maintained by the tight junctions that act as a diffusion barrierbarrier for these lipids in the outer leaflet (103). Cholesterol constitutes 30 mol% in both apicalapical and basolateral membranes. Its transbilayer distribution is unclear. The phospholipids PS,PS, PE, plus some PC (30 mol% in apical and 50 mol% in basolateral membranes) are situatedsituated on the cytosolic surface. Endosomes and lysosomes are similar to the plasma membranemembrane but contain significant concentrations of the unique phospholipid LBPA in their internalinternal vesicles (55). Hardly any sphingolipids and cholesterol are found in the ER, and the GolgiGolgi contains intermediate concentrations. Peroxisomes and mitochondria have ER-like lipid compositions,compositions, but mitochondria contain the unique phospholipid cardiolipin in their inner membranemembrane (6). The transbilayer distribution of lipids in these organelles is unknown. The distributiondistribution of membrane proteins (triangles, circles, stars, and diamonds) is often restricted to oneone organellar membrane.

MembraneMembrane tension and curvature

Endocytosiss is inhibited during mitosis. Apart from the inactivation of essential proteins (63), thee membrane tension increases. Release of the tension restores endocytosis, suggesting that endocytosiss is regulated via membrane tension (64). In addition, vesicle budding requires membranee bending and the generation of a lipid imbalance across the bilayer. At the level of thee headgroups, the outer leaflet of a 60 nm diameter vesicle contains 1.5 times the number of lipidd molecules of the inner leaflet. This may not be a problem in a membrane where lipids can freelyy cross the bilayer, like the ER and presumably the cis Golgi. In these flexible membranes thee assembly of the COP-coats appears to be sufficient for budding. However, the situation is differentt in membranes where lipids do not experience free transbilayer movement, like the plasmaa membrane, the TGN or endosomes. It has been proposed that the ATP-consuming aminophospholipidd translocator, a flippase, drives vesicle budding from the plasma membrane byy expanding the surface area of the cytoplasmic at the expense of the non-cytoplasmic bilayer leaflett (65). Clearly, this poses a problem at the level of late endosomes. Here, vesicles bud towardss both the cytosol and the endosomal lumen (45). The driving force behind this lumenal buddingg event is unknown but it involves membrane sorting (66). Possibly, at some stage of endosomall maturation budding towards the cytosol is blocked, allowing budding in the oppositee direction. At a later stage cytoplasmic budding would resume. Some cases of lipid storagee may be due to obstruction of these transitions.

FigureFigure 2: Vesicular transport pathways

TheThe organelles along the exocytic and endocytic transport routes are connected by carrier vesiclesvesicles budding from one organelle and fusing with the next. Transport can be visualized as twotwo recycling pathways, one between ER and cis Golgi, and one between endosomes and the plasmaplasma membrane. The latter has a bidirectional connection with the TGN. Vesicles move up

andand down the Golgi stack. Each arrow may represent more than one pathway mediated by differentdifferent coats and, therefore, may be regulated independently. Examples are COPI and COPII-mediatedCOPII-mediated pathways out of the ER (104), AP-1 and AP-3 mediated pathways from the TGNTGN to the endosomal system (105), and AP-2/clathrin, caveolin and

non-clathrin/non-caveolincaveolin mediated pathways from the plasma membrane (106).

Vesiclee budding requires membrane bending. Besides the positive curvature of the forming bud,, an extreme, negative curvature must be generated at the site of membrane fission (Box 3).

Thee cone shape of some lipids makes them ideally suited for fitting in the area of constriction. Thee budding of various types of transport vesicles in the Golgi involves the conversion of an invertedd cone, lysoPA (LPA), into a cone, PA, by acylation utilizing acyl-CoA (67, 68). PA is hydrolyzedd by a phosphohydrolase to DAG with an even more conical shape (69). LPA formationn involves the action of phospholipases D and A2 (Box 2). A phospholipase D is activatedd during vesicle budding (70). The activity of a phospholipase A2 appears required for tubulationn (71), which may have a similar function as vesicle percolation (72, 73).

Sortingg by

coatss bilayer thickness lipid anchors oligomerization

FigureFigure 3: Lateral sorting of membrane proteins

ProteinsProteins can be sorted via address labels in their cytosolic tails that interact with a protein coatcoat directing the resulting vesicle towards a specific target organelle. The interaction may be directdirect as in the case of COP-coats, or indirect via adaptor proteins (APs) as in the case of clathrinclathrin coats. Proteins may also be sorted by the length of their hydrophobic domain, whereby thethe incorporation of lipid domains of a certain thickness into distinct transport carriers may be determineddetermined by physical properties of the membrane or via address labels on proteins present in eacheach domain. Lipid-anchored proteins are sorted via the partitioning properties of their anchor:anchor: (I) Proteins anchored via a glycosylphosphatidylinositol partition into liquid-ordered domains.domains. (II) Acylation of proteins with myristoyl- and palmitoyl chains tails is a signal for incorporationincorporation into the same part of the membrane that contains sphingolipids and cholesterol, suggestingsuggesting that the cytosolic surface of these sphingolipid/cholesterol domains has a special lipidlipid composition as well. (Ill) Prenylated proteins, which are anchored by farnesyl- or

geranylgeranyl-tails,geranylgeranyl-tails, are excluded from the sphingolipid/cholesterol domains. Finally, proteins maymay be sorted via binding to another membrane protein. Oligomerization may be a physical

determinantdeterminant in localization of proteins to the Golgi.

PhosphoinositidesPhosphoinositides in membrane budding

Originally,, the signaling functions of phosphoinositides (Pis) were thought to reside in the breakdownn products of PI(4,5)P2, that were generated by stimulus-activated Pi-specific

phospholipasee C. DAG activates protein kinases C and inositoltrisphosphate opens Ca channelss in the ER. Only later, it was realized that cells use the Pis themselves for regulatory functions.. They form membrane binding sites for soluble proteins and, as such, recruit proteins too membranes, stabilize protein complexes on membranes, or activate membrane proteins. Theyy are involved in signaling, cytoskeleton-membrane interactions, and in membrane vesicle buddingg (74). Specificity of each PI is based on its structure, its location and the timing of its synthesis,, modification and hydrolysis. Still, the same PI is found to complex with different proteinss on different membranes, showing that additional binding sites for the proteins must be presentt on their target membrane. Pis are products of and substrates for various families of kinases,, phosphatases, and phospholipases C and D. These enzymes rapidly alter the levels of particularr Pis in specific regions of the membrane, and the regulated activity of these enzymes providess the basis for the efficient spatial and temporal regulation of vesicular traffic.

Thee structures of the phosphoinositides can be found in Box 1 and 2. PI(3)P is found in endosomes,, and in their internal membranes, in yeast (75) and animal cells (76). In yeast it is essentiall for sorting proteins to the vacuole (75). In mammalian cells it is involved in various aspectss of endosomal transport. PI(3)P is recognized by FYVE domains in a wide variety of proteinss (77) and its synthesis by the 3'-kinase and hydrolysis in the endosome/vacuole have beenn characterized in detail (78, 79). However, the multiple mammalian PI 3-kinase isoforms phosphorylatee also other Pis (80). PI(3,4)P2 and PI(3,4,5)P3 are recognized by pleckstrin

homologyhomology (PH) domains on a range of proteins and function in plasma membrane to endosome transportt (81) and in signaling (80). Other phosphoinositides are recognized by PH domains

withh different binding specificities (82). They recruit and activate proteins in the COPI-mediatedd pathways between Golgi and ER and within the Golgi (Figure 2) starting with the firstt proteins required for COPI-coat assembly ARF1 and its guanine nucleotide exchange factorss (83). ARF1 also recruits the coats in AP-1, AP-3, AP-4 and GGA-dependent transport pathwayss (GGA; Golgi-localized, Gamma ear-containing, ARF-binding proteins; 84). Also dynamin,, which is involved in the final step of budding, vesicle fission, carries a PH domain (85).. These examples illustrate the wide-spread involvement of Pis in transport vesicle formation.. The functions of the same Pis in a variety of signaling cascades suggest that vesicle formationn in some complex way depends on the signaling state of the cell.

Finally,, also the lipid composition of the Golgi may modulate vesicle traffic. A membrane-boundd PI/PC binding protein has been proposed to regulate the interface between lipid metabolismm and Golgi secretory function by means of a Golgi DAG pool (34). Such a dual-specificityy lipid binding protein has also been found in the Golgi of mammalian cells where it seemss to be involved in metabolism (86).

Lipidd and protein transport

Overr the past 25 years, we have learned a great deal about the lipid composition of membranes, , lipidd transmembrane asymmetry, and lately on their heterogeneous lateral distribution. Also, wee have a basic understanding of the molecular mechanisms by which the cellular lipids are transported.. Novel protein players are being identified, often in the study of metabolic disorders,, and insights from the mode of action of related proteins sheds new light on lipid behaviorr in cells. Still, a number of fundamental questions remain. We have only sketchy informationn on the lipid composition of organellar membranes. In addition, our insights in the intracellularr dynamics of lipids is very limited. Finally, we know little about lipid-protein interactionss at the molecular level, let alone, the lipid-lipid interactions in complex mixtures. Solvingg the questions at the molecular level will depend on the application of novel biophysicall techniques, for example the study of the behavior of single molecules in

membranes.. At the cellular level, the new developments in high sensitivity mass spectrometry alloww the quantitative analysis of the lipids not only of cellular organelles but even of transport vesicles.. These novel techniques will have to be combined with careful biochemical approaches,, as was recently demonstrated by the broad application of novel cholesterol acceptorss in cholesterol transport studies. Comparable assays need to be developed for other membranee lipids.

Besidess the identification of novel lipid metabolic enzymes and transporters, genetic approachess reveal the fact that many of these belong to large families. The specific properties off individual family members and the effect of their combined actions on cellular lipid compositionn or dynamics will provide novel insights in how cells deal with lipids. Cells make usee of the possibilities offered by the tremendous heterogeneity of membrane lipids to modulatee their membrane proteins. It is a challenge to integrate the new information in our vieww of how the cell works as a living entity.

Outlinee of this thesis

Thee central aim of the research described in this thesis was to find out whether glycosphingolipidss are involved in the sorting of membrane proteins. A first step was to obtain moree insight in the organization of glycosphingolipid metabolism in cells. We studied the cellularr localization of glycosphingolipid synthesizing enzymes, particularly the UDP-galactosexeramidee galactosyltransferase (chapter 2 and 3). We learned that the synthesis of galactosylceramidee is dependent on an accessory protein, the UDP-galactose transporter (chapterr 4). By transfecting cells with the cDNA of glycolipid synthesizing enzymes we changedd the glycosphingolipid levels, and studied the effect on intracellular sorting of proteins (chapterr 5).

Chapterr 2

Analysiss of galactolipids and UDP-galactose:ceramide

galactosyltransferase e

Introduction n

Glycosphingolipidss form a highly polymorphic class of lipids andd several hundreds of the more thann 2000 possible molecular species (107) have been characterized (108). There are at least 20 differentt ceramide (Cer) backbones due to differences in sphingoid base, mostly sphingosine (4-sphingenine)) and phytosphingosine (4-hydroxysphinganine), and acyl chain. The headgroupss can vary from 1-60 sugars. Glycosphingolipids in mammals can be subdivided into twoo major classes, galacto- and glucosphingolipids, based on the presence of Gal or Glc as the firstfirst sugar moiety. Most complex glycolipids are based on Gal Bl-4 Glc Bl-1 Cer, lactosylceramide.. Galactosylceramide (Gal Bl-1 Cer or GalCer) serves as a precursor for few simplee glycolipids: the sulfatide SGalCer (S03-3 GalCer), galabiosylceramide (Gal a 1-4

GalCer,, and the ganglioside sialo-GalCer (I3NeuAc-GalCer or GM4). Gal and SGal are also foundd on diglycerides: Gal Bl-3 diacylglycerol (GalDAG), Gal 61-3 alkyl-acyl-glycerol (GalAAG),, digalactosyldiglyceride, and seminolipid: SGV3 GalAAG (109).

Glycosphingolipidss are enriched in the outer leaflet of the plasma membrane of most eukaryoticc cells where they are thought to be involved in cell recognition and signaling (107). Whilee glycosphingolipids constitute only a few mol% of the lipids in most membranes, they aree major components of the myelin sheath (110), where GalCer and SGalCer are involved in axonall insulation, myelin function, and stability (111, 112). The apical plasma membrane of epitheliall cells in the gastro-intestinal and urinary tracts is enriched in glycosphingolipids. In rodentss these are typically glucolipids (18, 108), whereas in humans most are galactolipids (113-115).. Glycosphingolipids play a structural role in rigidifying and protecting the apical cell surface.. Their role in sorting lipids and proteins to various membranes along the exocytotic and endocytoticc transport routes is not fully understood (18, 116).

Thee foremost enzyme involved in the biosynthesis of galactosphingolipids is the UDP-galactosexeramidee galactosyltransferase, GalT-1 (117). GalT-1 catalyses the transfer of galactosee from UDP-galactose (UDP-Gal) to Cer yielding GalCer (118) and has a relatively promiscuouss substrate specificity. Whether there are one or more GalT-1 enzymes with distinct specificityy and cellular localization has been a controversial issue (119-123). Importantly, knock-outt mice do not make GalCer (111, 112), showing there is only one GalT-1. In vitro studiess demonstrated that partially purified GalT-1 from brain has >15 fold preference for 2-hydroxyy fatty acid- over non-hydroxy fatty acid containing Cer (118, 124). This has been confirmedd for GalT-1 after transfection into GalT-1-negative cells (123, 125). In vivo, however,, GalT-1 is responsible for the galactosylation of 2-hydroxy fatty acid- as well as non-hydroxyy fatty acid containing Cer. The GalT-1 activity, specific for non-hydroxy fatty acid containingg Cer, which was previously found in the Golgi (120, 122, 123), has now been demonstratedd to be an in vitro activity of the Golgi UDP-glucose:ceramide glucosyltransferase (CGlcT;; chapter 3). GalT-1 is also responsible for the galactosylation of diglycerides (123, 126). .

Thee localization of GalT-1 has long been enigmatic (119, 127-130). Recently, we showed that thee enzyme was exclusively localized to the ER by immunogold electron microscopy on ultrathinultrathin cryosections (chapter 3). GalT-1 is a high-mannose type glycoprotein that is N-glycosylatedd at Asn 78 and Asn 333 (131) and contains a putative carboxy terminal Lys-Lys-Val-Lyss ER-retrieval signal (125, 132, 133). Surprisingly, the conceptual translation product exhibitss no amino acid sequence similarity with other glycosyltransferases. Instead, GalT-1 is relatedd to the superfamily of UDP-glucuronosyltransferases.

Thus,, while most glycosylation steps of sphingolipids occur in the Golgi complex, the GalT-1 enzymee activity resides in the lumen of the endoplasmic reticulum (Figure 1; chapter 3). Cer is

synthesizedd at the cytosolic surface and is sufficiently hydrophobic to diffuse freely across cellularr membranes. How the other substrate, UDP-Gal, reaches the active center of GalT-1 is unclear.. CH01ec8 cells, which are deficient in UDP-Gal import into the Golgi apparatus (134), aree also impaired in UDP-Gal import into the endoplasmic reticulum (chapter 3 and 4). Whetherr UDP-Gal import in the ER and in the Golgi complex is mediated by the same or distinctt UDP-Gal importers remains to be resolved. GalCer is converted to galabiaosyl-ceramidee (135) and sulfatide (136) in the lumen of the Golgi, from where these products cannot reachh the cytosolic surface (122). In contrast, GalCer can translocate from the lumenal to the cytosolicc leaflet of the ER membrane (122), where it may interact with cytosolic galactose bindingg lectins (137), or, in contrast to present dogma, may oligomerize and form microdomainss in the cytosolic leaflet.

UDP-Gal l

FigureFigure 1: Schematic organization of GalCer synthesis in the ER membrane

ForFor details see text.

Detectionn of GalT-1 by its products

Untill recently, the presence of the GalT-1 could only be assessed via the presence of its productss or by enzyme assay. GalCer and S-GalCer were originally discovered as major lipids inn human brain by Thudichum in 1884 (138), whereas glycerol-based galactolipids were discoveredd by Carter et al. (139).

ChemicalChemical detection of galactolipids

Tissuee can be analyzed for galactolipids chemically. Routinely, lipids are first extracted in chloroform/methanoll (one-phase) at elevated temperatures for maximal yield. For sphingolipid analysis,, glycerolipids are removed by alkaline hydrolysis, and acidic and neutral sphingolipids aree separated by a DEAE column. Non-polar lipids and sphingomyelin can then be removed by acetylation,, column chromatography, and deacetylation. Next, the glycosphingolipids are subfractionatedd by TLC. Including dialysis steps and additional columns, this procedure may takee two weeks (140). A simplified analysis starts with a two-phase extraction (141), after whichh the more polar lipids like sulfatides, which partition to some extent into the aqueous phase,, can be recovered by adsorption to a reversed-phase cartridge. Lipids can be separated by two-dimensionall TLC (123, 142). Separation of GalCer from GlcCer requires the use of

borate-impregnatedd Whatman paper or TLC plates (20, 117, 123, 142-144). Spots are classicallyy visualized by charring or staining by a variety of reagents (117,140,144).

Galactolipidss can be conveniently radiolabeled by using galactose, acetate, fatty acid, and sulfate,, whereas the sphingolipids will be efficiently labeled also by serine, palmitate, sphingosine,, sphinganine or a ceramide containing a C6(2-OH) chain (Figure 2A, B).

Fluorescentt galactolipids can be produced from NBD-Cer, but more efficiently from (2-OH)NBD-Cer(123,, 145), while NBD-DAG can be used to obtain NBD-GalDAG (Figure 2C). Radiolabelss and fluorescence are detected and quantitated by phosphorimaging or fluorography andd scintillation counting, and fiuorimaging or fluorometry (118, 123, 142).

Originally,, galactolipids on TLC plates were identified by chemical determination of the sphingoidd base or glycerol, fatty acid, galactose, or sulfate (139). Often, sufficient information iss obtained from co-migration with standards, sensitivity of the lipid to enzymes like a- or 6-galactosidase,, and in cell lines, after radiolabeling with specific precursors or treatment of the cellss with inhibitors of glycolipid synthesis or sulfation (123, 142). The precise structure of a galactolipidd can be obtained with mass spectrometry in combination with NMR spectroscopy (146).. While even one 2D-TLC separation of total lipids may yield galactolipid spots of sufficientt purity to allow identification by mass-spectrometry (123), HPLC remains the method off choice for this purpose (147). Amounts in the pmol range can now be quantified with nano-electrosprayy tandem mass spectrometry (148). Often, a combination of the methods described heree is required to define the precise galactolipid content of a sample (123, 142, 149).

ImmunologicalImmunological detection of galactolipids

Somee lipids can be identified by antibody-overlay techniques (150). Antibodies are available thatt recognize GalCer, GalDAG, GalAAG, galabiaosylceramide, and their sulfated forms (151-163),, with a degree of specificity (164, 165). A variation on this theme is the use of bacterial toxinss recognizing GalCer (166), the ectodomain of of human immunodeficiency virus gpl20 thatt recognizes GalCer and sulfatides (167-174), or mammalian proteins that recognize sulfatidess (175-179). A common problem of these assays is their lack of specificity.

Expressionn patterns of galactolipids may be established by immunolabeling methods. For light-microscopy,, a primary, galactolipid-binding protein is visualized with fluorescently or otherwisee labeled antibodies. For electron microscopy, protein A conjugated with colloidal goldd is the detection method of choice. Because of the potential cross-reactivity of the galactolipidd binding protein, morphological techniques must always be confirmed by lipid analysis.. Immunolabeling of (glyco)lipids is hampered by artefacts that include relocation and solubilizationn of the antigen during fixation with organic solvents and permeabilization with detergents.. Immunolabeling of thawed cryosections may also result in redistribution of lipid molecules.. The best method so far is freeze-substitution (20, 60). Glycolipids are thought to be enrichedd in patches in the membrane (18, 60, 116). However, antibody labeling may cluster glycolipidss artificially, even after fixation. This can only be prevented by a second round of fixationn after binding of the first antibody (180).

AssaysAssays for GalT-1 enzyme activity

Thee enzyme activity producing GalCer was first demonstrated by Morell and Radin (118) and, sincee then, it has been characterized under numerous conditions. A technical problem is the difficultyy to control the Cer concentration in the membrane containing GalT-1 as Cer is tightly regulatedd in the ER membrane in vivo (181). Moreover, natural ceramides do not efficiently exchangee between membranes in vitro, limiting the possibilities to manipulate Cer levels of isolatedd ER membranes. Cer has been efficiently supplied in detergent (117, 124, 182).

Detergentt assays test enzyme activity under standard but non-physiological conditions, as the ERR membrane has been dissolved. Moreover, enzyme activity is reduced many-fold. Cer has alsoo been presented from Celite (118) or phosphatidylethanolamine "membranes" (183). Disadvantagess are the low efficiency, undefined local ceramide concentrations, and, in some cases,, uncontrolled effects on the GalT-1-containing membrane (by fusion for example). As an alternative,, short-chain ceramides provide a very efficient assay for the enzyme activity in the ERR membrane (123, 125, 133, 184, 185). However, they yield indirect data on kinetics and substratee specificity.

Assayy for GalT-1 activity in cells using short-chain ceramides

Thee method to detect GalT-1 enzyme activity is based on measuring the incorporation of

fluorescentfluorescent or radioactive short-chain Cer into GalCer. Because of the short fatty acyl chain thesee ceramides and their products will display a higher off-rate from membranes than the

naturall membrane lipids. For that reason, short-chain lipid analogs can be efficiently presented too or depleted from membranes by a back-exchange against liposomes or BSA, in the absence off detergent (142, 186). The reaction requires UDP-Gal, which, for in vitro studies, must be addedd exogenously. Lipids are extracted, separated by 2D-TLC, and quantitated by fluorescencefluorescence of radioactivity.

Reagents Reagents

Reactionn mixture: HB containing 2% w/v BSA, 4 mM UDP-Glc, 4 mM UDP-Gal, 4 mM MgCl2,, 4 mM MnCl2, 1 ug/ml protease inhibitors, and 50 uM of

NBD-Cerr or NBD-DAG, or 35 nM of C6OH-[3H]Cer.

Celll incubation mixture: Hanks' Balanced Salt Solution, 20 mM Hepes-NaOH, pH 7.2, 1% w/v bovinee serum albumin (BSA; fraction V from Sigma, St. Louis, MO), andd 35 nM of C6OH-[3H]Cer.

Homogenizationn buffer (HB): 250 mM sucrose, 10 mM Hepes-NaOH, pH 7.2,1 mM EDTA. Ceramides:: Fluorescent N-6(7-nitro-2,l,3-benzoxadiazol-4-yl)-aminohexanoyl-ceramide (NBD-Cer)) was obtained commercially (Molecular probes, Eugene, OR). The radiolabeled short-chainn ceramides hexanoyl-[3H]Cer (C6-[3H]Cer) and 2-hydroxyhexanoyl-[3H]Cer (C6

OH-[3H]Cer;; 800 Mbq/umol), were synthesized according to Ong and Brady (142,187). Ceramides weree dried from stock solutions in chloroform/methanol (2:1, v/v) under nitrogen, dissolved in ethanoll (final concentration less than 0.2% v/v) and injected into BSA-buffer under vortexing too yield the reaction mixture. This was incubated 30 min on ice allowing BSA-complexes of thee ceramides to be formed prior to addition of the enzyme source.

Fluorescentt 1 -palmitoyl-2-6(7-nitro-2,1,3-benzoxadiazol-4-yl)-aminohexanoyl-diacylglycerol (NBD-DAG)) was prepared from NBD-phosphatidylcholine (Avanti Polar Lipids, Alabaster, AL)) using phospholipase C (123). TLC-plates (Si60, Merck, Darmstadt, FRG) were dipped in 2.5%% w/v boric acid in methanol (144), and dried prior to usage. Borate treatment is required to separatee GlcCer from GalCer analogs. All reactions and lipid extractions were performed in Corexx or Pyrex glassware. Chromatography solvents were of Pro Analyse quality. All lipid stockss are stored in chloroform/methanol (2:1, v/v) at -20°C. Solutions are stored under nitrogenn and should be checked routinely for concentration and purity.

GalT-1GalT-1 source

Chinesee hamster ovary (CHO) cells transfected with GalT-1 (GalT-1-CHO cells; 123) were culturedd in Eagle's minimum essential medium (MEM)-alpha with nucleotides, 10% fetal calf serum,, 10 mM Hepes, and 500 ug/ml G418. To prepare a postnuclear supernatant (PNS), a 10 cmm diameter dish of GalT-1-CHO cells is washed twice with ice-cold PBS, and gently scraped inn 1 ml ice-cold HB. Cells are pelleted and resuspended in 400 ul HB. The cells are

homogenizedd by 12 to 14 passages through a 25 Gauge needle and centrifuged for 15 min at 375gg at 4°C to remove nuclei and unbroken cells. Protein in the PNS was measured using the BCAA assay (Pierce, Rockford, IL) and adjusted to 2 mg/ml with HB. In some cases, saponin is addedd to 0.4% w/v to the PNS to permeabilize membranes during an incubation of 30 min on icee prior to the experiment. Madin-Darby canine kidney type II (MDCK II) cells were grown as monolayerss in MEM with 10 mM Hepes and 5% FCS.

Incubation Incubation

AA 3 cm dish of GalT-1-CHO cells or a 24 mm filter with MDCK cells is incubated with 1 ml celll incubation mixture. When PNS is used, one volume of reaction mixture is added to the PNSS and the samples are incubated for 1 or 2 h at 37°C. The reaction is stopped by transferring thee samples to an ice bath and by starting the lipid extraction.

LipidLipid analysis

Lipidss from cells, media or PNS are extracted by a two-phase extraction (141). The aqueous solutionn used for the phase separation contains 20 mM acetic acid and (for radiolabeled lipids) 1200 mM KC1. An additional chloroform wash of the upper (aqueous) phase is performed. The organicc (lower) phase is dried under N2 at 37°C and the lipids are applied to borate-treated

TLCC plates using chloroform/methanol (2:1, v/v). TLC plates are developed in the first dimensionn using chloroform/methanol/25% v/v NH4OH/ water (65:35:4:4, v/v), and in the

secondd dimension in chloroform/acetone/methanol/acetic acid/water (50:20:10:10:5, v/v). Fluorescentt spots are quantitated using a STORM 860 imager (Molecular Dynamics, Sunnyvale,, CA) using ImageQuant software. Alternatively, spots are detected under UV, scrapedd and extracted from the silica in 2 ml chloroform/methanol/20 mM acetic acid (1:2.2:1, v/v)) for 30 min. After pelleting the silica for 10 min at 2,000g, NBD-fluorescence in the supernatantt is quantified in a fluorometer at 470 nm/535 nm using the appropriate controls and afterr calibration of the fluorometer using the Raman band of water at 350 nm/397 nm. Radiolabeledd spots are detected by fluorography after dipping the TLC plates in 0.4% v/v 2,5 diphenyl-oxazolee in 2-methylnaphthalene with 10% v/v xylene (188). Preflashed film (Kodak X-Omatt S, France) is exposed to the TLC plates for several days at -80°C. The radioactive spotss are scraped from the plates and the radioactivity is quantified by liquid scintillation countingg in 0.3 ml Solulyte (J.T. Baker Chemicals, Deventer, The Netherlands) and 3 ml of Ultimaa Gold (Packard Instrument Company, Downers Grove, IL, USA).

Results Results

Thee results of this assay are highly reproducible. In dog kidney MDCK cells C60H-[3H]Cer is convertedd to GalCer, galabiaosylceramide, and SGalCer, while also GlcCer and sphingomyelin aree being formed (Figure 2A). In contrast, transfection of CHO cells with rat GalT-1 results in aa shift from incorporation into GlcCer and sphingomyelin to production of C60H-[3H]GalCer (Figuree 2B). In homogenates from both cell types, GalT-1 has a great preference for ceramides containingg a 2-OH fatty acid (118, 123, 124). Interestingly, tissues expressing high GalT-1 activityy also contain high levels of 2-OH fatty acids. GalCer produced in GalT-1-CHO cells containedd exclusively non-hydroxy fatty acids (123), which suggests that in the genome GalT-11 and the enzymes responsible for the synthesis of 2-hydroxy fatty acids are coordinately controlled.. This is apparently also the case for the a 1-4 galactosyltransferase responsible for thee synthesis of galabiaosylceramide and the sulfotransferase synthesizing SGalCer. In contrast too the parental CHO cells, GalT-1-CHO cells synthesized GalDAG from NBD-DAG (Figure 2C),, and a mixture of GalDAG and GalAAG from [3H]galactose (Figure 2D).

Itt should be noted that cellular factors may influence the GalT-1 activity measured. For example,, the synthesis of GalCer is dependent on UDP-Gal import into the lumen of the ER. Somee cell lines, such as CH01ec8 cells, have an impaired UDP-Gal import. A PNS of GalT-1-CH01ec88 cells displayed low GalT-1 activity. This activity could be restored by permeabilizing membraness prepared from GalT-l-CH01ec8 cells with saponin suggesting that the ER in CH01ec88 cells does not import UDP-Gal. These cells are known to lack the Golgi UDP-Gal transporterr (134), suggesting that the two transporter activities may reside within the same protein. . Cer r

B B

tt

9 9

FigureFigure 2: Lipid synthesis in cell lines expressing GalT-1

TLCTLC analysis of the lipid products synthesized A: during 1 h at 37°Cfrom C6OH-[3H]Cer in

dogdog kidney MDCKII cells, B: in Chinese hamster ovary cells transfected with GalT-1 (GalT-1-CHO),CHO), and C: during 2 h from NBD-DAG in GalT-1-CHO cells. D: Panel shows the fluorographfluorograph of GalT-1-CHO lipids after an overnight incubation with [3 H]galactose. FFA,

NBD-fattyNBD-fatty acid; GalDG, sum of GalDAG and GalAAG; Ga2Cer, galabiaosylceramide; MAG,

monoacylglycerol.monoacylglycerol. See also: Abbreviations. For solvents and further details, see text. Panels A, CC and D were reproduced with permission (123, 142).

Enzymee assays have suggested the existence of two GalT-ls with different intracellular locationss (see above). In our own studies this finding was caused by an artefact of the GalT-1 assay.. After the observation that the second GalT-1 activity had many properties in common withh CGlcT in the Golgi (122, 123), a comparison between GalT-1 negative cells that did or didd not express CGlcT demonstrated that CGlcT can synthesize GalCer when assayed in the absencee of UDP-Glc (chapter 3). Similar observations were made using a purified CGlcT (189).. In the presence of UDP-Glc (as in living cells) UDP-Gal was essentially competed out. Alternatively,, GalCer synthesis by CGlcT can be inhibited by a specific CGlcT inhibitor, such ass D-threo-l-phenyl-2-decanoylamino-3-morpholino-l-propanol (190).

Detectionn of GalT-1 protein in cells

Untill recently, the characteristics of the GalT-1 could only be addressed by measuring its activityy in isolated subcellular fractions (122, and references therein). Although antibodies havee been available for some time (185), only the recent antibodies raised against recombinant GalT-11 have facilitated analysis of the protein. Histidine-tagged fusion proteins representing differentt regions of rat GalT-1 were used to generate rabbit polyclonal antisera that specifically recognizee different lumenal regions of rat GalT-1 (chapter 3). The GalT-1 antisera work well forr Western blotting, immunoprecipitation, and immuno-fluorescence microscopy. Cross reactivityy in other species has not been tested yet.

Too study the properties of GalT-1 in cultured cells, newly synthesized proteins are metabolicallyy labeled with radioactive amino acids and chased with unlabeled amino acids for variouss time periods. Now different aspects of GalT-1 can be studied in more detail, such as its biosyntheticc maturation and its membrane topology. Assays for analysis of its co- and post-translationall modifications can also be found elsewhere (132). Radiolabeled GalT-1 is isolated byy immunoprecipitation, followed by separation by sodium dodecyl sulfate-polyacrylamide gel electrophoresiss (SDS-PAGE) and analysis by phosphorimaging.

Reagents Reagents

Depletionn medium: Cysteine- and methionine-free minimum essential medium (MEM alpha, Sigma,, M3786), 20 mM Hepes pH 7.3, at 37°C.

Pulsee medium: Depletion medium containing 250 uCi/ml Tran[35S]label (> 1,000 Ci/mmol;; ICN, Costa Mesa, CA), at 37°C.

Chasee medium: MEM supplemented with 5 mM methionine, 5 mM cysteine, and 20 mM Hepess pH 7.4, at 37°C.

StopStop buffer: PBS, 20 mM N-ethylmaleimide, ice-cold. An alkylating agent, such as N-ethylmaleimidee or iodoacetamide, should be included in the stop and lysiss buffer to prevent artificial formation of disulfide bonds.

Lysiss buffer: PBS, 0.5% v/v Triton X-100 (TX-100), 1 mM EDTA, 20 mM N-ethyhnaleimide,, 1 mM phenylmethylsulfonyl fluoride, and lug/ml of aprotinin,, chymostatin, leupeptin, and pepstatin A, ice-cold. Because alkylatingg agents and protease inhibitors have short half lives in aquous solutions,, they should be added to buffers immediately prior to use. Washh buffer: 150 mM NaCl, 2 mM EDTA, 100 mM Tris-HCl pH 8.3, 0.1 % w/v SDS,

0.5%% w/v Nonidet P40,0.5% w/v sodiumdeoxycholate. HB:: Homogenization buffer: see above.

TE:: 20 mM Tris-HCl pH 6.8,1 mM EDTA.

4xx sample buffer: 800 mM Tris-HCl pH 6.8, 12% w/v SDS, 40% v/v glycerol, 4 mM EDTA,, 0.01% w/v bromophenol blue, 300 mM dithiothreitol.