Tropomyosin-containing Actin Cables Direct the Myo2p-dependent Polarized Delivery of Secretory Vesicles in Budding Yeast

Standard rich and synthetic media used for growing yeast are described by Sherman (1991). For temperature control experiments, 1-ml culture volumes were transferred to prewarmed or precooled glass tubes for various times, then fixative was added directly to the medium. Yeast transformations were performed using the Frozen-EZ Yeast Transformation Kit (ZYMO Research, Orange, CA) or using a lithium acetate protocol (Ito et al., 1983). Escherichia coli strains DH5α and DH10B were used for all bacterial manipulations. Restriction enzymes and T4 DNA ligase (Life Technologies, Inc., Gaithersburg, MD) were used following standard protocols as was Taq DNA polymerase (Boehringer Mannheim Corp., India-napolis, IN), except where described below. Except where noted, primers for PCR were from Life Technologies, Inc.

Temperature-sensitive alleles of TPM1 were generated by mutagenic PCR. The wild-type TPM1 gene from pRS314 (Sikorski and Hieter, 1989) was amplified (primers: GGGGTCGATGTATAGTCTAAG and GGG-GTCGACACATATATCTTACCCG; Cornell Biotechnology Analytical/Synthetic Facility) using conditions favorable for misincorporations, specifically: 50 mM KCl, 10 mM Tris-HCl, 0.5 mM MnCl₂, 7 mM MgCl₂, 0.2 mM each dGTP and dATP, and 1 mM each dCTP and dTTP (Cadwell and Joyce, 1992). Mutant alleles were transformed into ABY335 (tpm1Δ tpm2Δ pRS316[TPM1]) by cotransformation of the PCR products with a gapped plasmid (Muhlrad et al., 1992), consisting of pRS314[TPM1] linearized with Bpu1102I and StyI. Trp⁺ transformants were yeast in which the gapped plasmid (deleted for nucleotides +21 to +488 of TPM1) had been repaired by homologous recombination with either PCR-derived tpm1 or pRS316[TPM1], thus transformants carry pRS316[TPM1] and pRS314[tpm1]. When tested, none showed dominant cold or temperature sensitivity. All clones were then grown on medium containing 5-fluoroorotic acid (United States Biological, Swampscott, MA) to select for the loss of the wild-type pRS316[TPM1], leaving pRS314[tpm1] as the sole copy of TPM1. Surviving clones were screened for recessive temperature and cold sensitivity: of 594 Trp⁺ transformants, 215 were 5-fluoroorotic acid resistant, and two of those were recessively temperature sensitive (alleles tpm1-1 and tpm1-2) while none were cold sensitive. The tpm1-1 and tpm1-2 coding regions were recloned into ABY335 to verify that the conditional mutations were in the coding region. They were then sequenced (Cornell University Biotechnology Resource Center). Resultant strains were ABY932 (tpm1Δ tpm2Δ pRS314[tpm1-1]) and ABY933 (tpm1Δ tpm2Δ pRS314[tpm1-2]).

The genotypes of all yeast strains used in this study are described in Table I. Strains with genomic tpm1-2 tpm2Δ were generated in the following manner: TPM1 in a plasmid-borne 2.2-kb genomic insert (Liu and Bretscher, 1989b) was digested with Bpu1102I/StyI to replace bases +21 through +488 with that region of tpm1-2. LEU2 was amplified by PCR from pRS315 (Sikorski and Hieter, 1989) to introduce flanking NheI sites (oligonucleotide primers GGGCTAGCGTGGTAAGGCCGT and GGGCTAGCGGTCGAGGAGAAC), then cloned into an NheI site at position −236 relative to tpm1-2. The tpm1-2::LEU2 insert was released by digestion with Alw26I/BcgI and transformed into ABY946 (TPM1 tpm2Δ) to replace TPM1 with tpm1-2::LEU2. Integration of tpm1-2 into Leu⁺ transformants was verified by testing for temperature sensitivity and sequencing the tpm1 locus. The resultant strain is ABY944 (tpm1-2 tpm2Δ). Homozygous diploids ABY971 (tpm1-2/tpm1-2 tpm2Δ/tpm2Δ) and ABY973 (TPM1/TPM1 tpm2Δ/tpm2Δ) were generated through back-crossing ABY944 to ABY945, sporulating the heterozygous diploid, and remating appropriate spores. ABY950 (tpm1-1 tpm2Δ) was generated in the same way.

A hemagglutinin (HA)¹ epitope-tagged MYO5 allele, (MYO5:HA3) was introduced into the ABY971 and 973 backgrounds in the following manner: PCR amplification of the plasmid pCS124 (a gift from Caroline Shamu, Harvard Medical School, Boston, MA) using primers having homology to the 3′ end of MYO5 (TAGAGAGTGATGACGAGGAGGCTAACGAAGATGAAGAGGAAGATGATTGGGTATTCACC-ATGGCCTACCC and TACTCTATTTGCTCGTATAGAGTATATACTCGCTAAATACATTTTGATTATGGTGCACTCTCAGTACAAT) yielded linear DNA that upon transformation into yeast converts MYO5 into MYO5:HA3::TRP1. Trp⁺ transformants of ABY971 were verified for production of tagged Myo5p by Western analysis using anti-HA mono-clonal 12CA5 (Boehringer Mannheim Corp.). The resultant heterozygous diploid was sporulated and appropriate spores remated to generate homozygous MYO5:HA3/MYO5:HA3 in the ABY971 background (ABY990) or crossed to ABY945 and again sporulated and backcrossed to generate MYO5:HA3 homozygous in a TPM1 background (ABY989).

SEC8:HA3 allele was introduced in the same manner, using primers with homology to the 3′ end of SEC8 (TTGGAAAACTTAAAAGCAAATTGAATGCTGTCCATACTGCAAACGAAAAAGTATTC-ACCATGGCCTACCC and TTTTCATTCATTTATTTATCAAATTATTTTTACACAAACTAAAAATGTCATGGTGCACTCTCAGTAC-AAT), resulting in homozygous SEC8:HA3::TRP1 in the ABY973 and ABY971 backgrounds (ABY987 and ABY988, respectively).

A green fluorescent protein-tagged CAP2 allele (GFP:CAP2) was introduced using a modified version of plasmid pBJ646 (Waddle et al., 1996), kindly donated by J. Cooper (Washington University School of Medicine, St. Louis, MO). A PstI/EcoRV fragment from pBJ646 was cloned into either pRS304 or pRS306 (Sikorski and Hieter, 1989), yielding pDP122 and pDP124, respectively. Linearization of either plasmid with EcoRI yields DNA competent to replace the endogenous CAP2 allele with cap2 with the terminal 31 residues replaced with 52 residues derived from the plasmid multiple cloning site, followed by a stop, then either TRP1 or URA3, and, finally, GFP:CAP2 behind the CAP2 promoter. In short, transformants bear either cap2::TRP1::GFP:CAP2 or cap2::URA3:: GFP:CAP2. Both alleles were transformed into ABY971 and ABY973 to generate homozygous tpm1-2/tpm1-2 tpm2Δ/tpm2Δ GFP:CAP2/GFP: CAP2 (ABY992) and TPM1/TPM1 tpm2Δ/tpm2Δ GFP:CAP2/GFP: CAP2 (ABY991), respectively.

Triple mutant myo2-66 tpm1-2 tpm2Δ was generated in two steps. First, ABY946 (tpm2Δ::HIS3) was crossed to NY1396 (myo2-66), then sporulated to isolate myo2-66 tpm2Δ. This was then crossed to a clone bearing tpm1-2 tpm2Δ SEC8:HA3::TRP1 and the diploid sporulated to isolate myo2-66 tpm1-2 tpm2Δ SEC8:HA3 (ABY1100). Triple mutant sec6-4 tpm1-2 tpm2Δ was generated by the same procedure except using the initial strains ABY945 (tpm2Δ) and ABY703 (sec6-4) and as a final step, crossing two haploid sec6-4 tpm1-2 tpm2Δ SEC8:HA3 spores to generate the homozygous diploid (ABY999).

Tpm2p-specific antibodies were prepared from crude rabbit antiserum B43 that recognizes both Tpm1p and Tpm2p (Drees et al., 1995). Tpm2p was purified as described (Drees et al., 1995) and 2.8 mg coupled to CNBr-activated Sepharose 4B (Sigma Chemical Co., St. Louis, MO). Tpm1p was purified as described (Liu and Bretscher, 1989a) and 1.5 mg coupled to CNBr-coupled Sepharose 4B. Tpm2p-specific antibodies were generated by affinity purification of B43 against coupled Tpm2p followed by immunodepletion by coupled Tpm1p. Affinity pure Tpm1p-specific antibodies were prepared from either crude rabbit antiserum 138 (Liu and Bretscher, 1989a) or from IgY yolk preparation C37 (from yolks from chickens immunized with Tpm1p, as described by Gassmann et al., 1990). Antibodies for Myo2p were raised in rabbits immunized with recombinant peptide encompassing residues 784–1118 of Myo2p (the IQ–repeats and coiled–coil region), then affinity purified with coupled recombinant peptide.

Immunofluorescence and fluorescence microscopy were performed as described by Pringle et al. (1989). Staining with anti-Act1p, -Tpm1p, -Tpm2p, and -Myo2p was done after MeOH/acetone postfixation (Pringle et al., 1989), whereas staining with anti-Sec4p or anti-HA was after 5 min of postfixation in 0.1% SDS in PBS, followed by 10 washes in PBS. For double-labeling with anti-Sec4p and anti-Tpm1p, with anti-Sec4p and anti-Myo2p, or with anti-Myo2p and 12CA5, cells were postfixed for 30 s in −20°C acetone, dried, then incubated for 5 min in 0.025% SDS/PBS followed by 10 washes in PBS. Double-labeling with rhodamine-phalloidin and anti-Tpm1p or with rhodamine-phalloidin and anti-Myo2p were after postfixation for 30 s in −20°C acetone. All cells were blocked 30 min in PBS/BSA. Antibody dilutions into PBS/BSA were: anti-Act1p (1:25), rabbit anti-Tpm1p (1:50), chicken anti-Tpm1p (1:100), anti-Tpm2p (1:100), anti-Myo2p (1:20), C.1.2.3. (anti-Sec4p [Walch-Solimena et al., 1997]; 1:50; kindly donated by P. Novick [Yale University School of Medicine, New Haven, CT]), 12CA5 (anti-HA; 1:75; Boehringer Mannheim Corp.), goat anti–rabbit IgG FITC (1:75; ICN Biochemicals, Inc., Aurora, OH), goat anti–rabbit IgG TRITC (1:100; ICN Biochemicals, Inc.), goat anti–mouse IgG FITC (1:300; Organon Teknika Corp., West Chester, PA), and donkey anti–chicken IgY TRITC (1:250; Jackson ImmunoResearch Laboratories, Inc., West Grove, PA). Incubations were for 1.5 h at room temperature for primary antibodies and 1 h at room temperature for secondary antibodies. All secondaries were preincubated with fixed spheroplasts for 1 h at room temperature before use. Staining of actin with rhodamine-phalloidin (Molecular Probes, Eugene, OR), DNA with 4′, 6′-diamidino-2-phenylidole dihydrochloride (Sigma Chemical Co.), and chitin with calcofluor (Sigma Chemical Co.) were performed as described by Pringle et al. (1989). Fluid-phase endocytosis was assayed as described by Riezman (1985) using Lucifer yellow CH (Molecular Probes).

Live cells were placed under 2% agarose in synthetic medium, with appropriate amino acids, in a ΔT dish (Bioptechs, Inc., Butler, PA) and visualized using differential interference contrast (DIC) microscopy. Temperature was controlled using a ΔT dish and objective controllers, stage adaptor, and objective heater (Bioptechs, Inc.). Live cells expressing GFPCap2p were washed with water and placed under 2% agarose in non-fluorescent synthetic medium for observation (Waddle et al., 1996).

DIC and fluorescence images were acquired by a RTC/CCD digital camera (Princeton Instruments, Inc., Trenton, NJ) using a Zeiss Axiovert 100 TV microscope (Carl Zeiss, Inc., Oberkochen, Germany) and then processed using the Metamorph Imaging System (Universal Imaging Corp., West Chester, PA). Photographs of Western blots were acquired through FOTO/Analyst Archiver (FOTODYNE Inc., Hartland, WI). All digital images were processed through Adobe Photoshop 3.0 (Adobe Systems, Inc., Mountain View, CA).

Intracellular membranes were visualized by electron microscopy of cells prepared using the permanganate fixation procedure (Kaiser and Schekman, 1990). 50-ml cultures of cells (OD₆₀₀ = 0.4) were grown under described conditions and fixed for 10 min by direct addition of 10-fold concentrated fixative 20% glutaraldehyde (EM grade; Electron Microscopy Sciences, Fort Washington, PA), 20% formaldehyde (EM grade; Electron Microscopy Sciences), 0.4 M KPi (pH 6.7), then pelleted gently (700 rpm, 5 min) and resuspended in fixative for 1 h at room temperature. Cells were then washed five times with water, resuspended in 5 ml prefiltered 4% aqueous KMnO₄, and incubated for 2 h at 8°C. Cells were then washed 10 times with water, resuspended in 10 ml 1% aqueous uranyl acetate, and incubated for 16 h at 8°C in the dark. These were then washed five times with water, dehydrated in graded ethanol (50, 70, 80, 85, 90, 95, and 3 × 100%), and embedded in Spurr resin (Polysciences, Inc., Warrington, PA): 1 h in 2:1 EtOH/resin, 2.5 h in 1:1 EtOH/resin, overnight in 1:1 EtOH/resin (permitting evaporation of EtOH), 2 h in 100% resin, followed by overnight baking at 65°C. Hardened samples were thin-sectioned, and pale gold sections were mounted on 300 hex mesh copper grids and stained for 1 min in ReynoldÕs lead citrate (Dykstra, 1993). Sections were viewed and photographed in a Phillips-301 electron microscope at 60 kV, using SO-163 film.

SDS-PAGE was performed as described by Laemmli (1970). For probing of internal and external Bgl2p, yeast were fractionated into internal and external proteins. In brief, yeast were killed by addition of 40 mM NaF and 10 mM NaN₃, washed twice in the same, and converted to spheroplasts (1.4 M sorbitol, 25 mM KPi, pH 7.5, 25 mM β-mercaptoethanol, 5 mM NaN₃, 20 mM NaF, 25 μg/ml zymolyase; ICN Immunobiologicals, Lisle, IL) for 1 h at 37°C. Spheroplasts were centrifuged to separate internal protein from external protein. Denaturing sample buffer was added to both fractions. All other cell extracts for Western analysis were prepared as described by Horvath and Riezman (1994). After electrophoresis, proteins were transferred by electroelution to nitrocellulose (Schleicher and Schuell, Keene, NH) and blocked with 10% milk for 30 min. Incubations with primary and secondary antibodies were all for 1 h at room temperature at the following dilutions: anti-Bgl2p (1:10,000; Mrsa et al., 1993), 12CA5 (1:1,000; Boehringer Mannheim Corp.), B43 (anti-Tpm1p/Tpm2p, 1:2,000; Drees et al., 1995), anti-Tpm2p (1:100), anti-Tpm1p (1:200), HRP-conjugated goat anti–rabbit IgG (1:10,000; Organon Teknika Corp.), and HRP-conjugated goat anti–mouse IgG (1:5,000; Life Technologies, Inc.), all in the presence of 1% milk in rinse buffer. Blots were visualized using an enhanced chemiluminescent kit (Amersham Life Science, Little Chal-font, United Kingdom). To assay invertase secretion, invertase was induced as described by Ballou (1990) and cells were fractionated as described above into internal and external proteins, with the exception that nondenaturing sample buffer was added to fractions before loading for electrophoresis in a 7.25% native acrylamide gel. Invertase activity was detected within the gel as described (Ballou, 1990).

Tpm1p specifically localizes to actin cables. To determine whether this is a common, and possibly essential, feature of tropomyosins in yeast, we examined the localization of the minor tropomyosin isoform, Tpm2p, in both wild-type and tpm1Δ cells. Although Tpm1p and Tpm2p are 64.5% identical in primary sequence, antibodies specific for each isoform have now been generated (Fig. 1). Antibodies specific for Tpm1p fail to stain tpm1Δ cells (Fig. 2 D), and antibodies specific to Tpm2p fail to stain tpm2Δ cells, thereby demonstrating their specificity for immunofluorescence.

In wild-type cells, Tpm2p was detectable along actin cables like Tpm1p (Fig. 2, compare C and E). Further, in large-budded wild-type cells, both Tpm1p and Tpm2p were present as a bar at the mother/bud junction (Fig. 2 C, arrowhead). In favorable views, the tropomyosin bar resolved as a ring, reminiscent of the bud neck F-actin ring (Chant and Pringle, 1995; Epp and Chant, 1997; Lippincott and Li, 1998), suggesting that tropomyosins in S. cerevisiae may play a role in cytokinesis, as has been shown for the tropomyosin encoded by cdc8 of Schizosaccharomyces pombe (Balasubramanian et al., 1992).

Loss of Tpm1p results in viable cells with no detectable actin cables and partial loss of polarization of the actin cortical patches, cell wall deposition, and growth (Liu and Bretscher, 1989b). In tpm1Δ cells, Tpm2p was restricted to the vicinity of sites of active growth: nascent bud sites in unbudded cells, bud tips in small-budded cells, and a diffuse distribution in the buds of medium-budded cells (Fig. 2 F). Although this distribution overlapped the distribution of actin cortical patches, the two did not colocalize, suggesting Tpm2p is not incorporated into patches. Tpm2p did not appear in extended cables either, which is consistent with the finding that tpm1Δ cells lack detectable actin cables. Rather, the appearance of Tpm2p, particularly in small buds, resembled the staining seen for tropomyosin in the small buds of wild-type yeast (for comparison to Tpm1p see Fig. 2 C, insets). The overlapping distribution of cortical patches with Tpm2p in tpm1Δ cells prevented determination of whether there is F-actin specifically associated with Tpm2p in these cells. Tpm2p also appeared as a bar between the bud and mother cell in large-budded tpm1Δ cells (Fig. 2 F, arrowhead), again suggesting a role in cytokinesis.

The small Rab-GTPase, Sec4p, is an essential component of secretory vesicles involved in their polarized delivery (Goud et al., 1988; Walch-Solimena et al., 1997). The unconventional type V myosin, Myo2p, has also been suggested to be involved in the polarized delivery of secretory vesicles (Johnston et al., 1991; Govindan et al., 1995). The enrichment of both Myo2p and Sec4p near regions of cell growth in wild-type cells is consistent with these proposals (Fig. 2, G and I; Brennwald and Novick, 1993; Lillie and Brown, 1994). Since tpm1Δ cells are viable and able to bud, albeit with reduced polarity, the localization of Sec4p and Myo2p was examined in these cells; both showed a polarized distribution, similar to that found in wild-type cells (Fig. 2, H and J).

To examine the short-term effects of the loss of all functional tropomyosin, we generated conditional tpm1 mutations (tpm1-1 and tpm1-2) in a tpm2Δ background (Fig. 3 A). Both tpm1-1 tpm2Δ cells and tpm1-2 tpm2Δ cells were found to grow at temperatures below 34°C, but were inviable at 34.5°C or greater. Both alleles could be suppressed by either TPM1 or TPM2, as expected, but were unable to suppress each other. The alleles were also suppressed by expression of rat skeletal muscle tropomyosin (data not shown).

When cells expressing Tpm1p from either tpm1-1 (ABY 950) or tpm1-2 (ABY944) were shifted to restrictive temperatures for up to 4 h, the amount of tropomyosin present did not decrease relative to total protein (Fig. 3 B). Therefore, the proteins are not degraded at the restrictive temperature, but become nonfunctional.

When observed at permissive temperatures, the tpm1-1 and tpm1-2 alleles displayed different phenotypes. Cells bearing tpm1-1 tpm2Δ (ABY950) resembled tpm1Δ TPM2 cells morphologically and in actin distribution, suggesting that the protein has only partial function at permissive temperatures. By contrast, tpm1-2 tpm2Δ (ABY944) cells were indistinguishable from wild-type cells in terms of growth rates and morphology and had actin cables, though fainter than those seen in wild-type cells (Fig. 4, A, C, and E), indicating that the tpm1-2 gene product is more fully functional at the permissive temperature. For this reason, all further work was carried out using tpm1-2 tpm2Δ cells.

The viability of tpm1-2 tpm2Δ cells was determined after shifting to the restrictive temperature of 34.5°C for varying lengths of time and then plating on rich medium at room temperature. For the first 2 h, the cells retained full viability, but then viability declined such that after 5 h at 34.5°C only 34% of cells plated were viable. This loss in viability is at least partially attributable to a time-dependent increase in the fraction of lysed cells observed microscopically.

When tpm1-2 tpm2Δ cells were shifted to the restrictive temperature of 34.5°C, growth continued but was completely depolarized. For small-budded cells, all growth occurred in the mother rather than in the bud, resulting after 4 h in huge round cells (Fig. 4 B). These large cells sometimes retained the small bud remnant, although in suspension the remnant often detached and floated away. Large-budded cells instead showed a thickening of the bud neck and also gradually grew into completely rounded cells. This depolarization of growth was much more complete than that seen in tpm1Δ cells, indicating that polarized growth is strictly dependent upon the presence of functional tropomyosin. Deposition of chitin was also completely depolarized. After 4 h, calcofluor stained uniformly over the entire cell surface with bud scars no longer visible (Fig. 4 D). This depolarized growth still depended upon an intact secretory pathway as sec6-4 tpm1-2 tpm2Δ cells showed no cell enlargement or morphological changes at restrictive temperatures (Fig. 4, G and H).

The actin cytoskeleton in tpm1-2 tpm2Δ cells completely depolarized after shifting to 34.5°C. When actin was examined, cables were no longer visible and the distribution of cortical patches became isotropic (Fig. 4 F). This demonstrates that tropomyosin is required for correct polarization of the actin cytoskeleton and that the remaining polarization seen in the tpm1Δ TPM2 cells depends upon Tpm2p.

It has been shown that shifting wild-type cells to 37°C results in transient depolarization of the actin cytoskeleton (Lillie and Brown, 1994); both our wild-type and tpm1-2 tpm2Δ strains also lost cables and depolarized when shifted to 36°C, with wild-type cells recovering their polarity after ~20 min. However, 34.5°C, a restrictive temperature for tpm1-2 tpm2Δ, has little effect on the polarity of the actin cytoskeleton of our isogenic control strain (ABY973 TPM1/TPM1 tpm2Δ/tpm2Δ) (Fig. 5, A–D, and Fig. 6 A).

Whereas budding and polarized growth had ceased in tpm1-2 tpm2Δ cells at 34.5°C, mitosis still occurred. Staining of nuclei revealed a gradual increase in binucleate cells over a 5-h period. Before the temperature shift, no cells contained >1 nucleus, while after 5 h at 34.5°C, 40% of unlysed cells were binucleate and 1% had >2 nuclei. However, the onset of mitosis was much slower than expected. While the doubling time for the isogenic TPM1 strain is 1.5 h at 34.5°C, only 15% of tpm1-2 tpm2Δ cells had undergone nuclear division after 1.5 h, suggesting a cell cycle checkpoint had been activated to delay mitosis, perhaps the actin cytoskeleton-dependent checkpoint described previously (Lew and Reed, 1995; McMillan et al., 1998).

Since after 4 h at 34.5°C tpm1-2 tpm2Δ cells lack both actin cables and cortical patch polarity, we examined the speed of these cytoskeletal changes by staining actin in cells fixed at earlier time points. Remarkably, actin cables were not detectable in these cells within 1 min after shifting to 34.5°C (Fig. 5, E–H, and Fig. 6 B), whereas TPM1 tpm2Δ control cells retained cables (Fig. 5, A–H, and Fig. 6 A). We worried that actin cables might still be present at 1 min but that fixation was too slow to preserve them against subsequent disassembly. To rule out this possibility, tpm1–2 tpm2Δ cells were subjected to various temperature shift and fixation protocols and stained for actin. Actin cables were seen when cells were fixed at room temperature for 1.5 h (Fig. 7 A), but not when cells were prewarmed for 1 min at 34.5°C before fixation at room temperature (Fig. 7 B). However, when cells were fixed for 10 s at room temperature and then shifted to 34.5°C for 1 min followed by 1.5 h at room temperature, cables were clearly evident (Fig. 7 C). Thus, 10 s of fixation is sufficient time to preserve cables through a shift to 34.5°C for 1 min. We conclude that detectable actin cables disassemble in tpm1–2 tpm2Δ cells within 1 min of shifting to 34.5°C, demonstrating that the conditional phenotype appears very rapidly and that cables are highly unstable in the absence of functional tropomyosin.

Loss of tropomyosin localization was equally rapid (Fig. 5, I–L, and Fig. 6 B). Whereas tropomyosin-staining cables were evident in tpm1-2 tpm2Δ cells at permissive temperatures, Tpm1p staining was diffuse within 1 min of shifting to 34.5°C. TPM1 tpm2Δ control showed virtually no loss of cable staining through 15 min at 34.5°C (Fig. 6 A). We confirmed the rapidity of fixation of Tpm1p-staining cables using the same control as described for actin above. The Tpm1p-staining ring noted at the necks of large-budded cells also vanished within 1 min at 34.5°C. Thus, although the tropomyosin protein in tpm1-2 tpm2Δ cells is stable at the restrictive temperature (Fig. 3 B), it cannot stabilize actin cables or assemble at the bud neck at restrictive temperatures.

In contrast, the distribution of actin cortical patches remained polarized and unperturbed for the first 5 min after shifting to 34.5°C, and then gradually became depolarized over the next 10–20 min (Figs. 5 H and 6 B). For the first 5 min, clustering of cortical patches did not appear to be disturbed (sample small buds are depicted in Fig. 5, E–G, insets). Another cortical patch component, the unconventional type I myosin, Myo5p, was examined in tpm1-2 tpm2Δ cells by tagging the chromosomal MYO5 locus with a triple-HA epitope; COOH-terminal tagging of this gene does not interfere with its function (Goodson et al., 1996). After shifting to 34.5°C, Myo5pHA3 remained colocalized with actin cortical patches throughout a 15-min time course (data not shown). Similarly, a GFP-tagged capping protein, GFPCap2p, which also has been shown to be functional and to colocalize with cortical patches (Waddle et al., 1996), gradually assumed a depolarized distribution in tpm1-2 tpm2Δ cells in a manner indistinguishable from that observed for actin. Thus, cortical patches appear to initially retain their polarized distribution after loss of tropomyosin function, but gradually depolarize in the absence of tropomyosin-containing cables.

Fluid-phase endocytosis has been shown to depend upon an intact actin cytoskeleton, and to be abolished in cells lacking functional components of cortical patches, such as actin (Act1p), fimbrin (Sac6p), and cofilin (Cof1p) (reviewed in Geli and Riezman, 1998; Wendland et al., 1998). To determine whether fluid phase endocytosis still occurs in the absence of functional tropomyosin, tpm1-2 tpm2Δ cells were incubated at 36°C for 1 h, then Lucifer yellow was added as an endocytic tracer for another hour at 36°C. The tropomyosin double mutant accumulated the dye to the same extent as wild-type control cells (data not shown).

Post-Golgi trafficking of secretory vesicles has also been shown to be affected in cells conditionally defective in actin function (Novick and Botstein, 1985). Recently, it has been found that two secretory markers, invertase and the cell wall endoglucanase encoded by BGL2, are transported from the Golgi apparatus to the plasma membrane by separate vesicle populations (Harsay and Bretscher, 1995). Therefore, we examined whether either of these markers accumulated in tpm1-2 tpm2Δ cells at their restrictive temperature. After induction of invertase at 36°C for 1 h, cells were fractionated into external (cell wall and periplasmic) and internal protein and assayed for invertase activity. For both the tpm1-2 tpm2Δ mutant and TPM1 tpm2Δ control, all glycosylated invertase produced was efficiently exported, while a secretion-defective control strain (sec6-4) retained all glycosylated invertase internally (data not shown). To examine whether Bgl2p accumulated internally, tpm1-2 tpm2Δ and TPM1 tpm2Δ cells were incubated at 36°C for 1, 2, 3, or 4 h, fractionated into external and internal fractions, and assayed for Bgl2p by Western blot. The distribution of internal versus external protein was identical between the TPM1 tpm2Δ control and tpm1-2 tpm2Δ cells (data not shown).

Since vesicles resembling post-Golgi secretory vesicles accumulate in a fraction of tpm1Δ TPM2 cells (Liu and Bretscher, 1992), we examined tpm1-2 tpm2Δ cells by thin section electron microscopy for accumulation of secretory membranes. TPM1 tpm2Δ, tpm1-2 tpm2Δ, and sec6-4 strains were shifted to 36°C for 20 min, fixed, and processed for electron microscopy. A subset of tpm1-2 tpm2Δ cell profiles showed an accumulation of membrane-bound structures in their cytoplasm identical to those accumulated in the sec6-4 strain at 36°C (Fig. 8 A, compare arrows in panel b to panels c and d). However, only 15% of the tpm1-2 tpm2Δ cells showed such structures, which was much less than the sec6-4 control (85%). Further, the accumulation was not temperature dependent (Fig. 8 B). Therefore, both bulk secretion and fluid phase endocytosis occur efficiently in the absence of functional tropomyosin.

Since membrane trafficking remains efficient in the absence of functional tropomyosin, we wished to determine whether spatial targeting of secretion was affected by tropomyosin defects. As previously discussed, Myo2p and Sec4p distributions correlate with directed growth. The localization of Myo2p and Sec4p was examined after shifting tpm1-2 tpm2Δ cells to 34.5°C.

Myo2p became delocalized rapidly, appearing as a diffuse stain after 2 min (Fig. 5, M–P, and Fig. 6 B), whereas in control cells (TPM1 tpm2Δ) the distribution of Myo2p did not change (Fig. 6 A). Sec4p also rapidly delocalized in tpm1-2 tpm2Δ cells at 34.5°C (Fig. 5, Q–T, and Fig. 6 B). Again, TPM1 tpm2Δ control cells were not perturbed significantly (Fig. 6 A).

The effect on the polarized distribution of Sec8p was also examined. Sec8p is a component of the exocyst, a complex of eight proteins necessary for the fusion of secretory vesicles to the plasma membrane (Novick et al., 1980; TerBush et al., 1996). This complex also colocalizes with regions of cell growth (TerBush and Novick, 1995). Sec8p was examined in tpm1-2 tpm2Δ cells by replacing the endogenous SEC8 locus with COOH-terminally HA epitope–tagged SEC8. Since this is the sole copy of the essential SEC8 and replacement conferred no deleterious phenotype, SEC8:HA3 must provide functional Sec8p. Immunofluorescence microscopy revealed that shifting tpm1-2 tpm2Δ cells to 34.5°C initially had no effect on the localization of Sec8pHA3. With longer incubations, the polarized staining decreased significantly, although some bud tip enrichment of Sec8pHA3 was evident in tpm1-2 tpm2Δ cells even after 15 min at 34.5°C (Fig. 5, U–X, and Fig. 6 B), indicating Sec8p is able to remain localized independent of actin cables. Prolonged incubation at 34.5°C (1 h) eventually delocalized Sec8pHA3. Conversely, although TPM1 cells had a modest decrease in Sec8pHA3 polarization at 34.5°C over a 15-min time course (Fig. 6 A), by 1 h they had reestablished a strong polarized distribution of Sec8 pHA3.

To examine the effects of the restoration of functional tropomyosin, we studied recovery of the tropomyosin double mutant from the restrictive temperature. tpm1-2 tpm2Δ cells were incubated at 34.5°C for 1 h to completely depolarize the actin cytoskeleton, then cooled to 26°C for various times before fixation and localization of actin and Tpm1p. Astonishingly, tpm1-2 tpm2Δ cells showed restoration of Tpm1p-containing cables within 1 min, changing from a diffuse Tpm1p staining to a filamentous one (Fig. 9 A, a and b). In ~37% of cells examined after just 1 min of recovery, cables visibly emanated from a single focus within the cell (Fig. 9 A, arrowheads in a, and Fig. 10). Actin-staining cables were not readily apparent (using actin antibodies) in most cells at early recovery times (for example, Fig. 9 C, m–o). However, we assume tropomyosin is associating with F-actin for several reasons. First, tropomyosin is only known to assemble into filaments under physiological conditions in the presence of F-actin. Second, double-labeling of recovering cells for Tpm1p and actin (by rhodamine-phalloidin staining) showed colocalization of Tpm1p with a focal point of actin (Fig. 9 D, q and r). Third, although the conditions used for double-labeling using rhodamine-phalloidin were not optimal for preserving nascent cables, when Tpm1p-positive cables were noted under those conditions, they also stained with rhodamine-phalloidin (Fig. 9 D, q and r). We attribute the lack of cable staining with actin antibodies to high background fluorescence from abundant cortical patches.

Myo2p, Sec4p, and Sec8pHA3 rapidly repolarized at 26°C (Fig. 9 A, c–h, and Fig. 10), although recovery of Sec8pHA3 appeared delayed by ~1 min relative to Sec4p and Myo2p (Fig. 10, arrowhead). Double-labeling for Myo2p and Sec4p (Fig. 9 D, u and v) and for Myo2p and Sec8pHA3 (Fig. 9 D, w and x) showed these proteins all polarized to the same location in recovering cells. Further, double-labeling for Tpm1p and Myo2p (data not shown) and for Tpm1p and Sec4p (Fig. 9 D, s and t) showed that Myo2p and Sec4p both repolarized at the convergence of the tropomyosin-containing cables. The recovery of these markers was even faster when the recovery temperature was 11°C, with significant polarization of cables (40% cells viewed) and Myo2p visible in as little as 10 s, showing that both cable reformation and cable-dependent transport are extremely rapid.

With repolarization of the secretory pathway, budding resumed in tpm1-2 tpm2Δ cells. Tropomyosin double mutant cells were placed under 2% agarose/synthetic medium and incubated for 4 h at 35°C, resulting in large, round un-budded cells (Fig. 9 B, i). The cells were then cooled to 26°C over a 5-min period and permitted to recover. Bud emergence resumed rapidly, in that within 5 min of reaching 26°C, new growing buds were visible (Fig. 9 B, j–l). The recovering cells did not resume budding in a uniform manner. While some established new buds quickly, others delayed formation of a new bud for 1 h or more. This correlates with the observation that not all cells showed a rapid repolarization of Tpm1p, Myo2p, Sec4p, or Sec8p during recovery (Fig. 10).

The overall distribution of cortical patches did not change during the short time points at 26°C during which Myo2p, Sec4p, and Sec8p repolarized (Fig. 9 C, m–o). Rather, between 15 and 30 min were required for patches to resume an overall polarized distribution (Fig. 9 C, arrows in p show cells with repolarized cortical patches). Observation of Myo5pHA3 and Cap2pGFP yielded identical results, suggesting that cortical patches repolarize slowly with the restoration of actin cables. However, double-labeling of actin with Tpm1p (Fig. 9 D, q and r) and with Myo2p (data not shown) showed a localized concentration of actin at the focal points for those two proteins at early recovery times, but we have not determined whether that actin corresponds to locally clustered cortical patches.

To determine whether the rapid recovery of Sec4p and Sec8p depends upon Myo2p, those markers were examined in a myo2-66 tpm1-2 tpm2Δ SEC8:HA3 strain. At the permissive temperature, Myo2p, Sec4p, and Sec8p all showed a wild-type distribution, and after 1 h at 34.5°C they all stained diffusely. When permitted to recover at 26°C for 5 min, however, only a small percentage of the cells showed any recovery of Myo2p to a single focal point, in contrast to the 77% recovery seen in the MYO2 tpm1-2 tpm2Δ SEC8:HA3 control (Fig. 11). This demonstrates that in the myo2-66 mutant, Myo2p remains nonfunctional for longer than 5 min after returning to the permissive temperature. Under these conditions, polarized cables detected with Tpm1p antibodies reformed in the myo2-66 tpm1-2 tpm2Δ cells to the same extent as in the MYO2 control. After 5 min of recovery, myo2-66 tpm1-2 tpm2Δ cells also showed very little repolarization of either Sec4p or Sec8pHA3 (Fig. 11). Therefore, polarized actin cables can reform despite reduced Myo2p function, but Sec4p and Sec8pHA3 do not repolarize without functional Myo2p.

This study establishes that actin cables in budding yeast target the delivery of secretory vesicles by Myo2p, and thus direct the polarity of bud growth. We have isolated yeast with conditionally defective tropomyosin, and demonstrated that several proteins involved in polarizing secretion respond very rapidly to the presence of functional tropomyosin and actin cables, but are unaffected by the overall distribution of cortical actin patches.

Tropomyosin is essential to budding yeast, since deletion of both tropomyosin genes, TPM1 and TPM2, is lethal (Drees et al., 1995). A recent report has suggested that tropomyosin may not be essential (Kagami et al., 1997); however, further analysis of strains thought to lack tropomyosin revealed that they still express Tpm2p. Additional experiments have confirmed that the loss of all tropomyosin is indeed lethal (our results and Kagami, M., A. Toh-e, and Y. Matsui, personal communication).

Actin cables require functional tropomyosin for stability. In wild-type cells, both Tpm1p and Tpm2p localize specifically to actin cables and, in large-budded cells, to a bud neck ring, but they are not associated with actin cortical patches. Although tpm1Δ cells lack detectable actin cables, Tpm2p is still localized to regions of cell growth and actin concentration. Furthermore, although Tpm2p staining in these cells overlaps cortical patches, it does not colocalize with patches. Rather, Tpm2p staining, particularly in small-budded tpm1Δ cells, resembles the appearance of tropomyosin in the small buds of wild-type cells, suggesting it may be associated with truncated cable-like structures within the bud. Since overexpression of Tpm2p in tpm1Δ cells produces extended actin cables, the absence of detectable cables may simply reflect the lower overall level of tropomyosin in tpm1Δ cells, which is about eightfold less than in wild-type cells (Drees et al., 1995). As Tpm2p binds avidly to F-actin (Drees et al., 1995), we suggest that Tpm2p in these cells is bound to truncated actin cables in the bud and possibly longer tenuous cables in the mother, too thin for detection by light microscopy. Work by Karpova et al. (1998) showing that actin cables can vary in thickness along their length supports this possibility. The presence of tenuous tropomyosin-containing actin cables in tpm1Δ cells would explain how Sec4p and Myo2p become polarized despite the absence of detectable actin cables in the mother cell.

To study the short-term effects of the loss of tropomyosin, we generated a conditionally defective tropomyosin mutant. Specifically, we isolated a temperature-sensitive tpm1-2 allele in a tpm2Δ background. Shifting the tpm1-2 tpm2Δ cells to 34.5°C results in the loss of tropomyosin function. A summary of the relationships between the presence of tropomyosin-containing actin cables and the polarity of several cell components is shown in Fig. 12.

The most rapid phenotype of the loss of functional tropomyosin is the disappearance of actin cables. Like wild-type tropomyosin, Tpm1p in tpm1-2 tpm2Δ cells localizes to actin cables as well as to a bud neck ring at permissive temperatures. However, within 1 min of shifting tpm1-2 tpm2Δ cells to 34.5°C, tropomyosin staining becomes diffuse, suggesting a rapid dissociation of the protein from F-actin structures (Fig. 12, step 1). At the same time, actin cables vanish, possibly reflecting either depolymerization of their actin or unbundling of the actin filaments to the point that they can no longer be resolved. In support of the actin depolymerization model, study of the actin-depolymerizing drug latrunculin-A shows that the F-actin of cables is capable of rapid turnover (Ayscough et al., 1997). Further, a recent report by Belmont and Drubin (1998) suggests that loss of tropomyosin from cables can lead to recruitment of the actin-depolymerizing protein cofilin (Cof1p), and that the F-actin of tropomyosin-free cables would be depolymerized.

The product of tpm1-2 is not degraded at high temperatures and the temperature-sensitive phenotype is rapidly reversible. When tpm1-2 tpm2Δ cells are restored to a permissive temperature, cables reappear within 1 min, suggesting that cables are highly dynamic structures that can assemble quickly (Fig. 12, step 5). Reassembled cables converge upon a single point, demonstrating that their polarity is reestablished as they reform. Again, two models of reassembly are possible. If cable disassembly reflected actin depolymerization, cable reformation would reflect a rapid polymerization event. If cable disassembly were due to unbundling, cable reformation may reflect rapid binding of tropomyosin to preexisting fibers and the recruitment of an actin bundling protein to the tropomyosin/actin, leading to consolidation into cables.

Loss of tropomyosin and detectable actin cables leads to the loss of polarized growth. That is, when small-budded tpm1-2 tpm2Δ cells are shifted to 34.5°C, bud growth ceases and the mothers become big and uniformly round. For large-budded cells, the tropomyosin neck ring disassembles and the bud neck thickens until distinction between the mother and daughter is lost, again generating uniformly round large cells. When tropomyosin function is restored, polarized growth resumes very rapidly.

This isotropic growth in tpm1-2 tpm2Δ cells reflects the loss of targeted secretion. An extremely tight correlation was seen between the presence of polarized cables and the polarized distribution of several proteins involved in targeted secretion, namely: the secretory vesicle-bound Rab-GTPase Sec4p, the unconventional type V myosin Myo2p, and the exocyst component Sec8p. Both Myo2p and Sec4p have been implicated in targeting secretory vesicles (Johnston et al., 1991; Govindan et al., 1995; Walch-Solimena et al., 1997). All three proteins are localized to regions of cell growth, paralleling the polarization of both actin cables and actin cortical patches (Adams and Pringle, 1984; Brennwald and Novick, 1993; Lillie and Brown, 1994; TerBush and Novick, 1995), although it had not been clear how they might interact with the actin cytoskeleton.

When tropomyosin-containing actin cables are lost from tpm1-2 tpm2Δ cells, Myo2p and Sec4p become delocalized within 2 min (Fig. 12, step 2). When cables are restored, Sec4p and Myo2p repolarize within minutes (Fig. 12, step 6), demonstrating a rapid response of those two proteins to the presence of tropomyosin-bound actin cables and indicating that the newly restored cables are functional for transport of secretory components. The swift recruitment of Myo2p in response to actin cables suggests that the protein acts as a motor to translocate along actin cables to regions of cell growth and that the actin cables are polarized with their barbed ends directed toward regions of cell growth. The repolarization of Sec4p in response to newly formed actin cables depends upon Myo2p, as when recovery is observed in a myo2-66 tpm1-2 tpm2Δ triple mutant, Sec4p does not repolarize despite the formation of cables. This result agrees well with a previous report that Sec4p localization is Myo2p dependent (Walch-Solimena et al., 1997) and is consistent with models whereby Myo2p binds secretory vesicles (with attached Sec4p) and ferries them along actin cables to sites of cell growth.

Sec8p redistributes more gradually than Myo2p or Sec4p after the loss of tropomyosin function, suggesting it binds to the plasma membrane at sites of cell growth by an actin cable–independent method (Fig. 12, steps 3 and 4). However, after extended periods in the absence of tropomyosin function, Sec8p also becomes delocalized, and, after the restoration of functional tropomyosin, Sec8p repolarizes very rapidly, though just delayed relative to Sec4p and Myo2p (Fig. 12, step 7). The delay may reflect a need for some factor delivered by cables, probably secretory vesicles, to growth sites for Sec8p recruitment, but, once Sec8p is bound there, it can remain for longer periods in the absence of both actin cables and nascent secretory vesicles. Consistent with this view is the report that Sec8p depends upon secretion in order to remain localized (Finger et al., 1998). As further support, Sec8p repolarization, like Sec4p, depends upon functional Myo2p and does not occur in the myo2-66 tpm1-2 tpm2Δ mutant.

Cortical actin patch distribution was also perturbed by the loss of tropomyosin function, but the response, to any degree that we could detect, was more than 10 times slower than that of any of the above markers (Fig. 12, step 3). While either Tpm1p or Tpm2p alone can maintain at least a partially polarized distribution of cortical patches, with the total loss of functional tropomyosin, cortical patches migrated to an isotropic distribution over the cell surface. The composition of the cortical patches under these circumstances appeared normal in that two known patch components, Cap2p and Myo5p, remained associated with the patches. Upon restoration of actin cables, the cortical patches gradually repolarized (Fig. 12, step 8), suggesting that the overall distribution of cortical patches depends somehow upon actin cables, possibly responding to some polarity cue delivered by cables to sites of growth. An overall polarized distribution of cortical patches is restored only after 10–15 min, although localized clustering of patches near the cable focal points occurs earlier.

Actin cortical patches are commonly thought to be the nucleators of actin cables. However, a recent report demonstrates that cables do not always terminate upon cortical patches (Karpova et al., 1998). Further, the report shows that actin cables can exist with cortical patches close to both ends, suggesting that association with patches is not an indicator of the inherent polarity of actin cables. Interestingly, Tpm1p-containing cables, Myo2p, Sec4p, and Sec8p all repolarize long before actin cortical patches resume an overall polarized distribution, suggesting cortical patches by themselves are not nucleators of actin cables. One possible explanation for this observation is that somehow a subset of cortical patches becomes established as nucleators of actin cables. An alternative explanation is that actin cables do not nucleate from cortical patches, but from some other site on the plasma membrane. It has been noted previously that Myo2p, Sec4p, and Sec8p all polarize similarly to cortical patches, but they do not colocalize with the patches (Brennwald and Novick, 1993; Lillie and Brown, 1994; Finger et al., 1998). The distribution of Myo2p, Sec4p, and Sec8p in wild-type cells might reflect their accumulation at a nucleation site where the barbed ends of actin cables converge that is distinct from cortical patches, explaining the lack of colocalization.

If cortical patches might not play a role in targeting secretion, what role might they play at regions of cell growth? One possibility is that cortical patches mediate some activity required to maintain efficient active growth. Actin patches are likely to function directly in endocytosis. There is a remarkable correlation between defects in components of cortical patches, such as Act1p/End7p, Arp2p, Cof1p, Sac6p, Dim2p/Pan1p, and Myo5p, and defects in endocytosis (reviewed in Geli and Riezman, 1998; Wendland et al., 1998). To maintain efficient secretion at the ends of cables throughout the cell cycle, actin cortical patches may be required to recycle membranes from those same locations to retrieve such components as v-SNARES and lipids for further rounds of exocytosis.

An interesting question is the nature of the polarizing signal that remains localized in tpm1-2 tpm2Δ cells to redirect actin cable assembly. Currently, the molecular nature of this polarizing cue is unknown. However, several proteins important to the establishment and maintenance of cell polarity and actin organization have been localized to the same regions toward which cables converge, notably the Rho-GTPases Cdc42p and Rho1p, as well as several proteins shown to interact with these GTPases, including the F-actin–binding protein Bem1p and the yeast formin Bni1p. Interestingly, Bni1p has been reported to bind two more actin-binding proteins, Bud6p/Aip3p and EF1α, as well as to yeast profilin, which could serve to locally stimulate F-actin formation and thereby nucleate cable reformation (reviewed in Tanaka and Takai, 1998). Further studies will be needed to determine which molecules remain polarized within cells lacking tropomyosin function. The use of appropriate mutants may then reveal which genes are required to generate polarized actin cables, and thereby establish polarity in budding yeast.

We are greatly indebted to Peter Novick (Yale University School of Medicine) for providing and Ruth Collins (Yale University School of Medicine) for transporting a generous supply of antibody to Sec4p, without which this study would not have been possible. We are also grateful to John Cooper (Washington University School of Medicine) for his donation of the GFP:CAP2 construct, pBJ646.