Over recent decades, diverse intracellular organelles have been recognized as key determinants of Ca2+ signaling in eukaryotes. In yeast however, information on intra‐organellar Ca2+ concentrations is scarce, despite the demonstrated importance of Ca2+ signals for this microorganism. Here, we directly monitored free Ca2+ in the lumen of the endoplasmic reticulum (ER) of yeast cells, using a specifically targeted version of the Ca2+‐sensitive photoprotein aequorin. Ca2+ uptake into the yeast ER displayed characteristics distinctly different from the mammalian ER. At steady‐state, the free Ca2+ concentration in the ER lumen was limited to ∼10 μM, and ER Ca2+ sequestration was insensitive to thapsigargin, an inhibitor specific for mammalian ER Ca2+ pumps. In pmr1 null mutants, free Ca2+ in the ER was reduced by 50%. Our findings identify the secretory pathway pump Pmr1, predominantly localized in the Golgi, as a major component of ER Ca2+ uptake activity in yeast.
All eukaryotic cells display some mechanism that exploits the unique chemical properties of the Ca2+ ion to transduce a cellular signal. Depending on cell type, local changes in the intracellular Ca2+ distribution trigger a variety of physiological responses, indicating the existence of distinct Ca2+ signaling pathways (Clapham, 1995). Due to the inherent stability of Ca2+ in any biological environment, access of this second messenger to the cytoplasm of cells must be strictly controlled. Thus, Ca2+ ions are actively transported out of the cell and into internal organelles, which in turn may release Ca2+ upon appropriate stimuli. Extensive studies in animal cells identified the endoplasmic reticulum (ER) as a major site of intracellular Ca2+ storage and release (reviewed by Berridge, 1994). To fulfill its role as a dynamic Ca2+ reservoir, the ER is suitably equipped with pumps for Ca2+ uptake, channels for Ca2+ release, and Ca2+‐binding proteins for Ca2+ buffering (see Pozzan et al., 1994).
The main Ca2+ store of the yeast Saccharomyces cerevisiae (at least with respect to overall Ca2+ content) is the vacuole, with a total Ca2+ concentration of ∼2 mM. Buffered by inorganic polyphosphates, the concentration of free Ca2+ in the vacuolar lumen is reduced to ∼30 μM (Dunn et al., 1994). Two vacuolar Ca2+ uptake systems have been identified: Vcx1, an H+/Ca2+ antiporter (Cunningham and Fink, 1996; Pozos et al., 1996) and Pmc1, a P‐type ATPase (Cunningham and Fink, 1994). A third yeast Ca2+ transporter, the secretory pathway ion pump Pmr1 (Rudolph et al., 1989), is predominantly localized in the Golgi (Antebi and Fink, 1992). Over the last decade, many components related to mammalian Ca2+ signaling pathways have been identified in yeast. Despite these efforts, functions of Ca2+ ions in the yeast ER and Ca2+ handling by this organelle remained elusive. Only recently, some progress was made with regard to possible functions. First, mutants in sphingolipid biosynthetic steps hosted in early secretory organelles (ER, Golgi) are sensitive to elevated Ca2+ levels (Zhao et al., 1994) and secondly, pmr1 mutants are sensitive to Ca2+ starvation and stabilize a misfolded luminal ER protein (Dürr et al., 1998), which normally undergoes retrograde transport from the ER into the cytosol for ubiquitination and subsequent degradation (reviewed by Sommer and Wolf, 1997). However, no Ca2+ transport systems have been identified in yeast for two major control sites of mammalian Ca2+ regulation, i.e. the ER and the plasma membrane. Moreover, no data are available about Ca2+ concentrations in the lumen of early secretory organelles in yeast. Evidently, methods to monitor free Ca2+ directly within the ER of living yeast cells could provide invaluable insights into the role of this organelle in Ca2+ homeostasis and signaling.
Here, we report the development of a yeast ER Ca2+ probe based on aequorin, a Ca2+‐sensitive photoprotein that emits light when exposed to Ca2+ in the presence of its prosthetic group coelenterazine (Allen et al., 1977; Inouye et al., 1985). Targeting of functional aequorin into the ER lumen of mammalian cells has been accomplished by several groups (Montero et al., 1995; Button and Eidsath, 1996; Montero et al., 1997; Alonso et al., 1998; Pinton et al., 1998). In yeast however, aequorin has only been used to monitor cytosolic Ca2+ (Nakajima‐Shimada et al., 1991). To construct a yeast ER Ca2+ probe, we fused aequorin onto Stt3, a subunit of the oligosaccharyl transferase complex in the ER membrane. This approach strictly localized aequorin to the ER lumen under all our experimental conditions. Using this probe, we measured a steady‐state concentration of 10 μM free Ca2+ for the ER of wild‐type yeast strains. The ER Ca2+ level was unaffected by treatment with thapsigargin, indicating that yeast lacks the sarco/endoplasmic reticulum (SERCA)‐type Ca2+ pump present in the ER of higher eukaryotes. However, ER Ca2+ was severely reduced in cells lacking the secretory pathway pump Pmr1, demonstrating that this ‘Golgi’ ion pump also controls, at least in part, the luminal Ca2+ concentration of the yeast ER.
A Stt3/aequorin fusion protein as a potential ER Ca2+ probe
Stt3 is a transmembrane protein with a hydrophilic C‐terminal domain, which protrudes into the ER lumen and tolerates additional sequences at the C‐terminus without abolishing Stt3 function (Zufferey et al., 1995). These facts, and its relative abundance make Stt3 an ideal candidate to anchor aequorin on the luminal side of the ER membrane. Expecting a high Ca2+ concentration (> μM) within the ER, we used an aequorin variant with reduced Ca2+ affinity (Kendall et al., 1992; Montero et al., 1995) and fused its cDNA to the entire STT3 coding region (Figure 1A), including a triple HA‐epitope tag at the Stt3/aequorin junction (Wilson et al., 1984). The entire construct was transplaced at the chromosomal STT3 locus under control of the STT3 promoter (see Materials and methods). Strains expressing ER–Aeq were viable and grew indistinguishably from their parental STT3 strains, demonstrating a functional replacement of the essential STT3 gene and suggesting a correct membrane topology/localization of the Ca2+ probe. A protease protection experiment (Figure 1B) confirmed that the aequorin domain of ER–Aeq was indeed located in the lumen of an organelle, presumably within the ER.
The intracellular localization of ER–Aeq was first examined by immunofluorescence microscopy. The staining pattern obtained with the anti‐HA antibody clearly coincided with that of the ER, as visualized by anti‐Kar2 antibody in the same cells (Figure 2A). A typical ring‐like staining pattern around the nucleus, plus some staining below the plasma membrane was observed, exactly as for the ER‐marker Kar2 (Rose et al., 1989). Strains lacking ER–Aeq did not stain with the anti‐HA antibody (data not shown). To verify ER–Aeq localization by a second method, cells expressing ER–Aeq were broken and fractionated by sedimentation on a sucrose gradient, essentially as described by Schröder et al. (1995). Gradient fractions were analyzed for the presence of marker enzymes by appropriate enzymatic assays and immunoblotting (Figure 2B). In excellent agreement with the immunofluorescence data, ER–Aeq co‐fractionated with all ER‐markers tested: Kar2, Wbp1 (te Heesen et al., 1992), Sec61 (Stirling et al., 1992) and Dpm1 (Preuss et al., 1991), all of which migrated together in the bottom fractions of the gradient. ER–Aeq was separated almost completely from the Golgi markers GDPase (Abeijon et al., 1989), Emp47 (Schröder et al., 1995) and Och1 (Nakayama et al., 1992) which occupied middle fractions of the gradient. Och1, the cis‐most Golgi marker functionally identified thus far, peaked in fractions 7–8 and was clearly displaced from the bulk of ER–Aeq. It is evident from the distribution of α‐mannosidase activity that vacuolar membranes hardly entered the gradient and were well separated from ER–Aeq. Taken together, our data clearly demonstrate that the steady‐state pool of ER–Aeq resides in the membrane of the ER, thus confining aequorin to the lumen of this organelle.
On denaturing gels, we observed a slightly faster mobility for ER–Aeq from cells treated with tunicamycin (see Figure 3A), an inhibitor of N‐linked protein glycosylation, suggesting that ER–Aeq was carrying carbohydrates at asparagine residues within the Stt3 moiety. This finding allowed us to examine further the potential overspill of ER–Aeq from the ER into a cis‐Golgi compartment, where α‐1,6‐mannosyltransferase activity is known to attach α‐1,6‐linked mannose residues to the carbohydrate chains of core‐glycosylated proteins (Nakanishi‐Shindo et al., 1993). To test whether ER–Aeq was exposed to the α‐1,6‐mannosyltransferase activity of early Golgi compartments, cells expressing ER–Aeq and HA‐tagged Och1 were radiolabeled briefly with [35S]methionine and chased for 30 min prior to immunoprecipitation with anti‐HA antibody. The immunoprecipitated proteins were eluted from the anti‐HA antibody and equal aliquots were subjected to a second immunoprecipitation with antibodies specific for α‐1,6‐mannose linkages and the HA epitope, respectively. As seen in Figure 3B, de novo‐synthesized ER–Aeq was not precipitable with the linkage‐specific antibody, indicating that ER–Aeq bore no α‐1,6‐mannose linkages. HA‐tagged Och1, a cis‐Golgi protein for which we found a similar fraction of α‐1,6‐linked mannose modification as reported in other studies (Harris and Waters, 1996), served as a positive control in this experiment. These results are consistent with our localization studies, i.e. immunofluorescence microscopy and subcellular fractionation found that ER–Aeq was restricted to the ER, and strongly suggest that ER–Aeq is virtually absent from Golgi compartments.
As a first indication of functionality of the Ca2+‐responsive photoprotein domain in ER–Aeq, we tested for Ca2+‐dependent chemiluminescence in crude cell extracts treated with coelenterazine to allow reconstitution of aequorin holoenzyme (Nakajima‐Shimada et al., 1991; Rizzuto et al., 1994). Crude lysates from cells expressing ER–Aeq, when incubated for 4 h on ice with coelenterazine in the presence of the chelator EDTA and 2‐mercaptoethanol (2‐ME), produced light upon addition of Ca2+ (5000 ± 250 counts; the background before adding Ca2+ was 5.5 ± 2.5 counts). In contrast, lysates made from a control strain lacking ER–Aeq produced no increase in light emission after addition of Ca2+ (background was 5 ± 2 counts). These experiments indicated that ER–Aeq was functional and prompted us to evaluate further its use as an ER Ca2+ probe in living yeast cells.
Reconstitution of ER–Aeq in vivo requires treatment with Ca2+ ionophore/EGTA
Reconstitution of active aequorin holoenzyme and subsequent light emission upon Ca2+ challenge of intact cells was previously shown for a yeast strain expressing cytosolic aequorin (Nakajima‐Shimada et al., 1991). Using our own expression system, we also observed efficient reconstitution of cytosolic aequorin (>107 counts per coverslip; J.Strayle and H.K.Rudolph, unpublished results). However, cells expressing ER–Aeq produced only a modest chemiluminescence signal (<105 counts per coverslip) upon cell lysis and addition of Ca2+, even when a divalent cation chelator (EGTA) was present during the entire procedure. To prevent premature aequorin consumption during the reconstitution period, we pretreated cells with the Ca2+ ionophore A23187 and EGTA to deplete intralumenal Ca2+ stores before addition of coelenterazine. A significant 10‐ to 15‐fold increase in total count yield (>106 counts per coverslip) was observed with this treatment (see Materials and methods).
To verify persistent ER localization of our Ca2+ sensor under the newly developed conditions, we performed immunofluorescence microscopy and subcellular fractionation studies on cells treated with A23187/EGTA/coelenterazine as for reconstitution. Curiously, the new protocol produced a somewhat altered distribution of Golgi markers (Figure 4B): whereas GDPase activity displayed a relatively sharp peak centered around fraction 7, the bulk of Emp47 was now more spread out and shifted towards lesser densities (fractions 3–7). However, as with untreated cells, ER–Aeq still co‐fractionated with several ER‐marker proteins (Figure 4B) and produced in immunoflourescence microscopy a typical Kar2‐like ER staining pattern (Figure 4A). These data demonstrate a firm association of the Ca2+ sensor with the ER, which persists even under ion depletion/reconstitution conditions and thus validates our protocol for reconstitution of ER–Aeq into a functional ER Ca2+ probe in living yeast cells.
Measuring the concentration of free Ca2+ in the ER
For measurements of ER Ca2+ with ER–Aeq, we used an experimental set‐up consisting of a perfusion chamber connected to a photon counting device (described by Rizzuto et al., 1994). In order to fix yeast cells in the perfusion chamber, we embedded the ion depleted/reconstituted spheroblasts in low melting agarose on a coverslip. In a typical experiment, the mounted spheroblasts were perfused first with an EGTA solution containing the lipophilic agent bovine serum albumin (BSA) to extract remaining traces of ionophore and to wash out loosely attached cells. Subsequently, a Ca2+‐containing standard yeast medium was applied to allow refilling of Ca2+‐depleted intracellular stores. The consumption of reconstituted ER–Aeq accompanying this process was quantitatively monitored by photon counting. At steady‐state, usually after 3–5 min, all cells were lysed by the addition of digitonin in the presence of excess Ca2+ to discharge all residual holoaequorin. Details regarding the conversion of aequorin chemiluminescence data into Ca2+ concentrations are described in Materials and methods.
Figure 5A (inset) shows the kinetics of chemiluminescence obtained with wild‐type yeast cells expressing ER–Aeq and reconstituted in the presence of 10 μM ionophore A23187. Photon emission, close to background levels during perfusion with EGTA‐containing medium, showed a rapid increase after switching to the Ca2+‐containing medium, and reached a plateau after ∼1–2 min. Subsequent lysis led to a transient increase in photon counts and the discharge of residual aequorin, demonstrating that ∼40% of reconstituted aequorin outlasted consumption during refilling. Conversion of the luminescence signal into Ca2+ values using the Ca2+ affinity for mutated aequorin determined by Montero et al. (1995), revealed a steady‐state concentration of ∼10 μM Ca2+ which was reached within 1–2 min after the switch to Ca2+‐containing medium and remained constant for at least 4 min (Figure 5A). We obtained very similar kinetics of chemiluminescence from two different wild‐type strains expressing ER–Aeq (Figure 5B), strongly suggesting that the steady‐state concentration of free Ca2+ in the ER lumen of wild‐type yeast cells is indeed ∼10 μM.
Since traces of A23187, if not adequately removed before the switch to Ca2+‐rich medium, could antagonize or limit refilling of the ER with Ca2+, we tested a range of different ionophore concentrations. As seen in Figure 5C, the use of 100 nM or 100 μM ionophore A23187 produced the same steady‐state Ca2+ concentration of ∼10 μM upon refilling, strongly arguing against any influence of A23187 on these measurements. Another potential artefact we considered was ‘wash‐out’ of spheroblasts during perfusion, which would reduce total counts and thus lead to an overestimation of Ca2+ levels. However, a comparison of the light emission during extended perfusion (5 min, 1.6×106 ± 2×104 counts/coverslip) with the total counts obtained during a short perfusion period (30 s, 1.6×106 ± 3×104 counts/coverslip) revealed no significant difference. Thus, ‘wash‐out’ of cells could be disregarded in our analysis.
To address whether a fraction of reconstituted ER–Aeq might escape from the ER into a potentially Ca2+‐rich post‐ER compartment, we introduced our Ca2+ probe into a ret2‐1 mutant with a conditional block in Golgi‐to‐ER retrograde transport due to a defective coatamer (COPI) (Cosson et al., 1996) which might enhance post‐ER accumulation of ER–Aeq and thus alter the refilling profile. ret2‐1 cells were grown at permissive temperature and processed for reconstitution as usual. The batch was then split into two aliquots: one, which was shifted to 37°C for 1 h before refilling, was monitored at this restrictive temperature, and the other aliquot was assayed at room temperature. As shown in Figure 5D, there was no remarkable difference in the refilling profiles of the COPI mutant at permissive or restrictive temperature. These findings corroborate our conclusion that the ER–Aeq probe senses ER Ca2+ levels without significant influence from post‐ER compartments.
The secretory pathway pump Pmr1 contributes to thapsigargin‐resistant, vanadate‐sensitive Ca2+ uptake into the ER
Thapsigargin, a potent and specific inhibitor of SERCA Ca2+ pumps, blocks Ca2+ uptake into the mammalian ER (Thastrup et al., 1989). As shown in Figure 5E, treatment of spheroblasts during reconstitution (1 h) with 5 μM thapsigargin, a concentration several magnitudes higher than that required to block SERCA pumps (Sagara and Inesi, 1991), had no effect on ER refilling to a steady‐state level of 10 μM Ca2+. We reported previously that expression of rabbit SERCA1a suppresses the EGTA‐hypersensitivity of mutants lacking the Golgi ion pump Pmr1 (Dürr et al., 1998). As seen in Figure 6, such a pmr1 mutant strain expressing SERCA1a is sensitive to thapsigargin, illustrating that thapsigargin can enter intact yeast cells and block an appropriate target. Thus our data imply that SERCA homologs are not involved in ER Ca2+ homeostasis in yeast, as previously suggested from the analysis of the yeast genome (Sorin et al., 1997).
All known ATP‐driven Ca2+ transporters, including the two Ca2+ pumps characterized in yeast (Pmr1, Pmc1), belong to the superfamily of P‐type ATPases and thus show some sensitivity to vanadate, which blocks formation of an auto‐catalytic phosphoenzyme intermediate. As illustrated in Figure 5F, the steady‐state Ca2+ concentration reached upon refilling was reduced by almost 50% when cells were preincubated with 1 mM vanadate. However, further addition of monensin to collapse H+ transmembrane gradients had no effect on refilling in the presence of vanadate. These results suggest that the ER membrane harbors at least one P‐type Ca2+ pump, but lacks a Ca2+/H+ antiport under our experimental conditions. Strikingly, pmr1 mutants displayed a reduced steady‐state ER Ca2+ level of ∼6 μM (Figure 5G), a value very similar to wild‐type assayed in the presence of vanadate. This result directly implicates Pmr1 in ER Ca2+ homeostasis. We verified that ER–Aeq was properly localized to the ER of pmr1 cells, well separated from Golgi markers (see Figure 7). The only noteworthy difference from wild‐type cells (see Figures 2 and 4) was altered fractionation profiles for α‐mannosidase and GDPase. Finally, inhibition of protein synthesis with cycloheximide, either during the entire reconstitution period (Figure 5H) or just before switching to Ca2+‐rich medium (data not shown), had no effect on ER refilling in wild‐type cells. Thus, the main contribution of Pmr1 to ER Ca2+ uptake can not be ascribed to newly synthesized protein, but rather reflects activity of a pre‐existing Pmr1 pool.
Here, we describe in vivo measurements of free Ca2+ in the lumen of the yeast ER using a protein chimera which consists of the Ca2+‐sensitive photoprotein aequorin fused onto Stt3, an ER‐resident oligosaccharyl transferase subunit. Measurements with this sensor reveal a steady‐state free Ca2+ level of ∼10 μM for the yeast ER, a concentration significantly lower than free Ca2+ in the mammalian ER. Our experiments identify the yeast ER as a thapsigargin‐resistant, but vanadate‐sensitive, Ca2+ store lacking the SERCA‐type Ca2+ ATPases known to control mammalian ER Ca2+. Most importantly, we demonstrate that the secretory pathway ion pump Pmr1, a P‐type ATPase prominently localized in the Golgi complex, is a major component of ER Ca2+ uptake activity in yeast.
ER–Aeq is a suitable ER Ca2+ indicator
Our approach to target aequorin to the ER lumen by virtue of an endogenous ER membrane protein (Stt3) avoids addition of an ER retention signal (i.e. the carboxy‐terminal tetrapeptide sequence His‐Asp‐Glu‐Leu HDEL) to the catalytically important C‐terminus of aequorin (Nomura et al., 1991) and should evade Kar2(BiP)‐dependent mechanisms for ER retention which may fail under Ca2+ limitation (Suzuki et al., 1991). Accurate ER localization of the probe was thoroughly evaluated by (i) subcellular fractionation, (ii) immunofluorescence microscopy and (iii) examination of ER–Aeq for the presence of Golgi‐specific carbohydrate modifications. By these criteria, ER–Aeq was strictly confined to the ER under all experimental conditions, even though the Ca2+‐depletion protocol required for holoaequorin reconstitution in vivo (see below) induced some restructuring (fragmentation) of the ER in pmr1 cells. This phenomenon could either reflect a stress response of pmr1 cells to some general experimental condition (fixation, perfusion) or result from specific, calcium‐dependent mechanisms for ER restructuring in yeast. In animal cells, ER vesicularization was observed upon expression of BiP variants defective for ATP hydrolysis (Hendershot et al., 1995) and in response to a persistent increase in cytosolic Ca2+ (Subramanian and Meyer, 1997). Ca2+ depletion also changed the distribution of Emp47, which is related to a lectin that binds mannose residues in a calcium‐dependent manner (ERGIC‐53; Itin et al., 1996). Thus, the observed redistribution of Emp47 might reflect an ion requirement for proper localization of this Golgi protein. These ambiguities, however, did not impinge upon the intracellular localization of ER–Aeq, which was strictly associated with ER markers under all experimental conditions.
Efficient reconstitution of ER–Aeq in vivo required Ca2+ depletion to prevent premature aequorin consumption, a problem frequently encountered with this photoprotein in Ca2+‐rich organelles. For the same reason, we employed an aequorin variant in our fusion with reduced Ca2+‐affinity due to a single amino acid exchange (Montero et al., 1995). With these precautions, we used our probe and found that refilling of the ER with Ca2+ reached a steady‐state level during a time period of ∼2 min, after which 40–50% of total reconstituted ER–Aeq was still uncharged as demonstrated by immediate cell lysis in a Ca2+‐rich digitonin buffer. Such a ratio of uncharged aequorin allows for conversion of luminescence into Ca2+ values based on the method of Allen et al. (1977). Thus, ER–Aeq is a suitable sensor of ER Ca2+ in yeast.
Characteristics of ER Ca2+ uptake in wild‐type cells
Using ER–AEQ in two different wild‐type strain backgrounds, we found that yeast cells at steady‐state maintain a free Ca2+ concentration of ∼10 μM within the ER. Studies on free Ca2+ in the mammalian ER report a range from 1 to 400 μM. Pinton et al. (1998) found ∼400 μM free Ca2+ using a soluble, ER‐resident aequorin and a synthetic coelenterazine analog. Similar values (40–400 μM) were recently obtained with a novel system wherein two fluorescent proteins were linked via calmodulin to allow Ca2+‐dependent fluorescence energy transfer (Miyawaki et al., 1997). In contrast, Button and Eidsath (1996) employed a membrane‐attached aequorin chimera to find a concentration of 1–2 μM, suggesting that at least a fraction of the ER membrane system might harbor a relatively low Ca2+ concentration. Despite the divergence in the data obained from animal cells (reviewed by Meldolesi and Pozzan, 1998), our measurements of the physiological steady‐state level of free Ca2+ in the yeast ER indicate a substantial difference in ER Ca2+ handling between yeast and animal cells. It is most likely that the lack of a specialized SERCA‐type Ca2+ pump from the yeast ER is the primary cause for this peculiarity, since Ca2+ uptake was completely resistant to thapsigargin. Moreover, the moderate level of free Ca2+ (10 μM) within the yeast ER might well limit the use of this compartment for intracellular Ca2+ signaling and thus implicate other organelles (Golgi, vacuole) as dynamic Ca2+ stores in this eukaryotic micro‐organism. It should also be noted that yeast cells can survive prolonged Ca2+‐depletion if sufficient Mn2+ is available in the growth medium. Under such conditions, yeast cells retain only 3% of their normal cellular Ca2+ content without apparent effects on growth (Loukin and Kung, 1995; Dürr et al., 1998).
Ca2+ refilling into the ER was unaffected by monensin, an ionophore capable of collapsing transmembrane H+ gradients which potentially energize a putative ER Ca2+/H+ antiport. Since Ca2+ uptake into purified yeast Golgi vesicles is markedly sensitive to the H+ ionophore (Sorin et al., 1997), presumably due to the presence of the vacuolar Ca2+/H+ antiport Vcx1 in these membranes, the monensin insensitivity we observed would underscore the localization of ER–Aeq in a compartment distinctly different from Golgi. However, Ca2+ uptake into mammalian Golgi in vivo, is insensitive to H+ dissipating agents (Pinton et al., 1998). Therefore, it remains to be seen whether Vcx1 is a genuine Ca2+ transporter of yeast Golgi, but under the conditions of this study we failed to detect such an activity in the ER membrane.
The role of Pmr1 in ER Ca2+ sequestration
A partial vanadate‐sensitivity of ER Ca2+ uptake and the low ER Ca2+ level of pmr1 cells together point to the P‐type ATPase Pmr1 as the main ER Ca2+ transporter in yeast. Loss of Pmr1 activity lowered free Ca2+ in the ER from 10 to ∼5 μM, providing direct biochemical support for our hypothesis that the ER‐associated defects observed in pmr1, as discussed in our previous study (Dürr et al., 1998), result from an altered ion content of this compartment. Although our new data strongly insinuate a direct role of the Pmr1 ion pump in ER Ca2+ sequestration, we can not rigorously exclude another interpretation. Pmr1 could serve just to stimulate an ‘elusive’ and as yet unidentified ER Ca2+‐transporter that would operate with reduced activity in pmr1 cells. ER fragmentation in the pmr1 mutant could also seclude a fraction of the ER–Aeq probe into vesicles lacking this transporter, thereby causing an artificial underestimate of free ER Ca2+ in pmr1 cells. It should be emphasized, however, that the ER‐related defects of pmr1 cells are already manifest in regular, Ca2+‐rich media, i.e. under conditions that do not cause ER fragmentation (Dürr et al., 1998).
Therefore, we conclude from the present study that yeast lacks a genuine ER Ca2+‐transporter specific to this compartment, but instead utilizes, at least in part, the secretory pathway pump Pmr1 for ER Ca2+ uptake. Undoubtedly, the main cellular pool of Pmr1 resides in Golgi compartments (Antebi and Fink, 1992; Schröder et al., 1995; Sorin et al., 1997). Nevertheless, subcellular fractionation experiments with cells grown in regular, Ca2+‐rich media consistently find a small fraction of Pmr1 (<<10%) associated with ER markers (Antebi and Fink, 1992; Sorin et al., 1997), and a non‐functional Pmr1 mutant protein was shown to enrich in ER fractions (Sorin et al., 1997). Unfortunately, rigorous biochemical analysis of the presumed ER pool of Pmr1 is severely compromised by the failure of available fractionation procedures to separate ER from early Golgi compartments completely. However, since we found that ER Ca2+ uptake proceeded unperturbed after cycloheximide treatment or a block in Golgi‐to‐ER retrograde transport, it seems likely that a fraction of Pmr1 is indeed retained in the ER to actively contribute to the ionic milieu of this compartment. In a similar fashion, putative ER pools of Pmc1, Vcx1 or both, could be responsible for residual Ca2+ uptake into the ER of pmr1 cells, since genetic studies show that yeast cells remain viable with any one of the three Ca2+ transporters as the sole intracellular Ca2+ transport system (Cunningham and Fink, 1996).
Future studies should address how yeast cells balance activity and spatial distribution of their endo‐membraneous Ca2+ transporters with the need for Ca2+ ions in the ER. As we have shown, ER–Aeq is a suitable probe to monitor free Ca2+ in this compartment. Together with probes targeted into other secretory organelles, this tool should permit the characterization of conditional transport mutants and thus ultimately allow the evaluation of individual components and their contributions to Ca2+ sequestration into the yeast ER.
Materials and methods
Yeast strains and growth conditions
Yeast strains used in this study were YS256 (MATα ade2 his3‐Δ200 leu2‐3,112 lys2‐Δ201 ura3‐52 STT3::HA::AEQ::LEU2), YS258 (the same as YS256 except pmr1‐Δ2::HIS3), YS368 (MATa leu2 ura3 his4 lys2 bar1 STT3::HA::AEQ::LEU2), YS371 (MATa leu2 ura3 his4 lys2 suc2 ret2‐1 STT3::HA::AEQ::LEU2; the parental strain is PC130 from P.Cosson), YS380 (MATα ade2 his3‐Δ200 leu2‐3,112 lys2‐Δ201 ura3‐52/br434 [CEN URA3 PMA1::SERCA1A::ADC1]) and YR663 (the same as YS380 except pmr1‐Δ1::LEU2) (Dürr et al., 1998). Standard yeast culture media were prepared as described by Sherman et al. (1986).
Construction of the STT3::aequorin fusion allele
The STT3::aequorin fusion construct, obtained in several cloning steps using different helper templates, can be summarized briefly as follows: a unique MluI site was introduced into the STT3 gene (at nucleotide + 2155 from the STT3 start codon) just before the STT3 stop codon, to allow in‐frame fusion of additional sequences onto the full‐length STT3 coding sequence. Using this site, a (3×HA)::aequorin cassette was inserted encoding three repeats of the nine amino acid HA‐epitope (81 bp in total), fused in‐frame with modified aequorin cDNA sequences (Montero et al., 1995) encompassing nucleotides from position +25 (eight codons downstream of the aequorin start) to +591 (after the aequorin stop codon). Thus the aequorin domain encoded by this cassette also lacks the first eight amino acids of aequorin, like the aequorin construct successfully used in yeast by Nakajima‐Shimada et al. (1991). As a 3′ terminator region, the STT3 HpaI–XbaI fragment (nucleotides from +2093 to +2509) was spliced behind the aequorin stop codon, thus duplicating a small segment of the STT3 open reading frame (ORF) (nucleotides from +2093 to +2154). Into this duplicated segment, we engineered a XhoI site to allow addition of LEU2 sequences (a 2.2 kb SalI–XhoI fragment), used as a selectable marker during transplacement of the entire construct as a SpeI–XbaI fragment to the endogenous STT3 locus by transformation. The correct integration of this allele at the STT3 locus was verified by Southern blotting analysis.
Subcellular fractionation and indirect immunofluorescence
Subcellular fractionation was carried out based on the protocol of Schröder et al. (1995). GDPase was measured as described by Abeijon et al. (1989). α‐mannosidase was detected according to the method of Opheim (1978). The activities of GDPase and α‐mannosidase are given as a percentage activity of the highest values measured.
For indirect immunofluorescence, cells grown to early log phase were fixed by adding concentrated formaldehyde solution and 1 M potassium phosphate pH 6.5 directly to cells in growth medium to a final concentration of 3.7% (w/v) and 100 mM, respectively. After 1 h fixation at 25°C, the procedure was carried out as described in Schröder et al. (1996).
Radiolabeling, immunoprecipitation and protease protection assay
Yeast strain YS256 expressing ER–Aeq and HA epitope‐tagged Och1 was grown in SC media lacking uracil. Metabolic labeling with [35S]methionine (Amersham), cell lysis and immunoprecipitation were carried out as described by Harris and Waters (1996). Labeling in the presence of 10 μg/ml tunicamycin was carried out for 20 min prior to incubation with [35S]methionine for 20 min.
To determine membrane topology, cells were spheroblasted and lysed at 4°C in a tissue grinder. After clearing, lysate was incubated at 4°C and treated with trypsin (500 μg/ml), trypsin (500 μg/ml)/Triton X‐100 (0.5%) or left untreated. After 20 min the reaction was stopped by addition of trichloroacetic acid. Following centrifugation and washes with acetone, the pellet was resuspended in sample buffer and subjected to SDS–PAGE.
Aequorin reconstitution in vitro
Cells were grown overnight in yeast extract/peptone/dextrose (YPD). Approximately 10 OD600 units of cells were harvested, followed by washing and resuspending in 400 μl TEP buffer (50 mM Tris–HCl, 5 mM EDTA, 1 mM PMSF pH 7.5). Cells were lysed by vortexing with glass beads. After a clearing spin, 300 μl of crude cell lysate was supplemented with 50 mM 2‐ME and 5 μM coelenterazine and incubated for 4 h on ice in the dark. The Ca2+‐dependent light emission was measured using a Packard Picolite luminometer by adding 10 mM CaCl2 to a 50 μl sample.
In vivo Ca2+ measurements
Cells were grown overnight to log phase and 20 OD600 units were harvested and washed with synthetic complete medium (Sherman et al., 1986) containing 1.2 M sorbitol buffered with 10 mM HEPES to pH 7 (SP). After resuspension in 1 ml of SP, 100 U of zymolyase 100T and 3 μl 2‐ME was added and incubated at room temperature. Spheroblasts were washed with SP containing 10 mM EGTA (SPE) and resuspended in 500 μl SPE (4°C) containing 10 μM Ca2+ ionophore A23187 and 14 mM 2‐ME. After addition of coelenterazine to a final concentration of 5 μM, the cell suspension was incubated for 1 h at 4°C in the dark. Cells were then harvested, washed twice with SPE containing 5% BSA (4°C) and incubated for 5 min in SPE (5% BSA) at 4°C. After centrifugation, cells were resuspended in 100 μl SPE (5% BSA), followed by addition of 100 μl low melting point agarose (1.5%, 37°C). Fifty microliters of this cell suspension was then placed onto 13‐mm round coverslips and stored at 4°C until the agarose solidified. The coverslips were transferred into a perfused, thermostated chamber (22°C; except ret2‐1 mutant at restrictive temperature), and placed in close proximity to a cooled, low‐noise photomultiplier with a built‐in amplifier–discriminator (EMI 9789, as described in Rizzuto et al., 1994). For Ca2+ refilling, the perfusion with SPE (5% BSA) was followed by perfusion with SP supplemented with 1 mM CaCl2. Permeabilization with 100 μM digitonin in the presence of 10 mM CaCl2 finished the experiment by discharge of residual aequorin. The aequorin luminescence data were captured by an EMI C660 photon counting board and stored in a personal computer.
For use of aequorin luminescence data in calculations of luminal Ca2+ concentrations during refilling, it is imperative that the entire refilling process consumes only a fraction of the total cellular pool of reconstituted holoaequorin. To meet this requirement in a potentially Ca2+‐rich environment (the ER), we employed an aequorin variant with reduced Ca2+ affinity due to a single amino acid exchange (Montero et al., 1995). Synthetic coelenterazine analogs, such as coelenterazine n (Shimomura et al., 1989), also allow modulation of the Ca2+‐sensitivity of aequorin sensors, but were not commercially available during the course of our experiments. In any case, the fraction of charged aequorin remaining after refilling can be determined by cell lysis in the presence of excess Ca2+. The total number of photons counted during refilling and subsequent cell lysis in high Ca2+ is then set to represent the total amount of reconstituted aequorin. To calculate Ca2+ concentrations from our light emission data, we used the method described by Allen and Blinks (1978) which correlates a given Ca2+ concentration to the ratio between initial light emission at this Ca2+ concentration and the amount of reconstituted aequorin initially present. Since holoaequorin is continuously consumed during a refilling experiment, the remaining pool of charged aequorin must be recalculated for each timepoint as the difference between the total amount of holoaequorin, as determined by cell lysis in saturating Ca2+, and the amount of consumed aequorin represented in the photons captured up to that timepoint. The mathematical algorithm used for these calculations is described in more detail in Brini et al. (1995) and we used the Ca2+ affinity constants of mutant aequorin as determined by Montero et al. (1995).
We would like to express our gratitude to Rosario Rizzuto and all members of the Pozzan laboratory for expert technical advice and generous hospitality. We thank P.Cosson, M.Rose, S.te Heesen, M.Aebi, G.Waters, Y.Jigami, S.Schröder‐Kühne, T.Rapoport and R.Schekman for providing antibodies, plasmids and strains, and S.Jäger and M.Hämmerle for helpful discussions. Initially, J.S. received a stipend from the Studienstiftung des Deutschen Volkes and was supported further by a short‐term EMBO fellowship. This work was also supported by a grant from the Deutsche Forschungsgemeinschaft to H.K.R.
- Copyright © 1999 European Molecular Biology Organization