A Role For Lte1p (a Low Temperature Essential Protein Involved in Mitosis) in Proprotein Processing in the Yeast Secretory Pathway*

We previously identified six single gene disruptions in Saccharomyces cerevisiae that allow enhanced immunoreactive insulin secretion primarily because of defective Kex2p-mediated endoproteolytic processing. Five eis mutants disrupted established VPS (vacuolar protein sorting) genes, The sixth, LTE1, is a Low Temperature (<15 °C) Essential gene encoding a large protein with potential guanine nucleotide exchange (GEF) domains. Lte1p functions as a positive regulator of the mitotic GTPase Tem1p, and overexpression of Tem1p suppresses the low temperature mitotic defect of lte1. By sequence analysis, Tem1p has highest similarity to Vps21p (yeast homolog of mammalian Rab5). Unlike TEM1, LTE1 is not restricted to mitosis but is expressed throughout the cell cycle. Lte1p function in interphase cells is largely unknown. Here we confirm the eis phenotype of lte1 mutant cells and demonstrate a defect in proalpha factor processing that is rescued by expression of full-length Lte1p but not a C-terminally truncated Lte1p lacking its GEF homology domain. Neither overexpression of Tem1p nor 13 other structurally related GTPases can suppress the secretory proprotein processing defect. However, overexpression of Vps21p selectively restores proprotein processing in a manner dependent upon the active GTP-bound form of the GTPase. By contrast, a vps21 mutant produces a synthetic defect with lte1 in proprotein processing, as well as a synthetic growth defect. Together, the data underscore a link between the mitotic regulator, Lte1p, and protein processing and trafficking in the secretory/endosomal system.

A genetic screen has previously been performed with yeast mutagenized by random gene disruptions (by insertion of lacZ/ LEU2) to identify genetically engineered cells exhibiting enhanced immunoreactive insulin secretion (eis, 2 derived from an integrated GAL1 promoter-driven insulin-containing fusion protein called ICFP, Ref. 1). The ICFP is comprised of the leader and propeptide of the yeast alpha-mating factor contiguous with a single chain insulin, separated by a Kex2p cleavage site. Cleavage at the Kex2p-processing site causes insulin to be delivered ultimately to the vacuole in a manner that requires specific insulin structural features (2). From 90,000 transformants that were originally screened, six independent gene disruptions were identified. Each of the six eis mutants were found to secrete slightly to dramatically increased amounts of unprocessed ICFP, and indeed, a kex2⌬ mutant exhibited maximal immunoreactive insulin secretion (1). Upon further analysis, five of the six eis mutants were found to be vacuolar protein sorting mutants vps8, vps35, vps13, vps4, and vps36, which affect protein trafficking and thereby impair protein processing in the Golgi % endosomal system (e.g. Ref. 3). Such an outcome can be explained because the ability of Kex2p and other secretory protein processing enzymes (such as Ste13p and Kex1p) to function properly requires their continuous recycling between Golgi and endosomal systems (4).
In this article, we report that eis4 is caused by disruption of the gene known as LTE1 (5). LTE1 is a low temperature essential gene product that is required at Ͻ13°C to allow yeast cells to advance through the spindle pole checkpoint during mitosis, in which the small GTPase Tem1p is essential (6). Lte1p is a positive regulator of Tem1p function (7) and overexpression of TEM1 rescues the cold-sensitive growth arrest of lte1 cells (8). Lte1p has a large, modular-appearing primary structure including small GEF homology domains embedded in the C-and N-terminal regions (9). Overexpression of lte1-mini (truncated to lack these domains) can suppress the cold-sensitive growth arrest of lte1D cells, raising the possibility that Lte1p might not be a GEF for Tem1p (10); however, other analysis supports the hypothesis that GEF function of Lte1p activates Tem1p (11). Importantly, while Tem1p expression becomes significant only during mitosis (telophase), Lte1p is stably expressed throughout the cell cycle (12), suggesting other possible roles for Lte1p in interphase cells.
Beginning at the G 1 3 S transition of the cell cycle, the secretory pathway focuses on the bud site and expands the bud that shall become a daughter cell (13). Lte1p localization (which is regulated via phosphorylation, Ref. 9) is normally confined to buds growing from secretory activity, at least in part by virtue of interactions with plasma membrane-associated Ras2p (10) and perhaps by being bound to other cell polarity proteins in the bud such as Kel1p and Kel2p (14). Even so, we found it surprising to discover lte1 in a screen revealing secretory protein processing mutants. In the present study, we have sought to confirm a secretory protein processing defect in lte1 mutant cells. The present work highlights the importance of the Lte1p C-terminal region (that includes a GEF domain) in proprotein processing. We also report genetic interactions of VPS21 (encoding the yeast homolog of mammalian Rab5) with LTE1, suggesting a new role for Lte1p in protein trafficking in secretory/endosomal systems.

EXPERIMENTAL PROCEDURES
Strains and Media-Standard yeast media and genetic manipulations were as described (15). Strain L5145/MI316 expresses ICFP from a single chromosomal locus driven from the GAL1 promoter (1). Strain lte1 773 is the eis4 mutant of L5145/pMI316 displaying enhanced immunoreactive insulin secretion after random mutagenesis and screening (1). The other strains employed in this study, unless stated otherwise, are isogenic to either W303 or L3852 (Table 1).
Filter Blot Assays-To assay secretion of insulin-containing peptides, cell patches were replica-plated onto nitrocellulose filters placed on top of plates containing 2% galactose as sole carbon source. After 24 -40 h, the filter was washed and processed for immunoblotting with guinea pig anti-insulin and a peroxidase-conjugated secondary antibody as previously described (1). CPY secretion from cells was detected by filter overlay and overnight growth of the cells as described (18) using mouse mAb anti-CPY (Molecular Probes, 1:4000 dilution) and a peroxidase-conjugated secondary antibody (Jackson Immu-noResearch, 1:10000 dilution).
Bioactive Alpha Factor and Proalpha Factor Secretion-Assay for secretion of bioactive alpha factor was done as described (19). MAT␣ cells (0.015 OD 600 /1.5 l) were spotted onto a thin lawn of MATa bar1 cells (LM23-3az; Ref. 20) on synthetic complete medium. After 1-2 days at 30°C, halos surrounding patches of cells were visualized by placing plates directly on a scanner and converting to digital images. Halo areas were quantified using Image Quant v5.2 (Molecular Dynamics, Inc.).
To examine secretion of proalpha factor, cells were grown to mid-log phase in synthetic complete medium without methionine and cysteine. Cells were harvested and resuspended at 0.8 OD 600 /ml in fresh medium with 1 mg/ml bovine serum albumin. 500 Ci of 35 S amino acid mixture (Expre 35 S 35 S, PerkinElmer Life Sciences) was added to each sample, and incubation continued for 50 min at room temperature with shaking. Medium was separated from cells by centrifugation at 19,500 ϫ g for 5 min. SDS was added to the medium to a final concentration of 1%, and samples were boiled for 3 min. Secretion of alpha factor-containing peptides was recovered by immuno- A, cells grown in liquid medium with 2% raffinose (to allow expression from the GAL1 promoter) were lysed and analyzed for expression of full-length mycLte1p or mycLte1 1012 by Western blot, normalized to equal total cell protein. Molecular mass as estimated by mobility of molecular weight markers is shown at the right. B, filter blot assay of immunoreactive insulin secretion on plates containing 2% galactose. The lte1 mutant shown in the figure is the eis4 mutant described previously (1) in which insertional mutagenesis results in Lte1p truncated after the first 773 residues. All eis mutants impair proprotein processing, resulting in enhanced immunoreactive insulin secretion (1), and expression of full-length mycLte1p restores the insulin secretion to levels seen in wild-type (wt) cells. FIGURE 2. Low temperature growth defect of lte1 1012 and lte1 null (lte1⌬) cells is rescued by expression of mycLte1p. A, wild-type (wt) and lte1 1012 cells were transformed with the empty pYES2 high copy expression plasmid; lte1 1012 mutant cells could not grow efficiently at 13°C on any carbon source. However, lte1 1012 cells expressing mycLte1p from the GAL1 promoter in pYES2 grew normally on galactose (in which the GAL1 promoter is strongly stimulated) or on raffinose (in which the GAL1 promoter is less strongly stimulated). B, quantitative growth assay on galactose plates in which wild-type (wt) yeast growth was compared with lte1⌬ yeast, or lte1⌬ bearing full-length mycLte1p (third line) or mycLte1p 1012 (bottom line). Each line contains an identical volume of yeast starting at an identical OD, spotted in sequential 10-fold serial dilutions from left to right. precipitation with the RW1 rabbit polyclonal anti-alpha factor antibody (kind gift of Dr. D. Shields, Albert Einstein College of Medicine, Bronx, NY) and analyzed by SDS-15%-PAGE and fluorography.
Western Blot-Steady-state levels of mycLte1p, Ste3-Myc, and Kex2p were analyzed in cell lysates prepared from mid-log cultures by vortexing with glass beads, as described previously (21). Sample loading was normalized to lysate protein as measured by BCA protein assay (Pierce). After SDS-PAGE and electrotransfer to nitrocellulose, blots were probed with anti-Myc (rabbit polyclonal A-14; Santa Cruz Biotechnology, Inc., Santa Cruz, CA) or anti-Kex2p (rabbit polyclonal M120D; gift of Dr. R Fuller, University of Michigan) and a peroxidase-conjugated secondary antibody (Jackson ImmunoResearch, 1:5000 dilution). Bands were visualized by enhanced chemiluminescence (Amersham Biosciences).
Metabolic Labeling and Immunoprecipitation of CPY-Metabolic labeling was performed using cultures grown to mid-log phase in synthetic complete medium without methionine and cysteine. Cells were resuspended at 1 OD 600 /ml, labeled with Expre 35 S 35 S at 0.1 mCi per 1 OD 600 cells for 5 min at room temperature, and chased in the presence of 10 mM unlabeled methionine and cysteine. At various chase times, aliquots were placed on ice in the presence of 10 mM sodium azide. Lysates were prepared and resuspended in radioimmune precipitation assay buffer (10 mM Tris, pH 7.5, 150 nM NaCl, 2 mM EDTA, 1% Nonidet P-40, 1% deoxycholate, 0.1% SDS) for immunoprecipitation with 1 l of mouse mAb anti-CPY and protein G-Sepharose beads.
Endocytic Degradation of Ste3p-As previously described (22), strains carrying pSL2015 (GAL-STE3-myc) were grown to mid-log phase in synthetic complete medium containing 2% galactose as the sole carbon source to stimulate the expression of Ste3p-myc. At time 0, glucose (3%) was added to arrest further Ste3-myc synthesis. At the indicated times after glucose addition, cells were harvested, and lysate was prepared for SDS-PAGE and Western blotting with anti-Myc antibody.

RESULTS
Identification of eis4 as lte1-The insulin secretory phenotype of the eis4 mutant was not as strong as some of the other eis mutants that proved to be vps mutants (1). By sequencing genomic DNA adjacent to the lacZ insertion, the eis4 mutant was found to be caused by disruption of LTE1. Full-length Lte1p is 1435 amino acids, and residues 1292-1315 are critical to the CDC25 homology domain that is thought to have guanine nucleotide exchange activity. We found that the lacZ/ LEU2 insertion of eis4 yielded the possibility of expression of an Lte1p gene product C-terminally truncated after residue 783. To test the effect of losing the C-terminal region of Lte1p on protein processing in the secretory pathway, we prepared several plasmid vectors, each bearing an N-terminally Myc-tagged version of Lte1p under control of the GAL1 promoter. We ultimately concentrated on two constructs, one encoding fulllength Lte1p and a second encoding an Lte1p 1012 that was C-terminally truncated after residue 1012. These forms were expressed in cells as stable proteins detected by Western blotting with anti-Myc antibody (Fig. 1A), and the functionality of these proteins was also tested (Fig. 1B, see below).
Haploid yeast on either glucose, galactose, or raffinose plates can grow at 13°C, whereas either eis4 cells (lte1 783 , data not shown) or lte1 1012 cells exhibit a significant low temperature growth phenotype regardless of carbon source (middle sector, Fig. 2A). Full-length mycLte1p could restore low temperature growth to lte1 1012 cells on galactose or raffinose plates, but not on glucose plates in which mycLte1p expression remains repressed (left sector, Fig. 2A). Using serial dilution, full-length mycLte1p expression restored quantitatively normal growth to lte1⌬ cells at 13°C, whereas mycLte1 1012 could not complement the growth defect (Fig. 2B). Likewise, enhanced secretion of exogenously expressed immunoreactive insulin in lte1 783 cells was complemented by mycLte1p (Fig. 1B). Thus, whereas C-terminal truncation of Lte1p supports neither low temperature growth nor normal handling of the insulin precursor, mycLte1p is fully functional in both capacities ( Figs. 1 and 2).
Defect in Proalpha Factor Processing in lte1 Mutant Cells-To test whether Lte1p influences processing of an endogenous The lte1⌬ cells exhibit a 37.7 Ϯ 1.8% reduction in alpha factor halo area. This experiment has been reproduced with replicates in five independent experiments. B, Western blot of steady state Kex2p levels, normalized to total cellular protein, in lte1 1012 mutant cells, vps21⌬ cells, or lte1 1012 vps21⌬ double mutant cells. The result from a kex2⌬ strain is provided as a negative control, at left. C, secretion of immunoreactive alpha factor from strains shown in B, as well as from vps8 cells. Each strain was grown to mid-log phase. Identical OD units of cells were labeled with 35 S amino acids as described under "Experimental Procedures," and the medium collected. Immunoprecipitates from the medium with an anti-alpha factor antibody were analyzed by SDS-PAGE and fluorography. A background band is seen in wild-type (wt) cells just below the proalpha factor migration position. A kex2⌬ strain was analyzed at the right; the presence of the high molecular mass proalpha factor band indicates deficiency of Kex2p activity. Other secretory protein processing activities may also be deficient in vps mutant strains. The appearance of a band at the bottom of the gel that comigrates with mature alpha factor is a useful control that total proalpha factor expression is the same in all strains, but it does not mean that all of this material is biologically active, as additional Kex2p, Ste13p, and Kex1p cleavages are required to produce bioactive alpha factor from other low molecular mass inactive intermediates (23) that are not resolved from the dye front under the SDS-PAGE conditions shown here.
secretory protein substrate, we checked the ability of lte1 1012 cells to produce bioactive alpha mating factor, as assessed by the ability to cause growth arrest of cells of the opposite mating type (20). When spotted on a lawn of Mata cells on galactose plates at 30°C, wild-type ␣ cells were surrounded by a large halo of growth-arrested Mata cells. By contrast, lte1⌬ cells always exhibited a decrease in halo area (Fig. 3A) that averaged a 35% reduction over numerous experiments. These data suggest that Lte1p assists in the production of bioactive alpha factor as well as in processing the insulin-containing fusion protein (1). Notably, both substrates require cleavage by Kex2p during their maturation within the secretory pathway while proalpha factor requires additional processing events to generate bioactive alpha factor.
Whereas a number of vps mutants impair Kex2p-mediated secretory proprotein cleavage via accelerated delivery of Kex2p to a degradative compartment resulting in decreased Kex2p levels, other vps mutants can more subtly influence processing enzymes without significant change in Kex2p levels (such as vps8 (3) or vps21 (this report)). It is therefore of interest to note that steady state Kex2p levels were not significantly affected by loss of expression of Lte1p (Fig. 3B). There was a small but reproducible increase in the secretion of unprocessed proalpha factor into the medium of lte1 1012 mutant cells, comparable to that of vps8 cells (Fig. 3C). (For comparison, Fig. 3, B and C also show that vps21 mutant cells exhibit more proalpha factor secretion without a significant decrease in Kex2p level, whereas kex2⌬ cells exhibit massive proalpha factor secretion and undetectable Kex2p.) Because Lte1p exhibits positive regulation of Tem1p, a small GTPase essential for mitosis (7), we wished to determine whether the secretory protein processing defect of lte1 cells might be secondary to an alteration in cell growth. To check this, we examined whether overexpression of Tem1p could suppress the protein processing defect of lte1 1012 cells. As shown in Fig. 4B, overexpression of Tem1p allowed lte1 cells to grow like wild type at 13°C, as previously reported (8). Importantly, however, Tem1p overexpression did not suppress the alpha factor secretion defect of lte1 (Fig. 4A, left). These data   suggest that impaired secretory protein processing in lte1 cells is not secondary to a defect in cell growth.

lte1 Mutant Cells Do Not Exhibit a vps Phenotype-LTE1
is not a known vacuolar protein sorting (VPS) gene (23), but as all other eis mutants were known vps mutants (1), we checked for secretion of CPY in lte1 cells (Fig. 5A). Unlike other established vps mutants (e.g. vps4 or vps23), lte1 mutant cells exhibited no augmented CPY secretion by filter overlay assay (Fig. 5A). A more sensitive immunoprecipitation assay from media bathing lte1 cells also demonstrated inconsequential secretion of newly synthesized CPY (Fig. 5B). We also considered the possibility of delayed delivery of precursor CPY to the endosomal/vacuolar system without frank CPY secretion in lte1 cells. To check for this, cells were pulse-labeled with 35 S amino acids for 5 min, washed, and chased for various times at 30°C (Fig. 5C). At 5 min of chase, the P1 form of CPY (in the endoplasmic reticulum) predominated with the P2 (Golgi) and mature (vacuolar) forms already apparent. By 10 min of chase, the majority of newly synthesized CPY had already been processed to the mature form and by 30 min of chase, processing appeared complete. In lte1 1012 mutant cells that exhibit a small alpha factor halo, there was no delay in the appearance of mature CPY, indicating that general protein trafficking within the endosomal system is not substantially perturbed (Fig. 5C). Further, we checked the endocytic internalization and degradation of the cell surface a-factor receptor, Ste3p, using a "GAL-shut-off " assay (22). Specifically, expression of Myc-tagged Ste3p under GAL1 control is shut off by addition of glucose, and vacuolar delivery and degradation of pre-existing cell surface receptors can be followed by Western blotting with anti-Myc. As shown in Fig. 6, Ste3p was degraded in lte1 1053 mutant cells with normal kinetics. Thus, lte1 is not a vps mutant, showing no global defects in endosomal trafficking.
Genetic Interactions between LTE1 and VPS21, the Yeast Homolog of Mammalian Rab 5-While functioning as a positive regulator of Tem1p, Lte1p has also been found to physically associate with Ras2p-GTP in a manner dependent upon the C-terminal GEF-like domain of Lte1p (10). To consider other related GTPases that might be involved in secretory protein processing, a BLASTp search of yeast proteins was performed using full-length Tem1p as the query sequence, yielding the list of small GTPases shown in Table 3. Overall sequence similarity is highest for Vps21p (the yeast homolog of mammalian Rab5), but because of the conserved GTPase domain itself, there is a high degree of similarity with all the GTPases listed. Plasmid vectors for overexpressing each of the yeast gene products shown in Table 3 were transformed into Mat␣ lte1 mutant cells to test for suppression of the alpha factor halo defect. As noted above, Tem1p overexpression could not restore a normal alpha factor halo to lte1 mutant cells (Fig. 4A), and indeed, neither could overexpression of Ras2p nor any of the dozen other GTPases listed in the lower portion of Table 3 (data not shown). However, VPS21 overexpression, either driven from its own promoter on a high copy plasmid (Fig. 7A) or a strong constitutive (TDH3) promoter on a centromeric plasmid (not shown), could rescue the alpha factor halo defect of lte1 mutant cells. The effect was specific, as overexpression of Vps21p could not increase the halo of wild-type yeast, nor could it restore the halo defect of an unrelated vps23 mutant (Fig. 7A). The abnormally enhanced immunoreactive insulin secretion of lte1 cells was also suppressed by overexpression of Vps21p (Fig. 7B). We then looked for synthetic interaction between vps21 and lte1 mutants. Indeed, in a side-by-side comparison on the same plates, while  Table 3). Note that VPS21 overexpression did not further enlarge the normal halo of wild-type cells, nor did it normalize the small halo of an unrelated (vps23) mutant. B, filter blot for immunoreactive insulin from wild-type (wt) and lte1 773 (eis4) mutant cells transformed with vector, or multicopy VPS21. Note that the abnormally eis phenotype of lte1 mutant cells is improved (diminished) upon overexpression of VPS21. a Plasmid sources shown in Table 2. JANUARY 19, 2007 • VOLUME 282 • NUMBER 3 vps21 mutant cells had an alpha factor halo that was smaller than that of wild-type cells, the lte1 vps21 double mutant clearly had a synthetic effect, with an alpha factor halo much smaller than either mutant alone (Fig. 8A). Such a synthetic effect could not be observed with most other vps mutants tested; e.g. lte1 and vps9 (an established GEF for the Vps21p GTPase, Ref. 24) showed no additive inhibition (reduction) of alpha halo (Fig. 8B). Notably, the lte1 vps21 double mutant cells have no major decrease in steady state levels of Kex2p (Fig. 3B), although they exhibit a synthetic increase in the secretion of unprocessed proalpha factor (Fig. 3C). Thus, the data support an interaction of VPS21 with LTE1 in secretory proprotein processing.

VPS21 Genetically Interacts with lte1 in Mitosis in a Different Way Than in Secretory Proprotein
Processing-Given the specificity of the Tem1p GTPase for the mitotic spindle pole checkpoint, we had no reason to expect that overexpression of Vps21p, which rescues the alpha factor halo defect of lte1 cells, would impact on low temperature growth of lte1 mutant cells. It was therefore surprising to discover that when VPS21 was expressed from its own promoter on a multicopy plasmid, lte1 1012 mutant cell growth was no longer arrested at 13°C (Fig.  9A). Also, when VPS21 was expressed from a strong constitutive promoter on a centromeric plasmid, it rescued the growth of lte1⌬ cells comparably to that of TEM1 expressed from the same promoter (Fig. 9B).
To explore further the interactions of VPS21 with lte1, we used a semipermissive low temperature of 17°C to examine growth of vps21 or lte1, or vps21 lte1 double mutant cells. At 25°C, each of the strains grew like the VPS21 LTE1 wild type (Fig. 10A, left). At 17°C, the mitotic blockade of lte1 cells was demonstrated as a partial growth defect, while growth of vps21 lte1 double mutant cells was profoundly inhibited (Fig. 10B). Growth of the vps21 mutant alone was not at all inhibited under these conditions (Fig. 10B). These data provide good genetic evidence that the presence of Vps21p, even at endogenous expression levels, limits the mitotic defect of lte1 mutant cells. By contrast, as shown in Fig. 10C, vps9 did not exert any appreciable growth phenotype when combined with the lte1 mutant, suggesting that the GEF activity of Vps9p is not important for Vps21p to play a role in Lte1p function.
To distinguish suppressor effects of Vps21p on the secretory protein processing defect of lte1 cells versus suppression of the low temperature growth arrest of lte1, we compared the effect of overexpressing inactive (persistently GDP-bound) Vps21p-S21L The results shown here were replicated in three independent experiments. Note that while the single lte1 or vps21 mutants had only modest deficiencies in bioactive alpha factor production; the lte1 vps21 double mutant exhibited a major reduction in halo size (28.3 Ϯ 11.3% of that in wild-type cells), indicating a synthetic defect. B, same as above, except combining lte1 and vps9 mutants. Although the single vps9 mutant has a small halo, its inhibitory effects on production of bioactive alpha factor are not additive to the inhibitory effect of the lte1 mutant. mutant versus constitutively active (persistently GTP-bound) Vps21p-Q66L mutant. As measured by Western blotting (not shown), the protein expression level for each recombinant Vps21p mutant was identical, and was equivalent to that in cells overexpressing wild-type Vps21p. Using these strains, it was apparent that rescue of the alpha factor halo defect of lte1 1012 cells was achieved by overexpressing the persistently GTP-bound Vps21p-Q66L mutant, whereas the GDP-bound Vps21p-S21L mutant was devoid of such activity (Fig. 11, left panel).
In contrast with this behavior, overexpressing the persistently GDP-bound Vps21p-S21L mutant exhibited a similar degree of suppression of the growth arrest phenotype of lte1 cells at 13°C as that of overexpressing wild-type Vps21p or the constitutive active Vps21p-Q66L mutant (Fig. 11, right panel). These data strongly suggest differences in the VPS21-mediated suppression of mutant phenotypes in lte1 cells: suppression of the secretory protein processing defect specifically requires an active GTP-bound Rab, whereas rescue of low temperature growth arrest does not require GTP-loading of overexpressed Vps21p.

DISCUSSION
Lte1p is a low temperature essential gene product that allows yeast cells to advance through the end of mitosis regulated by the small GTPase Tem1p. It is debated whether the CDC25 homology (GEF) domain contained in the C-terminal region of Lte1p is normally required for its mitotic function (10) although this is a currently favored view (11). TEM1 was originally discovered in a screen for high copy suppressor genes that rescue cold-sensitive growth arrest of lte1 cells (8). Tem1p expression is turned on during mitosis, in which it acts as an essential controller of the mitotic exit network. The timing of Tem1p activation by Lte1p occurs when the spindle pole body migrates into the bud. Lte1p localization at the bud appears to be regulated by its interactions with Ras2p (10), its phosphorylation state (11), and interactions with cell polarity proteins (14). However, Lte1p function during S-phase is largely unknown.
We identified lte1 as eis4, from a screen for yeast mutants that exhibit enhanced insulin secretion because of defects in secretory protein processing (1). The eis group of mutants appears to work by impairing the recycling of secretory protein processing enzymes (e.g. Kex2p, Ste13p, and Kex1p) between Golgi and endosomal systems (1). In this study we confirm our original observations on eis4 (which hypothetically produces a protein containing only the first 783 residues of Lte1p). Specifically, we demonstrate that the eis4 phenotype is complemented by full-length mycLte1p (Fig. 1). Moreover, we demonstrate a proprotein processing defect in lte1 mutant cells ranging from a complete null to the lte1 1012 mutant that lacks the C-terminal GEF domain. This truncation mutant also exhibits classic low temperature growth arrest, and under conditions where the truncated Lte1 1012 protein is stably expressed, it cannot rescue low temperature growth of lte1⌬ cells (Fig. 2). The lte1 1012 mutant produces less bioactive alpha factor as judged by a reproducibly smaller alpha factor halo and an increase in the secretion of unprocessed proalpha factor. The data suggest that the C-terminal GEF domain of Lte1p may be involved in its role in secretory protein processing. The phenotype is mild, resulting in no significant change in intracellular Kex2p level, comparable to certain vps mutants (Fig. 3). Nevertheless, lte1 is not a vps mutant, exhibiting no defect in the trafficking of vacuolar carboxypeptidase Y (Fig. 5) nor any defect in endocytic internalization and degradation of cell surface Ste3p (Fig. 6). These results are compatible with a growing list of mutants in trafficking genes that affect Golgi-early endosomal trafficking and secretory proprotein processing yet do not produce a vps phenotype, such as tlg2 (and to a large extent, tlg1 (25), chc1 (26), as well as inp53 (27,28)). As one intracellular site of Vps21p function involves Golgi-early endosome trafficking (29), Lte1p may facilitate Vps21p activity at this site; thus lte1 should be considered a candidate for addition to the foregoing list of mutants.
During S-phase, Lte1p is unlikely to work via Tem1p activation because Tem1p expression is limited during this phase of the cell cycle (12). Indeed, unlike the low temperature growth defect, the secretory protein processing phenotype of lte1 cannot be rescued by overexpression of TEM1 (Fig. 4). One might argue that even after multicopy expression (Fig. 4), TEM1 cannot suppress the alpha factor halo defect of lte1 mutant cells because cell cycle-dependent regulation limits Tem1p availability. However, when the TEM1 promoter is replaced with the strong constitutive TDH3 promoter, TDH3-TEM1 also cannot rescue the alpha factor halo defect of lte1 cells (not shown). Moreover, neither overexpression of Ras2p, Ras1p, nor 11 other GTPases has any effect on secretory protein processing ( Table 3), suggesting that Lte1p is unlikely to work with any of these GTPases in secretory protein processing.  By contrast, VPS21, when overexpressed in lte1 mutant cells, restored production of bioactive alpha mating factor (as measured by halo assay) and restored to normal the amount of immunoreactive insulin secretion (Fig. 7). This effect was quite specific, insofar as VPS21 overexpression could not increase secretory protein processing in wild-type yeast or in vps mutants bearing unrelated genetic defects (Fig. 7). Moreover, vps21⌬ exhibited a synthetic defect in secretory protein processing with lte1, and such a synthetic defect could not be mimicked by deletion of one particular Vps21p GEF, vps9 (Fig. 8).
The results provide genetic evidence that LTE1 interacts with VPS21 in a manner that is independent of VPS9. (Intriguingly in lte1 mutant cells, the synthetic defect exhibited by vps21⌬ could be mimicked by several other vps mutants tested, including vps4, vps23, and pep12 (not shown), all implying activity in recycling of proprotein processing enzymes through the endosomal system.) Whereas several explanations are possible, a simple hypothesis is that Lte1p might function as a GEF for Vps21p within one endocytic or secretory subcompartment in cells during S-phase. Certainly there is a rapidly expanding number of potential GEFs being identified that appear to support the Vps21p GTPase or its mammalian homolog, Rab 5 (24).
A finding that was surprising (and we believe, quite distinct from that described above) was that overexpression of VPS21, which has not previously been implicated in mitosis, nevertheless rescued low temperature growth in lte1 mutant cells (Fig.  9). This was true both for lte1⌬ cells and cells bearing the lte1 1012 truncation. Furthermore, vps21⌬ exhibited a synthetic growth defect with lte1 at 17°C, and this genetic interaction with lte1 in mitosis could not be mimicked merely by cutting off VPS9-mediated GEF activity to Vps21p (Fig. 10). We propose that these unusual mitotic effects might derive from structural similarities between Vps21p and Tem1p (Table 3). Whereas we have no reason to believe that Vps21p plays a specific role in mitosis, we hypothesize that Vps21p might compete for certain interaction partners, such as GAP (GTPase-activating) proteins that may tonically inhibit Tem1p by converting the protein from an active GTP-bound form to an inactive GDP-bound form. The recruitment of such proteins to Vps21p might thereby increase the activity of Tem1p, facilitating low temperature mitotic progression in an lte1 background. Alternatively, VPS21 activity in endosomal trafficking might theoretically facilitate the localization of one or more anchoring proteins or regulators of the mitotic exit network, thereby enhancing Tem1p activation at the spindle pole body checkpoint. However, we favor the former hypothesis because rescue of low temperature growth of lte1 mutant cells is also achieved by overexpressing the inactive, persistently GDP-bound, Vps21p-S21L mutant (Fig. 11). Altogether, these findings underscore a secretory protein processing defect of lte1 mutant cells, the rescue of which specifically requires overexpressing an active Vps21p that can achieve the GTP-bound state (Fig. 11).