Cell, Vol. 61, 709-721,

May 16, 1990, Copyright

0

1990 by Cell Press

SNAPs, a Family of NSF Attachment Proteins Involved in Intracellular Membrane Fusion in Animals and Yeast Douglas 0. Clary:t Irene C. Griff: and James E. Rothman’ * Department of Biology Princeton University Princeton, New Jersey 08544 f Department of Biochemistry Stanford University Stanford, California 94305

Three new and likely related components of the cellular fusion machinery have been purified from bovine brain cytosol, termed a-SNAP (35 kd), P-SNAP (36 kd), and y-SNAP (39 kd). Transport between cisternae of the Golgi complex measured in vitro requires SNAP activity during the membrane fusion stage, and each SNAP is capable of binding the general cellular fusion protein NSF to Golgi membranes. The SNAP-NSFmembrane complex may be an early stage in the assembly of a proposed multisubunit Fusion machine” on the target membrane. SNAP transport factor activity is also found in yeast. Yeast cytosol prepared from a secretion mutant defective in export from the endoplasmic reticulum (secl7) lacks SNAP activity, which can be restored in vitro by the addition of pure a-SNAP, but not p- or y-SNAPs. These data suggest that the mechanism of action of SNAPs in membrane fusion is conserved in evolution. Introduction Intracellular membrane fusion requires both catalysis and specificity. Despite the wide variety of membranes that are programmed to fuse in pairwise fashion, recent findings suggest that a common mechanism involving the fusion protein NSF (N-ethylmaleimide-sensitive fusion protein) may underlie diverse types of fusion events in all eukaryotes. NSF was first identified and purified according to its ability to restore the capacity for transport vesicle fusion to N-ethylmaleimide (NEM)-treated Golgi membranes (Glick and Rothman, 1987; Block et al., 1988; Malhotra et al., 1988; Orci et al., 1989). Subsequent studies revealed that NSF is also required to drive fusion of endoplasmic reticulum (ER)-derived transport vesicles with the Golgi stack, for transport between the trans Golgi and the trans-Golgi network, and for at least one fusion step on the pathway of receptor-mediated endocytosis (Beckers et al., 1989; Rothman, 1987; Diaz et al., 1989). Cloning and sequencing of the mammalian NSF gene (Wilson et al., 1989) revealed a striking homology to a gene from yeast, SEC78 (Eakle et al., 1988). The protein product of yeast SEC78 (SEC18p) in fact contains NSF activity that will catalyze the fusion reaction between mammalian transport vesicles and Golgi membranes. In all likelihood, ATP hydrolysis is involved in the action of NSF, as NSF has a ca-

nonical ATP binding site in its sequence and rapidly inactivates in the absence of ATP How does NSF interact with membranes to promote fusion? Although NSF can be isolated in a membranebound form, its amino acid sequence indicates that it is intrinsically water-soluble, and it is purified in such a state (Wilson et al., 1989; Block et al., 1988). This suggests that NSF operates in conjunction with other components to catalyze membrane fusion. Identification of these would constitute a major step forward toward elucidating the mechanism of fusion. More particularly, Weidman et al. (1989) have shown that, by itself, pure NSF will not bind to Golgi membranes. However, addition of crude cytosol (containing a factor(s) we have termed SNAP for soluble NSF attachment protein) enables NSF to bind to Golgi membranes (Weidman et al., 1989). Binding of NSF is saturable both with respect to an integral membrane receptor in the Golgi membranes and with respect to the SNAP components in the cytosol fraction, suggesting that a stoichiometric complex is formed. Here we report that each of a family of three related peripheral membrane proteins (of 35, 36, and 39 kd) have both SNAP and transport factor activities. These three proteins (now called a-SNAP, f3-SNAP, and ?-SNAP respectively) were previously purified according to their transport factor activities (Clary and Rothman, lQQO), without knowledge of their SNAP activities. We find that the formation of coated transport vesicles is independent of SNAPS and that the transport activity of SNAP proteins is first needed at the same step as NSF-the first discernable stage in the NSF-dependent fusion pathway. It appears that the same pathway operates in yeast during fusion of ER-derived transport vesicles with the Golgi stack, as the yeast sec77 mutant is deficient in an activity that can be replaced with a-SNAP Results Background An in vitro assay has been devised (Balch et al., 1984a) that measures the transport of the vesicular stomatitis virus-encoded glycoprotein (VSV-G) from the cis cisternae of Golgi stacks in fractions prepared from a mutant Chinese hamster ovary (CHO) cell line (defective in N-acetylglucosaminyl [GlcNAc] transferase 1) to the medial cisternae of wild-type CHO Golgi stacks; the addition of tritiated GlcNAc from its sugar nucleotide donor marks the arrival of VSV-G at the acceptor compartment and serves as a quantitative measure of the entire transport reaction. Analysis of kinetic intermediates formed during the in vitro transport pathway has revealed several sequential stages in the maturation of a transport vesicle prior to fusion (Figure 1; Balch et al., 1984b; Wattenberg et al., 1988; Malhotra et al., 1988; Orci et al., 1989). Inhibition of transport with the nucleotide analog GTPyS blocks the maturation of vesicles early in the pathway, resulting in an accumulation of coated vesicles that have transferred from donor to

Cell 710

NSF SNAPS

ATP

BuddIng

Figure

Targeting

1. Summary

IJncoatlng

of the Pathway

Based on data presented et al., 1990.

Leading

here and results

I-

to the Fusion from Balch

Maturathl

of Transport

Vesicles

et al.. 1984b: Wattenberg

acceptor cisternae, but whose cytoplasmic coats have failed to be removed (Melancon et al., 1987). These “Golgi”-coated vesicles lack clathrin, but contain instead several other prominent protein species (Malhotra et al., 1989). A subsequent intermediate, now uncoated, is accumulated by prior treatment of the Golgi membranes with NEM under conditions that inactivate the fusion protein NSF (Malhotra et al., 1988). A later intermediate in the fusion pathway, termed the LCRI (low cytosol-requiring intermediate), requires a higher concentration of cytosol to be produced than to be consumed and occurs just before the last step, which is an ATP-dependent bilayer fusion reaction (Balch et al., 1984b; Wattenberg et al., 1986). The transport block imposed by the inactivation of NSF with NEM can be overcome by addition of a peripheral membrane protein extract prepared by incubating Golgi membranes with ATP and KCI (Glick and Rothman, 1987) and NSF was purified to homogeneity based on this activity assay (Block et al., 1988). Kinetic and morphological data (Malhotra et al., 1988; Orci et al., 1989) have confirmed the requirement for NSF at an early stage in vesicle maturation, showing that it is not needed until after the GTPyS block, but is needed to form the LCRI. We have recently described a fractionated transport system that reconstitutes intercisternal vesicular transport using CHO Golgi membrane fractions that have been stripped of many peripheral membrane proteins by extraction with 1 M KCI (termed “K Golgi”; Clary and Rothman, 1990). We noted that although a standard preparation of yeast cytosol is sufficient to drive transport with untreated CHO Golgi membrane fractions (Dunphy et al., 1986; Clary and Rothman, 1990) yeast cytosol is not sufficient for reconstitution of transport with the salt-extracted K Golgi membranes. Evidently, certain peripheral transport activities, removed by the salt wash, cannot be supplied by yeast cytosol. They can, however, be supplied by animal cytosol. Fractionation revealed that in addition to mammalian NSF, two resolved pools of bovine brain cytosol, with average molecular sizes of 500 kd and 40 kd (termed Fraction 1 [Frl] and Fraction 2 [Fr2], respectively),

-------+

with the Acceptor

Membrane

Fusion

Cisterna

et al., 1986; Malhotra

et al., 1988; Orci et al., 1989; and Pfanner

are needed for reconstitution. Yeast cytosol thus acts as a “biochemical mutant” for these activities and can reconstitute K Golgi membranes for transport only when all three animal activity pools are also present. Using this assay, a purification protocol was developed for the complementing activities in the Fr2 pool (Clary and Rothman, 1990). Fr2 activity was found to consist of any of three likely related proteins with apparent sizes of 35, 36, and 39 kd, known as FrP-a, -6, and-y, respectively. These three proteins have similar molecular weights, chromatographic properties, and in vitro activities, but largely different peptide maps. Each has a different specific activity and plateau value when reconstituting transport with K Golgi membranes. No synergy of the three Fr2 proteins was seen in mixing experiments, but complex formation between them has not been ruled out. Assay of Fr2 Activities in a Homologous System The fractionated K Golgi transport system as originally described contain K Golgi membranes, NSF, yeast cytosol, and two pools of bovine brain cytosol termed Frl and Fr2, which were resolved from each other by gel permeation chromatography (Clary and Rothman, 1990). We sought to revise the FrSdependent transport assay so that the activities of the Fr2 proteins could be measured in an assay containing only mammalian components, to study their function in the absence of yeast homologs. To accomplish this, we obtained crude fractions of bovine brain cytosol that lacked Fr2 activity but could efficiently reconstitute transport when mixed with both Frl, Fr2, and NSF Based on the chromatographic properties of Fr2 (Clary and Rothman, 1990) two pools of bovine brain cytosol were prepared as a replacement for yeast cytosol: Fr3, the brain cytosol flowthrough fraction of Q-Sepharose (50 mM KCI), and Fr4, brain cytosol depleted of Fr2 proteins by passage over phenyl-Sepharose. Together, these two pools could efficiently substitute for yeast cytosol in driving transport in assays containing K Golgi, Frl, crude or purified Fr2, and NSF (Figure 2A, bar 4). Under these conditions, transport is dependent upon the additions of both

NSF Attachment 711

Proteins

in Animals

and Yeast

10

Fr2 protein

a

P

-

66-

.

-39

=3t

21.5-

. .

1

2

3

(ng)

Figure 2. Assay of the Fr2 Family Golgi Transport System

Y

M,x10-3

97

added

of Proteins

in a Homologous

Intra-

(A) Development of a FR-dependent transport assay using only mammalian components. Transport with K Golgi can be reconstituted with purified NSF and Fr2 proteins and three cytosolic pools prepared from bovine brain cytosol. K Golgi donor and acceptor fractions were incubated in assays containing 23 ug of Frl, 6.5 ug of Fr3, 3.1 trg of Fr4, 5.6 ng of NSF, and 12 ng each of FrP-a, -8, and -y. The reactions were performed using standard conditions, and transport was monitored by the incorporation of tritiated GlcNAc into VSV-G. Reactions containing all five cytosolic pools (complete) as well as reactions lacking the Frl pool (no Frl), Fr2 protein mixture (no Fr2), or CHO NSF (no NSF) are shown. Background subtracted was 156 cpm-the result of an incubation in the absence of all cytosolic fractions. All points were determined in duplicate. (B) Each Fr2 protein has activity in the homologous K Golgi transport assay. Reactions contained K Golgi donor and acceptor fractions, Frl, Fr3, Fr4, and NSF, in the concentrations described in (A). Each Fr2 species (a, open squares; 8, solid circles; and y, open triangles) was titrated individually from 0 to 16 ng. Also shown is the analysis of an equal weight mixture of FR-a, -5. and -y, titrated at the same total concentration of added protein (solid squares). The values expected for the mixture of Fr2 proteins if their activities were additive are shown with open circles and were calculated as averages of the individual activities. Data points are the average of duplicates. The background incorporation measured in the absence of exogenous Fr2 protein (588 cpm) has been subtracted. (C) Electrophoretic analysis of the three Fr2 proteins. One microgram of each Fr2 protein was subjected to electrophoresis on a 10% SDS-polyacrylamide gel and stained with Coomassie brilliant blue R-250. Lanes l-3 contain FR-a (35 kd), FrP-8 (36 kd), and FrP-Y (39kd), respectively. The relative molecular sizes of the marker proteins are shown at left: phosphorylase b (97 kd), BSA (66 kd). ovalbumin (43 kd). carbonic anhydrase (31 kd), and soybean trypsin inhibitor (21.5 kd).

Cell 712

Frl and Fr2 (Figure 2A, bars 1 and 2) and NSF (Figure 2A, bar 3). Transport is reduced approximately 500/o, but not eliminated, upon omission of either Fr3 or Fr4 (not shown), most likely because their respective transport activities are not completely resolved from each other. Titration of Fr2-a, -b, and -y shows that the specific activities and plateau values of the three Fr2 proteins (Figure 26) are largely the same as those found with the yeast cytosol-based assay (Clary and Rothman, 1990); therefore, the presence of yeast cytosol did not activate or inhibit the individual Fr2 proteins, which might have resulted from the formation of heterologous protein complexes of different efficacies. An approximately equimolar mixture of the three Fr2 proteins (Figure 26) has an activity on a weight basis that is similar to the average expected for the three proteins assayed individually, so no synergy between them is apparent. Electrophoretic analysis of the purified Fr2-a, -b, and -y proteins is shown in Figure 2C.

ciated with acceptor stacks (Orci et al., 1989) upon inhibition of transport with the analog. Therefore, vesicle budding has occurred (and vesicle transfer and targeting as well; Orci et al., 1989), but later events in the vesicle fusion process, such as removal of the coat, are blocked by GTPyS (MelanGon et al., 1987; Orci et al., 1989). Subsequent intermediates, such as the NSF-dependent intermediate and the LCRI, are part of a series of subreactions that results in bilayer fusion (Malhotra et al., 1988). If the requirement for Fr2 activity can be localized after the GTPyS block, it would imply that Fr2 is part of the fusion pathway; if Fr2 is required before the GTPyS block, it would be implicated in the budding or targeting reactions. Figure 3 shows the results of a two-stage incubation experiment designed to determine the stage of action of Frl, Fr2, and NSF relative to the GTPyS block. In the first stage we performed either a complete K Golgi transport reaction or reactions that were missing any one of Frl, Fr2, or NSF to see which were needed to drive the transport reaction past the point of GTPyS blockade. In the second stage, bovine brain cytosol and NSF were added to saturate the requirements for Frl, Fr2, and NSF, and a GTPyS block was imposed. UDPj3H]GlcNAc was added to measure the fusion of VSV-G-containing transport vesicles with the acceptor cisternae. The transport measured in the complete reaction indicates the maximum amount of VSV-G that could have progressed past the GTPyS block in stage 1 (Figure 3). Reactions lacking either NSF or Fr2 in the first stage were still able to drive over 70% of the VSV-G protein

Fr2 Proteins Are Not Needed for Coated Vesicle Budding or Attachment to the Acceptor Cisterna At what stage in vesicular transport are the Fr2 family of proteins required? Preincubation experiments can define whether a component is needed for vesicle formation, consumption, or both. One clear landmark in the transport pathway is the point of inhibition by GTPyS, mediated by an as yet unidentified GTP binding protein(s) (MelanCon et al., 1987; Figure 1). Coated vesicles carrying the VSV-G protein accumulate (MelanGon et al., 1987) and are asso-

Figure 3. K Golgi + Fr3. + Fr4 f Frl fFRCI f NSF

45

FrZ-a and NSF Act after the GTPyS

Block

Preincubate:

Add: UDP-[3H]-GlcNAc some Brain cytosol NSF GTPS min

At

/

w J

/

stop

Complete Premcubatlon

/

Preincubation

lacking FrPa

Stage 1 incubation (preincubation): The complete reaction (solid squares) was performed using standard assay conditions and contained 5 PI (2 pg) of mixed K Golgi donor and acceptor, 5 pl of 1 M sucrose, 23 pg of Frl, 12 ng of Fr2-a. 4.3 pg of Fr3, 3.1 pg of Fr4, and 5.6 ng of NSF in a volume of 50 ~1. Preincubations in which Frl (open circles), FrP-a (open triangles), or NSF (solid circles) was omitted were also performed. The reactions were preincubated for 45 min at 30°C. Stage 2 incubation: The following addittons were made to each reaction: 224 vg of bovine brain cytosol, 5.6 ng of NSF, and 0.3 ICi of

UDPj3HJGlcNAc; 20 PM GTPyS was added to

Preincubation

0

45 Time (At ) of incubation

90 (minutes)

lading

Frl

block transport. The reactions were allowed to incubate a further 45 or 90 min at 30°C, and then the VSV-G protein was immunoprecipitated as usual. The amount of tritiated GlcNAc Incorporated into VSV-G protein during the second stage is plotted. The background of incorporation during stage 2 was determined from assays in which a complete preincubation was performed at 0%. These second-stage backgrounds were 85 cpm (0 min), 142 cpm (45 min), and 330 cpm (90 min). and have been subtracted. All points were done in duplicate.

NSF Attachment 713

Proteins

in Animals

and Yeast

into a GTPyS-resistant state (compared with the complete incubation), while in the reaction lacking Frl in the first stage little GTPyS-resistant intermediate was formed. Omission of GTPyS from the second stage allowed complete recovery of transport activity after each of the first stage conditions (data not shown). These results imply that the Fr2 proteins, like NSF, are first required in a step subsequent to the GTPyS block and therefore at some point during the fusion pathway. It appears that an activity provided by Frl is needed during an early stage in transport, possibly vesicle production, and thus the majority of the VSV-G protein in preincubations lacking Frl could not progress to a GTPyS-resistant state. Fr2 Proteins Act in Parallel with NSF in the Fusion Pathway Several intermediates have been detected during maturation in the fusion pathway (Figure 1; Balch et al., 1984b; Wattenberg et al., 1986; Malhotra et al., 1988; Orci et al., 1989). The VSV-G protein passes through the following states after the GTPyS block: an NSF-dependent intermediate, an intermediate that is resistant to cytosolic dilution (LCRI), and a prefusion complex resistant to NEM treatment at 37% but requiring ATP to progress to the glycosylation step. This last stage includes the process of bilayer fusion. To determine when Fr2 is first required along the fusion pathway, we asked whether Fr2 (like NSF) is required to form the LCRI. Complete reactions or reactions lacking either Fr2 or NSF were incubated in stage 1 to allow VSV-G protein to progress to the LCRI at the acceptor cisternae (Figure 4). In stage 2, NSF or Fr2 was added to those reactions that lacked them in the first stage. The reactions were then diluted 5-fold with either a mixture containing UDP-[3H]GlcNAc, sucrose, salts, nucleotides, and all cytosolic fractions (each at the standard assay concentrations) as acontrol, or an identical mixture, but lacking the cytosolic fractions to measure the consumption of the dilution-resistant LCRI selectively. The result of the dilution is that the concentrations of the cytosolic fractions either remain constant (control) or drop to 20% of their original levels. At these low concentrations of cytosol, the assay measures exclusively the consumption of the LCRI (Wattenberg et al., 1986; Wattenberg and Rothman, 1986). Control reactions that contained all required cytosolic pools during the preincubation were about 50% resistant to the effects of dilution in the second stage (Figure 4, bars 1 and 2, and 5 and 6). Therefore, about half of the VSV-G protein could progress to the late stages of fusion in 45 min (at 30%). Reactions lacking Fr2 (bar 4) or NSF (bar 8) during the preincubation generated only small levels of the dilution-resistant LCRI intermediate. Thus, it appears that Fr2 (like NSF; Malhotra et al., 1988) is required to progress to the LCRI. Together, the GTPrS and dilution experiments imply that Fr2 is first required after uncoating and prior to the LCRI. Because NSF and Fr2 are qualitatively and quantitatively indistinguishable in the staging assays, it is likely that they both act at a similar point in the pathway, possibly at the same step.

Each of the Fr2 Proteins Can Bind NSF to Golgi Membranes Using in vitro binding assays, a cytosolic factor (SNAP) has been defined that enables NSF to bind to an integral membrane receptor in Golgi membranes (Weidman et al., 1989). SNAP seemed to have the properties of a peripheral membrane protein, and assays of SNAP in crude cytosol suggested that NSF and SNAP form stoichiometric complexes of high affinity. The native molecular size of SNAP is about 35 kd based on gel filtration of crude cytosol. Given that the Fr2 proteins are of this size and that at least FrPa acts in concert with NSF in the fusion process, we wondered whether FrBa (and possibly Fr2-8 or -y) might be SNAP We tested which of the Frl, Fr2, Fr3, or Fr4 pools of bovine brain cytosol contains SNAP activity as measured in the in vitro NSF binding assay (Figure 5). As these pools together contain all activities needed to reconstitute transport with K Golgi membranes (except NSF), SNAP activity should be present in at least one if SNAP is needed for transport. To measure NSF binding, carbonate-extracted Golgi membrane fractions are incubated with purified CHO NSF and a cytosolic fraction to be assayed for SNAP activity at 0%. The membranes are then pelleted, and the amount of membrane-bound NSF in the pellet is measured in a second stage, using an NSF-dependent transport assay (NEM-treated Golgi fractions and NSF-free CHO cytosol; Block et al., 1988). Figure 5 shows that SNAP activity indeed was restricted to the crude Fr2 pool and was also found in an approximately equimolar mixture of the purified FR2-a, -8, and -y proteins. A standard yeast cytosol preparation failed to bind NSF, as reported previously (Weidman et al., 1989). Which of the Fr2 proteins have SNAP activity? Remarkably, each Fr2 species has binding activity, although their relative specific activities are quite different (Figure 6). The plateau value of binding for brain cytosol was slightly lower than that of Fr2-a; while the plateau of the mixture of the three purified Fr2 proteins was approximately the same as FR-a alone (data not shown). Mixing experiments using subsaturating concentrations of the three SNAPS failed to reveal any synergy in binding (not shown). Importantly, the concentrations of each Fr2 protein needed to fulfill the transport requirement (Figure 28) were very similar to the concentrations needed to bind NSF to membranes. This implies that SNAP (and membrane binding) are essential components in transport in the fusion pathway. Prior to this, there had been no direct evidence that an association of NSF with membranes is essential for its action in fusion. The specific activities of brain cytosol and the Fr2 proteins were compared in the FrP-dependent transport assay and the NSF binding assay (Table 1). Purification of the FrZ-a protein resulted in a 580-fold enrichment of the Fr2 activity over bovine brain cytosol and a 1900-fold enrichment of SNAP activity. The three Fr2 species show the same relative order of specific activity in both assays, with the FrP-a protein having the highest activity and FR-y protein having the weakest. The extent of purification of FrP-a based on transport activity has been judged to be an underestimate of the true value (Clary and Rothman, 1990)

Cell 714

K Golgi + Frl, Fr3, Fr4 f NSF fFRU

dilute 1:5 in buffer containing all cytosol fractions

q

)

stop

)

stop la

60 min 1

45 min Add t\ NSF or FR a

dilute 15 in buffer without cytosol fracUons 60 min

2

1

0 1

Preincubetion: Figure

4. FR-a

2

3 4

complete

noFRa

and NSF Act before

5

6

complete

7

8

no NSF

the LCRI

(A) The Fr2 requirement precedes the LCRI. Stage 1 incubation: All reactions contained 2 jrg of K Golgi donor and acceptor fractions, 23 trg of Frl, 4.3 pg of Fr3, 3.1 ug of Fr4, and 5.6 ng of NSF in a volume of 25 sl and were performed under standard assay conditions. Twelve nanograms of FrP-a was either present (complete reaction, bars 1 and 2) or absent (no Fr2 a. bars 3 and 4). The reactions were incubated for 45 min at 3O%, and then 12 ng of FR-a was added to those reactions from which it had been omitted. Stage 2 incubation: All assays were diluted with the addition of 100 pl of a dilution mix either containing cytosol fractions (dark cross-hatching, bars 1 and 3) or lacking cytosol fractions (light cross-hatching, bars 2 and 4). Both dilution mixes contained 0.2 M sucrose, 12 pCi/ ml UDP-[3H]GlcNAc, and the standard assay concentrations of potassium chloride, magnesium acetate, HEPES (pH 7.0) palmitoyl-CoA, DTT, and the ATP regenerating system. The dilution mix containing cytosol fractions also included per 100 ul: 92 ug of Frl, 48 ng of FrP-a, 17.2 ug of Fr3, 12.4 fig of Fr4, and 22.4 ng of NSF. Transport that was completed during the second incubation (60 min, 30%) is measured by the incorporation of tritiated GlcNAc into VSV-G protein. All determinations were made in duplicate. The assay background, determined from a reaction in which FrP-a was not present in either the first or second stages of incubation (454 cpm), has been subtracted. (B) The NSF requirement precedes the LCRI. Incubations were done essentially as described in (A), except that NSF was present (complete, bars 5 and 6) or absent (no NSF, bars 7 and 8) during the first stage incubation. Afterward, NSF was added to those reactions lacking it, and the reactions were diluted with a mix containing the cytosol fractions (dark cross-hatching, bars 5 and 7) or lacking them (light cross-hatching, bars 6 and 8) as described above. All reactions were done in duplicate. The background (345 cpm) was determined from reactions that lacked NSF in both stages of incubation and has been subtracted.

to the low yield of activity (2%). One likely explanation for the low yield is the presence of transport factors other than Fr2 in the starting material that stimulate the FR-dependent transport assay and cause an overestimation of the Fr2 concentration in the homogenate. The owing

higher plateau value of bovine brain cytosol compared with a crude pool of Fr2 activity (such as that used in Figure 5) in the transport assay indicates the presence of such extrinsic activators (data not shown). Since the binding assay exhibits a similar plateau value for brain cytosol

NSF Attachment 715

Proteins

in Animals

and Yeast

2-l

I

Figure 5. The Family Binding) Activity

of Fr2 Proteins

Is Responsible

for SNAP

Yeast cytosol, brain cytosol, and the Frl-Fr4 cytosolic pools were tested for the ability to bind NSF to membranes. A modified NSF binding assay was used (see Experimental Procedures) in which the association of purified NSF with sodium carbonate-extracted Golgi membrane fractions is measured. Each pool was tested in the following amounts: yeast cytosol, 21 ng; bovine brain cytosol. 56 pg; Frl. 23 ng; Fr3,4.3 Kg; Fr4, 3.1 ng; crude Fr2 pool (prepared by chromatography of bovine brain cytosol on Q-Sepharose), 3.2 hg; and FrP-uf3y, 16 ng each of FrP-a, -8, and -y. The amount used of each pool is sufficient to saturate its requirement in the in vitro K Golgi transport assay. The amount of NSF bound is measured with an NSF-dependent transport assay (Glick and Rothmann, 1987; Block et al., 1988) and is plotted as units of NSF (Weidman et al., 1989). The background of NSF associated with the membranes in the absence of added cytosol (0.24 U) has been subtracted; all points were determined in duplicate.

and crude Therefore, the binding The Fr2

or pure Fr2, irrelevant stimulation is less likely. the 1900-fold purification of activity based on assay is probably a more accurate value. protein family, purified on the basis of its re-

Table 1. Comparison FrP-Dependent

of Transport

Transport

and NSF Binding

brain cytosol

NSF Binding

Bovine FrP-a FrP-8 FrP-y

of Bovine

Figure

6. Each Fr2 Species

Can Bind NSF to Membranes

FrP-a, -6, and -y and bovine brain cytosol were individually titrated in the NSF binding assay. Assay conditions were as described in Figure 5. The activity of each protein is plotted as units of NSF bound. The background of the assay (0.21 U of NSF bound in absence of added protein) has been subtracted. All points were determined in duplicate.

quirement in fusion of transport vesicles, is necessary to attach the fusion protein NSF to a membrane receptor. The members of the Fr2 family will thus be renamed a-SNAP (FR-a, 35 kd), f%SNAP (Fr2-f3,36 kd), and y-SNAP (Fr2-y, 39 kd). The SNAPs Form Complexes with NSF in the Absence of Membranes Although SNAPS are necessary for NSF attachment to Golgi membranes, they need not bind to NSF directly.

Brain Cytosol

and the Fr2 Proteins

Assaya Specific (cWn9)

Bovine FrP-a Fr2-8 FrP-y

Activities

FrZlSNAP protein added (ng) BoVltw BraIn CytOMl added (MO)

(NSF

Activity

Specific (Relative

Fold Enrichment

Activity to FrP-a)

111

0.533 308 114 86

578 214 161

Specific Activity (Units NSF Bound per ng [ x 104])

Fold Enrichment

2.22 4250 1450 420

1910 653 189

Ill 0.37 0.28

Assayb

brain cytosol

Specific (Relative

Activity to FrP-a)

Ill

a Bovine brain cytosol or purified FR-a, -8, or -y proteins were titrated into the yeast cytosol-based man, 1990). Specific activities are based on an estimate of the initial slope of the titration. b Bovine brain cytosol or pure FrP-a, -8, or -y proteins were titrated into the NSF binding assay mined from an estimate of the initial slope.

il 0.34 0.10 FrP-dependent (see Figure

transport

assay (Clary

6). The specific

activities

and Rothare deter-

Cell 716

Figure7. A Solid Phase Binding Assay Detects Interaction between NSF and Each SNAP Species

B E 2 d c’ B D s 3 3

0.0

0.6

0.4

0.2

1

2

3

4 60

5

6

7

20

60

60

60 -

ng SNAP added

-

-

bg brain cytosol

20 60

20

---

---

---

a SNAP

p SNAP

y SNAP

6

-+-

9

60

BSA pretreatment of tube

a SNAP

SNAP could bind to the membrane receptor, for example, thereby increasing the receptor’s affinity for NSF. We took advantage of SNAP’s ability to bind plastic surfaces to set up a solid phase binding assay that has in fact shown that each SNAP can, by itself, form a complex with NSF. Microcentrifuge tubes were preincubated with dilutions of the purified SNAP proteins and then saturated by incubation with a solution of bovine serum albumin (BSA). NSF was added to the tubes, incubated for 10 min on ice, and removed. The amount of NSF that had bound to the tube was measured by using an NSF-dependent transport assay. The association of NSF with the plastic surface was strongly dependent on preincubation of the surface with SNAP proteins (Figure 7). Each of the three SNAPS could facilitate the NSF binding (bars l-7). Preincubation of the tubes with BSA eliminated NSF association, presumably by blocking association of the SNAP with the plastic surface (bar 8). Bovine brain cytosol with an equivalent amount of SNAP activity could not promote NSF binding (bar 9), likely owing to the high level of nonspecific protein present compared with purified SNAP Since we cannot easily measure the amount of SNAP that has bound the tube, we do not know the relative efficiency of each SNAP protein in binding NSF, but clearly each is capable of binding NSF in the absence of the membrane receptor. a-SNAP (FrS-a) Activity of Yeast Is Defective in the sec77 Mutant NSF and SNAP are both required for reconstitution of vesicular transport in the homologous animal cell system described so far. We wanted to establish conditions in which the presumed yeast homologs of SNAPS could be assayed so as to screen for mutants in the existing yeast secretion mutant collection (Novick et al., 1980). Although yeast does contain a homolog of NSF (SEC18p) that will

added

The binding of NSF to SNAP was investigated with a solid phase assay. Microcentrifuge tubes were incubated with 20 or 60 ng of pure SNAP protein in a 20 PI volume (a-SNAP, bars 1, 2, 7, and 6; D-SNAP. bars 3 and 4; y-SNAP, bars 5 and 6) for 15 min at 0% and removed. The tubes were incubated with BSA to block any remaining binding sites and then with 14 ng of NSF (in 20 ~1 of NE) for 10 min. After removmg the NSF solution, the tubes were washed once with NB. and then the amount of NSF associated with the plastic surface was quantitated with an NSF-dependent transport assay. Each pure SNAP protein could cause NSF to associate with the plastic. If the BSA treatment preceded incubation with a-SNAP (bar 6), or if bovine brain cytosol (containing an equivalent amount of SNAP activity but lOOO-fold more protem, bar 9) was used in the first incubation, no association of NSF with the plastic occurred. The amount of NSF detected is plotted as units of NSF bound; all assays were performed in duplicate, and the backgroundof binding in the absence of protein in the first incubation (0.076 U) has been subtracted.

rescue Golgi transport in reactions lacking mammalian NSF (Wilson et al., 1989), yeast cytosol prepared in the absence of ATP (as is standardly done) loses all endogenous NSF activity (Wilson et al., 1989). Standard yeast cytosol prepared without ATP also does not possess a SNAP activity that will bind Cl-l0 NSF to CHO Golgi (Figure 5). We found, however, that a Fr2 activity (presumably a SNAP) could be demonstrated in yeast cytosol provided that excess yeast NSF (SEC18p) was used as the source of NSF activity in the transport reaction. As no purification currently exists for SEC18p, we used cytosol prepared from a SEClBp-overproducing yeast strain (Wilson et al., 1989) as a crude source of SEC18p. Addition of this cytosol to the K Golgi assay negated the requirement for exogenous CHO NSF (not shown) and also relieved the requirement for bovine brain Fr2 proteins (SNAPS) in the transport process (Figure 8, bar 5). The requirement for Frl addition is unaffected by the presence of SEC18p (bar 4). The addition of yeast cytosol that lacks active SEC18p to the K Golgi transport assay does not affect requirements for Frl or SNAP activity (Figure 8, bars l-3). This indicates a direct need for yeast NSF (SEC18p) in the process by which yeast Fr2 activity is utilized in transport, consistent with the idea that yeast Fr2 activity, like mammalian Fr2 or SNAP, interacts with NSF. Preliminary data from mixing experiments with resolved fractions confirm that the yeast Fr2 can function in the presence of yeast NSF, but not CHO NSF (not shown). We interpret this to mean that yeast SNAP will not bind animal NSF to membranes, but that yeast SNAP present in the cytosol can bind yeast NSF (SEC18p) to the membranes to form an active complex. We therefore have a means to test for possible Fr2 (putatively SNAP) mutations in yeast. To screen set mutants for possible defects in Fr2 activities, we employed a simplified activity assay consisting of

NSF Attachment 717

Proteins

in Animals

and Yeast

yeast cylosol added (no acllve SEC16p)

I

0 1 Frl added

-++

a-SNAPadded

+

2

3

-

+

4

5

6

-++

Figure 8. Yeast Cytosol That Contains Possesses SNAP (Fr2) Activity

+-+ Endogenous

NSF Activity

Also

The effect of yeast NSF (SEClBp) on the K Golgi transport assay was measured by the inclusion of yeast cytosol that either contained or lacked active SEC16p. All assays were done in duplicate and contained K Golgi donor and acceptor fractions, Fr3, Fr4. and mammalian NSF in the concentrations described in Figure 2A. Half of the reactions (bars l-3) were supplemented with 50 ug of yeast cytosol from the strain SEY2102.1 prepared in the absence of ATP to inactivate endogenous SEC18p; the other half of the reactions (bars 4-6) were supplemented with 50 trg of cytosol from the same strain of yeast harboring the SEClBp-overproducing plasmid prepared in the presence of ATP; 23 ng of Frl pool (bars 2 and 3, and 5 and 6) or 8 ng of purified a-SNAP protein (bars 1 and 3. and 4 and 6) or both were also added. Backgrounds subtracted from bars l-3 (321 cpm) and bars 4-6 (574 cpm) were the result of incubations lacking both Frl and a-SNAP addition. The arrows above bars 2 and 5 denote K Golgi transport reactions that lack only the addition of a mammalian SNAP (Fr2) activity pool. The addition of yeast cytosol containing active SEC16p alleviates the requirement for the addition of mammalian SNAP

K Golgi, bovine brain Frl, and yeast cytosol prepared from the set mutant yeast strains. Each cytosol preparation was done in the presence of ATP to stabilize endogenous SEC18p activity. This is in fact an adaptation of the original K Golgi transport assay (Clary and Rothman, 1990) with the yeast cytosol also contributing NSF and SNAP activities as well as the cytosolic transport components. Using this simplified assay, yeast secretion mutants that contain temperature-dependent blocks in transport between the ER and the Golgi were screened for defects in in vitro transport. Cytosol preparations were made from a wildtype yeast strain, as well as from the following secretion mutants: secl6, secli: sec20, sec21, sec22, and sec23 (Novick et al., 1980), all after growth at the permissive temperature. Each cytosol preparation could drive transport in the K Golgi assay with the exception of sec77 (Figure 9). Several different preparations of sec77 cytosol were found to be defective, including a preparation from a sec77 strain carrying the SEC18p-overproducing plasmid (seclir SEC18p). All of the cytosols tested had saturating levels of endogenous NSF activity (measured in the NSF assay of Wilson et al., 1989) and were fully active in driving transport with Golgi membranes that had not been salt ex-

Figure 9. The Yeast Secretion (Fr2) Activity

Mutant

secl7

Cannot

Provide

SNAP

Cytosol was prepared from the following yeast strains: wild type (SEY2102.1) sec76, secli: sec20, sec27, sec22, and sec23. ATP was present at all times to preserve SEC18p activity (Wilson et al., 1989); 44 trg of each cytosol preparation was tested for activity in assays containing 2 trg of K Golgi donor and acceptor fractions, 23 pg of Frl pool, and 5.6 ng of CHO NSF Duplicate assays were performed using standard conditions. The amount of transport reconstituted with each preparation is indicated by the amount of tritiated GIcNAc incorporated into VSV-G protein; background (159 cpm) has been subtracted-the result of a reaction lacking the addition of yeast cytosol. Each yeast cytosol preparation had saturating levels of NSF (SEC18p) activity and could drive transport using unextracted Golgi membrane fractions that contain endogenous SNAP activity (data not shown).

tracted (Dunphy et al., 1986; Clary and Rothman, 1990; data not shown). All of this strongly suggests that the sec77 cytosol is deficient in a Fr2 activity, and this is likely to be responsible for the in vivo secretion defect. Attempts to show a temperature dependence of the sec77 defect were unsuccessful, as sec77 cytosol was inactive at all temperatures tested, and the defect was present when the sec77LSEC18p-overproducing cells were grown either at 23% or 17% (data not shown). The defect in secV/SEC18p cytosol could be reversed by the addition of a wildtype yeast cytosol preparation; this reversal did not require SEC18p activity in the complementing cytosol (data not shown). If the FrPlike defect in sec77 is due to a missing yeast SNAP homolog, the sec77cytosol might be complemented in vitro by one or more pure animal cell SNAPS. In fact, the defect in sec77cytosol was completely reversed by the addition of 5 ng of the purified bovine a-SNAP (Figure 10). Thus, the only defect in the sec77cytosol relevant to transport is in the SNAP activity. Interestingly, even 16 ng of the 6- or y-SNAP proteins failed to detectibly complement the sec77 defect in vitro (Figure 10). The transport assay in this case sharply discriminates among the three related proteins. This is especially important in the case of a-SNAP and B-SNAP, as these have very similar physical properties, transport activities, and peptide maps (Clary and Rothman, 1990). Because the sec77 defect can be over-

Cell 716

I 20

10

Bovine Brain SNAP added (ng) Figure

10. Only a-SNAP

(FrP-a)

Can Rescue

the sec77 Defect

Each purified bovine SNAP species was tested for the ability to complement the biochemical defect found in secl7 cytosol. The reactions contained 2 ftg of K Golgi donor and acceptor fractions, 24 pg of yeast cytosol prepared from the secl7strain overexpressing SEClEp, and 23 ng of Frl pool. The SNAP proteins, CL(open squares), 6 (solid circles), and y (open triangles), were individually titrated from 0 to 16 ng into the reaction. The amount of transport reconstituted is indicated as the amount of tritiated GlcNAc incorporated into VSV-G protein. A background of 142 cpm, determined from reactions lacking the addition of sec77cytosol. has been subtracted. All points are the average of duplicates

come with the addition of only the a-SNAP polypeptide, SEC17p is likely the homolog of a-SNAP in yeast. This conclusion is, however, tentative and requires a direct structural confirmation. Discussion NSF is a generalized fusion protein, and it has been postulated to act in a protein complex (a “fusion machine”) that ultimately causes bilayer fusion (Malhotra et al., 1988). According to this model, once a vesicle has targeted and become uncoated, proteins from the cytosol recognize the vesicle-cisternal junction and assemble in a stepwise fashion to build an enzymatic machinery that catalyzes fusion. This model was developed from kinetic and morphological experiments concerning the fusion stages of intraGolgi transport and the role NSF plays in the pathway (Balch et al., 1984b; Wattenberg et al., 1986; Malhotra et al., 1988; Pfanner et al., 1990). If NSF activity is lacking from a transport reaction, an intermediate accumulates in which the vesicles have targeted and uncoated, but have yet to fuse (Malhotra et al., 1988; Orci et al., 1989). However, following the NSF-dependent intermediate are several biochemically distinct and successive stages that must be traversed before the actual bilayer fusion can occur (Figure 1; Balch et al., 1984b; Wattenberg et al., 1986). Our finding that the attachment of NSF to membranes in the form of complexes with SNAPS is required for the fu-

sion process is a clear demonstration that NSF assembles into a multisubunit complex on the membranes as a prerequisite for fusion. Specifically, each of the three Fr2 family members appears to be a SNAP able to bind NSF to membranes. The NSF-SNAP-membrane complex formed in the in vitro NSF binding assay is likely to be relevant to the mechanism of fusion for the following reasons: both NSF and SNAP (Fr2) are required for in vitro transport activity, each acts during the same stage of the transport reaction, and gently isolated Golgi membrane fractions contain high levels of both NSF and SNAP in a membrane-bound state (Glick and Rothman, 1987; Clary and Rothman, 1990; Weidman et al., 1989). It is likely that other components forming additional subunits of this complex remain to be discovered. The finding that the defect of the secl7 mutant of yeast can be complemented in vitro by pure a-SNAP (but not b-or Y-SNAPS) provides a demonstration of both the physiological importance of SNAP activity in vivo and the remarkable evolutionary conservation of the pathway of membrane fusion. It also offers the second identification of a yeast secretion gene with a known function in the transport process, joining the linkage of SEC18 with NSF fusion protein activity (Wilson et al., 1989). Both sec77and sec78 mutants accumulate ER and small vesicles (presumably transport vesicles) at the restrictive temperatures (Novick et al., 1980) as would be predicted from the requirements for SNAP and NSF in fusion in vitro. It remains to be directly demonstrated that SEC17p binds SEC18p to Golgi membranes. C. Kaiser and Ft. Schekman (personal communication) have found that sec77 and sec78 interact genetically and have confirmed the conclusion that the sec77 and sec78 mutations result in the accumulation of transport vesicles as well as ER. These findings are consistent with and readily explained by our identifications of SEC17p as SNAP and SEC18p as NSF. Two extreme models, each of which can explain the existence of multiple species of SNAPS, are not distinguishable from the available evidence. In one view, the “adaptor” model, each species of SNAP would be specialized for a different subcellular compartment(s), thus enabling a single fusion protein to interact with a multiplicity of target membranes, as NSF must to promote fusion at multiple stations in the Golgi (Malhotra et al., 1988; Rothman, 1987; Beckers et al., 1989) and in the endocytic pathway (Diaz et al., 1989). In favor of the adaptor model is the finding that the various SNAP species seem to act independently in transport (e.g., in Fr2 assays) and in NSF binding assays. As an adaptor, only one of the SNAPS would promote cis to medial Golgi transport; the activity of the other SNAPS would have to be due to cross-talk between integral membrane receptors for SNAP-NSF complexes. A weakness of the adaptor model is that the degree of crosstalk between receptors must then be very high to explain the fact that all three SNAPS have specific activities Well within the same order of magnitude of each other (Table 1). Also, the sec77 mutation produces a defect in export from the ER, and based on our results this defect would lie in the failure of vesicles budded from the ER to fuse with the yeast Golgi complex. Yet a-SNAP was purified on

NSF Attachment 719

Proteins

in Animals

and Yeast

the basis of its need in intra-Golgi transport in animals. If SEC17 encodes a-SNAP in yeast, then a-SNAP/SEC17p (like NSWSEC18p) must function in fusions to at least two very different compartments, which would not be predicted if a-SNAP is adapted for a specific fusion event. The independence of action of the SNAPS in promoting transport (Figure 28) may be an illusion if the Golgi membranes contain low and different levels of each SNAP species. This would eliminate apparent synergy among exogenously provided SNAPS and may well occur as very high salt concentrations (about 1 M KCI) are required to remove Fr2 activities (Clary and Rothman, 1990). Without antibodies, it is impossible to directly assess the presence of SNAPS on the Golgi membranes. Another view (the “particle” model) is that the three SNAPS act cooperatively, forming a hetero a-P-y-SNAPNSF complex: a particle (perhaps with other subunits) that is a generalized fusion machine and catalyzes fusions at multiple compartments. In such a particle each type of SNAP would serve a distinct purpose in the fusion reaction, and each would be required at every compartment employing the NSF-dependent fusion pathway. Several lines of evidence favor the particle model. First, the ability of a-SNAP but not P-SNAP or Y-SNAP to complement thesec77defect in vitro establishes that a-SNAP can provide functions that the other SNAPS cannot. Second, mutation of the sec77 (putatively the a-SNAP gene) in yeast entirely blocks secretion in vivo. Assuming that yeast cells also possess homologs of S- and y-SNAP, this suggests that neither p- nor Y-SNAP can substitute for a-SNAP in vivo. A weakness of the particle model is the apparent lack of synergy of the various SNAPS in NSF binding and in transport (Fr2) activities. Lack of synergy in NSF binding can be explained in the particle model by assuming that any one SNAP subunit is sufficient for membrane attachment, even though the complete particle is necessary for fusion activity. The lack of synergy among the three SNAPS in transport assays is a more serious deficiency, although as noted above, this may well be an illusion. Clear distinctions between these and other less extreme models can be made from the in situ localization of SNAPS, from direct determinations of the complexes that combinations of SNAP species can form with NSF, and from seeking mutations in p-and Y-SNAP genes in yeast and examining their phenotypes. The identification of SNAPS as a novel set of molecules involved in the pathway of intracellular membrane fusion brings us one step closer to an understanding of the means by which a powerful physical chemical process, bilayer fusion, is utilized with great specificity and control. Experimental

Procedures

Preparation of Mammalian Membrane and Cytosoi Fractions The preparation of VSV-infected CHO 158 donor Golgi fractions and wild-type acceptor Golgi fractions has been described (Balch et al., 1984a). K Goigi were prepared by extracting donor and acceptor fractions with 1 M KCI at 3pC and reisolating them for use in transport reactions, as previously described (Ciary and Rothman, 1990). Golgi membranes used in the NSF activity assay were treated with 1 mM NEM to inactivate endogenous NSF (Block el al., 1988). Carbonatetreated membranes used in the in vitro NSF binding assay were pre-

pared by extracting wild-type Golgi fractions with 100 mM sodium carbonate and reisolating them by centrifugation (300,000 x g. 30 min) through a sucrose step gradient containing cushions of 35% and 25% sucrose in 10 mM Tris (pH 7.4). The membranes were recovered at the 35%/25% interface in the original volume (yielding a 0.6 mglml concentration of protein). Bovine brain cytosoi was prepared (Wattenberg and Rothman, 1986) with some modifications (Clary and Rothman, 1990); it was concentrated by precipitation with 60% ammonium sulfate and dialyzed into 50 KTD (50 mM KCI, 25 mM Tris [pH 7.8],1 mM dithiothreitol [DTT]) before use. The Frl pool was prepared, and the FrP-a, -B, and -y proteins were purified from bovine brain homogenates as described (Clary and Rothman, 1990). NSF was purified from CHO cell homogenates (Block et al., 1988). Fr3 was prepared by chromatography of bovine brain cytosol on Q-Sepharose (Pharmacia). Sixteen milliliters of bovine brain cytosol (after the ammonium sulfate concentration and dialysis, 7 mglml) were loaded onto a 7.5 ml Q-Sepharose column (9.6 x 1.0 cm) equilibrated in 50 KTD. The flowthrough peak of protein was collected and pooled, typically yielding a Fr3 pool of about 2 mglml. Under these conditions, Fr2 activity binds to the column and elutes at about 200 mM KCI and is therefore depleted from the Fr3 pool. The Fr4 pool was prepared by subjecting 10 ml of bovine brain cytosol (prepared as above, 11 mglml) to two rounds of chromatography over a 12 ml phenyl-Sepharose column (6.8 x 1.5 cm, Pharmacia) equilibrated in 50 KTD. In each round, the flowthrough fractions are pooled. Fr2 activity binds to the matrix and thereby becomes depleted from the Fr4 pool. This procedure yields a preparation of Fr4 of about 1.6 mglml. Yeast Strains and Cytosol Preparation See Table 2 for a list of strains used. Standard yeast cylosol was prepared as described (Dunphy et al., 1986). To obtain yeast cyiosol with active SEC18p, the cytosol was prepared and desalted in buffers containing ATP as described (Wilson et al., 1989). To obtain cytosol containing overexpressed SEC18p, yeast strains were transformed with the YEpl3/pSEC18-1 plasmid (a 2pmbased multicopy plasmid carrying the SEC78 gene under its own promoter (the kind gift of K. Eakle and S. Emr, California Institute of Technology) and cytosol prepared in ATP as above. Transport Assay Conditions The homologous K Golgi transport assay used in this study to assay for Fr2 activity is a modification of the K Golgi/yeast cytosol transport assay described earlier (Clary and Rothman. 1990). Typically, a reaction contains 5 bI(2 vg) of K Golgi membranes (in 1 M sucrose, 10 mM Tris [pH 7.41. mixed donor and acceptor), 23 ug of Frl pool, 4.3 pg of Fr3 pool, 3.1 w of Fr4 pool, 5.6 ng of NSF, and the Fr2 pool being tested, in a volume of 25 ~1. Under these conditions, reconstitution of transport is highly dependent on the addition of both the Frl and Fr2 activity pools. The concentrations of the salts, ATP regenerating system, and palmitoyl-CoA are as previously described (Clary and Rothman, 1990). Transport is measured by the addition of 0.3 WCi of UDP[3H]GlcNAc (lo-30 Cilmmol) per assay, and the VSV-G protein is immunoprecipitated as described (Balch et al., 1984a). Incubations are performed at 30X for 2 hr, unless stated otherwise. NSF Binding Assay The following solutions are used in the NSF binding assay (Weidman et al., 1989). NB (NSF binding buffer) contains 20 mM HEPES (pH 7.4), 2 mM EDTA, 100 mM KCI, 250 PM ATP, 1 mM DTT, 1% polyethyleneglycol4000, and 250 pglml soybean trypsin inhibitor (final concentrations). RB (rinse buffer) contains 80% NB, 20% 1 M sucrose. RB plus Mg is identical to RB, except that the 2 mM EDTA is replaced by 2 mM MgClp: Cylosoiic pools were tested for NSF binding activity in a modification of the in vitro NSF binding assay described in Weidman et al. (1989). The assay is performed in two stages. The first stage is the formation of a complex between carbonate-extracted Golgi fractions, purified NSF, and SNAP provided by a cyiosolic pool. The incubation contains, in a total volume of 20 pI,O.72 Fg of carbonate-extracted membranes, 14 ng of NSF, O-4 ~1 of a cytosolic pool to be tested for SNAP activity, and a final concentration of lx NB. The reactions are performed in 1.5 ml microtubes that have previously been incubated for 10 min with a

Cell

720

Table 2. Yeast Strain

Strains

Used

in This Study

Genotype

Source

SEY2102.1

(a; ufa3-52;

/eu2-3,7

RSY267

(a; sec76-2; (a; sec77-7;

ufa3-52; ura3-52;

WY269

WY275 RSY277 RSY279 RSY281 IGY267 IGY269

(a; (a; (a; (a (a (a;

IGY275

(a;

IGY277 IGY279 IGY281

(a;

(a; (a;

72; ade2-707;

suc2-A9;

ga12; ApepRtLEU2)

his4-679) his4-679)

sec20-7; ufa3-52; his4-679) sec27-7; ufa3-52) sec22-3; ura3-52; his4-619) sec23-7; ufa3-52; his4-679) sec76-2; ade2-707; Apep4::LElJP; sucBA9; sec7 7-7; his4-679; ade2-701; Apep4::LElJP sec20-7; ade2-707; Apep4::LEU2; suc2-A9; sec27-7; ade2-707; Apep4::LEUP; suc2-A9; sec22-3; ade2-707; Apep4::LEU2; suc2-A9; sec23-7; his4-679; Apep4::LEUP; suc2-A9;

ufa3-52)

suc2-A9;

ufa3-52)

ufa3-52) ufa3-52) ufa3-52)

ufa3-52)

S. Emr, California Institute R. Schekman, University R. Schekman. University R. Schekman, University R. Schekman, University R. Schekman, University R. Schekman, University This study This study This study This study This study This study

of of of of of of

of Technology California, California, California, California, California, California,

Berkeley Berkeley Berkeley Berkeley Berkeley Berkeley

The IGY strains were obtained by mating each set mutant strain (from R. Schekman) with the wild-type strain SEY2102.1. Spores were obtained from each mating, and haploids were isolated by scoring for the ade2 phenotype. Each strain was then screened for the temperature-sensitive phenotype (the set allele), suc2 deficiency (Goldstein and Lampen, 1975). and pep4 phenotype (Jones, 1977). Note that any chromosomal copy of leu2 is masked by the PEP4 insertion and that the gal2 phenotype was not scored.

BSA solution (10 mg/ml rn water) and rinsed with water, to block nonspecific binding sites. The reactions are incubated for 2 min at 0%. then pelleted to separate the membrane-bound NSF from the soluble NSF (12,000 x g, 4 min). The supernatants are carefully removed, and the pellets are washed by pipetting in and out 50 ul of RB. Ten microliters of RB plus Mg is added to each tube to overlay the membrane pellet. In the second stage, the amount of NSF bound to the membranes is quantified in an NSF-dependent transport assay. Forty microliters of a premix containing NEM-treated donor and acceptor membranes, NSF-free CHO cytosol. ATP regenerating system, UDP[sH]GlcNAc, and salts (assay conditions as described in Block et al., 1988) is added to each tube, the reactions are incubated for 2 hr at 30%, and the VSV-G protein immunoprecipitated. The incorporation of tritiated GlcNAc into VSV-G protein indicates the units of NSF activity that had associated with the membranes. NSF activity in the second stage of the reaction was dependent on the addition of carbonatetreated membranes and a SNAP activity pool in the first (binding) stage. The technique of measuring the NSF concentrations directly on the membrane pellets was found to yield the same values as the pellet homogenization protocol described in Weidman et al. (1989). Solid Phase NSF Binding Assay To test whether NSF could bind SNAP proteins in the absence of membranes, we used a solid phase binding assay where NSF could associate with SNAP proteins adsorbed to plastic. Pure SNAP proteins were diluted into 50 mM KCI, 25 mM Tris (pH 7.8). 1 mM DTT. The diluted SNAPS were incubated rn a volume of 20 ul in a 1.5 ml microfuge tube (Sarstedt) on ice for 15 min. The SNAP dilution was removed, and the tubes blocked by incubation with 100 nl of a 10 mglml BSA solution in the same buffer conditions. After 2 min. the BSA was removed, and 20 nl of an NSF solution (14 ng of NSF, lx NB) was incubated in the tubes for 10 min. After removing the NSF, the tubes were washed with 50 ul of NB. To measure the NSF associated with the tube, 10 ul of NB and 40 PI of NSF assay mix (identical to the one used in the NSF membrane binding experiments above) were added, and the reactions were incubated for 2 hr at 30%. The amount of tritium incorporated into VSV-G protein was measured as above. Miscellaneous Protein samples were analyzed by electrophoresis on 0.75 mm discontinuous SDS slab gels using the method of Laemmli (1970). Proteins were visualized by staining with Coomassie brilliant blue R-250. Protein concentrations were determined by the Bio-Rad protein assay, using BSA as a standard and a BSA correction factor (actual concentration of the sample = 2.5x measured concentration).

like to thank

Received

February

13, 1990.

References Balch, W. E., Dunphy, W. G., Braell, W. A., and Rothman, J. E. (1984a). Reconstitution of the transport of protein between successive compartments of the Golgi measured by the coupled incorporation of N-acetylglucosamine. Cell 39, 405-416. Balch, W. E., Glick, 8. S., and Rothman. J. E. (1984b). Sequential intermediates in the pathway of intercompartmental transport in a cellfree system. Cell 39, 525-536. Beckers. C. J. M., Block, M. R., Glick, B. S., Rothman, J. E., and Balch. W. E. (1989). Vesicular transport between the endoplasmic reticulum and Golgi stack requires the NEM-sensitive fusion protein. Nature 339, 397-398. Block, M. R.. Glick, B. S.. Wilcox, D. A., Wieland, F. T., and Rothman, J. E. (1988). Purification of an N-ethylmaleimide-sensitive protein catalyzing vesicular transport. Proc. Natl. Acad. Sci. USA 85, 7852-7856. Clary, D. O., and Rothman. J. E. (1990). Purification ripheral membrane proteins needed for vesicular Chem.. in press.

F? Weidman,

P Melancon,

and D. Wilson

for

of three related petransport. J. Biol.

Diaz, R., Mayorga, L. S., Weidman, f? J., Rothman, J. E., and Stahl, F? D. (1989). Vesicle fusion following receptor-mediated endocytosis requires a protein active in Golgi transport. Nature 339, 398-400. Dunphy, W. G., Pfeffer, S. R., Clary, D. O., Wattenberg, B. W.. Glick, B. S., and Rothman, J. E. (1986). Yeast and mammals utilize similar cytosolic components to drive protein transport through the Golgi complex. Proc. Natl. Acad. Sci. USA 83, 1622-1626. Eakle, K. A., Bernstein, M., Emr, S. D. (1988). Characterization component of the yeast secretion machinery: identification SEC18 gene product. Mol. Cell. Biol. 8, 4098-4109. Glick, B. S.. and Rothman, coenzyme A in intracellular Goldstein, hydrolase

of a of the

J. E. (1987). Possible role for fatty acylprotein transport. Nature 326, 309-312.

A., and Lampen. J. 0. (1975). 8D-Fructofuranoside from yeast. Meth. Enzymol. 42. 504-511.

fructo-

of Sacchafomyces

cerevisiae.

Laemmli. U. K. (1970). Cleavage of head proteins during of the head of bacteriophage T4. Nature 227, 880-685.

the assembly

Jones, E. W. (1977). Proteinase Genetics 85, 23-33.

Acknowledgments We would

helpful discussions, C. Kaiser and R. Schekman for strains and discussion, and Barbara Devlin for excellent technical assistance. This research was supported by a National Institutes of Health grant (DK27044) to J. E. R. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

mutants

NSF Attachment 721

Proteins

in Animals

and Yeast

Malhotra, V., Orci, L., Glick. 8. S., Block, M. R., and Rothman, J. E. (1988). Role of an N-ethylmaleimide-sensitive transport component in promoting fusion of transport vesicles with cisternae of the Golgi stack. Cell 54, 221-227. Malhotra. V., Serafini, T., Orci, L., Shepherd, J. C., and Rothman, J. E. (1989). Purification of a novel class of coated vesicles mediating biosynthetic protein transport through the Golgi stack. Cell 58, 329-336. Melancon. F?, Glick, 6. S., Malhotra, V., Weidman. P. J., Serafini, T., Gleason, M. L.. Orci, L., and Rothman. J. E. (1987). Involvement of GTP-binding “G” proteins in transport through the Golgi stack. Cell 51, 1053-1062. Novick, P, Field, C., and Schekman. R. (1980). Identification plementation groups required for post-translational events secretory pathway. Cell 21, 205-215.

of 23 comin the yeast

Orci. L., Malhotra, V., Amherdt, M., Serafini, T., and Rothman. J. E. (1989). Dissection of a single round of vesicular transport: sequential intermediates for intercisternal movement in the Golgi stack. Cell 56, 357-368. Pfanner, N.. Glick, 8. S., Arden, S. R., and Rothman, J. E. (1990). Fatty acylation promotes fusion of transport vesicles with Golgi cisternae. J. Cell Biol. 710, 955-962. Rothman, J. E. (1987). Bansport of the vesicular stomatitis glycopro tein to frans Golgi membranes in a cell-free system. J. Biol. Chem. 262, 12502-12510. Wattenberg, 6. W., and Rothman, J. E. (1988). Multiple cytosolic components promote intra-Golgi protein transport. Resolution of a protein acting at a late stage, prior to membrane fusion. J. Biol. Chem. 261, 2208-2213. Wattenberg, 8. W., Balch, W. E., and Rothman, J. E. (1988). A novel prefusion complex formed during protein transport between Golgi cisternae in a cell-free system. J. Biol. Chem. 267. 2202-2207. Weidman, P J., Melancon, P., Block, M. R., and Rothman. J. E. (1989). Binding of an N-ethylmaleimide-sensitive fusion protein to Golgi membranes requires both a soluble protein(s) and an integral membrane receptor. J. Cell Biol. 708, 1589-1596. Wilson, D. W., Wilcox, C. A., Flynn, G. C., Chen, E., Kuang, W.-J., Henzel, W. J., Block, M. R., Ullrich, A., and Rothman, J. E. (1989). A fusion protein required for vesicle-mediated transport in both mammalian cells and yeast. Nature 339, 355-359.