Traffic 2000 1: 769–782Munksgaard International Publishers

Proteomic Analysis of Two Functional States of theGolgi Complex in Mammary Epithelial Cells

Christine C. Wua, John R. Yates, IIIb,

Margaret C. Nevillec and Kathryn E. Howella,*

a Department of Cellular and Structural Biology, Universityof Colorado Health Sciences Center, 4200 E. 9th AvenueB-III, Denver, CO, 80262 USAb Department of Cell Biology, The Scripps ResearchInstitute, La Jolla, CA, USAc Department of Physiology, University of Colorado HealthSciences Center, Denver, CO, USA* Corresponding author: K. Howell,Kathryn.Howell@UCHSC.edu

Organellar compartments involved in secretion are ex-

panded during the transition from late pregnancy (basal

secretory state) to lactation (maximal secretory state) to

accommodate for the increased secretory function re-

quired for copious milk production in mammary epithe-

lial cells. The Golgi complex is a major organelle of the

secretory pathway and functions to sort, package, dis-

tribute, and post-translationally modify newly synthe-

sized proteins and membrane lipids. These complex

functions of the Golgi are reflected in the protein com-

plement of the organelle. Therefore, using proteomics,

the protein complements of Golgi fractions isolated at

two functional states (basal and maximal) were com-

pared to identify some of the molecular changes that

occur during this transition. This global analysis has

revealed that only a subset of the total proteins is up-

regulated from steady state during the transition. Iden-

tification of these proteins by tandem mass

spectrometry has revealed several classes of proteins

involved in the regulation of membrane fusion and se-

cretion. This first installment of the functional proteomic

analysis of the Golgi complex begins to define the

molecular basis for the transition from basal to maximal

secretion.

Key words: Functional screen, Golgi complex, pro-

teomics, tandem mass spectrometry, two-dimensional

gel electrophoresis

Received 6 June 2000, and accepted for publication 17

July 2000

Our ultimate goal is to understand the molecular changesthat occur within the Golgi complex as it changes in func-tional states from basal to maximal secretion. The mammaryepithelial cell provides an ideal system in which to study thistransition because, in response to the complex hormonalchanges at parturition, there is a rapid proliferation and ex-pansion of the organelles involved in milk secretion: 1) therough endoplasmic reticulum (RER) increases from �15 to�25% of total cell volume; 2) the Golgi complex increasesfrom �1–3% to �5–15% of the total cell volume and the

total number of cisternae in the Golgi stack increases morethan 2-fold (1–3); 3) secretory granules/vesicles increase innumber and size and in their content of casein micelles (2,4).These morphological changes are correlated with an increasein the activities of the enzymes involved in the synthesis ofcasein, lactose, and milk lipid (5). The whole process resultsin mammary glands capable of milk production within 1–4days after parturition, depending on the species (4).

Although it is clear that there is a dramatic morphologicalchange in the volume of the secretory organelles during thetransition from late pregnancy to lactation, there is littleinformation available on the molecular mechanisms by whichthese changes occur. There are three possibilities for theobserved expansion of the individual organelles relative tototal cell volume: 1) the organelles increase in size uniformlyover time during the transition (i.e. the production of allcomponents of the organelle are all uniformly up-regulated,thereby creating an identical organelle of larger size withincreased capacity, and steady-state distributions remain un-changed), 2) only a subset of components are up-regulatedabove steady-state distributions, and these components en-able the organelle to increase secretory capacity, or 3) sub-sets of components are up-regulated above steady-statedistributions, and the changes occur sequentially over time.To distinguish among these possibilities, global non-biasedstrategies of analyses are required.

The field of proteomics has been driven by technologicaladvances, most notably reproducible two-dimensional (2D)gel systems, user-friendly mass spectrometers, and theavailability of vast amounts of information through thegenome projects (6–9). The availability of these technologieshas stimulated cell biologists to begin to undertake organellarproteomic projects. Organelles such as the mitochondria,phagosome, and Golgi complex have been studied using 2Dgel analysis (10–13). Highly reproducible 2D gels allow theone-time identification of a protein spot from a complex ororganelle without the requirement of any prior knowledgeabout the protein being examined. Recently, rapid proteinidentifications obtained using the robust technique of high-throughput tandem mass spectrometry (MS/MS) make theglobal proteomics approach even more enticing because vastamounts of data can now be obtained in a short amount oftime (14,15). Furthermore, it is possible to begin to addressfunctional questions by comparative analyses of proteomesobtained for organelles during the transition between distinctfunctional states. A comprehensive review of proteomic ap-plications in cell biology by McDonald and Yates can befound elsewhere in this issue (61).

The Golgi complex is a major organelle of the secretorypathway (16). It is responsible for the sorting, packaging, and

769

Wu et al.

distribution of newly synthesized proteins and membranelipids, as well as recycling molecules. In addition, the post-translational modifications of newly synthesized proteins,lipids, and complex carbohydrates are carried out by theenzymes contained within the membranes of the variouscisternae (17–19). Given the many complex functions of theorganelle, some of which remain unknown, the Golgi com-plex has been an area of intense research in the last 20years. It is estimated that at least 1000 proteins make up theprotein complement of the Golgi complex, and, to date, lessthan 200 of these are known (10). Clearly, efficient screensrevealing novel Golgi proteins and protein interactions wouldenhance our present understanding of Golgi function.

Traditionally, researchers in membrane traffic have identifiednovel proteins of the Golgi through immunologic methods(20–25). Families of known Golgi proteins were subse-quently defined using similarity searches when the genomedatabases became available (26). In addition, many haveutilized genetics on eukaryotic organisms such as Saccha-romyces cerevisiae to screen for functional molecules. Con-ditional mutant yeast strains have provided powerful toolswith which to analyze and dissect secretory function (27).These methods, while successful, are limited. Proteomicsmay be a more comprehensive tool for the rapid identifica-tion of novel protein candidates, and by using a comparativeapproach, can provide a global functional screen.

In this study, we present the first installment of a compara-tive analysis between two functional states of the Golgicomplex in rat mammary epithelial cells. 2D gels were pre-pared to identify the major differences in the protein compo-sition of Golgi fractions isolated at the basal secretory state(late pregnant glands) and the maximal secretory state (lac-tating glands). Protein spots selected from these gels wereidentified using reverse phase liquid chromatography (LC)coupled with tandem mass spectrometry (8). Using this ap-proach, we have identified 30 proteins from the isolatedGolgi fractions that are up-regulated from the steady-stateprotein distribution in the mammary epithelial cell during thefunctional transition from basal to maximal secretion. In addi-tion, 15 proteins of unknown function were identified (notaddressed in this study). Knowledge of the identities of theseregulated proteins and further analyses of the unknowns willfuel the development of new hypotheses and further theunderstanding of the molecular mechanisms involved inGolgi expansion as well as basic Golgi functions.

Results

Golgi isolation from enriched acinar (alveolar)

populations

Fractionation of the Golgi complex from mammary epithelialcells is complicated because the mammary gland containsseveral other cell types in addition to the secretory epithelialcells, and the relative fraction of each cell type changes withdevelopmental stage (28). Therefore, acini (also referred toas alveoli), or enriched populations of epithelial cells, werefirst collected from the gland. This procedure was optimized

using the inguinal glands from lactation day 10 (L10) ratsbecause lactating glands provide a reasonably abundantsource of tissue. The glands were minced and treated withcollagenase and hyaluronidase using a modification of a pro-tocol originally developed for the isolation of pancreatic acini(29). Release of acini from the tissue was monitored by lightmicroscopy and an enriched acinar preparation was collectedby low-speed centrifugation. Subsequently, Golgi fractionswere isolated from this enriched epithelial preparation (30).

Biochemical/morphological evaluation of fractionated

Golgi

Extensive assays of enrichment and yield of various markersare routinely used to characterize each step in a fractionationprotocol (30). Each isolated Golgi fraction (GF) was character-ized biochemically by quantitative immunoblot, and morpho-logically by electron microscopy (EM). The enrichment ofrepresentative Golgi markers is shown in Figure 1. The iso-lated post-nuclear supernatant (PNS), an intermediate frac-tion (SII), and the Golgi fraction (GF) were analyzed andcompared to the GF isolated from the livers of the same

Figure 1: Golgi markers are enriched in the isolated Golgi

fraction from mammary acini. Golgi markers in the fractions ofthe isolation protocol were analyzed by immunoblot and quanti-tated by Protein A-I125 signal detection by Phosphorimager. Thestarting tissue was an enriched acinar preparation from controlL10 mammary glands. PNS (500 mg), SII (250 mg), and GF (25 mg)samples from the mammary acini fractionation protocol arepresented and compared to the GF (25 mg) fraction obtainedfrom liver. The fractions are labeled at the top of each lane(PNS-post nuclear supernatant, SII-intermediate fraction, GF-Golgi fraction), and the Golgi markers are labeled on the side.Note: It was necessary to increase the protein load of fractionsin which Golgi markers were present in low concentrations toobtain a reliable signal.

770 Traffic 2000: 1: 769–782

Functional Proteomics of Golgi Complex

Table 1: Enrichment of organellar markers in GF

Localization Enrichment (fold/PNS)Marker

P18 L10

Intermediate compart-P58 2.10 28mentCis-Golgi matrix NDGM130 ND

MG160 Medial-Golgi 180 220Trans-Golgi 30TGN38 85Endoplasmic reticulum 0.47KETE 1.50

ETF Mitochondrial matrix 1.60 6.00EDH Mitochondrial inner 1.17 1.72

membranePlasma membrane 1.02 2.28MUC4

ND, not detected; KETE, ER retrieval peptide; ETF, electrontransfer flavoprotein; EDH, ETF dehydrogenase; MUC4, mucin4.

TGN38 shows several distinct bands as compared to a broaddiffuse band (85–95 kDa) in the liver. These bands werethought to represent distinct glycosylated forms of theprotein, and neuraminidase digestions of immunoprecipitatedmammary TGN38 confirmed this hypothesis. All four bandscollapsed to a single broad diffuse band with a MW of 60–65kDa, suggesting that the bands are due to differential sialyla-tion (data not shown). 2) GM130 (a cis-Golgi matrix protein)(31) is present in the liver GF but is not detected in themammary GF. Markers of the endoplasmic reticulum (ER),mitochondria, and plasma membrane also were quantitatedto evaluate for contamination of the fraction with other or-ganelles. A summary of the enrichments of all markerstested is presented in Table 1. It is clear that the L10 GF isenriched from the PNS and that contamination from ER,plasma membrane, and mitochondria is present but minimal.

For morphological evaluation, the L10 Golgi fraction used forthe quantitative immunoblot analysis (Figure 1, Table 1) wasembedded in Spurr’s resin, sectioned, stained, and viewedby EM. Figure 2A and B show two different fields of view ina thin section of the plastic embedded L10 GF. The sample isrelatively homogeneous, showing membrane structures thatappear to be large vesicles containing casein in variousstages of aggregation (large arrowheads), single Golgi cister-nae (small arrowheads), and, in one instance, a coated bud

animals. Quantitation of the bands by phosphorimager analy-sis shows enrichment in the GF relative to the PNS: a220-fold enrichment of MG160 (a medial-Golgi marker) (22),an 85-fold enrichment of TGN38 (a trans-Golgi marker) (23),and a 28-fold enrichment of p58 (a marker for the intermedi-ate compartment) (21). Two major differences are apparentbetween the mammary and liver Golgi markers. 1) Mammary

Figure 2: Electron micrographs showing overviews of the GF isolated from control L10 mammary glands. (A) and (B) areimaged at 6,000× magnification. The sample is relatively homogeneous showing what appears to be large vesicles containing caseinin various stages of aggregation (large arrowheads) as well as single Golgi cisternae (small arrowheads). The star symbol in A indicatesa vesicular structure with a coated bud. A major presence of other organelles is not apparent.

771Traffic 2000: 1: 769–782

Wu et al.

(star). Fortuitously, casein stains well with traditional EMstaining protocols and provides a convenient marker to iden-tify compartments of the secretory pathway. The presenceof casein in various stages of maturation (submicelles ormicelles) identifies the secretory compartments as Golgi orGolgi-derived (3). Remarkably, in both views, the Golgi ap-pears to be almost completely unstacked. This morphology isdistinctly different from the nicely stacked Golgi previouslyisolated from rat livers (30), and, together with the absenceof the Golgi matrix protein GM130, suggests that the matrixmay be modified in mammary epithelial cells.

Evaluation of control and cycloheximide-treated Golgi

fractions

L10 rats were treated with cycloheximide (CHX) for 4 h priorto sacrifice to terminate protein synthesis thereby clearingthe Golgi of secretory and membrane proteins (cargoproteins) in transit. Isolated L10 GF from control (CTL) andcycloheximide (CHX) treated rats were biochemically frac-tionated into soluble and membrane fractions using high pH(32). Biochemical sub-fractionation has two major advan-tages. Proteins can be placed into distinct classes (the solu-ble or membrane categories), and minor proteins (e.g.regulatory proteins) are enriched and more easily identified.Analytical (silver stained) 2D gels were prepared from each ofthe sub-fractions resulting from high pH treatment of CTLand CHX Golgi fractions. Figure 3A and B show the mem-brane and soluble sub-fractions prepared from the CTL L10Golgi fraction, respectively. Figure 3C and D show the corre-sponding 2D gels resolving the membrane and soluble sub-fractions prepared from the CHX L10 Golgi fraction,respectively. The major soluble milk proteins, a and b casein,are visualized and labeled in the soluble fraction of the CTLsample (Figure 3B), but are not present in the soluble fractionof the CHX sample (circled locations in Figure 3B are trans-posed onto Figure 3D) or in the CTL or CHX membranefractions. The membrane cargo protein, polymeric IgA recep-tor (pIgAR), is visualized in the membrane fraction of the CTLsample (Figure 3A), but not in the membrane fraction of theCHX sample or the soluble fractions of the CTL and CHXsample (circled location in Figure 3A is transposed ontoFigure 3C,B,D). These data indicate that: 1) the high pHcarbonate treatment used to separate the membrane fromthe soluble proteins is efficient, and 2) the 4 h CHX treatmentof the rats prior to sacrifice almost completely eliminates thecargo proteins in transit. We proceeded next to the two-di-mensional analysis of the molecular differences betweenGolgi membranes isolated in the basal and maximal secre-tory states.

Comparison of Golgi membrane fractions isolated from

distinct functional states

Golgi high pH membrane fractions were prepared from CTLand CHX treated animals at pregnancy day 18 (P18) (32).Biochemical enrichments of the CTL P18 GF are presented inTable 1, and the GF fraction appears to be enriched in Golgimarkers. Analytical (silver stained) 2D gels were prepared

resolving the membrane sub-fractions, and Figure 4 showsthe comparison between P18 and L10 Golgi membrane frac-tions. Figure 4A and B show the 2D gels resolving CTL P18and L10 Golgi membrane sub-fractions, while Figure 4C andD show the 2D gels resolving CHX P18 and L10 Golgimembrane sub-fractions. Major proteins present in all fourgels are boxed for comparison. Several general observationscan be made. 1) The pattern of the 2D gel spots in all fourgels is similar, suggesting that the majority of the overallprotein distribution at steady state remains the same duringthe functional transition. 2) The CHX membrane samplesshow a more defined and less complex pattern than CTLsamples. Proteins endogenous to the Golgi are enriched inthe CHX samples because the proteins in transit have beenreduced (30). 3) Upon closer inspection, the pattern of theprotein spots of the P18 and L10 CHX gels, while similar, areclearly not identical. Analysis of the protein spots on the gelusing the Melanie II software (BioRad) is typically carried outat 2X magnification for better resolution of the spots. Wehave focused on spots that are visualized on the L10 CHX 2Dgel but are barely visible (or not visible) on the P18 CHX 2Dgel for the analysis of proteins that are up-regulated from thesteady-state protein distribution with the transition frombasal to maximal secretion. A subset of these spots wasselected for identification by MS/MS.

Protein identification by tandem mass spectrometry

Preparative 2D gels with increased protein load (Coomasiestained) were prepared from the P18 CHX and L10 CHXGolgi membrane fractions for analysis by MS/MS (33). Spotswere excised and in-gel digested with trypsin (8). Peptideswere extracted from the gel, separated by liquid chromatog-raphy, and analyzed by MS/MS (8). Identifications weremade off-line using Sequest software (34). In Figure 5, Pan-els A–G show enlarged regions from the gels presented inFigure 4C and D. The protein spots selected for MS/MSanalysis are labeled with identification numbers (Map IDs).Protein spots labeled with an asterisk (*) serve as fiducialmarkers or landmarks. The identifications of the numberedproteins in Figure 5 are presented in Table 2. In other exper-iments not shown, Golgi fractions were biochemically sub-fractionated using TritonX-114 (TX114) partitioning to yieldthree sub-fractions (35): 1) aqueous (soluble proteins), 2)detergent soluble (membrane proteins), and 3) detergentinsoluble (matrix-like proteins) (13). Symbols (c) next to theidentifications indicate those proteins enriched in the deter-gent insoluble fraction of TX114 extractions of the Golgifractions. Proteins listed without symbols (c) were enrichedin the detergent soluble fraction. All identifications from thevarious gels were made by MS/MS, and none were made byco-migration.

Table 2 represents the first installment of the comparisonbetween Golgi proteins isolated from the basal secretorystate (P18) and the maximal secretory state (L10). These datafocus on proteins that are up-regulated from the steady-stateprotein distribution during the transition between the func-

772 Traffic 2000: 1: 769–782

Functional Proteomics of Golgi Complex

Figure 3: Comparison of the protein composition of biochemical sub-fractions (membrane and soluble) of Golgi fractions

isolated from the mammary glands of control (CTL) and cycloheximide (CHX) treated animals by 2D gel electrophoresis. Thesub-fractions presented are as follows: (A) 100 mg CTL GF membrane, (B) 25 mg CTL GF soluble, (C) 100 mg CHX GF membrane, and(D) 25 mg CHX GF soluble. The soluble milk proteins, a and b casein, are only visualized and labeled in the soluble fraction of the CTLsample (B), but not in the soluble fraction of the CHX sample (circled locations in B are transposed onto D for reference) nor in theCTL or CHX membrane fractions. The membrane cargo protein, polymeric IgA receptor (pIgAR), is only visualized in the membranefraction of the CTL sample (A), but not in the membrane fraction of the CHX sample nor the soluble fraction of the CTL sample (circledlocation in Figure 4A is transposed onto C and B for reference). Note: It was necessary to load different amounts of proteins in themembrane and soluble sub-fractions because the silver consistently stains the soluble proteins more intensely than the membraneproteins.

tional secretory states. Future installments will provide listsof proteins that are down-regulated in addition to those thatremain the same. Together, these data will provide the data-base required to begin to dissect the mechanisms of Golgiexpansion.

Discussion

Lactating mammary epithelial cells have traditionally beenone of the model systems used by advocates of the cisternalmaturation model of trafficking (3). This cell type, studied

773Traffic 2000: 1: 769–782

Wu et al.

during lactation, seems to utilize both the constitutive and amodified regulated secretory pathway and does not adhereto the ‘classical rules’ of secretion (36). Pulse-chase experi-ments carried out in lactating mouse mammary alveoli inculture show that 30% of the labeled casein is secretedconstitutively, and the additional casein can be secreted byraising intracellular calcium levels (37,38). In addition, a sig-nificant amount of compound exocytosis is observed bymorphology (36). However, this is most likely due to themassive amounts of protein, lipid, and lactose being rapidly

secreted during this developmental stage of the gland (36).During pregnancy, the secretory epithelial cells are not mas-sively secreting milk components but, rather, are maintaininga basal secretory level (1,2,5). Using comparative proteomicsto analyze the molecular components of secretory organellesduring the transition between late pregnancy (basal secre-tion) and lactation (maximal secretion), we have the uniqueopportunity to investigate the molecular mechanisms of or-ganelle expansion and to evaluate the trafficking models ofcisternal maturation and vesicular transport.

Figure 4: Comparison of the protein composition of Golgi membrane sub-fractions at the basal secretory state (P18) and the

maximal secretory state (L10) by 2D gel electrophoresis. The sub-fractions presented are as follows: (A) 100 mg CTL P18 GFmembrane, (B) 100 mg CTL L10 GF membrane, (C) 100 mg CHX P18 GF membrane, and (D) 100 mg CHX L10 GF membrane. Majorproteins present in all four gels are boxed for comparison. The overall pattern of the 2D gel spots in all 4 gels is similar. The membranecargo protein, polymeric IgA receptor (pIgAR), is circled to provide a reference point from the previous Figure 3.

774 Traffic 2000: 1: 769–782

Functional Proteomics of Golgi Complex

Figure 5: Up-regulated protein spots of Golgi membrane during the functional transition are selected for protein identifica-

tion. Panels A-G show enlarged regions from the gels presented in Figure 4 C and D to provide better resolution of the spots selectedfor protein identification. The spots selected for analysis were those appearing to be up-regulated from the steady-state distributionsof the Golgi membrane protein complement during the functional transition from basal (P18) to maximal (L10) secretion. All spotslabeled with identification numbers (IDcs) were excised and identified by tandem mass spectrometry. Spots labeled with an asterisk(*) serve as fiducial markers or landmarks. The identifications of the numbered proteins are presented in Table 2.

775Traffic 2000: 1: 769–782

Wu et al.

Isolated Golgi are significantly enriched and of high

yield

The validity of all the proteomic analyses conducted heredepends on the quality of the subcellular fractions analyzed.Golgi fractions were prepared from the inguinal mammary

glands from groups of 12 L10 rats and groups of 18 P18 rats.The use of multiple rats for each sample preparation controlsfor variation between animals. Conditions for the fractiona-tions were first optimized in lactating glands because theyprovided a more abundant source of starting material. Subse-

Fig 5: (Continued)

776 Traffic 2000: 1: 769–782

Functional Proteomics of Golgi Complex

Fig 5: (Continued)

quent fractionations from pregnant glands were carried outusing the conditions optimized in lactating glands.

A major contaminant to be considered from this protocolwould be Golgi originating from cells present in the glandother than the mammary epithelial cells. To prepare Golgifractions from enriched epithelial cell populations, acini werefirst collected from finely minced mammary glands dissoci-ated from fibrous connective tissue by collagenase and hy-luronidase digestion (29). Golgi fractions were subsequentlyisolated from the enriched acinar preparation.

Since no fractionation protocol results in pure Golgi, contam-ination from other organelles was evaluated using severalmethods. The novel method of high-throughput protein iden-tification using 2D gels coupled with tandem mass spec-trometry offers a quick and ideal way to addresscontamination spot by spot. Proteins identified to be resi-dents of other organelles are not considered further, andthose spots are subtracted from subsequent 2D analyses.However, if the identified protein is one of unknown func-tion, then the only way to determine its endogenous or-ganelle of residence is by actual localization of the protein inthe cell. This can be done fairly quickly if the unknownprotein is cloned into a GFP fusion construct. Expression ofthe GFP fusion protein in the cell results in localization of thefluorescent protein by microscopy.

The classical assays of quantitative immunoblots and elec-tron microscopy were also carried out to quantitate contami-nation. From the data presented in Figure 1 and Table 1, itcan be concluded that contamination from ER, plasma mem-brane, and mitochondria is minimal in both the L10 and P18Golgi fractions. Electron micrographs taken of the Golgi frac-tion isolated from the lactating mammary glands showed

that most membrane profiles contain casein at variousstages of micelle formation. The presence of casein submi-celles and micelles is indicative of Golgi cisternae or secre-tory vesicles even though intact Golgi stacks are notobserved. Evaluation of secretory granules is more complexbecause it is unclear where the Golgi ends and secretorygranules begin. There are presently no unique markers avail-able for the detection of secretory granules in mammaryepithelial cells. In fact, supporters of the maturation model oftransport through the Golgi would argue that there are nosecretory granules and that the Golgi extends completely tothe apical plasma membrane. This is an important questionand we anticipate that it can begin to be addressed byproteomic analyses. Taken together, the Golgi fractions ob-tained from pregnant and lactating glands were significantlyenriched and of high enough yield to warrant functionalproteomic analysis.

Comparison of the 2D gel maps of CTL (Figure 3A and B) andCHX (Figure 3C and D) Golgi membrane subfractions gener-ated from P18 and L10 glands, respectively, shows that theoverall protein spot pattern on the gels are quite similar. Infact, the fiducial markers presented in each of the panels inFigure 5 demonstrate that many proteins remain the same inthe two samples analyzed. These data suggest that the twofractions are reasonably equivalent and the observedchanges (up-regulation of protein spots from the steady-statedistribution) are significant.

Significance of proteins identified

The identifications of the 30 up-regulated proteins at themaximal secretory state of the Golgi complex were amaz-ingly organized (Table 2). Multiple members of major func-tional classes of trafficking and regulatory proteins wereidentified using this screen and clearly demonstrate the

777Traffic 2000: 1: 769–782

Wu et al.

power of this approach. Six membrane fusion proteins (an-nexins I, III, IV, V, a-SNAP and synaptotagmin), six regulatoryproteins (rabs 2, 6, 11b, 11e, 14, 18), four microtubule motorproteins (dynein light and intermediate chain, kinesin, andmyosin heavy chain), and four actin remodeling proteins(rhoA, rhoGAP, cdc42, and hsp27) were all up-regulated.Importantly, four housekeeping proteins were identified toremain at the same level in both functional states (BiP,tubulin, actin, and GBB1), providing an ideal control for thescreen and further supporting the significance of the proteinsidentified to be up-regulated.

Six membrane fusion proteins were identified to be up-regu-lated from steady state during the transition from basal tomaximal secretion. Annexins I, III, IV, V, a-SNAP and synapto-tagmin have been well characterized, although the specificsof the mechanism of action of each in regulated secretion inresponse to increased calcium concentration is poorly under-stood. The annexins and synaptotagmin are mediators ofcalcium-dependent fusion events (39). The annexins com-prise a group of calcium-dependent membrane aggregatingproteins that can initiate contacts between secretory vesiclemembranes, which subsequently fuse (39). Synaptotagmins

Table 2: Protein identifications by mass spectrometrya

L10 Mr (kDa)Map ID Identification (abbreviation) P18 pI

Albuminb + ++ 68.7 6.202BiPb ++3 ++ 74.2 4.81Lactoferrin + ++ 76.3 7.126

6.7847.4 (65)+++O-acetyl GD3 ganglioside synthase745.0++− 6.30Arrestin 2c8

Tubulinb ++11 ++ 47.9 5.1712 Actinb ++ ++ 41.5 5.22

6.40 (6.1)38.7++−Somatin-like protein1338.7++++ 6.40 (6.1)GBB1b14

16 4.55 (6.0)29.1+++14-3-3 bc

17 14-3-3 oc + ++ 27.8 4.80 (6.1)− ++ 25.122 6.40 (7.0)gp25LP

5.0626 Ca2+ binding protein/p22 + ++ 19.04.63 (6.3)19.0 (29)− ++Rap 1B28

31 ++Rap 1A/B 4.63 (6.4)19.0 (28)+

Membrane fusion proteins++ 42.69 6.71Synaptotagmin −

Annexin I + ++ 35.0 7.7710Annexin Vc + ++ 35.6 4.7015

5.2333.2++−a-SNAP1819 Annexin IV − ++ 35.9 5.80 (8.0)

5.85 (8.0)32.5++−Annexin III20

Regulatory proteins24 Rab6 − ++ 23.5 5.42

23.0 5.24++−Rab18255.9824.5+−Rab11b29

− + 23.4Rab11e 6.4430Rab14 − ++ 23.8 5.85 (6.5)33Rab2 −34 ++ 23.5 6.08 (6.5)

Actin remodeling proteins6.16 (6.7)21.3+++G25K/cdc4221

27 RhoGAP − ++ 245.0 (28) 8.53 (6.2)6.12 (6.5)22.8+++Hsp27c32

20.5++− 4.99 (6.5)RhoA35

Microtubule motor proteins1 Myosin heavy chainc + ++ 226.0 (100) 5.49

++−Kinesinc4 7.36125.25 6.66Dynein intermediate chainc − 68.3 (80)++

28.6++−Dynein light chainc23 6.96 (6.1)

a All protein identifications listed were determined by tandem mass spectometry. b Identified proteins also used as fiducial markersor landmarks. c Proteins also identified from the TX114 detergent insoluble phase. Note: numbers in the Mr and pI columns aretheoretical. Numbers in parentheses are the actual and were given if the experimental number deviated substantially from thetheoretical. Deviations in Mr are most likely due to proteolysis (IDc1 and 27), glycosylation (IDc7), and interactions with thecytoskeleton (IDc5, 28, and 31). Deviations in pI are most likely due to post-translational modifications.

778 Traffic 2000: 1: 769–782

Functional Proteomics of Golgi Complex

are a family of proteins implicated in the control of Ca2+-de-pendent exocytosis. Synaptotagmin I has been shown toregulate secretion by potentiating and accelerating Ca2+-de-pendent exocytosis in mast cells (40). Alpha-SNAP has beenshown to be required for SNARE complex disassembly andshown to play a role in priming granules for release at theearly step of Ca2+-dependent exocytosis (41,42). Their con-current identifications here suggest that they might all inter-act in a coordinate manner.

The up-regulation of four microtubule motor proteins (dyneinlight and intermediate chain, kinesin, and myosin heavychain), and four actin remodeling proteins (rhoA, rhoGAP,cdc42, and hsp27) complement the observation that fusionevents are up-regulated. Secretory vesicles and/or entireGolgi cisternae would require transport on microtubules, andonce at the apical plasma membrane, actin reorganizationwould allow for fusion and exocytosis. The small GTP-bindingprotein RhoA, perhaps upon activation by RhoGAP (43), hasbeen shown to be essential for secretory function and isinvolved in the disassembly, relocalization, and polymeriza-tion of F-actin (44–46). Cdc42, a Rho family GTPase that alsoreorganizes actin, was originally discovered as an essentialgene in Saccharomyces cerevisiae that is required for bud-ding and establishment of cell polarity (57). Recent studiessuggest that activated Cdc42 competes with transmembranecargo proteins for binding on the g-subunit of the COPIcoatomer complex through the dilysine (KK) motif and regu-lates secretory traffic (58,59). In addition, activation of Cdc42in MDCK cells has been found to be essential for the deliveryof VSV-G protein to the basolateral plasma membranes andwas found to increase the rate of secretory protein transportby 1.5 to 2-fold (59,60).

Other interesting proteins were identified to be up-regulated.Two 14-3-3 proteins (b and o) are members of a family ofdimeric phosphoserine-binding proteins that participate insignal transduction and have been shown to be involved ininhibition of apoptosis (47). O-acetyl GD3 ganglioside syn-thase (GD3 synthase) is a member of the sialyltransferasefamily and localizes to the proximal side of the Golgi and theTGN (48). Up-regulation of all three of these proteins wouldimply involvement of signaling pathways as well as increasedGolgi function.

Secretory events in the mammary epithelial cell are clearlyvery highly ordered and initiated in response to the complexhormonal changes at parturition. The identification of theup-regulation of six members of the rab family of smallGTP-binding proteins (rabs 2, 6, 11b, 11e, 14, 18) furthersupports the high degree of regulation. Small GTP-bindingproteins have traditionally been shown to play a crucial role inregulating exocytosis in a variety of tissues (42,49).

Since the functions of the GTP-binding proteins are hypothe-sized to be related to their cellular locations, future directionsinclude experiments to begin to dissect the temporal andspatial distribution in which these molecules are regulated.

Are these molecules all up-regulated together, or is there aspecific order to the process? Using an in vivo mouseovariectomy model that mimics lactogenesis and initiates thetransition from basal to maximal secretion (50), this questioncan be addressed. We will first narrow the time frame of thetransition and then track some of the proteins identified herewith immunofluorescence in mammary tissue taken in atemporal series during the functional transition. We are cur-rently in the process of establishing this time frame, andinitial results reveal that the amplification of secretory or-ganelles occurs within a narrow 12-h window. Another majorsecretory process in mammary epithelial cells involves cyto-plasmic lipid droplets (CLDs) and is unique to this cell type.This process appears to be connected to the exocytic path-way and is also tightly regulated and initiated within thesame 12-h window. We have used the same cell fractiona-tion and comparative proteomic approach to study the mech-anism of secretion of CLDs (51).

Use of functional proteomics as a global screen

2D gel analysis is an unbiased approach in that spots fromfractions of interest can be quantitatively compared witheach other with no foreknowledge of which protein spots areknown or unknown. This technique is similar to microarrayanalysis/differential display, but has a number of importantadvantages. 2D gels allow for the analysis of a specificsubset of cellular proteins, e.g. those associated with theGolgi complex or other organelles. The focus is on thecompartment of interest and enhances our ability to identifyless abundant proteins. In addition, 2D gel maps of thebiochemically fractionated material provide basic biochemicalinformation about proteins of interest including the ability toanalyze post-translational modifications (13). Use of microar-ray analysis/differential display is useful but will not provideinformation on what type of protein is up-/down-regulated orwhat subcellular compartment(s) the protein is localized to.Finally, even when changes in the expression of mRNA aredetected, corresponding changes at the protein level muststill be demonstrated to hypothesize functional significance.Using a comparative analysis on the Golgi proteome at twofunctionally distinct states (basal and maximal secretion), wehave identified 30 proteins, many of which belong to majorfunctional classes of proteins known to be involved in theregulation of secretory events. The identities of most of theproteins revealed to be up-regulated during the functionaltransition of the organelle confirmed the high degree ofregulation involved in the active secretory state of the or-ganelle and is an added testament to the usefulness ofcomparative proteomics as a tool for functional screens.

Materials and Methods

Isolation of acini from rat mammary glands

Acini were first isolated from the total gland using a modification of themethod developed for isolation of pancreatic acini (29). A minimum of12 female Sprague–Dawley rats (first pregnancy and lactation) wereused for each preparation to eliminate the necessity to characterizerandom variations between animals. Inguinal mammary glands wereremoved from lactating (L10) or pregnant (P18) rats (9cycloheximide).

779Traffic 2000: 1: 769–782

Wu et al.

Cycloheximide blocks protein synthesis allowing proteins in transit toclear from the Golgi and reach their final compartments of residence.It was administered as published previously (30). The tissue is mincedand incubated at 37°C in 0.075% collagenase/0.15% hyaluronidase/HBBS containing 1 mg/ml each of proteolytic inhibitors chymostatin,leupeptin, antipain, and pepstatin. Release of acini is monitored by lightmicroscopy over a 90-min digestion period. Mammary epithelial cellsremain metabolically active for more than 3 h after collagenasetreatment (37,38). Acini are strained through cheesecloth and collectedby centrifugation at 1500×g for 10 min at 4°C. Morphological analysis(light and EM) reveal that the acini are intact and enriched (data notshown).

Fractionation of Golgi from enriched acini

The Golgi fractionation protocol is similar to that modified fromLeelavathi (52) by Taylor and colleagues (30). The livers of the sameanimals from which the mammary glands were harvested wereprepared in parallel. The liver was minced finely, and the mammaryglands were collagenase and hyaluronidase treated to produce acini.The mammary acini and the minced livers were lightly homogenizedusing a Polytron PT 10/35 (Brinkmann, Westbury, NY) in 100 mMKH2PO4/K2HPO4, pH 6.8, 5 mM MgCl2 with 1 mg/ml each of proteolyticinhibitors chymostatin, leupeptin, antipain, and pepstatin. All subse-quent sucrose solutions were prepared in this buffer. The homogenatewas subjected to low-speed centrifugation, 3000×g for 10 min at 4°C,and the resulting PNS was loaded in the middle of a sucrose stepgradient (1.3 M, 0.86 M, PNS, 0.25 M) and centrifuged at 100000×gfor 1 h at 4°C. The SII (intermediate fraction) was collected from thePNS/0.86 M interface. The SII fraction was adjusted to 1.15 M sucroseand layered at the bottom of a second sucrose step-gradient (1.15 MSII, 1.00 M, 0.86 M, 0.25 M) and centrifuged at 85000×g for 3 h at4°C to float the Golgi fraction (GF) collected at the 0.25M/0.86Minterface.

1D Gel Electrophoresis and Immunoblotting

SDS-PAGE was carried out using a 7.5% acrylamide gel and the Bioradbuffer system. For immunoblots, Immobilon-P filters were blocked for1 h in Blotto (5% nonfat powdered milk/PBS/0.02% sodium azide). Thefilters were incubated overnight in primary antibody/Blotto and washedthree times for 10 min in Blotto. When using a mouse primary antibody,the filters were incubated with rabbit secondary antibodies raisedagainst mouse IgGs for 2 h and detected using [125I]-protein A andphosphorimager analysis (Molecular Dynamics, Sunnyvale, CA).

Biochemical subfractionation of isolated Golgi

High pH carbonate treatment. High pH treatment was carried outaccording to Howell and Palade (32). Briefly, the GF was pelleted at150000×g for 30 min and resuspended at 1 mg/ml in 200 mM sodiumcarbonate, pH 11, for 30 min at 4°C. The suspension was then pelletedat 150000×g for 30 min. The supernatant was collected and analyzedas the ‘soluble’ fraction. The pellet was re-treated at 1 mg/ml in 200mM sodium carbonate, pH 11, for 30 min at 4°C. The suspension wasre-pelleted at 150000×g for 30 min. The supernatant was discarded,and pellet was resuspended in 100 mM KH2HPO4 (pH 6.8) andanalyzed as the ‘membrane’ fraction.

TX114 extraction. TX114 subfractions (aqueous, detergent, insoluble)were prepared following the method of Bordier (35). GF was pelletedin a microfuge at 16000×g for 20 min at 4°C. The pellet wasresuspended at 1 mg/ml in ice-cold 1% TX114, 150 mM NaCl, 10 mMTris, pH 7.4 containing proteolytic inhibitors using a glass and teflonhomogenizer and allowed to sit on ice for 1 h with occasional gentlemixing. The insoluble fraction was pelleted by centrifugation for 20 min

at 16000×g at 4°C. The supernatant and a wash of the pellet waslayered on top of 200 ml cushion containing 0.06% TX114, 6% sucrose,150 mM NaCl, 10 mM Tris, pH 7.4. The tube was incubated at 37°Cfor 3 min to allow phase separation to take place before centrifugationat 3000×g for 2 min at room temperature to sediment the detergentphase through the cushion. The aqueous phase was carefully collectedaway from the detergent phase.

2-D gel electrophoresis

Protein (20–50 mg) was precipitated using methanol and chloroform(53) for analysis on silver stained gels. Protein (1–2 mg) was precipi-tated for preparative Coomasie (R-250) stained gels used for subse-quent identification by mass spectrometry (55). Precipitated proteinswere solubilized in 8 M urea, 4% CHAPS, 1 M thiourea, 83 mMdithioerythritol, 18 mM Trizma base, 0.4% carrier ampholytes pH3.5-10 (BDH; Poole, UK), and 0.0025% bromophenol blue. Isoelectricfocusing was performed using Immobiline dry strips (nonlinear pHrange 3.5–10; Pharmacia). The dry strips were rehydrated with thesolubilized protein sample by in-gel reswelling (54) and were elec-trophoresed for 14 h. The second dimension was run on a 9–16%polyacrylamide SDS gel using a BioRad Protean XiII cell (BioRad). Silverstaining was performed according to SwissProt 2D PAGE procedures(54). All gels were scanned using a flatbed scanner, and the resultingimages were analyzed by the Melanie II software (BioRad).

In-gel digestion of 2D spots

In-gel digests of Coomasie blue stained 2D spots were performed asdescribed (56). Briefly, spots were cut out with a scalpel and washedin 100 mM NH4HCO3 for 10 min. Proteins were reduced with 3 mMDTT/100 mM NH4HCO3 for 20 min at 60°C. After cooling to roomtemperature, iodoacetamide was added to 6 mM final concentrationand incubated in the dark for 20 min at room temperature. The aqueoussolution was discarded, and the gel slice was washed in 50% aceto-nitrile/100 mM NH4HCO3 for 20 min. The gel slice was cut into 1 mm3

pieces and transferred into new 0.5 ml low adhesion Optimum® tubes(Life Science Products, Inc.). Gel pieces were dehydrated with theaddition of 100% acetonitrile for 15 min. The solvent was removed,and the gel pieces were dried. Gel pieces were reswelled with 0.2 mgmodified trypsin/25 mM NH4HCO3 overnight at 37°C. The supernatantwas removed and placed into new 0.5 ml Optimum® tubes. Peptideswere extracted from the gel pieces with 100 ml 60% acetonitrile/0.1%trifluoroacetic acid (TFA) for 20 min. The supernatant was removed andadded to the previous extract. This extraction was repeated twice. Thepooled extract was lyophilized to dryness and reconstituted in 10 ml 5%formic acid immediately before analysis by mass spectrometry.

Identification of protein peptides by tandem mass spectrometry

Peptide samples were centrifuged for 10 min at 15000 RPM in amicrofuge to remove any particulate matter. Samples were thenloaded onto a 100 mm fused silica column with a 5 mm tip. Flow rateswere adjusted to �300–500 nl/min on an HP1100 pump and posi-tioned in front of the heated capillary opening on a Finnigan LCQ.Peptides were eluted with an acetonitrile gradient from 98% BufferA/2% Buffer B to 40% Buffer A/60% Buffer B where Buffer A is 5%acetonitrile/0.5% acetic acid and Buffer B is 80% acetonitrile/0.5%acetic acid. Spectra were analyzed using Sequest software (7,8,34).Positive identifications were made with the criteria of three separatespectra with high correlation scores. Factors included in the calculationof the correlation score are part of the Sequest software and arediscussed elsewhere (34).

Electron microscopy

Golgi samples (50 mg) were pelleted and fixed in 2% glutaraldehyde/2% sucrose/100 mM sodium cacodylate, pH 7.35 overnight at 4°C. The

780 Traffic 2000: 1: 769–782

Functional Proteomics of Golgi Complex

tissue was post-fixed in 2% OsO4 in 0.8% potassium ferrocyanidebuffered with 100 mM sodium cacodylate, pH 7.35. The tissue blockswere washed with water, dehydrated with increasing EtOH steps(50%, 75%, 90%, 98%, 100%), and embedded in Spur’s resin.Sections were post-stained in 5% uranyl acetate in MeOH andReynold’s lead citrate. Samples were imaged using a Phillips CM10microscope.

Acknowledgments

We thank our many colleagues for their generous gifts of antibodies(Marilyn Farquhar/p58, Nicholas Gonatas/MG160, George Banting/TGN38,Graham Warren/GM130, Frank Frerman/ETF, EDH, and Kermit Carraway/MUC4). We especially thank Lara Hays and Jimmy Eng for sharing theirexpertise in mass spectrometry and Sequest. We thank Randy Taylor forhis advice on the presentation of these data. This work was supported byNIH grants cRO1 GM42629 (Kathryn E. Howell) and cR37 HD19547(Margaret C. Neville).

References

1. Hollmann KH. Cytology and fine structure of the mammary gland. InLarson BL, Smith VR (Eds.), Lactation I. New York and London:Academic Press, 1974: 3–95.

2. Wooding FB. Comparative mammary fine structure. In Peaker M (Ed.),Comparative Aspects of Lactation. London: Academic Press, 1977:1–41.

3. Clermont Y, Xia L, Rambourg A, Turner JD, Hermo L. Transport ofcasein submicelles and formation of secretion granules in the Golgiapparatus of epithelial cells of the lactating mammary gland of the rat.Anat Rec 1993;235: 363–373.

4. Wooding FB. Formation of the milk fat globule membrane withoutparticipation of the plasmalemma. J Cell Sci 1973;13: 221–235.

5. Baldwin RL, Yang YT. Enzymatic and metabolic changes in thedevelopment of lactation. In Larson BL, Smith VR (Eds.), Lactation I.New York and London: Academic Press, 1974: 349–411.

6. Yates JR, Eng JK, McCormack AL. Mining genomes: correlatingtandem mass spectra of modified and unmodified peptides tosequences in nucleotide databases. Anal Chem 1995;67: 3202–3210.

7. Dongre AR, Eng JK, Yates JR. Emerging tandem mass spectrometrytechniques for the rapid identification of proteins. Trends Biotechnol1997;15: 418–425.

8. Gatlin CL, Kleemann GR, Hays LG, Link AJ, Yates JR. Proteinidentification at the low femtomole level from silver-stained gels usinga new fritless electrospray interface for liquid chromatography-mi-crospray and nanospray mass spectrometry. Anal Biochem 1998;263:93–101.

9. Yates JR, Morgan SF, Gatlin CL, Griffin PR, Eng JK. Method tocompare collision-induced dissociation spectra of peptides: potentialfor library searching and subtractive analysis. Anal Chem 1998;70:3557–3565.

10. Taylor RS, Fialka I, Jones SM, Huber LA, Howell KE. Two-dimensionalmapping of the endogenous proteins of the rat hepatocyte Golgicomplex cleared of proteins in transit. Electrophoresis 1997;18:2601–2612.

11. Scianimanico S, Pasquali C, Lavoie J, Huber LA, Gorvel JP, DesjardinsM. Two-dimensional gel electrophoresis analysis of endovacuolarorganelles. Electrophoresis 1997;18: 2566–2572.

12. Rabilloud T, Keiffer S, Procaccio V, Louwagie M, Courchesne PL,Patterson SD, Martinez P, Garin J, Lunardi J. Two-dimensionalelectrophoresis of human placental mitochondria and protein identifi-cation by mass spectrometry: toward a human mitochondrial pro-teome. Electrophoresis 1998;19: 1006–1014.

13. Taylor RS, Wu CC, Hays LG, Eng JK, Yates JR, Howell KE. Proteomicsof rat liver Golgi complex: Minor proteins are identified throughsequential fractionation. Electrophoresis 2000; In press.

14. Ducret A, Van O, Eng JK, Yates JR, Aebersold R. High throughputprotein characterization by automated reverse-phase chromatogra-

phy/electrospray tandem mass spectrometry. Protein Sci 1998;7:706–719.

15. Yates JR, Carmack E, Hays L, Link AJ, Eng JK. Automated proteinidentification using microcolumn liquid chromatography-tandemmass spectrometry. Methods Mol Biol 1999;112: 553–569.

16. Palade G. Intracellular aspects of the process of protein synthesis.Science 1975;189: 347–358.

17. Rothman JE. Mechanisms of intracellular protein transport. Nature1994;372: 55–61.

18. Farquhar MG, Hauri HP. In Berger E, Roth J (Eds.), From the GolgiApparatus. Basel: Birkhauser Verlag, 1997: 63–129.

19. Duden R, Schekman R. In Berger E, Roth J (Eds.), From the GolgiApparatus. Basel: Birkhauser Verlag, 1997: 219–246.

20. Burke B, Griffiths G, Reggio H, Louvard D, Warren G. A monoclonalantibody against a 135-K Golgi membrane protein. EMBO J 1982;1:1621–1628.

21. Saraste J, Palade GE, Farquhar MG. Antibodies to rat pancreas Golgisubfractions: identification of a 58-kDa cis-Golgi protein. J Cell Biol1987;105: 2021–2030.

22. Gonatas JO, Mexitis SGE, Steiber A, Fleischer B, Gonatas NK.MG-160. A novel sialoglycoprotein of the medial cisternae of the Golgiapparatus. J Biol Chem 1989;264: 646–653.

23. Luzio PJ, Brake B, Banting G, Howell KE, Braghetta P, Stanley KK.Identification, sequencing and expression of an integral membraneprotein of the trans-Golgi network (TGN38). Biochem J 1990;270:97–102.

24. Schindler R, Itin D, Zerial M, Lottspeich F, Hauri HP. ERGIC-53, amembrane protein of the ER-Golgi intermediate compartment, carriesan ER retention motif. Eur J Cell Biol 1993;61: 1–9.

25. Subramanian VN, Krijnse-Locker J, Tang BL, Ericsson M, YusoffArbM, Griffiths G, Hong W. Monoclonal antibody HFD9 identifies anovel 28 kDa integral membrane protein on the cis-Golgi. J Cell Sci1995;108: 2405–2414.

26. Waters MG, Pfeffer SR. Membrane tethering in intracellular transport.Curr Opin Cell Biol 1999;11: 453–459.

27. Schekman R. Genetic and biochemical analysis of vesicular traffic inyeast. Curr Opin Cell Biol 1992;4: 587–592.

28. Houdebine LM, Djiane J, Dusanter-Fourt I, Martel P, Kelly PA, DevinoyE, Servely JL. Hormonal action controlling mammary activity. J DairySci 1985;68: 489–500.

29. Amsterdam A, Jamieson JD. Structural and functional characteriza-tion of isolated pancreatic exocrine cells. Proc Natl Acad Sci USA1972;69: 3028–3032.

30. Taylor RS, Jones SM, Dahl RH, Nordeen MH, Howell KE. Character-ization of the Golgi complex cleared of proteins in transit andexamination of calcium uptake activities. Mol Biol Cell 1997;8:1911–1931.

31. Barr FA, Nakamura N, Warren G. Mapping the interaction betweenGRASP65 and GM130, components of a protein complex involved inthe stacking of Golgi cisternae. EMBO J 1998;17: 3258–3268.

32. Howell KE, Palade GE. Hepatic Golgi fractions resolved into mem-brane and content subfractions. J Cell Biol 1982;92: 822–832.

33. Bjellqvist B, Sanchez JC, Pasquali C, Ravier F, Paquet N, Frutiger S,Hughes GJ, Hochstrasser D. Micropreparative two-dimensional elec-trophoresis allowing the separation of samples containing milligramamounts of proteins. Electrophoresis 1993;14: 1375–1378.

34. Yates JR, Carmack E, Eng JK. Direct database searching usingtandem mass spectra of peptides. In Celis JE (Ed.), Cell Biology: ALaboratory Handbook. San Diego: Academic Press, 1998: 539–546.

35. Bordier C. Phase separation in integral membrane proteins in TritonX-114 solution. J Biol Chem 1981;256: 1604–1607.

36. Dylewski DP, Keenan TW. Compound exocytosis of casein micellesin mammary epithelial cells. Eur J Cell Biol 1983;31: 114–124.

37. Blatchford DR, Hendry KA, Turner MD, Burgoyne RD, Wilde CJ.Vectorial secretion by constitutive and regulated secretory pathwaysin mammary epithelial cells. Epithelial Cell Biol 1995;4: 8–16.

38. Turner MD, Rennison ME, Handel SE, Wilde CJ, Burgoyne RD.Proteins are secreted by both the constitutive and regulated secre-tory pathways in lactating mouse mammary epithelial cells. J CellBiol 1992;117: 269–278.

39. Zaks WJ, Creutz CE. Evaluation of the annexin as potential media-tors of membrane fusion in exocytosis. J Bioenerg Biomembr1990;22: 97–120.

781Traffic 2000: 1: 769–782

Wu et al.

40. Baram D, Linial M, Mekori YA, Sagi-Eisenberg R. Ca2+-dependentexocytosis in mast cells is stimulated by the Ca2+ sensor, synapto-tagmin I. J Immunol 1998;161: 5120–5123.

41. Xu T, Ashery U, Burgoyne RD, Neher E. Early requirement foralpha-SNAP and NSF in the secretory cascade in chromaffin cells.EMBO J 1999;18: 3293–3304.

42. Arvan P, Castle D. Sorting and storage during secretory granulebiogenesis: looking backward and looking forward. Biochem J1998;332: 593–610.

43. Vincent S, Settleman J. Inhibition of RhoGAP activity is sufficient forthe induction of Rho-mediated actin reorganization. Eur J Cell Biol1999;78: 539–548.

44. Norman JC, Price LS, Ridley AJ, Koffer A. The small GTP-bindingproteins, Rac and Rho, regulate cytoskeletal organization and exo-cytosis in mast cells by parallel pathways. Mol Biol Cell 1996;7:1429–1442.

45. Sander EE, Collard JG. Rho-like GTPases: their role in epithelialcell-cell adhesion and invasion. Eur J Cancer 1999;35: 1905–1911.

46. Doussau F, Gasman S, Humeau Y, Vitiello F, Popoff M, Boquet P,Bader MF, Poulain B. A Rho-related GTPase is involved in Ca(2+ )-dependent neurotransmitter exocytosis. J Biol Chem 2000;275:7764–7770.

47. Xing H, Zhang S, Weinheimer C, Kovacs A, Muslin AJ. 14-3-3proteins block apoptosis and differentially regulate MAPK cascades.EMBO J 2000;19: 349–358.

48. Daniotti JL, Martina JA, Giraudo CG, Zurita AR, Maccioni HJ. GM3alpha2,8-sialyltransferase (GD3 synthase): protein characterizationand sub-Golgi location in CHO-K1 cells. J Neurochem 2000;74:1711–1720.

49. Watson EL. GTP-binding proteins and regulated exocytosis. CritRev Oral Biol Med 1999;10: 284–306.

50. Nguyen DD, Neville MC. Tight junction regulation in the mammarygland. J Mammary Gland Biol Neoplasia 1998;3: 233–246.

51. Wu CC, Howell KE, Neville MC, Yates JR, McManaman JL. Pro-teomics Reveal A Link Between The Endoplasmic Reticulum AndLipid Secretory Mechanisms In Mammary Epithelial Cells. Elec-trophoresis 2000; In press.

52. Leelavathi DE, Estes LW, Feingold DS, Lombardi B. Isolation of aGolgi-rich fraction from rat liver. Biochim Biophys Acta 1970;211:124–138.

53. Wessel D, Flugge UI. A method for the quantitative recovery ofprotein in dilute solution in the presence of detergents and lipids.Anal Biochem 1984;138: 141–143.

54. Pasquali L, Fialka I, Huber LA. Preparative two-dimensional gelelectrophoresis of membrane proteins. Electrophoresis 1997;18:2573–2581.

55. Neuhoff V, Arnold N, Taube D, Ehrhardt W. Improved staining ofproteins in polyacrylamide gels including isoelectric focusing gelswith clear background at nanogram sensitivity using CoomassieBrilliant Blue G-250 and R-250. Electrophoresis 1988;9: 255–262.

56. Shevchenko A, Keller P, Scheiffele P, Mann M, Simons K. Identifica-tion of components of trans-Golgi network-derived transport vesi-cles and detergent-insoluble complexes by nanoelectrospraytandem mass spectrometry. Electrophoresis 1997;18: 2591–2600.

57. Johnson DI. Cdc42: an essential Rho-type GTPase controlling eu-karyotic cell polarity. Microbiol Mol Biol Rev 1999;63: 54–105.

58. Goldberg J. Decoding of sorting signals by coatomer through aGTPase switch in the COPI coat complex. Cell 2000;100: 671–679.

59. Wu WJ, Erickson JW, Lin R, Cerione RA. The g-subunit of thecoatomer complex binds to Cdc42 to mediate transformation. Na-ture 2000;405: 800–804.

60. Kroschewski R, Hall A, Mellman I. Cdc42 controls secretory andendocytic transport to the basolateral membrane of MDCK cells.Nat Cell Biol 1999;1: 8–13.

61. McDonald WH, Yates JR III. Proteomic tools for cell biology. Traffic2000;1: 747–754.

782 Traffic 2000: 1: 769–782

.