Accurate measurement of intracellular pH in unperturbed cells is fraught with difficulty. Nevertheless, using a variety of methods, intracellular pH oscillations have been reported to play a regulatory role in the control of the cell cycle in several eukaryotic systems. Here, we examine pH homeostasis in Schizosaccharomyces pombe using a non-perturbing ratiometric pH sensitive GFP reporter. This method allows for accurate intracellular pH measurements in living, entirely undisturbed, logarithmically growing cells. In addition, the use of a flow cell allows internal pH to be monitored in real time during nutritional, or growth state transition. We can find no evidence for cell-cycle-related changes in intracellular pH. By contrast, all data are consistent with a very tight homeostatic regulation of intracellular pH near 7.3 at all points in the cell cycle. Interestingly, pH set point changes are associated with growth state. Spores, as well as vegetative cells starved of either nitrogen, or a carbon source, show a marked reduction in their internal pH compared with logarithmically growing vegetative cells. However, in both cases, homeostatic regulation is maintained.
Research in a wide variety of experimental organisms has lead to the proposition that intracellular pH fluctuations play a regulatory role in the eukaryotic cell cycle. For example, reproducible increases in intracellular pH (pHi) are associated with the fertilization of sea urchin eggs (Johnson and Epel, 1976), with mitogenic stimulation of mammalian cells (Schuldiner and Rozengurt, 1982; Moolenaar et al., 1983) and with passage through `start' in budding yeast (Anand and Prasad, 1989). Periodic cycling of intracellular pH over successive cell divisions has also been observed in the protozoan Tetrahymena pyriformis, as well as the slime moulds, Dictyostelium discoideum and Physarum polycephalum (Gillies and Deamer, 1979; Aerts et al., 1985; Gerson and Burton, 1976). In Dictyostelium, a maximal pHi is temporally correlated with the onset of DNA synthesis, and persists through M phase. In Physarum, one alkaline shift per cell cycle is observed with a maximal value at mitosis, whereas two alkaline shifts associated with both S phase and cytokinesis are observed in Tetrahymena.
Intriguingly, artificially raising pHi in Dictyostelium results in an increased rate of both protein and DNA synthesis (Aerts et al., 1985). This has led to the suggestion that cytoplasmic alkalinization serves as an on/off trigger for passage through `start'. In mammalian cells, however, artificially raising the pHi of mitogen-deprived cells does not by itself trigger division, although blocking the rise can stop cellular proliferation upon exposure to mitogens (Moolenaar, 1986; L'Allemain et al., 1984; Pouyssegur et al., 1985; Lucas et al., 1988).
Typical methods for intracellular pH measurement have included the use of H+-selective microelectrodes (Gerson and Burton, 1976), 31P-NMR (Gillies et al., 1981; Civan et al., 1986; Barton et al., 1980; Navon et al., 1979), radiolabeled membrane-permeable weak acids (Gillies and Deamer, 1979; Schuldiner and Rozengurt, 1982; L'Allemain et al., 1985; Anand and Prasad, 1989) and pH-sensitive fluorescent dyes (Grinstein et al., 1984; Gillies et al., 1990; Haworth and Fliegel, 1993; Pena et al., 1995). In all of these cases the methods used have required the extensive manipulation of cells. This has ranged from centrifugation and re-suspension at high density in non-physiological buffers, to the electroporation (or de-energizing) of cells to allow entrance of fluorescent dyes. The discrepancy in the timing and duration of pHi fluctuations reported for different organisms thus raises the question of the possible consequences of such pre-treatments on the accurate estimation of pHi.
Furthermore, experiments addressing the role of pHi in cell-cycle progression have called for the synchronization of cultures through starvation, heat shock, or the use of temperature-sensitive mutants. These experiments have thus failed to address the simple question of the status of intracellular pH in an unperturbed exponentially growing culture. In this report we have used a non-invasive method of intracellular pH measurement to definitively determine the limits of pH homeostasis and whether or not pH fluctuations are associated with progression through the cell cycle under physiologically relevant conditions.
The method chosen involves the use of a ratiometric pH-sensitive GFP (as opposed to non-ratiometric alternatives; Kneen et al., 1998; Llopis et al., 1998; Robey et al., 1998; Elsliger et al., 1999). This engineered derivative of the wildtype GFP (referred to as `phluorin') has a bimodal excitation spectrum, but unlike wildtype GFP, the relative emission intensities at the excitation wavelengths used show striking pH dependence (Miesenbock et al., 1998). By expressing phluorin in fission yeast cells, and examining them microscopically under native conditions in a flow cell, we have an exquisitely sensitive and stable measure of pHi. We are thus able to examine wild-type and mutant cells either during growth or following various experimental treatments.
The use of S. pombe offers a distinct advantage for these studies. Since S. pombe grows only by extension in length at its growing tips, it is possible to quickly and easily determine cell-cycle position by the measurement of cell length (Mitchison, 1957; Mitchison, 1990). Thus pHi and cell length can be measured in a random population of unsynchronized, logarithmically growing cells, and the individual cell data easily correlated with cell-cycle position.
MATERIALS AND METHODS
Strains, media and growth conditions
All Schizosaccharomyces pombe strains used in this study (leu1-32 h+; leu1-32 h90; cdc2-1w leu1-32 h-; leu1-32 ade6-210 h-; cdc10-129 leu1-32 h+; cdc22-M45 leu1-32 h-; cdc25-22 leu1-32 h+; nda3-KM311 h-) were derived from wild-type strains 972 (h-), 975 (h+), or 968 (h90). Cells were grown in thiamine-free Edinburgh Minimal Media (Alfa et al., 1993) that had been pH-adjusted with either HCl or NaOH. EMM(minus nitrogen) and EMM(minus glucose) were made by omitting NH4Cl and glucose, respectively. EMM and EMM(minus glucose) were supplemented where indicated by the addition of 100% ethanol to a final concentration of 1%. EMM(proline) and EMM(glutamine) were made by omitting NH4Cl and adding either 10 mM L-proline or 10 mM L-glutamine, respectively. Nutritional and temperature shifts were performed using conditioned growth media collected from early log phase cultures in order to minimize any possible variation upon shifts in growth conditions. High pH media was buffered by the addition of 40 mM MES (pH 5.7, 6.0, 6.3) or 20 mM PIPES (pH 6.5 and above). All cultures were grown with shaking at 25°C to early log phase (<1×106 cells/ml) unless otherwise indicated in the text. At these cell densities the pH of the external medium was sufficiently buffered (against the tendency of the cells to acidify their environment) to maintain its initial value. Cells were transferred from EMM to EMM(minus glucose), or EMM(minus nitrogen) by filtration (Nurse et al., 1976). The pH of all media was measured before and after growth experiments.
The ratiometric pH-sensitive GFP gene (Miesenbock et al., 1998), a gift from J. E. Rothman, was PCR amplified (forward primer: 5′-ggg gga tta ata tga gta aag gag aag aac ttt tca ctg g-3′; reverse primer: 5′-ggg ggg tcg act tat ttg tat agt tca tcc atg cca tgt g-3′) and cloned into the unique NdeI and SalI sites of pREP1, or pREP41 (Maundrell, 1993; Basi et al., 1993), using standard techniques (Maniatis et al., 1989). Nuclear-targeted phluorin was constructed in the same way, but with PCR primers (forward: 5′-gga gga tta ata tga gta aag gag aag aac ttt tca ctg g-3′; reverse: 5′-ggg ggg tcg act tag acc tta cgc ttc ttc tta ggt ttg tat agt tca tcc atg cca tgt g-3′) incorporating the SV40 large T antigen nuclear localization signal, PKKKRKV (Gorlich and Mattaj, 1996), to the C-terminus. All plasmids were transformed into fission yeast strains through electroporation (Toone et al., 1998).
A standard curve of the fluorescence of ratiometric phluorin at different pH values was generated by growing a leu1-32 h+ strain, expressing phluorin behind the thiamine-repressible nmt1 promoter, in thiamine-free media to a concentration of ∼1×107 cells/ml. These phluorin-expressing cells were then permeabilized to H+ ions in highly buffered media at various pHs as follows. 1 ml aliquots of culture were collected and washed three times (3000 rpm, 5 minutes) with 1 ml of either succinate pH 5.6, MES pH 6.0, MES pH 6.4, PIPES pH 6.6, PIPES pH 6.8, MOPS pH 7.0, MOPS pH 7.15, MOPS pH 7.3, HEPES pH 7.45, HEPES pH 7.6, or Tris pH 8.0. All buffers were at a concentration of 200 mM and pH-adjusted with NaOH or HCl. Protease inhibitors were added (final concentrations 15 μg/ml pepstatin A, 1 mM PMSF, 1 μg/ml o-phenanthroline, 10 μg/ml leupeptin, 10 μg/ml aprotinin, 50 μg/ml antipain), followed by addition of the protonophore CCCP (final concentration 50 μM), sarcosine (final concentration 0.05%) and toluene (final concentration 0.1%) to aid in permeabilization. Samples were then incubated for 20 minutes on a rotary shaker, collected, and again washed three times with 1 ml of the appropriate pH-adjusted buffer. 40 μl aliquots were then placed onto poly-L-lysine coated coverslips, and the permeabilized cells allowed to settle for 40-60 minutes. Coverslips were inverted onto a depression slide filled with the appropriate buffer and sealed with vaseline before image acquisition (see below). Differences of temperature in the range of 20-36°C did not alter phluorin calibration curves.
Image acquisition and analysis
All cells were prepared for image acquisition by aliquoting 40 μl of an early log phase culture to coverslips that had been treated with 0.1% poly-L-lysine (Pringle et al., 1991). A 40-minute incubation was sufficient time for the cells to settle and adhere to the poly-L-lysine coated surface. Coverslips were subsequently inverted onto depression slides, sealed with vaseline, and the cells examined. Under these conditions randomly sampled individual cells monitored under the microscope grew at a rate of 1.61±0.28 μm/hour. This correlates favourably with the calculated rate of 1.73 μm/hour based on a generation time of 260 minutes (Table 1) and 7.5 μm of extension growth per generation. All solutions and glassware were kept at 25°C (with the exception of experiments dealing with the effects of temperature, where solutions and glassware were maintained at the temperature being assayed). Images were acquired using a Leitz DMRB fluorescence microscope (Leica Microsystems), a Lambda 10-2 filter wheel controller (Sutter Instruments), and a high performance cooled CCD camera (Cooke SensiCam) operated by Slidebook image analysis software (Intelligent Image Innovations). Some variation in excitation intensity and/or spectra was noted with the Hg light source depending on the age of the bulb. This necessitated repeating the standard curve every time an absolute measure of internal pH was being made. Excitation was through D420/30X (`low') or D460/20X (`high') excitation filters (Chroma Technology Corporation) and emission monitored using a JP3 BS dichroic and D535/50M emission filter (Chroma Technology Corporation). Emission intensities were determined from the acquired images (using Slidebook's masking and statistical functions) by randomly sampling a block of 24 to 112 pixels per cellular compartment (i.e. cytoplasm, nucleus, spore or vacuole) and subtracting background. Cell lengths could be calculated directly from images using the Slidebook ruler function. Images were taken using either a 100× or 40× objective (Leica).
Flow chamber experiments
80 μl aliquots of an early log phase culture were transferred to poly-L-lysine-coated 25 mm circular coverslips. After 40-60 minutes, to allow adherence of cells, coverslips were fitted to the imaging chamber/chamber heater platform (RC-21BR/PH-2, Warner Instrument Co.), which contained∼ 400 μl of the same media. The chamber was subsequently sealed with vacuum grease and fitted to the microscope stage. All experiments (unless otherwise noted in the text) were carried out at 25°C and preceded by a 30-60 minute equilibration time during which the initial media was perfused. Temperature was controlled using a heater controller (TC-344B, Warner Instrument Co.) and in-line heater (IA SF-28, Warner Instrument Co.). Changes in media were controlled using a perfusion valve controller (VC-6, Warner Instrument Co.). Flow rates were set with a peristaltic pump (P-3, Pharmacia). The pump was halted during image acquisition. Each image was acquired from a different field of view to eliminate photobleaching as a variable.
FACS analysis was performed as described (Alfa et al., 1993).
Cells were grown in EMM pH 5.5 to early logarithmic phase (<1×106 cells/ml) and subjected to the following stresses: (1) centrifugation (3000 rpm, 5 minutes) followed by resuspension in the same media at the same cell density; (2) centrifugation (3000 rpm, 5 minutes) followed by resuspension in the same media at a density of 1×108 cells/ml for 3 hours; (3) incubation for 1 hour at 4°C; (4) centrifugation (3000 rpm, 5 minutes) followed by resuspension at the same density in sterile water buffered to pH 5.5 with 20 mM MES for 1 hour; or (5) electroporation as described (Pena et al., 1995). Cells were then transferred to poly-L-lysine-coated coverslips, incubated for a minimum time (5-10 minutes), and inverted onto depression slides before images were acquired. The effects of heat shock were assayed in the flow chamber by shifting the temperature from 25°C to 36°C. The effects of hyper-osmotic shock were also assayed in the flow chamber by shifting the growth medium from EMM pH 5.5 to EMM pH 5.5 supplemented with 1 M sorbitol. Cells were also examined by transferring 2 μl aliquots to a conventional slide and overlaying with a coverslip. Unperturbed controls were prepared as described in `Image acquisition and analysis'.
Characterizing the system
Phluorin was cloned into the fission yeast pREP1 vector under control of the thiamine-repressible nmt1 promoter (see Materials and Methods). Expression under this promoter had no observable effects in terms of cell-cycle distribution, generation time, general morphology, or the ability to regulate cell size at division (Fig. 1A,B; Table 1). Phluorin gave a strong signal for both excitation channels and was localized to the cytoplasm, the nucleus, and was excluded from the vacuolar compartment (Fig. 1). To make a phluorin standard curve, wild-type cells were permeabilized in the presence of various strong buffers adjusted to pHs at or near their pKa values. After image acquisition the emission intensities at the high and low excitation wavelengths were calculated and expressed as a ratio. A typical standard curve is presented in Fig. 1C.
Intracellular pH of wild-type cells was then determined as a function of the extracellular pH of liquid minimal growth media (EMM). Cultures were first grown to early logarithmic phase in the appropriate pH-adjusted media and the images acquired as described (Fig. 1D). Adherence to the coverslip through poly-L-lysine did not appear to have any adverse effects on the cells as judged by their ability to grow normally on the coverslip surface for extended periods of time.
Since fluorescent intensity varied greatly from cell to cell (Fig. 1D) it was critical to ensure that any observed measurements were in fact independent of phluorin levels. We thus expressed the data normalized to the most weakly fluorescing cell in each individual data set (Fig. 1E). Emission intensity varied up to eightfold, but the majority of the data points (89%) were within a fourfold intensity range of the weakest expressing cell. As expected (based on the ratiometric nature of our reporter) pHi values were independent of intensity of signal.
When expressed as a function of external pH, pHi showed a statistically significant decrease of only 0.14 pH units (from 7.31 to 7.17) over a change in external H+ ion concentration of three orders of magnitude (degrees of freedom (df)=4, T=6.93, P<0.05; Fig. 1F). These estimates are very similar to the internal pH estimated for higher eukaryotes and are generally greater than previous estimates for the budding yeast Saccharomyces cerevisiae (which have ranged anywhere from between 6 and 7.5; see Discussion). This estimate is also in a general agreement with the only other available estimate in S. pombe, an internal pH value of 7.0 estimated using carboxy-seminaphthorhodafluor-1 (SNARF) (Haworth and Fliegel, 1993).
Since our pHi measurements showed remarkable homeostasis, and to test that the system could in fact detect pHi fluctuations in vivo, wild-type cells were prepared as above and fitted into a flow chamber (see Materials and Methods). EMM was used as the initial perfusion medium and thereafter a simple valve switch allowed the medium to be changed to EMM (adjusted to a final pH of 5.5) containing either 15 mM acetic acid, or 200 mM bicarbonate (to lower and raise internal pH, respectively). As shown in Fig. 1G, positive or negative changes of 0.2-0.3 pH units were easily resolved by the system. In both cases the response was transient and the cells regulated back towards their homeostatic set point of 7.2-7.3 while remaining in the acetic acid or bicarbonate media.
Intracellular pH and the cell cycle
To determine whether pH fluctuations were associated with cell-cycle progression we first attempted to follow the internal pH of single cells over successive cell cycles. Not surprisingly, we found that repeated exposure to the high and low excitation wavelengths resulted in a cessation of growth within 4-5 hours, making the methodology unsuitable. We also observed that fission yeast cells were exquisitely sensitive to even mild stresses. Treatments such as centrifugation, temperature shifts, prolonged suspension at high density, or changes in medium osmolarity, all resulted in statistically significant changes in pHi (Fig. 2; t-tests assuming unequal variance all gave P-values <0.05). Since typical methods of syncronization (e.g. elutriation, lactose gradients, temperature-sensitive mutations) would require such treatments, we were concerned about their possible artefactual effects on pHi.
As an alternative we made use of the fact that cell-cycle position in S. pombe can be easily monitored by the measurement of cell length (Mitchison, 1957; Mitchison, 1990). This allows us to photograph populations of cells with only one exposure at each wavelength, and then to determine pHi in cells of different size. Examples of S. pombe cells in different phases of the cell cycle can be easily distinguished in Fig. 1D (see legend).
As seen in Fig. 3 (upper left panel), a typical scatter plot of an individual replicate (n=1000, mean pHi=7.32, s.d.=0.05) clearly shows that pHi is constant irrespective of cell length. Using the non-parametric Spearman's rank correlation test (since cell length is not itself normally distributed), no association between cell length and pHi was detected (df=998, R=0.008, P>0.05). The slope of the line of best fit was -0.0002 and attempts to fit 2nd to 6th order polynomials to the data did not produce curves that deviated significantly from a straight line. Furthermore, upon classifying the cells into arbitrary 1μ m size classes (Fig. 3, lower left panel), analysis of variance demonstrated that the means of the populations were not significantly different (ν1=9,ν 2=990, F=1.61, P>0.05).
Since S. pombe has an unusually long G2 phase comprising ∼70% of the cell cycle, and a G1 comprising only∼ 10% it was conceivable that a fluctuation of pHi in G1 might be difficult to resolve, and thus formally possible that a pHi change at the G1/S boundary might be missed. To examine these possibilities the same methods were applied to a strain carrying the cdc2-1w mutation (Carr et al., 1989). Strains carrying this mutation maintain a smaller cell size and have an extended G1 phase (Fig. 1B). The scatter plot of an individual replicate of this experiment (n=998, mean pHi=7.29, s.d.=0.05) is shown in Fig. 3 (upper right panel). Just as in the wildtype, a linear relationship between cell length and internal pH is observed. Spearman's rank correlation test showed no association between pHi and length (df=996, R=-0.002, P>0.05), and upon classification into 1 μm size classes (Fig. 3, lower right panel), analysis of variance showed no significant difference between the means (ν1=6, ν2=991, F=1.58, P>0.05).
Having shown that pHi changes were not associated with normal progression through the cell cycle, we next asked if an association existed between internal pH fluctuations and discrete delays, or advancements, in cell-cycle progress. To do this we made use of the fact that entry into mitosis in S. pombe is governed by a nutritionally sensitive cell size control. For example, upon shift from a growth media containing a relatively rich nitrogen source to a media containing a relatively poor nitrogen source, cells will reduce the size threshold they must attain to initiate division. In the converse shift the opposite is true (Fantes and Nurse, 1977). Glutamine (generation time (tgen)=309 minutes) and proline (tgen=475 minutes) were chosen as the rich and poor nitrogen sources, respectively, as both are non-ionizable polar amino acids and would not be expected to acidify or alkalinize the cytoplasm per se.
The results of an EMM(glutamine) to EMM(proline) nutritional shift are shown in Fig. 4 (left panel). In EMM(glutamine) cell length ranges from approximately 7-15.5 μm. Forty to eighty minutes after the shift one can see a stimulation of division as shown by the increase in the proportion of binucleate cells (open diamonds). This is a result of the cells in the upper portion of the size distribution already being over the new, reduced threshold for mitotic initiation. This is followed by a corresponding shift to the left of the cell size distribution at later time points.
In the converse shift (Fig. 4, right panel), cell size in EMM(proline) is initially distributed from approximately 6-13.5 μm. Forty to eighty minutes after the shift a delay in division is visible as the proportion of binucleate cells (open diamonds) falls to zero. This is accompanied by a rightward shift of the cell size distribution at later time points as the cells grow until they attain the new, increased size threshold for mitotic initiation. Notably, internal pH showed no significant changes in the short- or long-term during shift from glutamine to proline (ν1=13, ν2=378, F=1.14, P>0.05) or from proline to glutamine (ν1=13,ν 2=378, F=1.69, P>0.05).
Although concerned about the artefactual effects caused by shifts in temperature, as well as the possible confounding effects caused by loss of gene function, we ultimately measured pHi in various cell-cycle mutants arrested at their restrictive temperatures. These mutants included cdc10-129 (G1 arrest prior to `start'; Aves et al., 1985), cdc22-M45 (S-phase arrest; Fernandez-Sarabia et al., 1993), cdc25-22 (G2 arrest prior to mitosis; Russell and Nurse, 1986) and nda3-KM311 (mitotic arrest; Hiraoka et al., 1984). As seen in Fig. 5A, temperature-sensitive mutants as well as wild-type cells shifted to 36°C demonstrated statistically significant decreases in internal pH (analysis of variance for each strain gave P-values <0.05). However, wild-type cells maintained at 25°C showed no significant changes in pHi (ν1=6, ν2=143, F=0.138, P>0.05). Thus, the observed pHi changes are consistent with the effects of heat shock, as opposed to arrest at different points in the cell cycle. Wild-type and cold-sensitive nda3-KM311 cells maintained at 36°C and then shifted to 20°C for 8 hours showed no statistically significant differences in pHi (Fig. 5B; df=38, T=0.449, P>0.05).
Intracellular pH in stationary phase and during sporulation
We next examined the status of pHi in cells in stationary phase, or G0 of the cell cycle. Cells were grown to early log phase in EMM and then transferred to media lacking either nitrogen or glucose to allow entrance to G0. Intriguingly, when starved of nitrogen, it was found that the phluorin signal moved from the cytoplasm and nucleus, to small spherical structures located in the cytoplasm. These structures were shown to be vacuoles (Fig. 6A) by starving phluorin-expressing ade6-210 cells for nitrogen, and subsequently localizing a fluorescent intermediate of adenine metabolism (rhodamine channel) that localizes to the vacuolar compartment and accumulates in the ade6-210 mutant (Szankasi et al., 1988).
In an attempt to prevent or delay this vacuolar localization the SV40 nuclear localization signal was fused to the phluorin C-terminus. The fusion protein (phluorin-NLS) was correctly targeted to the nucleus, but strong expression from the nmt1 promoter resulted in a slower growth phenotype. This necessitated placing the construct under control of the weaker nmt41 promoter (Fig. 6B). Expression from this promoter did not produce any observable phenotypes and the signal was restricted to the nucleus.
Although one would not expect nuclear pH to differ from cytoplasmic pH based on the size of nuclear pores, there have been reports asserting pH differences between the two compartments (Dubbin et al., 1993; Seksek and Bolard, 1996). To formally test this, phluorin and phluorin-NLS-expressing wild-type cells were grown concurrently to early log phase (i.e. as a mixed population in the same culture) in order to minimize any potential variation. This was possible since phluorin and phluorin-NLS-expressing cells are easily distinguished upon image acquisition based on the localization of the GFP signal. Under these conditions no statistically significant difference between nuclear and cytoplasmic pH was observed (Fig. 6C; df=78, T=0.529, P>0.05). We thus continued with the use of the phluorin-NLS construct to provide us with an estimate of internal pH.
Upon 36 hours of nitrogen starvation, the phluorin-NLS signal was distributed in the vacuole, as well as in the nucleus (Fig. 6D), allowing both nuclear and vacuolar pH measurements to be made. Interestingly, we found that nitrogen-starved cells maintained a reduced nuclear pHi relative to logarithmically growing cells (pH 6.6 versus 7.3; Fig. 6E). We also found that these cells were able to tightly regulate their internal pH in response to wide changes in external pH. Over a change in H+ ion concentration of three orders of magnitude from pH 6.5 to 3.5, nuclear pH dropped by only 0.13 pH units. Nuclear pH in logarithmically growing cells, although maintained approximately 0.60 to 0.65 pH units higher, decreased to a similar extent (0.15 pH units) over the same external pH range. Vacuolar pH was maintained approximately 0.4 units lower than nuclear pH and was also tightly controlled over wide changes in external pH (Fig. 6E).
We next nitrogen-starved a homothallic phluorin-NLS-expressing strain to induce mating, and the formation of asci (Fig. 6F). We found that quiescent spores, just as nitrogen-starved cells, maintained a lower pHi than logarithmic phase cells, and that, remarkably, pHi decreased to an extent similar (0.05 pH units) to that seen in logarithmically growing cells over the external pH range of 6.5 to 3.5 (Fig. 6E). To ensure the validity of these results, the viability of spores and of nitrogen-starved cells was determined by plating samples to rich media. Viability as measured by colony formation was >85% at all external pHs tested.
Intriguingly, glucose-starved cells demonstrated no translocation of the phluorin signal to the vacuoles, and an even greater reduction in pHi. These cells also demonstrated a compromised ability to regulate their internal pH in response to even moderate changes in external pH (Fig. 6G). Over an external pH range of 6.6 to 5.7 the internal pH of glucose-starved cells decreased by 0.46 pH units (from 6.21 to 5.75), whereas, over the same external pH range, exponentially growing cells did not show a statistically significant difference.
Since cells starved of glucose lack both a carbon and an energy source, we repeated the experiment by growing cells in EMM, and shifting them to EMM(minus glucose) in the presence of 1% ethanol. Although fission yeast cells are unable to utilize ethanol as a sole carbon source for growth (since they lack a functional glyoxylate cycle; Fiechter et al., 1981; Tsai et al., 1987) ethanol can be used as an energy source through its conversion to acetyl-CoA and entrance into the TCA cycle (Tsai et al., 1987). Remarkably, under these conditions, glucose-starved cells behaved similarly to cells starved of nitrogen. The phluorin signal moved to the vacuoles, and the cells demonstrated a reduced, but homeostatically maintained, internal pH (as measured using the phluorin-NLS construct). As in nitrogen-starved cells vacuolar pH was maintained ∼0.4 units lower, and was also homeostatically maintained (Fig. 6G,H). These results show that the loss of homeostasis upon glucose starvation is a result of the lack of an energy source.
Since wild-type cells show a cell-cycle response as well as a downregulation of pHi when starved of a nitrogen source, we performed a short-term nutritional shift in which logarithmically growing cells were transferred from EMM to EMM(minus nitrogen) (Fig. 7). This is a nutritional down-shift analogous to the EMM(glutamine) to EMM(proline) shift performed previously and causes cells to initiate mitosis at a smaller cell size. After 40-80 minutes, entry into mitosis is stimulated as demonstrated by the increase in the proportion of binucleate cells. This is followed by a subsequent shift in the cell size distribution to the left (i.e. a decrease in cell size) at later time points. Notably, no significant change in pHi was observed in this time frame (ν1=13,ν 2=378, F=1.65, P>0.05).
Longer term nitrogen starvation studies showed that pHi begins to decrease only between 6 and 12 hours of incubation in EMM(minus nitrogen) (data not shown). Thus the decrease in pHi seen after long-term nitrogen starvation is independent of the initial changes in the regulation of cell-cycle progression. Cells starved of glucose (in the presence of 1% ethanol) do not show any changes in the regulation of cell-cycle progression, but do show a similar pHi decrease starting between 6 and 12 hours (data not shown). These data thus strongly suggest that the drop in pHi is associated with metabolic quiescence as opposed to changes in the progression of the cell cycle.
In this report we have undertaken a systematic determination of intracellular pH homeostasis as a function of cell-cycle position and growth state. These studies were intentionally performed in an optimal, unperturbed environment, and with a system that minimizes any possible artefactual effects associated with the preparation, or handling of our cellular samples.
pHi is homeostatically maintained at a value around 7.3
These studies have revealed that under optimal growth conditions intracellular pH is tightly regulated at a value of approximately 7.25-7.35. This estimate is similar to estimates from higher eukaryotes (Madshus, 1988; Ober and Pardee, 1987; Schuldiner and Rozengurt, 1982; Moolenaar et al., 1983; Musgrove et al., 1987; Kneen et al., 1998; Llopis et al., 1998) and suggests that, like other fundamental biological processes, the regulation of intracellular pH at or around this value is conserved. This is not surprising considering the extensive similarity between S. pombe and higher eukaryotes, and the importance of pHi for protein structure, stability and function (Madshus, 1988). This is also supported by examples of cross-species complementation, which suggest similarity between intracellular environments (Lee and Nurse, 1987; Campbell et al., 1995; Rodel et al., 1997; Jimenez et al., 1990).
Comparisons of our pHi estimates to those in S. cerevisiae are difficult to interpret in this respect owing to the great variation in internal pH (from 6.0 to 7.5) estimated by different researchers using different methodologies. These methodologies have included incubation (2×108 cells/ml) in nitrogen, and glucose-free buffers at 2°C for periods of up to a week (Gillies et al., 1981), washes with ice-cold water (Anand and Prasad, 1989), centrifugation and suspension in nitrogen- and glucose-free buffers at temperatures ranging from 0 to 22°C (Navon et al., 1979; Carmelo et al., 1997; Haworth and Fliegel, 1993; Sychrova et al., 1999; Pena et al., 1995), incubation at densities up to 3×109 cells/ml (Greenfield et al., 1987; Gonzalez et al., 2000), and electric shocks of 1.5-2.0 V (Pena et al., 1995; Sychrova et al., 1999).
These methodologies, together with our own data, which demonstrate the artefactual effects caused by some of these treatments, strengthen the notion that differences in methodology can have serious consequences for the accurate estimation of pHi. We believe our method to be superior. No S. cerevisiae study employing a ratiometric GFP (or other probe) to monitor internal pH in unsynchronized cells as a function of bud size or cell volume has been reported.
Data also demonstrate that robust mechanisms exist to maintain pHi at a homeostatic value of 7.3 Fig. 1F shows that artificially raising or lowering pHi by 0.2-0.3 units elicits an immediate response that returns internal pH towards initial values within 10-20 minutes. Furthermore, intracellular H+ ion concentration varies only from 4.90×10-8 M (pH 7.31) to 6.76×10-8 M (pH 7.17) over a 1000-fold increase in external H+ ion concentration, a small but statistically significant (∼38%) change in concentration.
Not surprisingly, our data also reveals that the mechanism of pH homeostasis is dependent on the presence of an energy source, since cells starved of glucose (in the absence of ethanol) demonstrate a greatly compromised ability to regulate pHi in response to even minor changes in external pH. In the absence of an energy source, H+ ion concentration shows approximately a threefold increase over a tenfold rise in external H+ concentration.
pHi regulation during the cell cycle
To determine whether pH fluctuations were associated with progression through the cell cycle, we decided to use methods in which cultures did not need to be synchronized by any of the commonly used techniques (i.e. lactose gradients, elutriation, starvation), or even the use of temperature-sensitive mutants. This was done because relatively minor stresses and perturbations could have drastic effects on pHi homeostasis (Fig. 2). Instead we made use of the simple fact that since S. pombe cells grow only by extension at their tips, it is possible to monitor cell-cycle position by monitoring cell length (Mitchison, 1957; Mitchison, 1990).
Using large sample sizes in which the pHi and length of individual cells were measured, we found no relationship between intracellular pH and cell-cycle position in wild-type cells. Similar experiments in cdc2-1w mutants, which exhibit a relatively longer G1 and shorter G2 than wildtype, also revealed no relationship between pHi and cell-cycle position. This result further supports the notion that intracellular pH is tightly regulated during logarithmic growth independently of cell-cycle-related parameters. It is thus unlikely to play any signaling or regulatory role.
We next asked if pH fluctuations were associated with induced alterations in cell-cycle progression by employing the S. pombe nutritionally sensitive cell size control. These experiments (Fig. 4) also do not support the involvement of intracellular pH fluctuations in promoting (or being associated with) delays or advancements in cell-cycle progression. Furthermore (despite the artefactual changes in pHi that arose due to heat shock), experiments with temperature-sensitive cell-cycle mutants also showed no relationship between pHi and cell-cycle position. Taking all data together, we conclude that changes in intracellular pH are not part of, and play no role in, progression through the cell cycle during logarithmic growth. By contrast, all data is consistent with a very tight regulation of internal pH around a homeostatic value of 7.3 at all points in the cell cycle.
These results are in contrast to data from several other organisms that purport to show the existence of cell-cycle-related oscillations of internal pH (Anand and Prasad, 1989; Gillies and Deamer, 1979; Aerts et al., 1985; Gerson and Burton, 1976). However, it is important to note that these studies have employed synchronized, and therefore perturbed, cell populations. Considering the sensitivity of fission yeast to different stresses, the methods of pHi measurement also raise concerns as to the possible detrimental effects on normal cellular function (these methods included treatments such as centrifugation, suspension in nitrogen- and/or glucose-free buffers, repeated insertions of micro-electrodes, heat shock, and washes with ice-cold water). In this respect, it is interesting to note that relatively mild disturbances, such as gentle centrifugation (800 g, 5 minutes), initiate the stress response in S. pombe as assayed by the activation of the spc1/sty1 stress-induced MAPK pathway (Shiozaki et al., 1998).
Furthermore, an examination of these studies reveals several fundamental differences including discrepancies in the magnitude, timing, and duration of pHi changes. Considering the conservation of the fundamental mechanisms of cell-cycle control from yeast to humans (Nurse, 1990), we believe that our results probably represent the true behavior of pHi under these conditions.
pHi regulation during changes in growth state
We next attempted to investigate pHi in cells that had entered G0 of the cell cycle through starvation for nitrogen. However, under these conditions we found that phluorin became targeted to the vacuole. This result is not surprising, and probably represents an attempt by the cell to recycle proteins under starvation conditions. Although not studied in detail in fission yeast, the vacuolar targeting and proteolysis of both cytoplasmic and plasma-membrane-bound proteins has been well documented in budding yeast in response to starvation (Van den Hazel et al., 1996; Volland et al., 1994).
The construction of a phluorin-NLS fusion allowed sufficient GFP to remain in the nucleus after 36 hours of starvation to obtain an estimate of internal pH. A signal was also present in the vacuole. These two compartments were easily differentiated and demonstrated two clear and distinct signals with pH estimates of approximately 6.2 for the vacuole, and 6.6 for the nucleus (at an external pH of 5.5). Furthermore, pHi in these compartments was tightly regulated and not very responsive to large changes in the pH of the external media. Quiescent spores demonstrated a reduction in internal pH similar to that seen in cells starved for nitrogen. Moreover, cells starved of carbon, but not of an energy source, demonstrated a similar and homeostatically maintained decrease in pHi.
A reduction in pHi in starved relative to proliferating cells has been documented in both mammalian cells (Musgrove et al., 1987), and budding yeast (Navon et al., 1979; Gillies et al., 1981). In addition, mitogen deprivation (although physiologically distinct from nutrient starvation) has been well documented to result in quiescence and a lowering of pHi (Schuldiner and Rozengurt, 1982; Moolenaar et al., 1983; Busa, 1986; Moolenaar, 1986). In this respect, it is interesting to note research that suggests a correlation of pHi to simple metabolic changes in budding yeast. For instance, Portillo and Serrano demonstrated a correlation between reduced growth rate, lowered H+-ATPase activity, and intracellular pH (Portillo and Serrano, 1989). Furthermore, Gonzalez et al. have shown internal pH to decrease from 7.5 to 6.8 upon shift from aerobic to anaerobic growth (Gonzalez et al., 2000).
Intriguingly, our short-term experiments in which the phluorin-NLS construct was used to monitor intracellular pH upon shift from EMM to EMM(minus nitrogen) showed no statistically significant intracellular pH changes. This indicates that the drop in pHi seen upon nitrogen starvation is associated with the longer term transition in metabolic state to quiescence, as opposed to the shorter term changes in cell-cycle control that advance progression into mitosis, and thereby shunt cells into G1 to prepare for mating. This hypothesis is further supported by pHi measurements in cells starved of glucose (in the presence of ethanol), which do not advance progression into mitosis, but show a similar reduction in pHi.
Cytoplasmic, nuclear pH and vacuolar pH
The phluorin-NLS construct allowed a formal testing of the question of whether nuclear pH differs from cytoplasmic pH. Although one would not expect a difference based on the sheer size of nuclear pores (which allow entry of a 27 kDa protein such as phluorin itself), there have been reports asserting differences between the two compartments, at least under certain conditions (Dubbin et al., 1993; Seksek and Bolard, 1996). Others have found no difference (Bright et al., 1987). With respect to other small, charged ions, such as calcium, a large body of literature exists, both supporting and refuting the idea that the nucleus can control or modulate Ca2+ levels independently of the cytoplasm (Macdonald, 1998; Brown et al., 1997; Lui et al., 1998; Allbritton et al., 1994; Meyer et al., 1995; Badminton et al., 1998).
Under our experimental conditions, we could detect no difference in pH between the nucleus and cytoplasm during logarithmic growth. However, it should be noted that these steady-state experiments do not rule out the possibility that the nuclear membrane buffers changes in H+ ion concentration, as has been suggested for Ca2+ (Al-Mohanna et al., 1994).
With respect to vacuolar pH measurements, it is clear that this sub-cellular compartment is homeostatically regulated in response to wide changes in external pH, albeit at a lower set point than the surrounding cytoplasm. Although no estimates are available in S. pombe, our measured values are in general agreement with the literature in budding yeast, which has yielded values of 6.0 (Carmelo et al., 1997), 6.2 (Preston et al., 1989), between 5.5 and 6.0 (Plant et al., 1999), as well as between 6.2 and 6.4, depending on the composition of external buffers (Greenfield et al., 1987). Estimates in Candida albicans yielded estimates of between 5.7 and 6.3, depending on growth form (Cassone et al., 1983).
This research was supported by the Natural Science and Engineering Research Council of Canada through grants to P.G.Y.J.K. was supported by the Bauman Foundation and by an Ontario Graduate Scholarship. We wish to thank. J. Rothman (Sloan-Kettering) for the generous gift of the phluorin probe, and an anonymous reviewer for suggesting the experiment involving glucose starvation in the presence of ethanol.
- Accepted May 18, 2001.
- © The Company of Biologists Limited 2001