Myo4p and She3p are required for cortical ER inheritance in Saccharomyces cerevisiae

During cell division, organelles are duplicated or fragmented and then partitioned into daughter cells. The accurate segregation of organelles from mother to daughter cells during the cell cycle can occur either randomly or by an ordered process that requires cellular machinery. A growing body of data suggests that the latter is the more common mechanism (Warren and Wickner, 1996; Yaffe, 1999; Catlett and Weisman, 2000). Although the inheritance of some organelles, such as the mammalian Golgi, the yeast vacuole, and mitochondria, has been investigated in some detail, less is known about the inheritance of the ER.

In the budding yeast Saccharomyces cerevisiae the ER has been divided into two classes: perinuclear ER and cortical or peripheral ER (Rose et al., 1989; Preuss et al., 1991). The perinuclear ER surrounds the nucleus, while the cortical ER forms a highly dynamic network of interconnected tubules that line the cell periphery. Like peripheral ER in mammalian cells, cortical ER undergoes ring closure and tubule branching movements (Prinz et al., 2000). During cell division, the first ER elements observed in the growing bud are cytoplasmic ER tubules oriented along the mother–bud axis (Du et al., 2001). These tubules either associate with the prebud site and are passively pulled into the growing bud (Fehrenbacher et al., 2002) or migrate into the daughter cell very soon after bud emergence (Du et al., 2001). The appearance of tubules aligned along the mother–bud axis suggests that a polarized structure is used for their orientation. The actin cytoskeleton is a likely candidate for this structure, as actin filaments are required for the inheritance of other organelles and for ER dynamics and motility (Catlett et al., 2000; Prinz et al., 2000; Fehrenbacher et al., 2002; Wöllert et al., 2002).

Type V myosins are actin-associated motors required for the polarized distribution of organelles, proteins, and mRNAs. These highly conserved myosins contain a number of structural elements that include an NH₂-terminal motor domain that binds to actin filaments and a COOH-terminal tail domain that binds cargo (Cheney et al., 1993; Tabb et al., 1998). In the squid giant axon, ER vesicles are transported on actin filaments by myosin V (Tabb et al., 1998). Loss of myosin Va, encoded by the mouse dilute locus, leads to neurological disorders that may result from defects in the transport of smooth ER from the dendritic shaft to the dendritic spine. The lightened coat color of dilute mice reflects a defect in melanosome transport (Wu et al., 1997). In plants, the actin network has been implicated in the maintenance of ER organization (Foissner et al., 1996; Boevink et al., 1998), and in Characea algae, the binding and sliding of ER membranes occurs along actin filaments (Kachar and Reese, 1988).

In S. cerevisiae, there are two type V myosins, Myo2p and Myo4p. Myo4p is a nonessential myosin required for the asymmetric distribution of ASH1 and IST2 mRNAs (Long et al., 1997; Takizawa et al., 2000). ASH1 mRNA binds to She2p, and the resulting RNA protein complex then binds to the adaptor protein She3p. She3p associates with Myo4p, and the fully assembled complex moves along actin cables to the bud tip (Bertrand et al., 1998; Böhl et al., 2000; Long et al., 2000). Ash1p then acts to inhibit mating type switching in daughter cells by repressing HO endonuclease expression (Bobola et al., 1996; Sil and Herskowitz, 1996). A similar mechanism of transport involving the She proteins is thought to regulate the asymmetric distribution of IST2 mRNA (Takizawa and Vale, 2000). Myo2p orients the mitotic spindle during the cell cycle (Yin et al., 2000). It is also required for polarized secretion (Govindan et al., 1995; Pruyne et al., 1998), and it is essential for the delivery of late Golgi elements (Rossanese et al., 2001), the vacuole (Catlett and Weisman, 1998), and mitochondria (Itoh et al., 2002) into daughter cells. Although it is clear that the transport of these organelles into daughter cells is dependent on an actin track (Simon et al., 1997; Hill et al., 1996; Rossanese et al., 2001) and a Myo2p motor, the role of actin and myosin motors in the inheritance of ER has been elusive. Here we show that actin, Myo4p, and She3p are required for the inheritance of cortical ER.

To identify new genes whose products play a role in the inheritance of cortical ER, we initiated a screen of the yeast deletion library. The Hmg1p-GFP fusion protein (NH₂-terminal transmembrane domain of HMG-CoA reductase isozyme 1 fused to GFP), used as an ER marker in our studies, was described and characterized in an earlier report (Du et al., 2001). The gene encoding this fusion protein was cloned into a CEN (LEU2) plasmid and then transformed into the deletion library with the aide of a 96-well microdispenser. After screening 768 deletion mutants, a strain containing a deletion of YAL029C was found to be defective in cortical ER inheritance. PCR analysis confirmed that this strain lacked the myo4 gene. The remainder of this report focuses on the phenotypic characterization of cells that lack myo4 and the role of Myo4p in ER inheritance.

To verify that the ER inheritance defect was due to the loss of myo4 and to confirm the phenotype in our strain background, we disrupted the myo4 gene in a wild-type strain (SFNY1054). In myo4Δ cells, a dramatic defect in the delivery of cortical ER tubules from mother to daughter cells was observed (Fig. 1, compare B with A). Approximately 73% of the myo4Δ small-budded cells (bud diameter between 0.3 and 0.5 of the mother) displayed a defect in cortical ER inheritance, while 99% of the wild-type small-budded cells acquired cortical ER (Table I). A similar percentage (76%) of mutant cells at an intermediate stage of bud growth (bud diameter >0.5 of the mother, but before nuclear segregation) also failed to contain cortical ER, while 88% of wild-type buds of the same size contained ER tubules that were uniformly distributed to the cell periphery. In cells at a late stage of bud growth (after nuclear segregation but before nuclear division), 73% of the myo4Δ cells displayed a defect in cortical ER inheritance (Table I). Thus, at all stages of the cell cycle leading up to nuclear division, there was a dramatic reduction in the delivery of cortical ER into the bud.

To determine if the motor activity of Myo4p plays a role in cortical ER inheritance, we constructed a point mutation in the ATP-binding region of the motor domain of Myo4p (myo4-1) that converted the glycine at amino acid 171 to a glutamate. As shown in Fig. 1 C, the myo4-1 mutant displayed a defect in ER inheritance. This finding implies that the motor activity of Myo4p is required to deliver ER into daughter cells.

To confirm that the apparent defect in transport of cortical ER into daughter cells was not specific to Hmg1p-GFP, the localization of a second ER marker was examined. For these studies, a construct containing an NH₂-terminal signal sequence and a COOH-terminal ER retrieval HDEL signal, fused to YFP, was transformed into the myo4Δ mutant. In wild type, the localization pattern of YFP-HDEL was similar to that of Hmg1p-GFP (Fig. 1, compare D with A). Quantitation of the YFP-HDEL localization pattern revealed that >95% of buds in the different stages of growth contained cortical ER (Table I). In the myo4Δ mutant expressing YFP-HDEL, most of the buds exhibited defects in cortical ER inheritance. Approximately 75% of the small buds, 67% of the intermediate buds (before nuclear segregation), and 66% of the large buds (after nuclear segregation, but before nuclear division) failed to acquire cortical ER (Fig. 1 E and Table I). This was similar to the data obtained using Hmg1p-GFP as an ER marker protein (Table I, compare row 4 with row 2). Thus, using two different ER marker proteins, we have confirmed that cells lacking Myo4p motor activity display a dramatic, but still incomplete, defect in cortical ER inheritance.

To determine if Myo4p is specifically required for the inheritance of cortical ER, we examined the inheritance of other organelles in the myo4Δ mutant. Vacuolar inheritance was examined in wild-type and myo4Δ cells expressing Hmg1p-GFP using the lipophilic vital dye FM4-64, which stains the vacuolar membrane (Vida and Emr, 1995). In the myo4Δ mutant, 137 buds that contained little to no cortical ER were examined for vacuolar inheritance. A similar number (123) of wild-type buds were also analyzed. In both wild-type and myo4Δ cells, all buds inherited a labeled parental vacuole (Fig. 2 A). Thus, the inheritance of vacuoles is independent of Myo4p.

The inheritance of mitochondria and Golgi elements was also examined in cells that lack Myo4p. Mitochondrial inheritance was studied by expressing a plasmid (pSFNB784), containing a mitochondrial targeting sequence fused to RFP (Mozdy et al., 2000), in wild-type and myo4Δ cells that contain Hmg1p-GFP. In myo4Δ cells, mitochondrial tubules were transported into buds that failed to inherit cortical ER (Fig. 2 B). Mitochondrial tubules were observed in 92 of the 93 myo4Δ buds and in 97 of the 97 wild-type buds examined. Thus, the distribution of mitochondria in wild-type and myo4Δ buds was essentially indistinguishable.

Sec7p is a peripheral membrane protein that is specifically localized to late Golgi membranes (Franzusoff et al., 1991; Rossanese et al., 2001). To examine the inheritance of late Golgi, the SEC7 gene was replaced with a SEC7-GFP fusion in wild-type and myo4Δ mutant cells. The distribution of Sec7p-GFP was found to be comparable in wild type and the myo4Δ mutant (Fig. 2 D). In the myo4Δ cells, all mutant buds (122) contained late Golgi elements. The same result was observed in the 97 wild-type cells examined. Furthermore, mutant and wild-type buds contained approximately the same number of Sec7p-containing structures. These results are consistent with prior studies showing that the polarized transport of vacuoles, mitochondria, and the late Golgi requires the Myo2p motor (Hill et al., 1996; Rossanese et al., 2001; Itoh et al., 2002).

While the inheritance of the late Golgi is thought to be an actin-dependent process that requires Myo2p (Rossanese et al., 2001), less is known about the inheritance of early Golgi elements. It has been proposed that the early Golgi is derived from the ER (Rossanese et al., 1999; Bevis et al., 2002). Like the cortical ER, early Golgi membranes enter daughter cells at an early stage of bud growth (Preuss et al., 1992; Du et al., 2001; Rossanese et al., 2001). The inheritance of early Golgi membranes was examined by comparing the distribution of Och1p in myo4Δ and wild-type cells. Och1p is a carbohydrate-modifying enzyme that defines one of the earliest Golgi compartments in yeast (Nakayama et al., 1992). A plasmid expressing HA-tagged Och1p was transformed into wild-type and myo4Δ mutant cells, and the distribution of early Golgi elements was followed by indirect fluorescence using an anti-HA monoclonal antibody. The localization of early Golgi membranes in 107 myo4Δ mutant cells that displayed a defect in ER inheritance was indistinguishable from that of the 103 wild-type cells examined (Fig. 2 C). Thus, Myo4p is specifically required for the inheritance of cortical ER and not the inheritance of the vacuole, mitochondria, and early or late Golgi. These findings imply that the two type V myosins in yeast, Myo4p and Myo2p, have distinct and nonoverlapping functions in organelle inheritance.

Cytoplasmic ER tubules are readily observed in wild-type cells expressing Hmg1p-GFP. These tubules emanate from the perinuclear region, and some extend toward the bud (Fig. 3 A and Fig. 4 A). Time-lapse imaging revealed that a subset of these tubules appear to grow into the bud, suggesting that they give rise to cortical ER in the daughter cell. We found that tubule growth required polymerized actin, as cells treated with the actin-depolymerizing drug latrunculin A (Lat-A) failed to extend tubules into the bud (Fig. 3 B).

We estimated, from 10 movies of ER tubule tip movement (see Videos 1–10), that the average length of movement is 1.6 μm. The average rate is 0.013 μm/s with a standard deviation of 0.0015 μm/s. The tight clustering of the rates, as well as the vectorial nature of the extension, is consistent with a motor driven process. The rate of ER tubule tip movement is, however, substantially slower than the maximal rate observed for the Myo4p-driven movement of actin fibers in vitro (Reck-Peterson et al., 2001). These findings suggest that Myo4p may play a role in forming and extending ER tubules, and that the velocity of tubule extension is limited by tubule growth and not the motor activity of Myo4p. Interestingly, the rate of tubule extension that we have observed is similar to the elongation rate of ER tubules in animal cells (Waterman-Storer and Salmon, 1998). Because tubules are seen within the bud soon after bud emergence and are initially quite faint, we could not exclude the possibility that tubules below our limit of detection associate with the prebud site and are passively pulled into the bud as it grows, as suggested by prior studies (Fehrenbacher et al., 2002).

To determine if actin and Myo4p play a role in the formation or stability of cytoplasmic ER tubules, we examined cells expressing Hmg1p-GFP and quantified the number of ER tubules. We found that ∼90% of wild-type cells contained at least one ER tubule, with 30–40% containing more than one tubule (Fig. 4 B). The number of tubules formed was similar in small- and large-budded cells. In contrast, treatment with Lat-A reduced the percentage of cells with at least one tubule to ∼52% of small-budded cells and ∼23% of large-budded cells, demonstrating the importance of actin filaments in the formation of ER tubules (Fig. 4, A and B). In myo4Δ cells, the percentage of cells with at least one tubule was ∼40% in both small- and large-budded cells (Fig. 4 C). Furthermore, the percentage of myo4Δ cells with more than one tubule was <10%. An even more dramatic result was seen when we restricted our analysis to those tubules oriented along the mother–bud axis in small-budded cells. While 90% of small-budded wild-type cells contained a cytoplasmic tubule oriented along the mother–bud axis, only 9% of small-budded myo4Δ cells contained such a structure (for examples see Videos 11–20). Thus, the formation, as well as orientation, of cytoplasmic ER tubules is dependent on actin and Myo4p. These findings are consistent with previous studies that implicated the actin cytoskeleton in the maintenance of ER dynamics (Prinz et al., 2000; Fehrenbacher et al., 2002).

The binding of Myo4p to ASH1 mRNA requires both She2p and She3p (Böhl et al., 2000; Takizawa et al., 2000). To determine if She2p and She3p play a role in the inheritance of cortical ER, we examined the delivery of Hmg1p-GFP into daughter cells in she2Δ and she3Δ strains and compared their phenotype to wild type and the myo4Δ mutant (Fig. 5, A and B). A dramatic delay in the delivery of cortical ER from mother to daughter cells was observed in she3Δ (Fig. 5 D), but not she2Δ (Fig. 5 C and Table II), mutant cells. In small buds with a diameter between 0.3 and 0.5 of the mother cell, ∼88% of the she3Δ daughter cells displayed a defect in cortical ER inheritance (Table II). In wild-type buds of the same size, ∼94% of the cells acquired cortical ER. In large-budded she3Δ cells, without nuclei, ∼72% of the buds were defective for cortical ER inheritance. In larger buds in which the nucleus had not yet divided, a significant portion of the she3Δ buds (∼38%) also showed a defect in ER inheritance (Table II). Thus, the she3Δ and myo4Δ mutants have a similar phenotype. These findings indicate that She3p, but not She2p, is required for the inheritance of cortical ER. The defect in ER inheritance observed in the myo4Δ and she3Δ mutants is not a secondary consequence of disrupting the actin network, as the loss of Myo4p and She3p does not alter actin assembly (unpublished data).

One prediction of the finding that Myo4p and She3p are required for ER inheritance is that both of these proteins should associate with the ER. Subcellular fractionation was performed to localize Myo4p and its adaptor protein She3p on a sucrose gradient that resolves the ER from other intracellular membranes (Antebi and Fink, 1992; Barrowman et al., 2000). Strains expressing either She3p-myc (SFNY1264) or Myo4p-myc (SFNY1221) were used to detect She3p and Myo4p. The presence of a 13myc tag on either protein did not affect ER inheritance or ASHI RNA localization (unpublished data; Takizawa and Vale, 2000). Lysates were prepared from SFNY1264 and SFNY1221, and the distribution of She3p and Myo4p, relative to the ER marker proteins Sec61p and Dpm1p (Wilkinson et al., 1996; Preuss et al., 1991), was determined by Western blot analysis. In several gradients, a substantial amount of the She3p (26–46%) cofractionated with the ER (Fig. 6 A), while a smaller portion (∼5–10%) of the Myo4p was found in ER fractions (Fig. 6 B). The amount of She3p and Myo4p associated with the ER may vary, as peripheral membrane proteins tend to shear off membranes during lysis. The localization of She3p and Myo4p was unchanged when lysates prepared from she2Δ strains expressing either She3p-myc or Myo4p-myc were fractionated (Fig. 7, A and B). The same results were also obtained when lysates expressing She3p-myc (Fig. 7 C) or Myo4p-myc (not depicted) were treated with RNase. Thus, the interaction of the Myo4p–She3p motor complex with the ER is not mediated by an associated RNA.

As the ER cofractionates with the plasma membrane on this gradient, we stripped the ER of ribosomes by fractionating extracts on a sucrose density gradient in the presence of EDTA, as described before (Roberg et al., 1997). As shown in Fig. 6 C, the ER fractionated at the top of the gradient and was well resolved from the more dense plasma membrane. Furthermore, She3p cofractionated with the ER and not the plasma membrane. The same result was obtained when Myo4p was localized on this gradient (Fig. 6 C). As the adaptor protein She3p and the class V myosin Myo4p form a complex (Takizawa and Vale, 2000) and both proteins cofractionate with the ER, these findings imply that the Myo4p–She3p motor complex is directly associated with the ER.

The finding that the Myo4p–She3p complex is required for ER inheritance raises the possibility that the asymmetric distribution of mRNA is dependent on cortical ER inheritance. To address this question, the localization of IST2 mRNA (Takizawa et al., 2000), which localizes to the bud tip throughout the cell cycle, was examined in the aux1 mutant, which is defective in ER inheritance (Du et al., 2001). AUX1 encodes a hydrophilic protein of 668 amino acids. The COOH terminus of Aux1p contains a J-domain that is homologous to the J-domain of the clathrin uncoating protein auxilin (Holstein et al., 1996). Aux1p may be a bifunctional protein, as the loss of the J-domain leads to vacuole fragmentation and membrane accumulation but does not affect the segregation of ER into daughter cells (Du et al., 2001). We found that IST2 mRNA localized to the bud tip in wild-type cells and in an aux1Δ mutant, lacking the entire AUX1 gene (Fig. 8 A). Similar results were obtained when the localization of ASH1 mRNA was examined (not depicted). Thus, cortical ER inheritance is not required for the localization of mRNA to the bud tip.

To determine if cortical ER inheritance and mRNA transport are independent processes, we constructed a mutant in she3 that disrupts the She2p binding site. Previous studies (Böhl et al., 2000) have shown that She3p binds to She2p via the COOH terminus of She3p (amino acids 213–426). To disrupt the interaction of She3p with She2p, we deleted the last 176 amino acids of She3p (see diagram in Fig. 8 B). As anticipated, IST2 mRNA failed to localize to the bud tip in this mutant (Fig. 8 C). In contrast, truncation of the COOH terminus of She3p had no effect on ER inheritance (Fig. 8 C). These findings demonstrate that different domains of She3p are required for mRNA localization and ER inheritance.

There are two type V myosins in yeast, Myo2p and Myo4p. Myo2p binds to a variety of cargo. It has been implicated in the polarized delivery of secretory vesicles (Govindan et al., 1995; Pruyne et al., 1998) and in the inheritance of several organelles (Catlett and Weisman, 1998; Rossanese et al., 2001; Itoh et al., 2002). Myo2p also plays a role in orienting the spindle during the cell cycle (Yin et al., 2000). However, until this study, the only known role for Myo4p was in the localization of ASH1 and IST2 mRNAs to the bud tip.

Proper localization of ASH1 and IST2 mRNA requires the SHE genes (Bobola et al., 1996; Jansen et al., 1996; Takizawa et al., 2000). ASH1 and IST2 bind to She2p, while She3p links Myo4p to the She2p–mRNA complex (Böhl et al., 2000; Takizawa and Vale, 2000). Movement of the She2p–mRNA ribonucleoprotein complex along actin cables is myosin dependent, and the deletion of myo4, she2, or she3 prevents these mRNAs from localizing to the bud tip (Jansen et al., 1996; Long et al., 1997).

Cortical ER enters daughter cells in a polarized fashion. It is enriched at the presumptive bud site and accumulates at the bud tip in small- and medium-sized buds (Fehrenbacher et al., 2002). Myo4p, which is required for this process, localizes to the cortex and bud tip (Jansen et al., 1996; Münchow et al., 1999) and is enriched in the bud by a retention mechanism that involves the RNA binding protein She2p (Kruse et al., 2002). Here we have identified a role for Myo4p in cortical ER inheritance. We show that the role of Myo4p in organelle inheritance is distinct from that of Myo2p. While Myo4p is required for ER inheritance, it is not required for the inheritance of vacuoles, mitochondria, and late Golgi elements, events known to be dependent on Myo2p (Hill et al., 1996; Rossanese et al., 2001; Itoh et al., 2002). Our findings indicate that the deletion of myo4 and she3, but not she2, results in a partial block in the delivery of cortical ER from mother to daughter cells. Furthermore, deletion of myo4 or inhibition of actin polymerization with the drug Lat-A leads to a significant reduction in the number of ER tubules in the cytoplasm. Tubules oriented along the mother–bud axis in small-budded cells are most dramatically affected. Cytoplasmic ER tubules also fail to enter daughter cells subsequent to Lat-A treatment. Myo4p and She3p may be needed for the growth and orientation of cytoplasmic ER tubules into daughter cells.

Although the deletion of either myo4 or she3 leads to a dramatic loss of cortical ER in daughter cells, ∼25% of the buds still receive ER elements. Treatment of myo4Δ cells with nocodazole did not decrease the amount of cortical ER found in the bud (unpublished data), indicating that the ER that enters the bud in myo4Δ cells is not microtubule dependent. These findings indicate that a Myo4p/She3p-independent mechanism can function to transmit ER into daughter cells, albeit inefficiently. Additional studies will be needed to define the nature of this secondary pathway to ER inheritance.

In neurons, the ER is transported on actin by myosin V (Tabb et al., 1998), and myosin V drives ER network formation in Xenopus egg extracts (Wöllert et al., 2002). Thus the growth, formation, and movement of ER by myosin V along actin is a highly conserved process. Our finding that both Myo4p and She3p physically associate with ER membranes implies that both of these proteins play a direct role in the inheritance of ER. Interestingly, a larger fraction of the She3p cofractionated with the ER. This suggests that She3p may link Myo4p to ER tubules. The finding that she3Δ cells display a defect in ER inheritance is consistent with this proposal. Although the interaction between Myo4p and ASH1 mRNA is dependent on She2p (Münchow et al., 1999), the association of She3p and Myo4p with ER membranes is not. Consistent with this result, we find that the loss of She2p does not lead to a defect in cortical ER inheritance, and that cortical ER inheritance is independent of mRNA localization. Thus, the Myo4p–She3p complex must bind to the ER via a factor other than She2p. We are currently screening the yeast deletion library to identify this unknown factor as well as additional components of the ER inheritance machinery.

Plasmid SFNB1030 (CEN URA3) encodes the HDEL signal sequence from KAR2 fused to the COOH terminus of YFP. To generate this plasmid, a 0.864-kb fragment that encodes HDEL-YFP was amplified from pDN300 (a gift from David Ng, Pennsylvania State University, University Park, PA) with primers to introduce SalI and SpeI restriction sites. This PCR product was then cloned into pTEF416 (CEN URA3) between the SalI and SpeI sites.

SFNY1196 and SFNY1270 (Table III) were constructed by transforming NY1211 and SFNY1235, respectively, with plasmid SFNB1030 (Table III) and selecting for Leu⁺ transformants. Strains SFNY1224 and SFNY1225 (Table III) were constructed by transforming SFNY1235 with plasmids SFNB784 and SFNB637 (Table III), respectively, and selecting for Leu⁺ transformants.

SFNY1261, SFNY1262, SFNY1263, and SFNY1267 (Table III) were constructed by transforming BY4741, YKL130C, YAL029C, and YBR130C from the ResGen deletion library, respectively, with plasmid SFNB1000 and selecting for Leu⁺ transformants. All deletions were confirmed by PCR analysis.

Full-length SHE3, in vector pRS316, was introduced into SFNY1267 to generate SFNY1303 (Table III). A she3 truncation mutant (lacking 176 amino acids from the COOH terminus), in vector pRS316, was introduced into SFNY1267 to generate SFNY1310 (Table III), and Ura⁺ transformants were selected. Plasmid PT201 (CEN LEU2), carrying a point mutation in the ATP-binding region of the motor domain of Myo4p (G171E), was transformed into SFNY1218 (Table III) to generate SFNY1220. The myo4-1 mutant allele was obtained from Susan Brown when she was at the University of Michigan (Ann Arbor, MI).

To construct SFNY1305 (Table III), YIplac204/DsRed.T1-HDEL (a gift from Ben Glick, University of Chicago, Chicago, IL) was linearized with EcoRV and transformed into PT68, and Trp⁺ transformants were selected.

The yeast MATa haploid deletion library from Research Genetics (ResGen-Invitrogen Corporation) was screened to identify genes that display a defect in the inheritance of cortical ER. The library contains 4,848 ORFs that were disrupted with the KanMX module in the parent strain BY4741. Cells were transformed with plasmid SFNB1000 (CEN, LEU) using a Hydra 96 microdispenser (Robbins Scientific Corporation). Stationery cultures were generated in a 96-well assay block (Corning Inc.) by inoculating YPD (580 μl per well) with cells (10 μl) from a ResGen master plate. Glass beads (3.5 mm; Fisher Scientific) were placed in each well to facilitate airation of the cultures. The block was incubated for 3 d at 25°C.

Stationery cultures were diluted (1:1,000) in YPD and grown overnight at 25°C to an OD₆₀₀ of 1.0–1.5. The cells were transferred (390 μl per well) to a new assay block without beads and centrifuged at 2,000 rpm for 5 min. Pellets were washed with sterile water (580 μl per well), centrifuged, resuspended in LTE (100 mM LiOAc, 10 mM Tris pH to 8.0, 1 mM EDTA; 580 μl per well), and then incubated at room temperature for 1 h. The robot was sterilized after each step in the transformation protocol by washing the teflon needles and glass wells three times in 70% ethanol followed by three washes in sterile water. The cells were then centrifuged and the pellets were resuspended in 35% PEG/LTE (405 μl per well). Transformation was performed by adding 15 μl of a 2 mg/ml stock of carrier DNA (salmon testes from Sigma-Aldrich) and 15 μl of DNA (50 ng/μl stock) from plasmid SFNB1000 (HMG1-GFP CEN LEU2) into each well followed by a 1-h incubation at 30°C. The cells were heat shocked for 10 min in a 42°C water bath and centrifuged. The pellets were washed in sterile water (580 μl per well), centrifuged, resuspended in synthetic complete (SC) minus Leu media (580 μl per well), and transferred to an assay block containing beads. Transformed cells were grown for 3 d at 25°C and then replica plated onto a synthetic dropout minus Leu (QBiogen) agar plate and grown for 3 d to confirm transformation. Stationary cultures were grown from colonies on the agar plate as described above. The stationery cultures were then diluted and grown overnight at 25°C in SC minus Leu to an OD₆₀₀ of ∼0.3–0.5. A total of 768 deletion strains out of the 4,848 in the library (16%) were screened.

SFNY1218 and SFNY1235 were generated using a PCR-based gene deletion method (Longtine et al., 1998). The forward primer, MYO4-F1 (5′-AACACAAAAAAACAAAAAAAATCCTATAACCAGTTCTCCCGCGGTCGA-CGGATCCCCGGGTT-3′) contained 40 nucleotides of MYO4 gene-specific sequence located upstream of the start codon followed by 20 nucleotides of gene-specific sequence from plasmid pFA6a-His3MX6, which contains the his5⁺ gene from Schizosaccharomyces pombe (Longtine et al., 1998). The reverse primer, MYO4-R1 (5′-TATATACATATATACATATATGGGCGTATATTTACTTTGTTCGAATTCGAGCTCGTTTAAAC-3′) contained 40 nucleotides of MYO4 gene-specific sequence located downstream of the stop codon followed by 20 nucleotides of gene-specific sequence from plasmid pFA6a-His3MX6. The fragment was amplified, and the PCR product was transformed into SFNY1054 and NY1211 (Table III). His⁺ transformants were selected, and replacement of the MYO4 gene with his5⁺ was confirmed by PCR. SFNY1226 and SFNY1227 (Table III) were generated by the same method with the following primers. To disrupt SHE2, SHE2-F1 (5′-CCCTCCTTAATTTTCCTTTTGCATAATACCAGACACTTAAAAGGTCGACGGATCCCCGGGTT-3′) was used as a forward primer and SHE2-R1 (5′-AGTGGTACTTATTTGCTCTTTTTGAGCTAAAAACTGAAGGCCGAATTCGAGCTCGTTTAAAC-3′) as a reverse primer. To disrupt SHE3, we used SHE3-F1 as a forward primer (5′-TATCAAGCACGCCAAGGTTCAACGACACTACTTTTGTGTAAGGGTCGACGGATCCCCGGGTT-3′) and SHE3-R1 as a reverse primer (5′-CTAAATGAATCCTATATATATACTCCCTTGTGTCGGCATATTGAATTCGAGCTCGTTTAAAC-3′).

To visualize the ER or mitochondria, cells expressing DsRed, GFP, YFP, or RFP fusion proteins were grown overnight in minimal or SC media at 30°C to an OD₆₀₀ of 0.3, pelleted, and resuspended in 30 μl of growth medium. Then 3 μl of the cell suspension was mixed with an equal volume of growth medium containing 0.6% NuSieve GTG low melting temperature agarose (FMC BioProducts).

Indirect immunofluorescence was performed according to Du et al. (2001) with the following modifications. To remove the cell wall, cells were incubated for 9 min at 30°C with gentle shaking in 1 ml of 100 mM sodium phosphate/1.2 M sorbitol buffer (pH 6.6) containing 143 mM β-mercaptoethanol and 50 μg/ml Zymolyase 100T (Seikagaku America). Spheroplasts were then washed three times with 5 ml of 100 mM sodium phosphate/1.2 M sorbitol buffer (pH 6.6) and resuspended in 250 μl of the same buffer. Subsequently, 30 μl of cells were attached to eight-well slides coated with poly-l-lysine and permeabilized in 30 μl of P buffer (1× PBS containing 0.1% SDS and 10 mg/ml BSA) for 5 min.

To visualize vacuoles in wild-type and myo4Δ mutant cells, vacuole membranes were labeled with 80 μM FM 4-64 (Molecular Probes) according to methods previously described (Wang et al., 1996; Du et al., 2001).

Cells were observed with a Carl Zeiss MicroImaging, Inc. Axiophot fluorescence microscope using a 100× oil-immersion objective. Images were captured with an Orca ER digital camera (model no. C4742–95; Hamamatsu Photonics) and the OpenLab 3.08 imaging software (OpenLab Inc.). Images were further processed with Adobe Photoshop 7.0 and Illustrator 10.0.

SFNY1061 (Table III) was grown in 5 ml of YPD overnight at 30°C to an OD₆₀₀ of 0.15. Cells were harvested and resuspended in 200 μl of YPD, and 25 μl was subsequently mounted in an equal volume of medium containing 0.6% low melting temperature agarose and 200 μM Lat-A (Molecular Probes). Cells were observed with a Carl Zeiss MicroImaging, Inc. fluorescence microscope immediately after Lat-A treatment.

Time-lapse fluorescence microscopy of SFNY1061 was performed on cells grown at 30°C in YPD to an OD₆₀₀ of 0.15. Cells (untreated or treated with 200 μM Lat-A) were mounted in 0.4% agarose and viewed with a Carl Zeiss MicroImaging, Inc. fluorescence microscope. Images were captured using OpenLab software at 18-s intervals for 3 min and exported as a Quicktime file. Image stills were generated using Adobe Photoshop 7.0 and Illustrator 10.0.

The distance traveled was determined by measuring the tubule movement from the bud neck to the bud tip in time-lapse image stills. The ruler tool in the measurement module of OpenLab 3.1.3 was positioned on the tubule at the bud neck and was extended over the length of the tubule as it extended toward the bud tip. The distance traveled by the tubule was then recorded in micrometers. To determine the rate of movement, the distance traveled was divided by the time a tubule approached the bud tip. The time was recorded and displayed in each image still. Rate measurements are reported as micrometers per second.

Cells were grown, converted to spheroplasts, and lysed in the absence or presence of 100 μg/ml RNase A as described previously (Barrowman et al., 2000). 66 OD₆₀₀ units were then fractionated on sucrose velocity gradients according to Barrowman et al. (2000). Fractions (∼1 ml each) were collected from the top of the gradient, and the protein in each fraction was precipitated by adding 500 μl of cold 50% trichloroacetic acid. The protein pellets were collected by centrifugation at 14,000 rpm for 5 min at 4°C, resuspended in 66 μl of 1 M Tris, and solubilized in 99 μl of SDS sample buffer. The fractions (30 μl each) were then resolved by SDS-PAGE and subjected to Western blot analysis. The antibodies were subsequently detected by ECL (Amersham Biosciences), and the bands were quantitated using BioImage software. The data are plotted as a percent of the maximum antigenic material.

The ER and plasma membrane were fractionated on a linear sucrose density gradient in the presence of EDTA as described previously (Roberg et al., 1997). In brief, cells were grown to an OD₆₀₀ of 1.0, harvested, and resuspended in 0.5 ml of STE10 buffer (10% [wt/wt] sucrose, 10 mM EDTA, 10 mM Tris-HCl, pH 7.5) containing a protease inhibitor cocktail (10 μM antipain, 30 μM leupeptin, 30 μM chymostatin, 1 μM pepstatin A, 1 μM PMSF, and 1 μg/ml aprotinin). Cells were then lysed by vortexing with glass beads. After lysis, 1 ml of STE10 buffer was added, and the lysate was cleared of unbroken cells by centrifugation at 500 g. 15 OD₆₀₀ units (∼300 μl of the lysate) were layered on top of a 20–60% linear sucrose gradient (5 ml) prepared in TE (10 mM Tris-HCl, pH 7.5, 1 mM EDTA). Samples were then centrifuged at 100,000 g for 18 h at 4°C using an SW50.1 rotor (Beckman Coulter) as described by Roberg et al. (1997). Fractions (∼440 μl each) were collected from the top of the gradient, and the protein in each fraction was precipitated with 50% TCA and processed as described above. The data are plotted as a percent of the maximum antigenic material.

Localization of IST2 mRNA by in situ RNA hybridization using an IST2 antisense probe labeled with digoxigenin was performed as described previously by Takizawa et al. (1997)(2000).

Videos 1–10 show time-lapse movies, at one frame per 6 s, of ER segregation tubules aligning along the mother–bud axis and entering the bud in wild-type cells (SFNY1061) expressing the ER marker Hmg1p-GFP. Videos 11–20 depict time-lapse movies at one frame per 6 s of myo4Δ mutant cells (SFNY1218) expressing the Hmg1p-GFP fusion protein. All supplemental material is available.

We thank Flora Rafi for assistance in screening the yeast deletion library, Yunrui Du for technical advice and comments on the manuscript, Yueyi Zhang for technical assistance, and Ben Glick and David Ng for providing plasmids.

This work was supported by a program project grant from the National Cancer Institute.