A Heterodimer of Thioredoxin and I^B₂ Cooperates with Sec18p (NSF) to Promote Yeast Vacuole Inheritance

Reagents were purchased or prepared as described (Xu and Wickner, 1996; Mayer et al., 1996; Haas and Wickner, 1996). Yeast strains were described in Xu and Wickner (1996) and Muller (1995). The strains YS18 (Mata ura3-5 his3-11 leu2-3,-112 can^R) and YPS19 (Mata Δpbi2::URA3) were kindly provided by Dr. D.H. Wolf (Institut für Biochemie der Universitat Stuttgart, Germany).

Preparation of cytosol, vacuoles, salt-washed vacuoles, and purification of the p10 subunit of LMA1 were as described (Xu and Wickner, 1996). For salt-washing, freshly isolated vacuoles were diluted to 0.3 mg/ml with 0% Ficoll buffer (Haas et al., 1994). Equal volumes of BJ3505 and DKY6281 vacuoles were mixed and KCl and KOAc added (from 3M stocks) to 100 mM and 50 mM, respectively. Aliquots of 200 μl were transferred to Eppendorf tubes, incubated 10 min at 30°C, chilled at 0°C for 2 min, and centrifuged (10,000 g, 75 s, 4°C). Supernatants were removed and pellets covered with 160 μl of cold 0% Ficoll buffer. LMA1 and Sec18p were largely removed by salt-washing.

The 30-μl in vitro vacuole fusion reaction (Haas et al., 1994) contains 4.8 μg of mixed, salt-washed vacuoles from BJ3505 and DKY6281, 20 mM Pipes-KOH, pH 6.8, 200 mM sorbitol, 150 mM KOAc, 5 mM MgCl₂, an ATP regenerating system (1 mM ATP, 40 mM creatine phosphate, and 70 U/ml creatine phosphokinase), either 1 mg/ml cytosol (from yeast strain K91-1A), fractions from Sephacryl S100HR gel filtration or purified proteins as indicated, and 0.1× proteinase inhibitor cocktail (PIC; Xu and Wickner, 1996). Salt-washed vacuoles were used in all experiments. In vitro reactions were incubated at 25°C for 90–120 min and stopped by chilling on ice. Alkaline phosphatase activity was determined as described (Haas et al., 1994). 1 U of fusion activity corresponds to 1 μmol of p-nitrophenol produced at 25°C per minute per microgram of BJ 3505 vacuoles.

Cytosol (stored at −80°C) was thawed and mixed with PIC and ATP to final concentrations of 30× and 0.1 mM, respectively, and then incubated on ice for 20 min followed by centrifugation (14,000 g, 15 min, 4°C). A 200-μl aliquot of clarified cytosol was applied to a Sephacryl S100HR column (1 cm × 26 cm) equilibrated in buffer C (20 mM TrisCl, pH 8.3, 5 mM Mg(OAc)₂, and 0.1 mM MnCl₂). This buffer has been optimized in Mn concentration for LMA1 purification. The column was eluted (12 ml/cm²/h, 300 μl fractions) with buffer C at 4°C.

Yeast vacuoles were visualized with the vital fluorophores fluorescein-5isothiocyanate (FITC, Molecular Probes, Inc., Eugene, OR) or FM 4-64 as described (Vida and Emr, 1995; Xu and Wickner, 1996).

Protein concentration was determined with a protein reagent (Bio-Rad Laboratories, MA) with BSA as a standard. SDS-PAGE and “High Tris” SDS-PAGE (Xu and Wickner, 1996) and affinity purification of IgG and immunoblot analysis (Mayer et al., 1996; Haas and Wickner, 1996) have been described. Thioredoxin reductase was purified from baker's yeast as described (Gonzales et al., 1970). Reductase activity was determined with the 5,5′-dithiobis (2-nitrobenzoic acid) (DTNB) assay described in Luthman and Holmgren (1982).

The vacuole inheritance reaction can be assayed in vitro by incubating isolated, salt-washed vacuoles with ATP and cytosol at physiological temperatures. The reaction is performed with vacuoles purified from two yeast strains (Haas et al., 1994). One strain has normal vacuolar proteases but a deletion in the PHO8 gene encoding the vacuole proalkaline phosphatase. The other strain has a normal PHO8 gene but lacks vacuolar proteases A and B which are needed for processing the catalytically inactive pro-alkaline phosphatase to its mature, active form. Upon fusion of vacuoles isolated from these strains, the proteases from one fusion partner cleave the pro-alkaline phosphatase from the other to produce the catalytically active enzyme, allowing quantitative assay of vacuole fusion.

We have previously reported the purification of a low molecular weight activity, termed LMA1, which supports the in vitro reaction (Xu and Wickner, 1996). LMA1 is a 23-kD heterodimer with a thioredoxin subunit of 12 kD and a smaller subunit termed p10. Edman degradation of the purified p10 subunit showed that it has the sequence ThrLysAsnPheIleValThrLeuLysLysAsnThrProAspValGluAlaLysLys which is identical to a portion of the reported sequence of yeast proteinase yscB inhibitor I^B₂, encoded by the PBI2 gene (Schu et al., 1991). Antibody prepared to a peptide from the COOH terminus of this protein specifically decorated the p10 subunit of LMA1 (data not shown), confirming its identification from the NH₂-terminal sequence. To determine whether I^B₂ is a functional component of LMA1, cytosols were prepared from a pbi2 null mutant and its corresponding wild-type strain. Cytosolic components were resolved by Sephacryl S100HR chromatography and fractions were assayed for their ability to replace cytosol in the in vitro reaction of vacuole fusion. The wild-type cytosol showed both the previously reported (Xu and Wickner, 1996) high molecular weight activity peak, termed HMA, and LMA1 (Fig. 1,A). Deletion of the PBI2 gene left the HMA peak while obliterating activity from the LMA1 region (Fig. 1,B). The identity of the LMA1 peak from wildtype cytosol was confirmed by immunoblot with anti-TRX1p peptide antiserum, whereas monomeric thioredoxin, recovered from more included fractions of pbi2Δ cytosol (data not shown), had no activity in our assay (Fig. 1,B). Deletion of the PBI2 gene had no effect on the cellular content of thioredoxin (data not shown). Thus I^B₂ is clearly an essential component of LMA1. To test whether I^B₂ plays a role in vacuole inheritance in vivo, the PBI2 deletion mutant and wild-type cells were examined by fluorescence microscopy. The PBI2 deletion mutant showed a clear vac phenotype of buds which frequently had no vacuole (Fig. 2,B). While almost all wild-type cells (Fig. 2,A) with a bud diameter of at least half the maternal cell diameter had inherited a bud vacuole, 34% of PBI2 deletion cells had no vacuole in their buds (Fig. 2,B and Table I). Taken together, these in vitro and in vivo results demonstrate that I^B₂, previously identified as a cytosolic inhibitor of vacuolar proteinase B, is a novel factor required for efficient vacuole inheritance in yeast.

To understand how LMA1 participates in the in vitro reaction, we tested whether it requires the redox activity of its thioredoxin subunit or the proteinase B inhibitor activity of its I^B₂ subunit. Yeast thioredoxin reductase was assayed and purified from baker's yeast as described (Gonzales et al., 1970; Luthman and Holmgren, 1982). No stimulation of fusion was observed when both NADPH and purified thioredoxin reductase were added to fusion reactions supported by LMA1 (Fig. 3,A). To further examine the role of the thioredoxin redox activity in vacuole fusion, we used a mutant form of Trx1p in which both cysteinyl residues at its active site were replaced by serines (Muller, 1995). The redox activity of Trx1p is completely abolished by this means. The genes for mutant and wild-type forms of TRX1 were transformed into the trx1, trx2 double deletion strain and cytosols were prepared. As reported (Xu and Wickner, 1996), the fusion activity of TRX1,TRX2 wildtype cytosol (Fig. 3,B a) is resolved into HMA, LMA1, and LMA2 by Sephacryl S100HR gel filtration and the LMA1 peak is selectively lost in the trx1,trx2 double deletion strain (Fig. 3,B b). As expected, the LMA1 peak is restored when the wild-type TRX1p is expressed from a plasmid in the trx1,trx2 double deletion strain (Fig. 3,B c). Strikingly, the LMA1 peak is also restored in this trx1,trx2 double deletion strain by a plasmid which expresses the double Cys to Ser mutant form of Trx1p (Fig. 3 B d). Furthermore, the vac phenotype of the trx1,trx2 double deletion strain was rescued in vivo by introducing this double Cys to Ser mutant form of Trx1p (data not shown). These data demonstrate that the function of thioredoxin in vacuole inheritance and fusion does not employ a cyclic reduction-oxidation mechanism.

Since LMA1 does not act via redox, we asked whether it acts by inhibiting vacuolar proteinase B, the purported activity of its other subunit. Sec18p is one of the two necessary soluble factors, as shown below (Fig. 5), and functions with LMA1 to support vacuole fusion and is also protease sensitive. We therefore checked the stability of Sec18p during the fusion reaction in the presence, or absence, of LMA1. The amount of soluble and membrane-bound fulllength Sec18p remained unchanged (Fig. 4,A), indicating that LMA1 is not acting to prevent Sec18p degradation. To determine whether LMA1 functions in vacuole fusion via inhibition of proteinase B cleavage of some other, unidentified vacuole protein, we prepared vacuoles from the strain BJ3505 (Moehle et al., 1986) in which both proteinase A and B genes are deleted and examined the fusion of these vacuoles by fluorescence microscopy. While clustering of these vacuoles was promoted by Sec18p (Mayer and Wickner, 1997), the fusion of these protease-deficient vacuoles still requires both LMA1 and Sec18p (Fig. 4 B). Measurements of all vacuole diameters in random fields showed average diameters of 0.58 μm (0.25 SEM), 0.73 μm (0.24), 0.66 μm (0.22), or 1.29 μm (0.21) after incubation with ATP only, LMA1 only, Sec18p only, or LMA1 and Sec18p, respectively. Thus, LMA1 does not support vacuole fusion via a proteinase B inhibitor activity of its I^B₂ subunit. The function of LMA1 in vacuole fusion must employ novel mechanisms which are unrelated to the established functions of its subunits in other cellular processes.

Both LMA1 and Sec18p can support vacuole fusion (Xu and Wickner, 1996; Haas and Wickner, 1996), yet the degree of requirement for each of these pure proteins varies among fresh vacuole preparations (Xu, Z., A. Mayer, E. Muller, and W. Wickner, unpublished observations). We find a consistent requirement for both proteins when the vacuoles have been salt-washed. Thus, little fusion occurred with limiting concentrations of Sec18p or LMA1 alone (Fig. 5,A), but full activity was obtained when both proteins were added. Furthermore, the reaction supported by these two pure proteins occurred to the same extent as seen with cytosol (Fig. 5 B) and was still sensitive to Gdi1p, GTPγS, and microcystin-LR, known inhibitors of later stages of the reaction (Conradt et al., 1994; Haas et al., 1994, 1995). Our isolated vacuole preparations have variable amounts of Sec18p and LMA1 that are largely removed or inactivated during a salt-wash procedure, and these two pure proteins constitute a minimal set of soluble proteins needed for reconstitution of the fusion reaction.

Previous studies (Conradt et al., 1994) have established that our in vitro reaction of vacuole inheritance occurs in defined stages in an obligatory sequence. To establish the stage(s) where Sec18p and LMA1 act, vacuoles were incubated with ATP, LMA1, and Sec18p. At the indicated times (Fig. 6,A), vacuoles were collected by centrifugation and resuspended in either the same, complete reaction mix or in buffer and ATP alone before resuming the second incubation. To control for fusion that occurred at each time of centrifugation, an aliquot was also transferred to ice. After a 30-min incubation, the reisolated vacuoles can undergo fusion without added LMA1 or Sec18p in the second incubation (Fig. 6,A). As expected, the reaction became resistant to anti-Sec18p antibodies in the second incubation but was still sensitive to GTPγS and Microcystin LR (Fig. 6 B). Thus, LMA1 and Sec18p participate at an early stage of the reaction.

To determine whether Sec18p and LMA1 function in an obligatory sequence, we first tested whether LMA1 stimulated the Sec17p release reaction, driven by Sec18p and ATP. Vacuoles were incubated with either no addition, LMA1, Sec18p, or both proteins for various times, then reisolated and assayed by immunoblot for bound Sec17p. LMA1 does not stimulate Sec17p release in our assay (Fig. 7,A). We further examined Sec17p release from fresh vacuoles (not salt-washed) prepared from a trx1, trx2 double deletion strain and a pbi2 null strain and found that neither LMA1 subunit is required to stimulate Sec17p release (data not shown). We then determined whether LMA1 action can be completed before that of Sec18p. Vacuoles were incubated with ATP and LMA1 alone for 30 min, after which time LMA1 can normally be removed without loss of subsequent reaction (Fig. 6). After reisolation, vacuoles still required both LMA1 and Sec18p for fusion (Fig. 7 B). Thus, the requirement for LMA1 cannot be satisfied before the Sec18p-catalyzed release of Sec17p.

LMA1 cannot act long after Sec18p-mediated Sec17p release, as this reaction creates a labile state of the vacuoles in which the vacuoles remain intact (data not shown) but lose competence for the reaction which leads to fusion (Fig. 8,A). Vacuoles were pre-incubated with ATP alone or with ATP and either Sec18p, antibody to Sec18p, or LMA1. After various incubation times, vacuoles were collected and resuspended in reaction buffer with ATP and both LMA1 and Sec18p for a second incubation. In agreement with earlier observations (Conradt et al., 1994), the competence of vacuoles to undergo fusion was lost during pre-incubation in the absence of LMA1 (Fig. 8,A) or cytosol (data not shown); the latter may contain additional stabilizing factors as well as LMA1. This lability was exacerbated by added Sec18p and partially relieved by an antibody to Sec18p and thus was largely due to Sec18p action. This lability limits the interval after Sec18p-mediated Sec17p release in which LMA1 can act. Can the Sec18p/ Sec17p reaction be separated from the LMA1 requirement, despite their close kinetic relationship? We incubated vacuoles with Sec18p for varying intervals, mixed them with an antibody to Sec17p, and added LMA1 (Fig. 8,B). Upon addition at 0 time, the antibody effectively blocked the reaction as there had been insufficient time for Sec17p release (Mayer et al., 1996). If the vacuoles were instead incubated for a few minutes before addition of antibody to Sec17p, the release reaction occurred with little vacuole inactivation and LMA1, when added, “captured” the activated vacuoles for the subsequent fusion reaction. After longer incubations, however, the labile vacuoles (Fig. 8,A) became inactive. This delay in addition of LMA1 resulted in loss of subsequent fusion capacity (Fig. 8 B). Thus LMA1 must act during, or shortly after, the Sec18p/ 17p reaction to permit the later steps of the reaction.

Our in vitro reaction measures the last step of inheritance, the fusion of vacuoles, but relies on a sequence of coupled steps which represents earlier stages in the pathway as well. Thus, the first step of inheritance to be observed in vivo, the creation of a segregation structure which projects into the bud, is also seen in our in vitro reaction (Conradt et al., 1992) and requires the vac-encoded proteins (Conradt et al., 1992; Haas et al., 1994; Nicolson et al., 1995). Recent studies (Fanton, V., and W. Wickner, unpublished) indicate that structure formation requires Sec17 and Sec18 proteins, further connecting it to the early stages of the in vitro reaction as assayed biochemically (Mayer et al., 1996). Analysis of the in vitro reaction thus offers insight into several stages of the organelle inheritance process.

Analysis of this inheritance reaction initially established four subreactions which occur in an obligate order: (I) A brief salt-dependent step, (II) a cytosol-dependent step, (III) an ATP-dependent step, and (IV) a step sensitive to microcystin-LR (Conradt et al., 1994). These reactions culminate in the fusion of vacuoles and mixing of contents, assayed by proenzyme maturation. Subsequent studies have identified some of the proteins which support these reactions. These include Sec17p (α-SNAP) and Sec18p (NSF) (Haas and Wickner, 1996), LMA1 (Xu and Wickner, 1996), and the Ras-like GTPase Ypt7p (Haas et al., 1995). As isolated, the vacuoles bear on their surface many or, in some vacuole preparations, all of the requisite proteins. Nevertheless, after “salt-wash,” both Sec18p and LMA1 are needed to reconstitute the cytosol requirement. Like the earlier findings with cytosol, our current studies show that these proteins act at an early stage. LMA1 stabilizes the reaction competence of vacuoles and is required during or immediately after the Sec18p-mediated release of Sec17p. Docking also requires prior Sec17p release (Mayer et al., 1997) and is followed by the Stage IV (Conradt et al., 1994) microcystin LR and GTPγS-sensitive fusion reaction.

Despite the increasing resolution of this reaction into an ordered series of steps, the catalytic roles of the relevant proteins are not known. LMA1, the focus of the current report, was detected in a general scheme of cytosol fractionation and purified to homogeneity (Xu and Wickner, 1996). Each subunit, thioredoxin and I^B₂, is needed for normal vacuole inheritance in vivo (Fig. 2 and Table I; also, Xu and Wickner, 1996). However, LMA1 acts neither in a typical thioredoxin redox reaction (Fig. 3) nor solely to inhibit proteinase B (Fig. 4). Thioredoxin has previously been shown to be required in a nonredox fashion in T7 replication (Huber et al., 1986) where it confers processivity on the polymerase, in filamentous phage assembly (Russel and Model, 1986), and in the mammalian early pregnancy factor system (Tonissen et al., 1993). It is also reasonable that the I^B₂ subunit has a role in addition to that of proteinase B inhibitor, since I^B₂ is cytosolic and proteinase B is soluble in the vacuole lumen and thus these two proteins may only meet infrequently under normal growth conditions. Though our current studies indicate that LMA1 may stabilize the activity of vacuoles in early reaction steps, the LMA1 requirement cannot be satisfied before Sec18p action. LMA1 acts during or immediately after the Sec17p/Sec18p reaction; its relation to vacuole docking and to the Ypt7p-dependent events is still to be determined. Nevertheless, the observation of its coupling to the action of the yeast homologues of NSF and α-SNAP, which are required for almost every step of membrane trafficking, suggests that LMA1 or its functional equivalents may also have a necessary role in other intracellular trafficking events. A detailed comparison of homotypic fusion during vacuole inheritance and the heterotypic fusion events during inter-organelle vesicular traffic will require identification of any SNARE proteins which support vacuole fusion.

Further understanding of the mechanism of vacuole inheritance will require a full resolution of the proteins which support the process. This in turn raises the question of developing assays for these components. Some assays will arise from a continued definition of subreactions, such as searching for the early salt-modulated factor, the components needed for Sec17p release, the agents of docking, or the functional receptors for vacuole association of LMA1, Sec17p, Sec18p, and Ypt7p. Antibodies from the burgeoning yeast community may provide a second source of assays, just as they have allowed us to establish the roles of several components. A third source of assays is the growing collection of vac mutants (Wang et al., 1996) which may allow in vitro complementation. Continued fractionation of the cytosol will also yield components, from the HMA and LMA2 activities (Slusarewicz, P., Z. Xu, and A. Haas, unpublished) to kinases and phosphatases which may regulate inheritance components (Conradt et al., 1992, 1994). Finally, with as many individual components as possible in hand, the vacuole membrane will have to be solubilized and reconstituted into proteoliposomes which are competent for the individual subreactions.

We thank Dr. Dieter H. Wolf for generously providing yeast strains, Drs. Albert Haas, Pawel Slusarewicz, Robert Matts, and Charles Barlowe for discussions, and Marilyn Leonard and Barbara Atherton for expert assistance.

This work was supported by a grant from NIGMS.

NSF

N-ethylmaleimide–sensitive fusion protein

SNAP

soluble NSF attachment protein