SEARCH

SEARCH BY CITATION

Keywords:

  • glycolipid;
  • glycosphingolipid;
  • lumenal pH;
  • V-type ATPase

Abstract

  1. Top of page
  2. Abstract
  3. Results
  4. Discussion
  5. Materials and Methods
  6. Acknowledgments
  7. References
  8. Supporting Information

Sphingolipids are considered to play a key role in protein sorting and membrane trafficking. In melanocytic cells, sorting of lysosomal and melanosomal proteins requires the sphingolipid glucosylceramide (GlcCer). This sorting information is located in the lumenal domain of melanosomal proteins. We found that two processes dependent on lumenal pH, protein sialylation and lysosomal acid lipase (LAL) activity were aberrant in GM95 melanocyte cells, which do not produce glycosphingolipids. Using fluorescence lifetime imaging microscopy (FLIM), we found that the lumenal pH in the trans-Golgi network and lysosomes of wild-type melanocyte MEB4 cells are >1 pH unit lower than GM95 cells and fibroblasts. In addition to the lower pH found in vivo, the in vitro activity of the proton pump, the vacuolar-type H+-translocating ATPase (V-ATPase), was twofold higher in MEB4 compared to GM95 cells. The apparent Ki for inhibition of the V-ATPase by concanamycin A and archazolid A, which share a common binding site on the c-ring, was lower in glycosphingolipid-deficient GM95 cells. No difference between the MEB4 and GM95 cells was found for the V-ATPase inhibitors apicularen A and salicylihalimide. We conclude that hyperacidification in MEB4 cells requires glycosphingolipids and propose that low pH is necessary for protein sorting and melanosome biogenesis. Furthermore, we suggest that glycosphingolipids are indirectly involved in protein sorting and melanosome biogenesis by stimulating the proton pump, possibly through binding of GlcCer. These experiments establish, for the first time, a link between pH, glycosphingolipids and melanosome biogenesis in melanocytic MEB4 cells, to suggest a role for glycosphingolipids in hyperacidification in melanocytes.

Glycosphingolipids are essential in mammals, with a lack of these lipids being embryonically lethal in mice (1). Complex glycosphingolipids reach the cell surface where they act as receptors and receptor activity regulators (2). In the Golgi and endosomes, glycosphingolipids and cholesterol are believed to form microdomains that segregate membrane proteins and drive their sorting (3). The overall aim of our work is to understand the mechanistic role of glycosphingolipids in protein sorting and membrane trafficking at the molecular level.

The melanocyte cell line MEB4 and its derivative GM95 have become a model to study the role of these essential lipids in protein sorting and organelle biogenesis. GM95 cells have no active glucosylceramide synthase (GCS), the first enzyme in glycosphingolipid biosynthesis, which produces the simplest glycosphingolipid, glucosylceramide (GlcCer), resulting in a total lack of glycosphingolipids in these cells (4). GM95 cells do not assemble pigmented melanosomes and display defective sorting of melanosomal and lysosomal proteins (5,6). Transfection of GM95 with GCS and exogenous addition of glucosylsphingosine, precursor of GlcCer, restore protein sorting and melanosome formation, whereby the synthesis of higher glycosphingolipids is not required (6). Unexpectedly, the glycosphingolipid-dependent sorting signal in the melanosomal proteins tyrosinase and tyrosinase-related protein (Tyrp1) is located in their lumenal domain (6).

Interestingly, the lumen of endosomal compartments of melanocytes are exceptionally acidic (7). Melanocytes also display other distinct traits. Melanosomal proteins are transported via a direct intracellular route from the trans-Golgi network (TGN) via the endosomes to the melanosomes, whereas lysosomal proteins are rerouted over the plasma membrane (5,6,8,9). These observations suggest that pH could play a role in sorting melanosomal proteins into the intracellular route.

Here, we investigated whether there is a link between glycosphingolipids and the exceptionally low pH in several organelles of melanocytes. We compared two pH-dependent processes, protein sialylation and lysosomal lipase activity, in MEB4 and GM95 cell lines and measured lumenal pH and vacuolar-type H+-translocating ATPase (V-ATPase) activity to reveal the importance of glycosphingolipids and pH in melanosomal protein sorting and organelle biogenesis.

Results

  1. Top of page
  2. Abstract
  3. Results
  4. Discussion
  5. Materials and Methods
  6. Acknowledgments
  7. References
  8. Supporting Information

Protein sialylation depends on GlcCer synthesis

In our search for the glycolipid-dependent sorting signal in the lumenal domain of proteins, we compared lumenal characteristics of MEB4 and GM95 cells. In order to directly link the observed effects to glycosphingolipids, we also generated a GCS rescued line of GM95 in which the synthesis of glycosphingolipids was restored by transfection with wild-type GCS (GM95-GCS). Additionally, we generated MEB4 cells with a decreased ability to synthesize glycosphingolipids due to the knockdown of GCS (MEB4-aGCS) or the knockdown of lactosylceramide synthase (LCS; MEB4-aLCS). LCS is the second enzyme in complex glycosphingolipid biosynthesis and generates lactosylceramide (LacCer) by transferring a galactose moiety onto GlcCer (3).

Protein glycosylation plays a role in sorting (10) and sialylation affects the processing of the melanosomal protein PMEL (11). Therefore, to investigate the impact of glycosphingolipids on protein sialylation, we compared protein glycosylation patterns amongst the different cell lines using lectin blotting. We observed large differences in the pattern of proteins that are glycosylated in the MEB4 and GM95 cells. The proteins in GM95 cells displayed a strong reduction in their content of sialic acid-α2,3 galactose and sialic acid-α2,6 galactose/N-acetylgalactosamine terminal structures, visualized by probing total protein samples with the Maackia amurensis and Sambucus nigra lectins (Figure 1A, left and middle panel). Glycosylation of proteins was restored in GM95 cells upon transfection with the GCS (asterisks in Figure 1, left and middle panel). In some cases, glycosylation was not restored in GM95-GCS cells (solid arrows), which must reflect glycolipid-independent differences between GM95 and MEB4 cells. This is most likely caused by the procedure used to generate GM95 cells, which were selected by being resistant to killing by the antibody M2590 against the sialylated glycolipid GM3 and complement (4). Furthermore, sialylation of proteins was affected by a knockdown of GCS (aGCS), but not by a knockdown of the LCS (aLCS) (Figure 1A, right panel). These results imply that GlcCer, and not LacCer or complex glycosphingolipids, is required for proper α2,6-sialylation.

image

Figure 1. Glycolipid-dependent α2,3- and α2,6-sialylation defects in GM95 cells. A) Western blots of total protein samples of MEB4, GM95 and GM95 cells transfected with the GCS (GCS), knockdowns of MEB4 cells for lamin (aLamin), GCS (aGCS) and lactosylceramide synthase (aLCS) were probed with Maackia amurensis lectin (MAA) against sialic acid-α2,3 galactose and Sambucus nigra lectin (SNA), which binds preferentially sialic acid-α2,6 galactose. Asterisks indicate proteins that lost α2,3- and α2,6-sialylation in GM95 and MEB4-aGCS cells, but recovered glycosylation when GM95 cells regained glycosphingolipid synthesis (GCS). Solid arrows indicate proteins where glycosylation was not recovered by the restoration of glycolipid synthesis. Protein sialylation was not affected in the lamin knockdown control or in MEB4-aLCS. Equal amounts of protein had been blotted as determined by the similar signals for the melanosomal proteins Tyrp1 and PMEL (unglycosylated ER form of ∼85 kDa). Membranes from MEB4, GM95 and GM95-GCS cells contained similar amounts of LAL. B) Reduced sialylation in GM95 was not caused by a reduction in ST6Gal I mRNA, as measured by semiquantitative reverse transcriptase-PCR on equal amounts of RNA. Gapdh was used as loading control. C) The sialylation defect in GM95 cells was not caused by a difference in enzyme activity. The in vitro activity of the ST6Gal was determined in microsomal fractions from MEB4, GM95 and GM95-GCS cells. Activity is expressed as pmol CMP-[3H]-NeuAc/mg protein*min. D) The optimal pH of ST6Gal activity was determined in an in vitro activity assay on a detergent cell lysate of MEB4 cells. Activity is expressed as pmol CMP-[3H]-NeuAc/mg protein*min. E) Protein sialylation was dependent on glycosphingolipids and pH. Cells were labeled with Tran-35S-label for 6 h in the presence or absence of 20 nm ConcA to inhibit acidification. Samples of the incubation medium and cell lysates were either directly loaded on gel, or glycoproteins were first precipitated by agarose-conjugated SNA (recognizing preferentially sialic acid-α2,6 galactose and sialic acid-α2,6 N-acetylgalactosamine), eluted and analyzed by SDS–PAGE and autoradiography.

Download figure to PowerPoint

The reduction in glycosylation observed in GM95 cells was not due to a change in the concentration or enzymatic activity of α2,6-sialyltransferase (ST6Gal) I, the main sialyltransferase responsible for the α2,6-sialylation of terminal Gal residues (Figure 1B,C). In addition, no change was observed in its intracellular localization or solubility of the protein in detergent (Figure S1). In an in vitro assay using asialo-α1 acid glycoprotein as a substrate, the ST6Gal activity displayed a pH optimum close to 6. The in vitro activity of the transferase dropped steeply as the pH was moved toward neutral, with only 50% activity remaining at pH 6.5 (Figure 1D). Protein sialylation was also dramatically inhibited in vitro in cell lysates by the addition of an inhibitor of organelle acidification, concanamycin A (ConcA), which inhibits the proton pump V-ATPase ((12,13); Figure 1E, top panel). ConcA and the lack of glycosphingolipids also caused a loss in protein secretion, but transfection with GCS restored protein secretion in GM95 cells (Figure 1E, bottom panel). These results suggested that protein α2,6-sialylation is pH dependent as described previously (14). Furthermore, both the inhibition of the proton pump and lack of glycosphingolipid synthesis (GM95 cells) resulted in defective α2,6-sialylation in cells, which pointed to a possible link between glycosphingolipids and the pH in the TGN, which was pursued further.

Lysosomal lipase activity is reduced in the absence of glycosphingolipids

In order to ascertain whether other pH-dependent processes were also affected by a deficiency in glycosphingolipids, we investigated the lysis function of lysosomes through investigating lipid storage. When stained for lipid particles using Nile red, GM95 cells showed a punctate pattern that disappeared upon transfection with GCS (Figure 2A). Lipid analyses revealed that this staining pattern was due to enhanced levels of triacylglycerol (TAG) and cholesteryl esters (Figure 2B). MEB4 cells did not accumulate lipid particles unless they were pretreated with ConcA or with N-butyldeoxygalactonojirimycin (NB-DGJ), which inhibits GlcCer synthesis ((15); Figure 2A).

image

Figure 2. In vivo LAL activity is decreased in GM95 cells. A) A lack of glycosphingolipids and reduced acidification caused an accumulation of neutral lipids. MEB4, GM95 and GM95-GCS cells, as well as MEB4 cells, incubated with NB-DGJ or ConcA, were stained with Nile red to visualize neutral lipids. Scale bar = 20µm. B) Accumulation of cholesterol and cholesterol esters in GM95 cells was dependent on glycosphingolipids. Total cholesterol and cholesterol esters were quantified in MEB4, GM95 and GM95-GCS cells (left panel). Lipid extracts were separated on TLC and visualized with iodine staining (right panel). C) The activity of LAL was lower in GM95 cells. LAL activity was measured on purified membranes (left panel) and whole cells (right panel) using the lipase substrate C-POM. Kinetics of uptake was similar in MEB4, GM95 and GM95-GCS cells (data not shown).

Download figure to PowerPoint

The enzyme responsible for the degradation of TAG in lysosomes is the LAL. We investigated whether an altered expression or activity of this enzyme in GM95 cells could be the cause of the lipid storage phenotype. Similar levels of the LAL were found in total protein samples of both cell lines by western blotting (Figure 1A). The activity of the enzyme was measured using a fluorogenic substrate in purified membranes and intact cells. Although GM95 cells displayed a higher lipase activity in isolated membranes (Figure 2C, left panel), indicative of the capacity of the enzyme, activity in intact cells was lower than MEB4 cells and was restored upon addition of GCS (Figure 2C, right panel). These results indicate that although GM95 cells have a functional LAL, its in vivo activity is hampered by glycolipid deficiency. Because this lipase activity was affected by alkalinization and the lack of glycosphingolipids, we therefore postulate a relationship between organelle pH and the presence of glycosphingolipids.

GM95 cells have a higher pH in the TGN and lysosomes than MEB4 cells

Two pH-dependent features, protein sialylation and LAL activity, were affected in GM95 cells. In order to test whether a lack of glycosphingolipids causes an aberrant lumenal pH in GM95 cells, we measured lumenal pH in the TGN and lysosomes in GM95, MEB4, GM95-GCS and fibroblasts.

We determined the lumenal pH using fluorescence lifetime measurements of specifically targeted pH-sensitive fluorophores using the retrograde transport route. For the TGN, plasma membrane-localized chimeras of the TGN-resident protein TGN38 and the plasma membrane protein CD25 were targeted with fluorescein isothiocyanate (FITC)-conjugated anti-CD25 antibodies as described by Demaurex et al. (16). The chimeras and associated antibodies are internalized and localized to the TGN via retrograde transport. As shown in Figure 3A, the pH-sensitive fluorophore FITC was found in a similar compartment in MEB4 and GM95 cells. The staining partially overlapped with the trans-Golgi/TGN ST6Gal I consistent with localization in the TGN. In HeLa cells, the anti-CD25-FITC antibodies displayed an additional, punctate staining throughout the cell, as observed previously (16). The pH was determined by measuring the fluorescence lifetime of fluorescein using FLIM as described in the Materials and Methods. A linear relation between pH and lifetime was observed between pH 5 and 7 (Figure S2A) as described earlier for fluorescence ratio imaging (17). A similar calibration curve was found for all cell lines; therefore only one curve is shown for brevity.

image

Figure 3. The pH in the melanocyte TGN depends on glycosphingolipids. A) Localization of endocytosed anti-CD25-FITC in the Golgi area. Cells expressing CD25-TGN38 were incubated with rat anti-CD25-FITC antibody (TGN) overnight and co-labeled with mouse anti-myc-ST6Gal I, and visualized with Texas-red anti-mouse. Scale bar = 10µm. B) The pH in the TGN. Average lifetime values for anti-CD25-FITC in MEB4, GM95, GM95-GCS and HeLa cells, ±SD with n = 23, 23, 13 and 18 cells, respectively, were measured and the average pH value of the TGN in these cells was calculated from the pH calibration curve (Figure S2A).

Download figure to PowerPoint

When measured in living cells, the pH in the TGN in the wild-type MEB4 cells was lower (5.1 ± 0.3) than in the glycosphingolipid-deficient GM95 cells (6.5 ± 0.3; Figure 3B). When GM95 cells were retransfected with the GCS, thus restoring the ability to produce glycosphingolipids, the pH in the TGN was similar to that of the wild-type MEB4 cells (5.0 ± 0.3). The pH in the TGN of HeLa cells (6.6 ± 0.2) was similar to that found in GM95 cells, which shows that wild-type melanocyte MEB4 cells have an exceptionally low pH in the TGN.

In order to investigate whether the hyperacidification was only limited to the TGN or the same for other organelles, we measured the lumenal pH in the lysosomes as well. The lysosomal protein LAMP-1, which is recycled over the plasma membrane, was targeted with anti-LAMP-1 antibodies coupled to the pH-sensitive fluorophore Oregon Green (OG). MEB4 and GM95 cells, with mouse fibroblasts (MFs) as a control, were incubated with anti-LAMP-1-OG antibodies and the endocytosed antibody-coupled fluorophore showed complete colocalization with the endogenous lysosomal protein LAMP-2 (Figure 4A). The calibration curve was similar for the different cell lines and revealed a linear relationship between pH and lifetime between pH 3.5 and 5.5 (Figure S2B). The lysosomal pH in GM95 cells (5.4 ± 0.2) was again higher than in MEB4 cells (4.1 ± 0.1; Figure 4B). In GM95-GCS (GCS) cells the lysosomal pH was even lower than in MEB4 (3.5 ± 0.1), indicating that the very low pH in melanocyte lysosomes depends on the GCS. The fact that the control MFs again displayed a pH similar to the GM95 cells is indicative of an unusually low pH in melanocyte lysosomes in MEB4.

image

Figure 4. Glycosphingolipids also lower the pH in lysosomes. A) Intracellular localization of endocytosed OG-labeled αLAMP-1 in lysosomes. MEB4, GM95 and MFs were incubated with OG-αLAMP-1 for 5 h and co-labeled with mouse αLAMP-2, visualized with Texas-red anti-mouse. Scale bar = 10µm. B) Lysosomal pH. Average lifetime values for OG-αLAMP-1 in MEB4, GM95, GM95-GCS and MF cells, with n = 18, 18, 10 and 6 cells, respectively, were measured and the average pH of the lysosomes was calculated from the pH calibration curve (Figure S2B).

Download figure to PowerPoint

The fact that MEB4 cells have a lower lumenal pH in TGN and lysosomes could be due to a lower cytosolic pH in MEB4 cells. If the pH gradient and organelle acidification are similar in MEB4 and GM95 cells, a lower cytosolic pH can cause a lower pH in all organelles. Therefore, the cytosolic pH was tested using cell-permeant carboxyseminaphthofluorescein (carboxy SNAFL)-diacetate, which becomes trapped in the cytosol and fluorescent when cleaved by cytosolic esterases. The carboxy SNAFL-diacetate has been applied before as a pH-sensitive probe suitable for FLIM (18,19). SNAFL uptake resulted in cytosolic staining in MEB4, GM95 and HeLa cells with the probe prominently localized in the nucleus (Figure 5A). Calibration curves showed that the probe was pH sensitive over the range 6.8–7.8 (Figure S2C). Somewhat different calibration curves were recorded for the different cell lines. Using the appropriate calibration for each cell line, the pH in the cytosol was similar in MEB4 and GM95 cells (7.5 ± 0.1) and HeLa cells (7.6 ± 0.1; Figure 5B). Lifetime values for the probe localized in the nucleus did not differ significantly from measurements in the cytosol (data not shown).

image

Figure 5. The cytosolic pH is independent of glycosphingolipids. A) Intracellular localization of the cytosolic pH probe SNAFL after 10 min uptake of the non-fluorescent SNAFL-diacetate. Scale bar = 10µm. B) The pH in the cytosol. Average lifetime values for SNAFL in MEB4, GM95 and HeLa cells, with n = 7, n = 4 and n = 7 cells, respectively, were determined and converted to the average pH values of the cytosol in these cells using the corresponding pH calibration curve (Figure S2C).

Download figure to PowerPoint

In summary, a significant difference in pH was observed in the TGN and lysosomes between the glycosphingolipid-deficient GM95 and its parental MEB4 cell line but not in the cytoplasm. These results indicate that GCS activity is required for maintaining the unusual acidic pH in the TGN and lysosomes of melanocytes.

The lysosomal pH depends on glycosphingolipid synthesis

In order to find out whether glycosphingolipids directly influence lumenal pH, we measured the effect of the inhibition of glycosphingolipid synthesis on the lysosomal pH. We measured lysosomal pH after treatment with the GCS inhibitor d-threo-1-phenyl-2-decanoylamino-3-morpholino-1-propanol (PDMP, (20)). When glycosphingolipid synthesis was inhibited by addition of 10 µm PDMP for 17 h, the lysosomal pH in MEB4 cells increased toward the lysosomal pH observed in GM95 cells (Figure 6A). PDMP did not affect the pH in GM95 cells, which do not synthesize glycosphingolipids, confirming that it exerted its effect on the pH in MEB4 cells via the inhibition of glycolipid synthesis. The treatment of living cells with PDMP indeed dramatically decreased the content of GlcCer as was determined in a total cell lipid extract, and in time also decreased the more complex glycolipid monosialoganglioside GM3 (NeuAcα2-3Galα1-4Glcα1-1Cer) as shown by the acidic orcinol staining of glycosphingolipids in Figure 6B. In this set of experiments, different calibration curves (Figure S2D) and lower pH values were obtained. This result is most likely caused by a combination of factors. The total uptake of antibodies, the required measurement time and the coupling efficiency of OG to the LAMP-1-antibodies were variables that were difficult to control precisely. However, the observed differences in lysosomal pH are even larger (2 pH units) than those found previously, which further support an argument for the involvement of glycosphingolipids in controlling lumenal pH.

image

Figure 6. Glycosphingolipid synthesis is required for the low lysosomal pH in melanocytes. A) Inhibition of glycolipid synthesis with PDMP increased lysosomal pH. Cells were incubated with 10 µm PDMP for 17 h prior to and during OG-αLAMP-1 uptake. The lifetime values for OG-αLAMP-1 were measured in MEB4 and GM95 in the absence and presence of PDMP. Gaussian distributions were fitted to histograms (bin size 0.01 nanoseconds) of lifetime values, obtained in three independent experiments. The measurements for MEB4 cells, − and + PDMP, and GM95 cells, − and + PDMP, were from n = 21, 18, 19 and 19 cells respectively. The bin with the highest occurrence (number of pixels) was set at 100% and the histogram was expressed in relative occurrence (%). B) Glycosphingolipid levels were greatly decreased after PDMP treatment. MEB4 cells were treated with PDMP, and lipids were separated and stained with orcinol/H2SO4 (left panel) and iodine (right panel). An equal amount of total phospholipids was loaded. C) The V-ATPase activity, defined as the ConcA-inhibitable ATPase activity, was twice as low in GM95 total membranes than in MEB4 membranes, both in the absence and presence of 10 µm of the ionophore CCCP. The assay was done using Method II (see Materials and Methods). The V-ATPase activity is expressed in nmol Pi/mg total protein per min. D) Equal amounts of V-ATPase were present in the activity assay. Proteins of the total membranes, input of the assay, were separated on a 17% SDS–PAGE gel. The blot was decorated with rabbit anti-subunit B (Atp6b1) of the V-ATPase and mouse anti-p23 as a loading control. E) PDMP pretreatment of the cells for 17 or 36 h did not affect the in vitro V-ATPase activity of MEB4 and GM95 membranes. Values represent the average of two independent experiments in triplicate in the presence of CCCP. The assay was done according to Method II. The V-ATPase activity is expressed in nmol Pi/mg total protein per min.

Download figure to PowerPoint

In conclusion, the lumenal pH of lysosomes in MEB4 cells is increased upon inhibition of glycosphingolipid synthesis by PDMP, whereas the lysosomal pH in GM95 cells is unaffected. These results provide a direct link between glycosphingolipids and lumenal pH.

The activity of the V-ATPase is two times lower in the GM95 mutant melanocytes

Next, we investigated which pH determinant might be a factor in this mechanism. We turned our attention to the main determinant of lumenal pH in eukaryotes, the proton pump or the V-ATPase. Another reason to investigate the V-ATPase was the fact that ConcA, a specific inhibitor of the V-ATPase, caused similar effects as glycosphingolipid deficiency. The V-ATPase is a large transmembrane protein complex that acidifies organelles of the late secretory and endocytic pathways. The enzyme complex consists of a transmembrane Vo domain and a cytosolic V1 domain. The V1 domain is able to detach from the Vo domain thereby inactivating the V-ATPase, which is the main regulatory mechanism of V-ATPase activity (21).

The ATPase activity of the V-ATPase in GM95 and MEB4 membranes was investigated in an in vitro assay on total cellular membranes using a colorimetric phosphate assay. V-ATPase activity was defined as the difference in absorbance of phosphate–molybdate–malachite green complex in the absence and presence of the specific V-ATPase inhibitor ConcA (12,13). Strikingly, we found a twofold difference in V-ATPase activity between MEB4 and GM95 (Figure 6C) cells, while the amount of associated, and therefore active, enzyme complex is similar in total membranes isolated from both MEB4 and GM95 cells as determined by western blotting using an antibody that recognizes the cytosolic B subunit (Figure 6D).

The activity of the V-ATPase is inhibited by the proton gradient and membrane potential built up by proton pumping (22,23). The fact that the pH in the lumen of organelles in MEB4 and GM95 cells is different indicate that distinct proton gradients exist in the different cell lines and might explain the observed differences in V-ATPase activity. Therefore, removal of the inhibitory proton gradient and membrane potential would enable us to measure the maximum activity of the V-ATPase present in membranes from both cell lines. We used the protonophore carbonyl cyanide m-chlorophenylhydrazone (CCCP; (22)) to dissipate the proton gradient and membrane potential built up by proton pumping. The activity was stimulated fourfold by the protonophore CCCP (Figure 6C) showing that the V-ATPase activity was indeed inhibited by the proton gradient and membrane potential. Strikingly, the difference in V-ATPase activity found between total membranes from MEB4 and GM95 cells was unaffected by the addition of the protonophore while similar amounts of assembled, active V-ATPase was found in both cell lines (Figure 6D). These findings suggest that the difference in activity is caused by differences in proton pumping capacity of the V-ATPase.

Although the V-ATPase activity in MEB4 and GM95 cells correlated with the presence of glycosphingolipids, the removal of glycolipids over a time course of 30 h by PDMP had no effect on the V-ATPase activity as measured in vitro (Figure 6E). The activity found here may not reflect the in vivo situation in lysosomes as the V-ATPase activity was measured in total membranes, and not in isolated lysosomal membranes. Unfortunately, attempts to measure V-ATPase activity in isolated lysosomes have failed in our hands. The fact that PDMP did cause a less acidic pH in lysosomes of MEB4 cells in vivo, but could not significantly reduce the V-ATPase activity in vitro, indicate there may be other possible explanations for the glycosphingolipid-dependent differences observed in V-ATPase activity. First of all, the localization of the V-ATPase may be different. Because of the high abundance of V-ATPase along the secretory pathway, only quantitative electron microscopy would address the question sufficiently; however, we have not yet been able to obtain suitable antibodies. Although we could not address the possibility that the V-ATPase may be localized differently in GM95 cells, the enzyme behaved similar in both cell lines when isolated by detergent-resistant membranes (Figure S1) indicating there is no difference in membrane environment and hence localization. Second, sialylation is affected in GM95 cells, which is necessary for the proper function of the regulatory subunit Ac45 of the V-ATPase (24). We have investigated the role of sialylation in V-ATPase activity in Chinese Hamster Ovary (CHO) cells that lacked a CMP-sialic acid transporter necessary for protein sialylation (CHO lec2 cells; (25)). We did not observe an effect on the V-ATPase activity in these cells (data not shown).

In conclusion, the activity of the V-ATPase was twofold higher in the presence of glycosphingolipids. This difference in activity is due to an altered proton pumping capacity of the V-ATPase as the twofold difference was unaffected by addition of a protonophore. There are no indications that the V-ATPase is localized differently in the GM95 cells or that sialylation plays a role in V-ATPase activity. The fact that PDMP did affect lysosomal pH in vivo but not V-ATPase in vitro indicates there are possibly other pH controlling factors that are also glycosphingolipid dependent.

The affinity of the V-ATPase for ConcA is higher in GM95 cells

The difference in V-ATPase activity is independent of proton gradient or membrane potential built up by proton pumping. Another possibility is that binding of ATP is affected in GM95 cells, thereby limiting the maximum amount of hydrolysable ATP hence lowering total activity. Therefore, we tested the affinity for ATP of the V-ATPase in total membranes isolated from GM95 and MEB4 cells by measuring the ATP concentration dependence of the reaction. The maximum amount of hydrolyzed ATP was twofold higher in MEB4 membranes, which is comparable to the results shown in Figure 6; however, the Km for ATP was similar for both the wild-type cell line MEB4 and mutant GM95 cells (Figure 7A,B). This finding indicates that ATP binding was not affected by glycosphingolipid deficiency. We also tested the affinity for different inhibitors of the V-ATPase, which bind at different positions within the enzyme, to search for any indications for changes in protein structure. The results are summarized in Table 1.

image

Figure 7. The affinity of the V-ATPase for ConcA, but not for ATP, is much higher in GM95 membranes. A and B) The affinity of the V-ATPase for ATP was similar in MEB4 and GM95 cells. The V-ATPase activity of total cellular membranes isolated from MEB4 (A) and GM95 cells (B) was assayed according to Method I described in Materials and Methods, in the presence of increasing concentrations of ATP. The amount of released phosphate (Pi) from ATP hydrolysis was determined colorimetrically at OD570 for each ATP concentration, and plotted for MEB4 and GM95 cells. Absolute OD570 values at 1 mm ATP were − ConcA: 0.59 ± 0.02 and + ConcA: 0.49 ± 0.02 for MEB4; ΔOD570 = 0.10 and − ConcA: 0.40 ± 0.01 and + ConcA: 0.35 ± 0.00 for GM95; ΔOD570 = 0.05. The data points were fitted according to first order kinetics: V = (Vmax×S)/(Km + S), and the Km (ATP) was calculated to be 0.33 ± 0.14 nm for MEB4 and 0.33 ± 0.06 nm for GM95 for three independent experiments. A representative experiment is shown. C and D) The affinity of the V-ATPase for ConcA was higher in GM95 cells. The apparent Ki for ConcA of the V-ATPase was determined in membranes from MEB4 (C) and GM95 cells (D). The amount of released Pi was determined for each ConcA concentration at 2 mm ATP. MEB4: ΔOD570 = 0.23 ± 0.03; Ki = 13.8 ± 6.3 nm; GM95: ΔOD570 = 0.17 ± 0.02; Ki = 3.7 ± 1.1 (n = 3 independent experiments). A representative experiment is shown. The apparent Ki (ConcA) was calculated to be 13.6 ± 6.3 nm for MEB4 and 3.7 ± 1.1 nm for GM95 cells.

Download figure to PowerPoint

Table 1.  Apparent Ki of different inhibitorsa
InhibitorActivity (%)MEB4GM95
  1. aThe inhibitors archazolid A, apicularen A, ConcA and salicylihalamide were tested on membranes isolated from MEB4 and GM95 cells. ‘Activity’ is the activity of GM95 membranes expressed in % relative to the activity in MEB4.

  2. bn = 1 experiment in triplicate.

  3. cn = 3 experiments in triplicate.

Archazolid Ab9626.3 ± 13.68.4 ± 2.7
Apicularen Ab740.8 ± 0.61.2 ± 0.6
ConcAc7413.6 ± 6.33.7 ± 1.1
Salicylihalamideb5656.2 ± 16.370.0 ± 13.6

The apparent Ki for the inhibitor ConcA, which binds to the rotor ring of c-subunits in the transmembrane domain Vo of the V-ATPase (25,26), was much lower in GM95 membranes (3.7 nm) than in MEB4 membranes (13.6 nm; Figure 7C,D). This implies that the affinity of the V-ATPase for ConcA was three- to four-fold higher in GM95 membranes in the absence of glycosphingolipids. Both Ki values are much higher than the value of 0.2–0.3 nm reported previously, which may be due to the instability of ConcA (27).

Differences in lipid composition between MEB4 and GM95 cells may influence the partitioning of ConcA, a lipophilic molecule, into the membrane, and thereby change the concentration of ConcA to which the V-ATPase was exposed. If membranes could scavenge a significant fraction of the lipophilic ConcA, this would increase the apparent Ki of the V-ATPase for the inhibitor. We therefore tested whether the addition of equimolar amounts of MEB4 or GM95 lipids to the assay would affect the low apparent Ki of the V-ATPase in GM95 membranes. The apparent Ki for ConcA of the V-ATPase in GM95 membranes in the presence of MEB4 lipids (3.3 ± 0.9 nm; n = 3) and GM95 lipids (3.1 ± 0.8 nm; n = 1) was similar to the apparent Ki in GM95 membranes without liposomes (3.7 ± 1.1 nm). Therefore, the concentration of inhibitor available for binding was likely similar for both membranes indicating that glycosphingolipids probably interfered with inhibitor binding to the V-ATPase in MEB4 membranes.

In order to test whether the effect of glycosphingolipids on the apparent Ki was specific for ConcA, two other inhibitors of the V-ATPase that bind to subunit c in the transmembrane domain, archazolid A and apicularen A, were investigated. It has been shown that archazolid A (partly) shares the same binding site in V-ATPase with ConcA (28). The apparent Ki of archazolid A was higher in MEB4 membranes (26.3 nm) than in GM95 membranes (8.4 nm), comparable to ConcA. The precise binding site of the benzolactone enamide apicularen A has not been identified, but the compound has been shown to bind within the c-subunit at a distinctly different site compared with ConcA and archazolid A (28,29). The apparent Ki of apicularen A was similar in membranes from MEB4 (0.6 nm) and GM95 cells (0.8 nm). We also investigated salicylihalamide, another benzolactone enamide that has been postulated to bind at the neck region of the V-ATPase between the V1 and the V0 complex (30,31). The apparent Ki of salicylihalamide was slightly higher in GM95 membranes (70.0 nm) compared to MEB4 membranes (56.2 nm), but the difference was not significant (Table 1).

In summary, glycosphingolipids probably affect the inhibition of the V-ATPase activity by ConcA and archazolid A, that share a similar binding site located at subunit c in the transmembrane domain of the V-ATPase. The inhibition of apicularen A that binds to the same c-subunit, but at a different site, was unaffected, suggesting that glycosphingolipids may compete with ConcA and archazolid A for binding to the V-ATPase.

Discussion

  1. Top of page
  2. Abstract
  3. Results
  4. Discussion
  5. Materials and Methods
  6. Acknowledgments
  7. References
  8. Supporting Information

In this article, we report that the pH in the lumen of the TGN and lysosomes of MEB4 cells is significantly lower than in the glycosphingolipid-deficient GM95 cells and that this hyperacidification requires glycosphingolipids. The fact that a knockdown of the GCS, but not the LCS, induced the pH-dependent sialylation defect (and also the protein sorting and pigmentation defects, (5,6)), implies that GlcCer is necessary for the low pH in MEB4 cells. Consistent with the finding of an acidic lumenal pH, the activity of the main determinant of lumenal pH, the activity of the V-ATPase, was twofold higher in MEB4 cells compared to the glycosphingolipid-deficient GM95 cells. Therefore, we propose that GlcCer plays a role in lowering lumenal pH by stimulating proton pumping by the V-ATPase. This study describes a novel function for glycosphingolipids in controlling an important cellular parameter, pH, in MEB4 cells. Additionally, an acidic lumenal pH in the TGN was identified as a possible requirement for the formation of melanosomes and a role for pH in protein trafficking in general.

We observed that GM95 cells display highly reduced levels of protein sialylation (Figure 1). For some proteins, sialylation in MEB4 was reduced by knockdown of GCS (but not the LCS) and the sialylation of exactly these proteins was restored in GM95 by transfection with the GCS. Similarly, we discovered that cells with a reduced or absent GCS activity displayed a lipid storage phenotype that was cured by transfection with the GCS (Figure 2). The induction of both the sialylation and lipid storage phenotypes by ConcA, which inhibits the proton pump, and the pH-dependent activity of both the Golgi sialyltransferase ST6Gal I (Figure 1D) and the LAL (32,33) suggested a reduced acidity in the organelles of cells with low or no GCS activity. Direct pH measurements confirmed this notion (Figures 3 and 4). It remains unclear why sialylation and lipase activity are normal in MEB4, which have an unusually acidic pH, and not in GM95, which have a pH in their TGN and lysosomes that is similar to the pH found in HeLa and MF cells.

Compared to non-melanocytic cells, such as HeLa and MFs, the MEB4 cells have an unusually low pH in the TGN and the lysosomes (Figures 3 and 4). Whether a low lumenal pH is a typical trait of melanocytes remains to be determined. However, it has been found previously that melanosomes are acidic and an unusual acidic pH is found in vacuolar early endosomal compartments in human melanocytes (7,34). Initial results also indicate a similarly low lysosomal pH in another mouse melanocyte cell line, Melan-a (data not shown). In addition, others have observed that inhibition of lysosomal degradation in melanocytic cells by traditional compounds such as methionine methyl ester and ammonium chloride did not work indicating that a higher pH barrier must be overcome (M. Marks, personal communication). The precise function of the low pH is unknown, but proteolytic cleavage of PMEL, necessary for melanin production, may require a low pH environment (35). Here, we also describe an unusual acidic pH in the TGN in melanocyte MEB4 cells indicating that an unusually low pH may be required for sorting melanosomal proteins.

The observed hyperacidification in MEB4 cells required the presence of glycosphingolipids, as lysosomal pH increased when the synthesis of glycosphingolipids was inhibited by PDMP in MEB4 cells, but not in glycosphingolipid-deficient GM95 cells (Figure 6A). The fact that glycosphingolipids and lumenal pH are interlinked properties of melanocytes complicates the separation of the contribution of these two factors in protein sorting and melanosome biogenesis. Earlier observations from GM95 cells and fibroblasts, however, may give us some clues. As the TGN and lysosomes of GM95 cells and fibroblasts have a similar pH (Figures 3 and 4), and fibroblasts like HeLa cells are known to contain glycosphingolipids, we are able to segregate the effects of GlcCer and lumenal pH. In the absence of glycosphingolipids, tyrosinase accumulates in the Golgi complex in GM95 cells, while in fibroblasts tyrosinase is sorted to the late endosomes/lysosomes (5,36). These results suggest that glycosphingolipids play a direct role in protein sorting.

Interestingly, when tyrosinase was able to escape the sorting block in the Golgi in GM95 cells, pigment production occurred, but no mature melanosomes were formed (6). Similarly, fibroblasts transfected with tyrosinase can produce low levels of pigment, but the cells did not form melanosomes (5,36). One may argue that other determinants for melanosome formation are absent in fibroblasts, but all the necessary factors are present in the GM95 cells, suggesting that lumenal pH is necessary for melanosome biogenesis.

Melanosome formation is preceded by the accumulation and processing of PMEL, a structural component of melanosomes in endosomal structures (37,38). Analogous to the requirement of acidification in the TGN for formation of regulatory secretory granules (39), the extremely low lumenal pH in the TGN of melanocytes may be necessary for segregation of PMEL from other proteins in the TGN and for formation of specialized endosomes destined to develop into melanosomes. In this process, glycosphingolipids and the lowered lumenal pH may act together. We have observed before that melanosomal protein sorting via the intracellular transport route required an adaptor protein, AP-3, in MEB4 cells (6). The fact that the GlcCer-dependent sorting signal is located in the lumenal domain of melanosomal proteins suggests that pH may play a role in cargo recognition. Glycosphingolipids are required to acidify the lumen, but may also be required to provide the optimal membrane environment, possibly facilitating oligomerization of melanosomal proteins in order to be recognized by AP-3.

Our findings presented in this study indicate that glycosphingolipids lower pH via stimulation of proton pumping through the activity of the main pH determinant in organelles, the V-ATPase (Figure 6). The pH gradient in the secretory pathway is regulated at multiple levels (39,40). The lumenal pH in organelles is determined by the balance among the active pumping of protons, the passive leak of protons and the conductance for counterions. It is unlikely that the reduced acidification in GM95 cells is due to the latter two factors. First of all, the presence of sphingolipids was found to increase proton permeation of membranes (41). Second, ion conductance seems to play only a minor role in determining organelle acidity (40,42). More importantly, removal of the proton gradient and membrane potential by the protonophore CCCP, which also removes the influence of proton leakage and counterion influx on proton pumping, did not affect the difference in V-ATPase activity found in closed membranes derived from GM95 and MEB4 cells while the amount of assembled, active V-ATPase was comparable in both cell lines (Figure 6C,D). This result indicates that an intrinsic property of the V-ATPase is affected by a lack of glycosphingolipids. We have found no indications that an aberrant protein sialylation that occurs in GM95 cells would affect V-ATPase activity or that the localization of the V-ATPase is different in GM95 cells compared to MEB4 cells (Figure S1B).

The question arises how glycosphingolipids can stimulate proton pumping. It has been reported previously that sphingolipids with a C26 acyl group are required for generating an active V-ATPase in yeast (43), and that the activity of the purified tonoplast proton pump of plants is increased by steryl glucoside (43–45). These findings suggest that the activation of the V-ATPase by specific lipids may be a general phenomenon across all eukaryotes. Direct addition of GlcCer to GM95 membranes would test whether a comparable mechanism exists in mammalian cells. Unfortunately, we were unable to reproducibly stimulate the V-ATPase by direct addition of GlcCer to isolated GM95 membranes, possibly due to experimental difficulties in delivering glycosphingolipids to membranes. However, the glycolipid-dependent inhibition of ConcA and archazolid A suggests changes in their common binding site in the transmembrane domain of the V-ATPase (Figure 7C,D and Table 1). Glycosphingolipids are known to affect the physical properties of a membrane, and may have affected the structure of the transmembrane domain of the V-ATPase and therefore the ConcA binding site. However, the fact that the inhibition of apicularen A, which binds the same transmembrane subunit of the V-ATPase, was unaffected suggests that glycosphingolipids may compete with ConcA and archazolid A for their common binding site and directly bind the V-ATPase.

The V-ATPase activity in MEB4 cells required glycosphingolipids; however, it is unlikely that a twofold increase in V-ATPase activity is solely responsible for a pH drop of 1 pH unit, as this represents a tenfold increase in proton concentration. This result suggests there may be other factors that are glycosphingolipid dependent. Supporting this notion is the fact that inhibition of glycosphingolipid synthesis by PDMP increased the lysosomal pH in vivo, but did not decrease V-ATPase activity in vitro. Additionally, we have found that the transfection of GCS in GM95 cells did not consistently increase V-ATPase activity. The generation of a stable GM95 cell line uniformly expressing GCS, which is required for V-ATPase activity measurements in total membranes, was not possible as GM95 cells gradually loose the acquired phenotypes. Other melanocyte-specific factors that may influence lumenal pH include the oculocutaneous albinism type 2 (OCA2) protein (46). Mutations in OCA2 protein cause the most common hypopigmentation disease OCA2 in Caucasians (47). The function of OCA2 has not been characterized in detail, but resembles an anion channel (48). This protein may play a role in pH homeostasis in vivo by an anion influx counteracting the charge build-up of protons, enabling the V-ATPase to pump more protons into the lumen. However, the fact that counterion influx does not play a major role in organelle acidification (40,42) seems to contradict this notion. Further research is necessary to identify all glycosphingolipid-dependent factors in pH regulation in melanocytes. The presence of other melanocyte-specific factors in lowering lumenal pH by glycosphingolipids may explain the fact that melanocytes display hyperacidification compared to fibroblasts, which do contain glycosphingolipids.

The stimulatory effect of glycosphingolipids on the activity of the V-ATPase presents a compelling regulatory mechanism. The ConcA binding site resides in the membranous c-subunits (26,49). This may explain the lack of glycosphingolipids also affecting the pH of the lysosomes (Figure 4), where the V-ATPase has the same c-subunits. The ConcA binding site resides on the cytosolic side of the c-subunit (26,49). GlcCer is synthesized on the cytosolic surface of a Golgi membrane (50–52). At the same time, GlcCer is continuously removed from that surface moving to the lumenal side of cellular membranes for the synthesis of complex glycosphingolipids by various mechanisms, details of which are not well understood (52–54). This scenario suggests that ongoing GlcCer metabolism may be required for maximal stimulation of V-ATPase activity.

Materials and Methods

  1. Top of page
  2. Abstract
  3. Results
  4. Discussion
  5. Materials and Methods
  6. Acknowledgments
  7. References
  8. Supporting Information

Materials

Chemicals, unless stated otherwise, were from Sigma-Aldrich and used in the highest purity available. Cell culture media, reagents, antibiotics and fetal bovine serum were from PAA laboratories. All lipids and lipid standards were from Matreya, and were stored as stock solutions in CHCl3/MeOH at −20°C. Phosphatidylcholine (PC) in CHCl3, isolated from egg yolk (Grade I), was from LipidProds. ConcA was from Wako and the inhibitors Apicularen A, Archazolid A and Salicylihalamide were kind gifts from H. Wieczorek and M. Huss (University of Osnabrueck).

DNA constructs

The CD25-TGN38-pCDM8.1 construct, encoding the lumenal domain of human CD25 and the transmembrane and cytosolic domains of TGN38, was a gift from F. Maxfield (Cornell University). Myc-tagged ST6Gal-pCB7 and GCS-KKVK-pCB7 were made as described previously (5).

Cell culture, transfection and fluorescent labeling

GCS-deficient GM95 and their parental MEB4 cells were from RIKEN Cell Bank and were grown in high glucose DMEM containing 10% FBS at 37°C with 5% CO2. HeLa cells (G. Warren) and MFs WT1.2 cells (J. Wijnholts, The Netherlands Cancer Institute) were grown under the same conditions as GM95 and MEB4 cells. GM95 cells were stably transfected with GCS-KKVK-pCB7 using lipofectamine 2000 (Invitrogen Corporation). Selection was done with hygromycin B (0.6 mg/mL) and individual clones were selected using limiting dilution subcloning. Stable transfectants were screened by assaying the GCS activity as previously described (5). For pH measurements in the TGN, cells were grown on 3 cm glass bottom dishes (Mattek Corporation) and transiently transfected with CD25-TGN38-pCDM8.1 using lipofectamine 2000. After 5 h, transfection medium was changed to normal growth medium with anti-CD25-FITC (5 nm) and the incubation continued for 16 h. For lysosomal pH measurements, anti-LAMP-1 labeled with OG-514 carboxylic acid succinimidyl ester (anti-LAMP-1-OG) was added to untransfected cells and incubated for 5 h in the presence of 10 µg/mL leupeptin. For pH measurements in the cytosol, cells were incubated for 20 min with 5 µm carboxy SNAFL-diacetate in Hank's buffered salt solution (HBSS), pH 7.4. For FLIM, cells were measured in DMEM high glucose without phenol red, 25 mm HEPES pH 7.4 without FBS or in HBSS pH 7.4 for measurements of cytosolic pH.

Antibodies and immunofluorescence

The rat-conjugated anti-CD25-FITC antibody against human CD25 was from Serotec. The polyclonal rabbit anti-human CD25 antiserum was a gift from M. Marks (University of Pennsylvania. Rabbit antiserum anti-PEP1 against the cytoplasmic tail of Tyrp1 and the anti-PEP13 antibody against the cytosolic tail of PMEL were a gift from V. Hearing (NIH). The monoclonal anti-c-myc 9E10 antibody, the rat 1D4B antibody against LAMP-1 and mouse monoclonal anti-HA were from Santa Cruz Biotechnology. The monoclonal anti-LAMP-2 antibody ABL-93 was from Developmental Studies Hybridoma Bank. OG and 5-(and 6-)carboxy SNAFL-1 diacetate were from Molecular Probes. OG was coupled to anti-LAMP-1 antibody (1D4B) following the instructions by Molecular Probes. The rabbit polyclonal antibody recognizing subunit B of the V-ATPase was a gift from M. Forgac (Tufts University) and the mouse monoclonal anti-p23 (JJ3) antibody was from Abcam. Horseradish peroxidase (HRP)-conjugated secondary goat anti-rabbit and mouse IgGs were from Dako. The secondary antibody rat-FITC was from Santa Cruz, other secondary antibodies for immunofluorescence were Alexa-conjugated from Molecular Probes. The immunofluorescence procedure was performed as described before (6).

SDS–PAGE and western blotting

Aliquots of membrane preparations were mixed with SDS sample buffer [200 mm Tris–HCl pH 6.8, 3% (w/v) SDS, 12% (v/v) glycerol, 1 mm ethylenediaminetetraacetic acid (EDTA), 0.003% (w/v) bromophenol blue and 1% 2-mercaptoethanol, final concentrations], boiled for 30 seconds at 95°C and separated by SDS–PAGE. For western blotting, polyvinylidene fluoride transfers were blocked for 1 h with 2% hen egg albumin and 0.01% (w/v) Tween in PBS. Tyrp1, PMEL, the V-ATPase B subunit (Atp6b1) or p23 were detected by appropriate primary antibodies. HRP-conjugated secondary antibodies and enhanced chemiluminescence (Amersham) were used for detection. Digoxigenin-labeled lectins were from Roche. The lectins were visualized using anti-digoxigenin antibodies (Roche) coupled to alkaline phosphatase followed by diaminobenzidine staining. For the flotation of detergent-resistant membranes, MEB4 and GM95 cells, transiently transfected with the myc-ST6Gal I, were lysed in 1% cold Lubrol-WX, followed by flotation of detergent-resistant membranes in a sucrose step gradient as described before (5).

Reverse-transcriptase polymerase chain reaction

Cells grown on 10 cm dishes were washed with PBS, harvested using trypsin/EDTA and pelleted in PBS. RNA isolated from cell pellets was stored at −80°C using the RNAII isolation kit from Machery-Nagel (Bioké). Isolated RNA (1–2 µg) from cells or mouse brain or testis was then subjected to reverse transcription in the presence of oligodeoxythymidilic acid12-18 primer (First strand cDNA synthesis kit; Amersham Biosciences) for cDNA synthesis in a final volume of 33 µL according to the manufacturer's instructions. The oligonucleotides used as primers for the polymerase chain reactions (PCRs) to study the expression of mST6Gal I and Gapdh are described by Takashima et al. (33). Fragments were amplified as already described using Taq & Go mastermix (MP Biomedicals); the amplified fragments were analyzed by electrophoresis on a 1.5% Tris-borate-EDTA (TBE) agarose gel containing ethidium bromide.

Sialyltransferase and LAL activity assays

ST6Gal activity was determined in vitro, using microsomal fractions from GM95, GCS and MEB4 cells as a source of enzyme, and asialo-α1 acid glycoprotein as a substrate, as this protein contains several terminal Gal residues on N-glycans which are substrates for ST6Gal. Sialyltransferase assays were performed as already described (55). Briefly, enzyme activity was measured in 0.1 m cacodylate buffer, 10 mm MnCl2, 0.2% Triton CF-54, 50 µm CMP-[14C]NeuAc (1.85 KBq), with 100 µg of asialo-α1 acid glycoprotein and 10 µL of the enzyme source in a final volume of 50 µL. The reactions were performed at 37°C for 4 h. The reaction was terminated by precipitation and filtration as previously described. The samples were separated by SDS-PAGE to analyze the sialic acid linkage after sialidase treatment.

The activity of the lysosome acid lipase was assayed in whole cells and purified membranes using the lipase substrate C-POM (O-pivaloyloxymethyl umbelliferone; Invitrogen Corporation) (56). Lipase activity produces a fluorescent hydrolysis product. The experiment was conducted according to manufacturer's instructions. Cells were washed extensively, scraped and cellular fluorescence was quantified in a spectrophotometer. Kinetics of uptake was assayed by the addition of Texas Red-labeled Ovalbumin and FITC-labeled dextran (both from Molecular Probes) to the medium (data not shown).

Lipid analysis

Cells were treated with 10 µm PDMP either for 17 h or 3 days in DMEM and 10% FBS. Cells were washed with PBS, harvested using trypsin/EDTA, pelleted and subsequently taken up in PBS. Lipids were extracted using the method described by Bligh and Dyer (57). In short, 3.2 volumes of methanol:chloroform (2.2:1) were added and left at room temperature for 30 min with occasional vortexing. For phase separation, 1 volume of 120 mm KCl with 10 mm acetic acid (no pH correction) and 1 volume of chloroform were added and centrifuged at 3000 ×g in a Beckman Coulter for 10 min at room temperature. The upper aqueous phase was isolated and washed with 2 volumes of chloroform. The organic phases were pooled. An equal amount of phospholipids, as determined by phosphate determination (58), was separated on a thin-layer chromatography plate using chloroform:acetone:methanol:acetic acid:H2O (50:20:10:10:5 v/v) as running solvent. Glycosphingolipids were visualized using 0.2% orcinol in 5 m H2SO4 and incubated in a 100°C oven (59). The bands at the height of the major phospholipids are probably due to a low level of direct staining of the phospholipids by orcinol. Cholesterol and cholesterolesters were quantified using the Amplex®Red Cholesterol Assay Kit according to manufacturers instructions (Molecular Probes).

FLIM and statistical analysis

Lifetime measurements were performed using a confocal laser scanning microscope (CLSM; Nikon PCM2000). For excitation, 460 nm pulsed light was used. Therefore, the CLSM was equipped with a Tsunami Titanium:Sapphire laser (Spectra-physics; Newport Corporation) that produced 2 picoseconds light pulses at 920 nm with a repetition rate of 82 MHz. For our purposes every 10th pulse was picked with a pulsepicker and the 920 nm was frequency-doubled to 460 nm using an lithium triborate (LBO)-crystal. The emitted fluorescence, selected with appropriate chromatic filters, was detected using a fast GaAsP photon-counting photomultiplier tube (PMT) (Hamamatsu; H7421-40, with a 450 picoseconds transient time spread). The output pulses from the PMT were coupled to a time-gated lifetime module, with four time-gates. The four time-gate widths were set to 2 nanoseconds each without delay between the time-gates. The start of the first time-gate was delayed until the intensity of a fast decaying dye in the first time-gate was reduced to 10% of the maximal obtainable signal (Rose Bengal, τ = 90 picoseconds; Sigma). In order to collect a stack of four time-gated intensity images of 160 × 160 pixels, cells were scanned at dwell times of 3 milliseconds per pixel using the 50 µm pinhole of the CLSM and a 60× water immersion objective (Nikon, PlanApo, NA 1.2). The fluorescence lifetime (τ) per pixel was determined by fitting the four time-gated intensities [I(t)] per pixel, corrected for the background, with a single-exponential decay function: I(t) = I(0) et/τ. These lifetimes were depicted in a lifetime image. The regions with the highest fluorescence intensity were selected by an intensity threshold. The intensity cut-off was set at the point where the fluorescence resembled a compact signal for the TGN and a punctate signal for lysosomes. The average lifetime of the fluorescence in these pixels was determined for an individual cell from a Gaussian fit of the lifetime histogram. Typically, some 200 pixels were measured per cell, and the average of the mean value of some 20 cells was calculated ±SD. No significant differences in average lifetimes were found when lifetimes of each pixel were weighted or unweighted with their pixel intensities; therefore, we omitted this weighing factor from our calculations. The data was either represented (1) in bar diagrams, where the average (±SD) is the average of all lifetime histograms, individually fitted to a Gaussian (Figures 3B, 4B and 5B) or (2) in a lifetime histogram, with bin size 0.01 nanoseconds, where the calculated lifetimes of the selected pixels of all experiments are taken together and fitted to a Gaussian (Figure 6B). For each cell line, lifetimes were measured on at least two independent experimental days. No statistically significant differences between days were found.

pH calibrations

The pH values were calculated using calibration of the fluorescence lifetime of the various probes versus the pH. These curves were generated using the high potassium/nigericin method (60). For measurements in the TGN, cells were washed twice in calibration buffer (100 mm KOAc, 50 mm KCl, 1 mm MgCl2 and 5 mm glucose) set at a pH value ranging from 5.0 to 7.0, followed by the addition of the H+ and K+ ionophore nigericin (13 µm final concentration). After 5 min, the fluorescence lifetime was measured in at least three different cells. Lysosomes were calibrated in the same calibration buffer with pH values ranging from 3.5 to 5.5. The cytosolic pH was calibrated in phosphate buffer (100 mm K2HPO4/KH2PO4, 50 mm KCl, 5 mm MgCl2 and 5 mm glucose) ranging from pH 6.8 to 7.8. Average lifetimes were plotted against pH. Error bars: sample standard deviation (SD).

Membrane preparation for ATPase assays

Cells were grown to 70% confluency on 15 cm dishes and treated with 10 µm PDMP for 0 h, 17 h or 3 days. Cells were washed, scraped and homogenized through a 23 Gauge needle in buffer (0.25 m sucrose, 20 mm HEPES–KOH, 1 mm EDTA, pH 7.4 with protease inhibitors: 1 µg/mL aprotinin, 1 µg/mL leupeptin, 1 µg/mL pepstatin, 5 µg/mL antipain and 1 mm benzamidine). Debris and whole cells were removed with a 1000 ×g spin for 5 min at 4°C. Total membranes were isolated from the post-nuclear supernatant using 20 000 ×g for 30 min at 4°C. The pellet was taken up in 150 µL buffer (20 mm HEPES–KOH, 150 mm K+-glutamic acid and 5 mm MgSO4, pH 7.4 with protease inhibitors) and protein content was determined with amidoblack using BSA as a standard (61). The samples were diluted to 0.1–0.25 mg/mL total protein.

ATPase assay

The assay was performed according to one of the following procedures. For all determinations of Ki and Km, we used Method I. The V-ATPase activity was determined of 4 µg total protein in 25 µL 10 mm HEPES–KOH pH 7.4, 75 mm K+-glutamic acid, 5 mm MgCl2, 2 mm ATP, 2 mm NaN3 and 0.1 mm vanadate, and incubated at 37°C for 30 min. The reaction was stopped with 175 µL 40 mM H2SO4 on ice. The amount of inorganic phosphate was determined by the addition of 0.7 m H2SO4, 0.03% malachite green and 0.3% ammonium molybdate, final concentrations, and reading of the optical density at 570 nm after 20 min. Phosphate concentrations were calculated from a NaH2PO4 calibration curve.

Method II used in the other determinations was adapted from Refs (62) and (29). Briefly, V-ATPase activity was determined using 4 µg total protein in 20 mm HEPES–KOH pH 7.4, 60 mm K+-glutamic acid, 2 mm MgCl2, 6.25% DMSO, 2 mm NaN3 and 0.1 mm vanadate, because activities were 1.3-fold higher when glutamate buffers were used for membrane isolation and the ATPase assay than in the original Tris–3-(N-morpholino)propanesulfonic acid (MOPS) buffer. The sample was incubated for 15 min at 37°C with 1 mm ATP (160 µL total volume) and frozen in liquid N2 to stop the reaction. For the determination of the amount of inorganic phosphate (Pi) generated, protein was first precipitated with 5% trichloroacetic acid and centrifuged 20 000 ×g for 1 min. One third of the supernatant was mixed with 24% H2SO4:200 mm Na-molybdate:H2O (1:3:6 v/v) to final concentrations of 1.5% H2SO4 and 37.5 mm Na-molybdate. Then, 1% polyvinyl alcohol with 0.074% malachite green was added and finally 7.8% H2SO4. Final concentrations were 15 mm Na-molybdate, 0.08% polyvinyl alcohol and 0.006% malachite green and 4.6% H2SO4. The absorbance at 625 nm was read after 90 min and quantified with a calibration curve of NaH2PO4. Generally, activity measurements were performed in the presence of 5 mm Mg2+ and 1–2 mm ATP, concentrations that are >5-fold higher than the respective Km of the V-ATPase (43). The V-ATPase activity is defined as the difference in activity in the absence and presence of the specific inhibitor ConcA: the ConcA-inhibitable activity (21,23). The V-ATPase activity was determined in the absence and presence of 10 µm of the protonophore CCCP. NaN3 and vanadate reduced background mitochondrial and transport ATPases by some 80–90%, allowing the assay to be performed in a more accurate OD range: in their absence, total ATPase activity was five times the V-ATPase activity. In order to determine the apparent Ki, the absorbance was plotted against inhibitor concentrations. The relation for first order kinetics according to Michaelis–Menten, V = (Vmax×S)/(Ki + S), was used to fit the datapoints.

Addition of MEB4 and GM95 lipids to GM95 membranes

Lipids were extracted from MEB4 and GM95 cells by a method adapted from Bligh and Dyer (57). After the extraction of the lipids in the organic phase, the waterphase was run over a SepPak C18 cartridge (Waters) in order to isolate the higher glycosphingolipids. The column was washed with water and eluted with methanol. The eluate was added to the organic phase of the extraction and the lipids were dried under a stream of N2. Liposomes were formed by hydration of the dried lipids in buffer (20 mm HEPES–KOH, 150 mm KCl and 5 mm MgCl2, pH 7.4), vortexing and intermittent sonication for 10 min on ice with a Branson Sonifier 450 (50% duty cycle, output 5) to homogenize the sample. Ten freeze/thaw cycles with liquid N2 and a waterbath at 40°C were used to generate larger vesicles. Finally, the liposomes were sized by 20 passages through a 0.4 µm polycarbonate filter mounted on an Avanti syringe-based extrusion device. An equimolar amount of MEB4 liposomes was added to GM95 membranes by phospholipid content as determined by phosphate determination according to Rouser et al. (58).

Acknowledgments

  1. Top of page
  2. Abstract
  3. Results
  4. Discussion
  5. Materials and Methods
  6. Acknowledgments
  7. References
  8. Supporting Information

We are grateful to Helmut Wieczorek, Markus Huss, Vince Hearing, Michael Marks and Michael Forgac for their generosity in providing reagents. We are grateful to Joep van den Dikkenberg for his excellent assistance in the lab. We thank our colleagues and Richard Newcomb for critical reading of the manuscript and helpful discussions. The majority of the work was performed in the laboratories of Membrane Enzymology, Utrecht University. This study was supported by grant 902-23-197 from ZonMW with financial support from NWO to J. W., S. G.-D. and G. M., and by ECHO grant 700.54.011 from CW-NWO to H. S. and G. M. with financial aid from the Netherlands Organization for Scientific Research. The authors declare no conflict of interest. We are grateful to the editor and the two anonymous reviewers for their helpful and constructive suggestions, which considerably improved the manuscript.

References

  1. Top of page
  2. Abstract
  3. Results
  4. Discussion
  5. Materials and Methods
  6. Acknowledgments
  7. References
  8. Supporting Information

Supporting Information

  1. Top of page
  2. Abstract
  3. Results
  4. Discussion
  5. Materials and Methods
  6. Acknowledgments
  7. References
  8. Supporting Information

Supporting Information

Additional Supporting Information may be found in the online version of this article:

Figure S1: Similar microscopic distribution and detergent solubility of ST6Gal I in MEB4 and GM95 cells. A) Cellular distribution of ST6Gal I was similar in MEB4, GM95 and GM95-GCS cells. MEB4, GM95 and GM95-GCS cells were transiently transfected with myc-tagged ST6Gal I, and ST6Gal I and the medial Golgi marker CTR433 were immunolocalized by confocal microscopy (5). Scale bar = 10µm. B) Detergent solubility of ST6Gal I was similar in MEB4 and GM95 cells. MEB4 and GM95 cells, transiently transfected with myc-tagged ST6Gal were lysed in 1% cold Lubrol-WX, followed by flotation of detergent-resistant membranes in a sucrose step gradient. Fractions of the gradient between the top (T) and bottom (B) were analyzed for the various proteins by western blotting. Although the raft marker Yes and the V-ATPase subunit Atp6b1 localized to the detergent-resistant membranes, Tyrp1 and ST6Gal I were detergent soluble.

Figure S2: Calibration curves of probes in TGN, lysosomes and cytosol using FLIM. A) The pH calibration of anti-CD25-FITC in MEB4 cells using FLIM in the TGN. Calibration measurements on cells were carried out after 10 min incubation in the presence of nigericin (13 µM) in 150 mM K+ buffers varying in pH between 5.0 and 7.0 at 37°C. Curves measured in MEB4, GM95 and HeLa cells were overlapping (not shown). For pH calculations, a linear fit was used. B) The pH calibration of OG-αLAMP-1 in MEB4 cells using FLIM in lysosomes. Calibration measurements were carried out in K+ buffers at pH values between 3.5 and 5.5 at 37°C. Calibration curves measured in MEB4, GM95 and MF cells were overlapping (not shown). The line is a linear fit used for pH calculations. C) The pH calibration of SNAFL in MEB4, GM95 and HeLa cells in the cytosol. Calibration measurements were done in phosphate buffers varying in pH between 6.8 and 7.8 at 37°C. The lines are linear fits of the data and used for pH calculations. D) The pH calibration curve in MEB4 and GM95 cells in lysosomes. The linear fit of the calibration curves was used to calculate pH in the presence and absence of PDMP (Figure 6A).

FilenameFormatSizeDescription
TRA_1263_sm_s1.pdf101KSupporting info item
TRA_1263_sm_s2.pdf60KSupporting info item

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.