The glycosphingolipid sulfatide in the islets of Langerhans in rat pancreas is processed through recycling: possible involvement in insulin trafficking

Glycobiology, Jan 2000

In previous studies we have shown that sulfatide (galactosylceramide-3-O-sulfate), in various species, is present in the insulin-producing cells in pancreatic islets of Langerhans. In this study the synthesis of sulfatide in the islets has been investigated by pulse chase labeling at varying glucose levels and in the presence or absence of the glycosphingolipid synthesis inhibitory agents, Brefeldin A, fumonisin B1 and chloroquine and the distribution of sulfatide by immune-electronmicroscopy. The data showed that (1) sulfatide was produced in islets of Langerhans, (2) the main pathway for synthesis was through recycling involving partial degradation in the lysosome, and that (3) high glucose levels, although not primarily reflected in an increased synthesis of sulfatide, lead to an increased expression of mRNA for the UDP-galactose:ceramide galactosyltransferase, producing the immediate precursor of sulfatide. Furthermore, mass spectrometry analyses revealed a high proportion of short chain fatty acids, C16:0 (50%) and no hydroxylated forms and thus special physicochemical properties, indicating important differences between pancreatic and brain/neural sulfatide. Immune electron microscopy revealed an intracellular expression of sulfatide in the secretory granules, the Golgi network and the lysosomes of the islets. These results indicate that sulfatide follows the same intracellular route as insulin and suggest a functional association between these molecules. We have raised the hypothesis that sulfatide possibly plays a role in the trafficking of insulin in the islets of Langerhans in rat pancreas.

A PDF file should load here. If you do not see its contents the file may be temporarily unavailable at the journal website or you do not have a PDF plug-in installed and enabled in your browser.

Alternatively, you can download the file locally and open with any standalone PDF reader:

https://glycob.oxfordjournals.org/content/10/1/39.full.pdf

The glycosphingolipid sulfatide in the islets of Langerhans in rat pancreas is processed through recycling: possible involvement in insulin trafficking

Pam Fredman 2 Jan-Eric Mnsson 2 Britt-Marie Rynmark 2 Knud Josefsen 1 2 Annette Ekblond 1 2 Linda Halldner 0 2 Thomas Osterbye 1 2 Thomas Horn 2 3 Karsten Buschard 1 2 0 Department of Physiology and Pharmacology, Section of Molecular Neuropharmacology, Karolinska Institute , SE-171 77 Stockholm, Sweden 1 Bartholin Instituttet, Kommunehospitalet , DK-1399 Copenhagen K, Denmark 2 Institute of Clinical Neuroscience, Section of Experimental Neuroscience, Gteborg University, Sahlgrenska University Hospital/Mlndal , SE-431 80 Mlndal, Sweden 3 Department of Pathology, Herlev Hospital, University of Copenhagen , DK 2730 Herlev, Denmark - In previous studies we have shown that sulfatide (galactosylceramide-3-O-sulfate), in various species, is present in the insulin-producing cells in pancreatic islets of Langerhans. In this study the synthesis of sulfatide in the islets has been investigated by pulse chase labeling at varying glucose levels and in the presence or absence of the glycosphingolipid synthesis inhibitory agents, Brefeldin A, fumonisin B1 and chloroquine and the distribution of sulfatide by immune-electronmicroscopy. The data showed that (1) sulfatide was produced in islets of Langerhans, (2) the main pathway for synthesis was through recycling involving partial degradation in the lysosome, and that (3) high glucose levels, although not primarily reflected in an increased synthesis of sulfatide, lead to an increased expression of mRNA for the UDP-galactose:ceramide galactosyltransferase, producing the immediate precursor of sulfatide. Furthermore, mass spectrometry analyses revealed a high proportion of short chain fatty acids, C16:0 (50%) and no hydroxylated forms and thus special physicochemical properties, indicating important differences between pancreatic and brain/neural sulfatide. Immune electron microscopy revealed an intracellular expression of sulfatide in the secretory granules, the Golgi network and the lysosomes of the islets. These results indicate that sulfatide follows the same intracellular route as insulin and suggest a functional association between these molecules. We have raised the hypothesis that sulfatide possibly plays a role in the trafficking of insulin in the islets of Langerhans in rat pancreas. 1To whom correspondence should be addressed Sulfatide (galactosylceramide-3-O-sulfate) is a glycosphingolipid that has attracted major interest as the dominating acidic glycosphingolipid in myelin in the nervous system (Norton and Autilio, 1966; Svennerholm et al., 1992). We have previously demonstrated the presence of sulfatide in islets of Langerhans of the pancreas and presented support for a possible role of sulfatide as an antigen in the development of insulin-dependent diabetes, diabetes type 1 (Buschard et al., 1993aa,b, 1994; Buschard and Fredman, 1996). The presence of sulfatide in pancreatic tissue and in isolated islets of Langerhans was consistent among various species including man, rat, mouse, pig, and monkey (Buschard et al., 1993bb, 1994). Immunohistochemical analyses of pancreatic tissue using the anti-sulfatide monoclonal antibody, Sulf I (Buschard et al., 1993bb, 1994), revealed selective staining in the islets of Langerhans, whereas the exocrine tissue was unlabeled. Biochemical analyses of pancreatic tissue and isolated a and b cells confirmed that among potential glycolipid antigens (sulfatide, sulfated lactosylceramide and seminolipid) sulfatide was the only antigen expressed (Buschard et al., 1994). Moreover, immune electron microscopy showed that sulfatide was present in secretory granules (Buschard et al., 1993bb). During the last few years it has become clear that membranebound lipids play a role in endocytosis (for review, see Riezmann et al., 1997) and in intracellular vesicle trafficking and sorting (van Meer, 1998). Among the these lipids are the glycosphingolipids, which have attracted special interest as components in caveolae (Parton, 1996; Simons and Ikonen, 1997) and in forming domains with specific proteins (Kasahara et al., 1997). Sulfatide is synthesized through a stepwise enzymatically catalyzed process starting in the endoplasmic reticulum and finally being sulfated in the Golgi (Benjamins et al., 1982), presumably the late Golgi or trans-Golgi network (Farrer et al., 1995), from which it is transported by membrane vesicle flow to the plasma membrane and possibly other intracellular organelles, like the lysosome. The presence and specificity of proteins in such vesicle remains an open question. The production of insulin starts in the endoplasmic reticulum with the synthesis of pre-proinsulin, which is converted to proinsulin and then transferred to the Golgi (for review, see Orci, 1986). It is also known that proinsulin is released from the trans-Golgi network in secretory granules. During the cytoplasmic transport of these secretory granules from the Golgi to a location close to the plasma membrane the proinsulin is transformed to insulin. Sulfatide and insulin thus both seem to be processed through the trans-Golgi network compartment and possibly sorted into and transported in the same granulae. Fig. 1. TLC-immunostaining of Sulf1 antigens in rat islets cells. The islets had been grown in media containing 11 mM glucose for 48 h. The extraction and isolated of the lipids as well as the TLC-ELISA procedure is described in detail in Materials and methods. Lane 1, seminolipid (100 pmol); lane 2, sulfated lactosylceramide (10 pmol), lane 3, sulfatide (20 pmol), lanes A and B, sulfatide fraction from islet cells (corresponding to 5.5 m g protein) before and after acid hydrolyses, respectively. Based on these previous findings together with immune electron microscopy investigations showing that sulfatide, like insulin, is present in secretory granules (Buschard et al., 1993b), we hypothesized that there might be a functional association between sulfatide and insulin, possibly related to the intracellular trafficking of secretory granulae and/or insulin processing. The aim of this study was to find evidence for synthesis of sulfatide in the islets of Langerhans, and if it exits, to elucidate the metabolic pathway of sulfatide in these cells. The study was performed using freshly isolated rat islet cells in culture and metabolic labeling with endogenous 35S-sulfate, 3H-galactose, and 14C-serine in the presence or absence of various inhibitors, Brefeldin A, fumonisin B1 and chloroquine, each described to inhibit different processes involved in synthesis of glycosphingolipids. The results showed that sulfatide was produced in the cells and that the main route of synthesis was recycling, where sulfatide was partially degraded in the lysosomes and then reutilized for synthesis in the Golgi. Together with the findings that sulfatide is present in the insulin secretory granules and that a major fraction of the insulin is degraded through the lysosomes (Halban and Wollheim, 1980), this lends support to a biologically relevant association between sulfatide and insulin in the cellular processing of insulin in islet cells in rat pancreas. Isolation and identification of sulfatide from islet cells With the lipid extraction and separation procedure used, sulfatide was found selectively in the C/M/W (65:25:4, by vol.) fraction and the recovery of 3H-labeled sulfatide was 95 97%. Thus, this fraction was the only one analyzed in the following experiments. Galactosylceramide was eluted in the same fraction. Isolation and purification of sulfatide and the mass spectrometry analyses were performed as described in Material and methods. The proportion corresponds to the relative intensities of [M-H] peaks with shown m/z recorded in the negative mode ES-MS. Only peaks with relative intensities >10% are shown. TLC-ELISA using the SulfI antibody showed two stained bands migrating (Figure 1) in the same region as the sulfatide standard isolated from pig brain. There were no stained bands migrating similarly to the other antigens, seminolipid and sulfated lactosylceramide. Acid hydrolysis of the sulfatide fraction abolished all staining with the antibody. The endogenous sulfatide content varied between the experiments from 3 to 5 nmol/mg protein. No difference was found between cells grown in 2.8, 11, and 20 mM glucose. The endogenous amount of galactosylceramide was estimated, by immunostaining, to be about half the amount of sulfatide. Autoradiography of lipids from 3H-galactose labeled islets (see control fractions in Figures 5B and 4B) revealed the presence of two bands migrating similarly to the galactosylceramide standard with a slightly higher staining intensity than sulfatide. Separation on borate-impregnated silica gel plates and liquid scintillation of this lipid fraction showed approximately equal amounts of galactosyl- and glucosylceramide. Mass spectrometry analyses of sulfatide isolated from islets of Langerhans from pancreas of Lewis rats confirmed that the fractions migrating similarly to brain sulfatide standard corresponded to sulfated monohexaosylceramide. Acid methanolysis of the isolated sulfated monohexaosylceramide fraction followed by analysis on borate-impregnated HPTLC plates showed only the occurrence of an orcinol-positive band migrating as reference galactosylceramide. Based on this result and the fact that the Sulf I antibody does not react with sulfated glucosylceramide (Iida et al., 1989), it is concluded that >90% of the isolated sulfatide molecules corresponded to galactosylceramide-3-O-sulfate, sulfatide. The mass spectra of Lewis rat brain sulfatide (Table I) showed that major molecular species contained the fatty acids C24:0 and C24:1 and also the corresponding hydroxy fatty acids. In the islet cells, the two dominating molecular species of sulfatide contained C16:0 and C24:0 fatty acids. Other prominent peaks corresponded to fatty acids C22:0 and C24:1. All molecular species contained 4-sphingenine as long chain base. Thus, in comparison with sulfatide isolated from rat Fig. 2. Autoradiogram of 14C-serin (lanes 1 and 2) or 35S-sulfate (lanes 3 and 4) labeled lipids from islets cells grown in media with low (2.8 mM) or high (20 mM) glucose. Isolation and culturing of the islets and the extraction and separation of the labeled lipid fraction is described in Materials and methods. Lipids corresponding to 100 m g islet cell protein were applied. Lane 5, standards of 14C-labeled sulfatide (Sul) and galactosylceramide (Gal-cer). The migration of ceramide (Cer) and sphingomyelin (Sphm) is indicated in the figure. brain, the islet cell sulfatide had a higher proportion of short chain fatty acids and in addition no measurable amounts of hydroxy fatty acids. No difference in the fatty acid composition of sulfatide from islet cells cultured in 2.8, 11, or 20 mM glucose medium was found. After a 24 h labeling period with 14C-serine, fractions migrating similarly to sulfatide were hardly detectable (Figure 2, lanes 1 and 2). Most intensely labeled was a fraction migrating as sphingomyelin. Liquid scintillation analysis showed that the counts found in the sulfatide bands at all glucose levels accounted for ~ 5% (~50 c.p.m.) of the total counts (9001300 c.p.m.). These results indicated that the de novo synthesis of sulfatide was very low during a 24 h period. Labeling with 35S-sulfate (Figure 2, lanes 3 and 4) for 24 h resulted in ~ 1000 c.p.m. (9301070) in the sulfatide bands. Although the counts cannot be directly compared between experiments, this result indicates that synthesis via recycling is more prominent than de novo synthesis. There was no measurable influence of the glucose level in the medium on the metabolic labeling, either in the 14C-serine or in the 35S-sulfate experiments. Determination of mRNA for UDP-galactose:ceramide galactosyltransferase and sulfotransferase in isolated rat islets cells mRNA for UDP-galactose:ceramide galactosyltransferase was detected in islet cells and the glucose level in the medium had a direct effect on its expression (Figure 3). When hybridized to actin and normalized with this signal, the expression was 2.6 0.64 (n = 5) more abundant (p = 0.0028, Mann Whitney U-test) in glucose stimulated (20 mM) than in glucose-deprived Fig. 3. Expression of ceramide galactosyltransferase in islets grown in media with low (2.8 mM) or high (20 mM) glucose levels. Northern blot of 10 m g total RNA from islets hybridised with UDP-ceramide galactosyltransferase (A) and actin (B). Lane L presents islets grown in 2.8 mM glucose and lane H islets grown in 20 mM glucose (2.8 mM) islets. There was no detectable sulfotransferase when using the cDNA probe provided by Dr. Honke (Honke, 1997). Effect on the metabolism of sulfatide in islets cells in the presence of Brefeldin A The two concentrations, 0.5 and 5.0 m g/ml medium, of Brefeldin A gave similar results (Figure 4). The islet cells were sensitive to this treatment and 6 h was the maximum incubation time that could be used without the cells disintegrating. Labeling with 3H-galactose indicated a slight increase in radioactivity in the lipid fractions migrating as monohexaosylceramide and sulfatide (Figure 4B). Further identification, using borate-impregnated plates, revealed that the increased labeling was associated with glucosylceramide (3-fold) and lactosylceramide (2-fold), respectively. The effect of Brefeldin A as observed by adding 14C-serine (Figure 4A) was a 3-fold increase of c.p.m. in the sphingomyelin fraction and a 6-fold increase of ceramide, which was identified by HPTLC and autoradiography. Labeling with 35Ssulfate (Figure 4C) was abolished in the presence of Brefeldin A, indicating that the recycling pathway was affected. Incubation of the islet cells in medium with Fumonisin B1 for 40 h markedly inhibited the incorporation of 14C-serine (Figure 5A). Two weak bands were seen of which the major one migrated similarly to sphingomyelin. However analyses on borate-impregnated plates showed that at most half the radioactivity corresponded to sphingomyelin. No further characterization was performed. In the presence of Fumonisin B1 the 3H-galactose labeling of sphingolipids was (Figure 5B) was reduced. The incorporation of 35S-sulfate was also inhibited, as revealed by a reduced amount of labeled sulfatide. Shorter incubation, 6 h, with Fumonisin B1 (Figure 6) had no effect on the 35S-sulfate labeling of sulfatide. These findings indicate that the effect of Fig. 4. Autoradiogram of 14C-serin (A), 3H-galactose (B) or 35S-sulfate (C) labeled lipids from islets cells grown in the presence or absence of Brefeldin A. Isolation and culturing of the islets and the extraction and separation of the labeled lipid fraction is described in Materials and methods. Lipids corresponding to 100 m g islet cell protein were applied. Standards applied to the plates were 14C-labeled Gal-cer (galactosylceramide) and Sul (sulfatide). The migration of standards of Cer, ceramide and Sphm, sphingomyelin are indicated in (A). The fraction marked with a in (B) was analyzed on borate impregnated plates and shown to be comprised of glucosylceramide and galactosylceramide, in the controls in the ratio 1.5:1 and in the Brefeldin treated 3:1. Fig. 5. Autoradiogram of 14C-serin (A), 3H-galactose (B), or 35S-sulfate (C) labeled lipids from islets cells grown for 40 h in the presence of fumonisin. Isolation and culturing of the islets and the extraction and separation of the labeled lipid fraction is described in Materials and methods. Lipids corresponding to 100 m g islet cell protein were applied. Standards applied to the plates were 14C-labeled Gal-cer, galactosylceramide; and Sul, sulfatide. The migration of standard Sphm, sphingomyelin and Cer, ceramide is indicated in (A). The fraction indicated with and seen in the fumonisin treated islets in (A). Fumonisin B1 on the recycling pathway is a later event and most likely a secondary effect of the inhibition of de novo synthesis of sphingolipids. Effect on the metabolism of sulfatide in islets cells in the presence of chloroquine In this experiment the cells were incubated in high glucose (20 mM) media to increase the activity of the cells and thereby possibly also the endocytosis and lysosomal degradation. Chloroquine (26 m g/ml) had, as expected, no apparent effect on the de novo synthesis of the major sphingolipids a shown by 14C-serine labeling (Figure 7, lanes 1 and 2). On the other hand, there was a marked reduction in the incorporation of 35Ssulfate into sulfatide (Figure 7, lanes 3 and 4), indicating an effect on recycling. Effect on the metabolism of sulfatide in islet cells in the presence of the monoclonal antibody Sulf I The presence of 40 m g/ml of the antisulfatide antibody Sulf I for 6 h in the culture medium with (11 mM) glucose had no Fig. 6. Autoradiogram of 35S-sulfate labeled lipids from islets cells grown for 6 h in the presence of fumonisin. Isolation and culturing of the islets and the extraction and separation of the labeled lipid fraction is described in Materials and methods. Lipids corresponding to 100 m g islet cell protein were applied. Standards applied to the plate were 3H-labeled Sul, sulfatide and Gal-cer, galactosylceramide. Fig. 7. Autoradiogram of 14C-serin (A) and 35S-sulfate (B) labeled lipids from islets cells grown for 6 h in the presence of chloroquine. Isolation and culturing of the islets and the extraction and separation of the labeled lipid fraction is described in Materials and methods. The islets were in this experiment grown in 20 mM glucose to increase the requirement for insulin secretion. Lipids corresponding to 100 m g islet cell protein were applied. Standards applied were 14C-labeled Gal-cer (galactosylceramide) and Sul (sulfatide). The migrations of ceramide (Cer) and sphingomyelin (Sphm) are indicated in the figure. effect on the incorporation of 35 S-sulfate into sulfatide in the islets cells as determined by autoradiography (Figure 8). Thus, cell surface located sulfatide is unlikely to be major pool for the recycling pathway. Ultrastructurally, immunogold labeling of sulfatide was found in secretory granules of beta cells; background staining was negligible (Figures 9, 10). Lysosomes and/or lysosomal-like Fig. 8. Autoradiogram of 35S-sulfate labeled lipids from islets cells grown for 6 h in the presence of the anti-sulfatide antibody SulfI. Isolation and culturing of the islets and the extraction and separation of the labeled lipid fraction is described in Materials and methods. Standards applied were 14C-labeled Galcer (galactosylceramide) and Sul (sulfatide). granules, having greater diameter than secretory granules and a more electron-dense and heterogeneous content, showed marked sulfatide staining (Figure 9). No labeling of the rough endoplasmic reticulum could be demonstrated whereas some of the Golgi profiles showed weak but distinct labeling (Figure 10). Quantification of the Golgi reaction showed a significantly (p < 0.02) higher number of labeled Golgi profiles in cells stimulated with 20.0 mM (46% of the lysosomes stained) than cells stimulated with 2.8 or 11.0 mM glucose (23 and 24% of the lysosomes stained, respectively). The present study has for the first time provided evidence for the cellular synthesis of sulfatide in islets of Langerhans isolated from rats and shown that the main metabolic pathway was recycling. Immune electron microscopy also showed sulfatide expression in Golgi, in secretory granules and in the lysosomes of the islet cells. Thus, the intracellular trafficking of sulfatide in the insulin-producing cells coincided with that described for insulin (Halban and Wollheim, 1980). These results and previous studies showing cell surface expression of sulfatide and insulin provide support for a possible functional association between these molecules and new insight into possible mechanisms for insulin secretion. The study was performed on isolated islets from rat pancreas which contain approximately 75% insulin-producing b -cells. The structure of the sulfatide present in the islets was confirmed but showed a fatty acid composition that was different as compared to sulfatide in the brain of Lewis rats and other mammalian species (Norton and Cammer, 1984; Vos et al., 1994; Ishizuka, 1997). The most striking difference was the high proportion of the short fatty acid C16:0 and the lack of Fig. 9. Ultramicrograph of a b cell showing immunogold labeling of sulfatide in secretory granules (thin arrows). In the center a larger electron dense and heterogeneous granule, most probably representing a lysosome, is seen to be heavily labeled (thick arrow). A crystal like structure as part of this granule indicates that insulin containing secretory granules are fused with lysosomes. Note the lack of background staining. Original magnification, 12,500 . (B) Higher magnification of part of (A) showing labeling with 10 nm immunogold particles in the above mentioned lysosome and in neighboring granules. Original magnification 12,500 ; magnified photographically 4 . hydroxy fatty acids. Similar results were recently obtained (Hsu et al., 1998) when analyzing islet cells isolated from pancreas from SpragueDawley rats. A relatively high proportion of nonhydroxylated fatty acids has also been reported to occur in sulfatide from mammalian kidney (Ishizuka, 1997). The fatty acid chain length, hydroxylation and degree of unsaturation affect the physicochemical properties of the molecule in the membrane (Stevensson et al., 1992; Tupper et al., 1992; Vos et al., 1994; Ishizuka, 1997). Sulfatide in myelin contains more saturated or monounsaturated long fatty acids (C22C26), both with and without 2-hydroxy groups, as compared to most other sphingolipids and phospholipids and spans more than half the membrane lipid bilayer and interdigitates into the opposite monolayer. The large proportion of shorter chain length (C16:0) fatty acid in sulfatide from islets cells does thus not have such interdigitation properties and Fig. 10. Ultramicrograph of a beta cell stimulated with 20mM glucose showing few immunogold labeled secretory granules and in the center labeling (arrows) of the Golgi apparatus. Original magnification 12,500 . (B) Higher magnification of (A) revealing labeling with immunogold particles of the Golgi apparatus and secretory granules. Original magnification 12,500 ; magnified photographically 4 . lateral migration in the membrane is therefore facilitated. It is beyond the scope of the experiments done so far to identify the precise role of the short fatty acid chains in the sulfatide molecule in the islets, but they show that sulfatide in these cells has different physicochemical properties than sulfatide in myelin. This in turn indicates that this molecule might have different functions in islets and myelin. It is also likely that molecular species with shorter and longer fatty acids have a different distribution and function in the insulin-producing cells. Another physicochemical property of sulfatide is that these molecules might associate and form patches in the membranes, a phenomenon that has been suggested to support close contact between membranes, as in fusion processes (Hakomori, 1991). Sulfatide is also a strong Ca2+ binder and binding induces a hydrocarbon chain disorder preferentially in nonhydroxylated molecular species. It is therefore tempting to speculate that the lack of hydroxylated fatty acids in islet cells favors its involvement in membrane fusion being influenced by Ca2+ alterations, which precede glucose-induced insulin secretion. Another effect of the fatty acid composition is that it influences the metabolic pathway (van Meer, 1998). Based on previous results, it is generally assumed that sulfatide is formed by sulfation of galactosylceramide (Vos et al., 1994; Farrer et al., 1995; Ishizuka, 1997). Farrer et al. (1995) showed that addition of Brefeldin A led to almost complete cessation of sulfatide synthesis in immortalized Schwann cells and, as in this study, inhibition was reached after 6 h treatment with 0.5 m g Brefeldin A/ml. Brefeldin A is known to mainly affect the transport between the ER and cis/ medial-Golgi (reviewed by Klausner et al., 1992) and inhibits glycosylation of complex glycosphingolipids (van Echten et al., 1995) and sulfatide (Farrer et al., 1995) which takes place in the trans-Golgi network. The results obtained in this study are in conformity with the described effect of Brefeldin A. Brefeldin A has also been shown to inhibit glycosylation of recycled glycosphingolipids (Gordon and Lloyd, 1994), which would then affect the major pathway for sulfatide synthesis in the islets of Langerhans. Moreover, Huang and Arvan (1994) showed that Brefeldin A inhibited the transfer of pro-insulin from the trans-Golgi network to granules. If sulfatide, as we hypothesize, is sorted into the same granules as pro-insulin the inhibited sulfatide synthesis might reflect the lack of proinsulin granule formation. In addition to the inhibition of sulfatide synthesis, Brefeldin A was found to increase glucosylceramide and lactosylceramide formation. This is in contrast to the study by Farrer et al. (1995), who instead found a concomitant increase of galactosylceramide. However, they studied another cell type, immortalized Schwann cells and glycosphingolipid synthesis is known to be cell type and cell differentiation specific (Hakomori, 1981). Brefeldin A was also found to induce a significant increase of formation of sphingomyelin and ceramide in the rat islets. The increase of cellular ceramide might be one factor that induced the disintegration seen in the islets of Langerhans when the incubation was prolonged for more than 6 h. Ceramide has been shown in many studies to induce apoptosis (Merill et al., 1997) and the 6-fold increase is likely to induce such a process. Fumonisin B1 is a fungal toxin that inhibits the acylation of sphingosine (sphingenine) by acylCoA to form ceramide and thereby the de novo synthesis of all products produced from ceramide, including sphingomyelin and glycosphingolipids. The weak labeling obtained with 3H-galactose might reflect the fact that the inhibition was not complete and/or that there was a re-use of partially degraded sphingolipid products from the recycling pathway. Although the latter pathway is not primarily affected by fumonisin, a marked reduction of 35S-sulfatelabeled sulfatide was seen after 40 h exposure. This might reflect the fact that during a 40-hour period the de novo synthesis is required to restore the endogenous pool and/or that the effect of fumonisin in the long run has a general effect on cellular metabolism. Islets exposed to fumonisin for only 6 h had no effect on 35S-sulfate-labeled sulfatide, supporting that recycling is the dominating pathway. The third glycosphingolipid inhibitor used in this study was chloroquine. This compound is an amphiphilic molecule known to accumulate in the lysosome (DeDuve et al., 1976) and possibly in the early endosomes, where it causes a rise in pH, which in turn effect the activity of the lysosomal enzymes having acidic pH optima. The inhibitory effects of chloroquine on lysosomal degradation of glycosphingolipids was reported many years ago (Klinghardt et al., 1981). The inhibitory effect of chloroquine on the 35S-sulfate labeling of sulfatide might thus be explained by a reduced activity of the sulfohydrolase in the lysosome and thereby reduced formation of product for the recycling pathway of sulfatide synthesis. Finally, we looked into the effect of a monoclonal antibody to sulfatide (Sulf I) with the aim of influencing recycling of sulfatide localized to the cell surface. However, there was no detectable reduction in 35S-sulfate-labeled sulfatide supporting that the main route for sulfatide biosynthesis is recycling of intracellularly localized sulfatide, i.e., in the secretory granules. The reason for using 3H-galactose in the labeling studies was to explore whether the recycling involved just desulfation or in addition removal of galactose to form ceramide. However, the incorporation of galactose was in most experiments too low to permit any firm conclusions. The most common way to increase labeling with exogenous substrate is to exclude the native molecule from the incubation medium. However, as the metabolism of islets is regulated by exogenous glucose, it is likely that the uptake of glucose competes with that of galactose and removal of glucose would create a nonphysiological environment. Biochemical analyses revealed the existence of endogenous galactosylceramide but the main route of synthesis remains to be resolved. The endogenous pool of insulin is large and a substantial fraction of insulin is never secreted but degraded within the b cell (Halban and Wollheim, 1980). Insulin-containing granules are fused with primary lysosomes and completely degraded. Sulfatide has previously been shown to be expressed in the secretory granules (Buschard et al., 1993b, 1994), and in this study also in Golgi and lysosomes. Some lysosomes were intensely labeled whereas others were not labeled at all. The varying labeling intensity of lysosomes might reflect different steps in the degradation of sulfatide and insulin. These findings support a fusion of sulfatide-containing secretory granules with lysosomes. There was no measurable change in the endogenous amount or the pulse chase labeling of sulfatide due to glucose levels in the media. As sulfatide is also located in the secretory granules, it is reasonable to assume that an increased secretion will reduce the proportion of secretory granula that enter the lysosome and thereby reduce the major pathway, recycling, for sulfatide synthesis. There was, however, increased sulfatide labeling in the lysosomes in islets stimulated with high glucose concentration, a finding that does not necessarily reflect increased fusion of secretory granules, but perhaps reduced degradation in stressed b -cells. High glucose levels (high glucose) might have induced a de novo synthesis of insulin (matured from pro-insulin, see below) which is not immediately followed by a corresponding increase of de novo sulfatide synthesis. This is supported by recent data (Buschard et al., in press) showing that the number of secretory granules expressing sulfatide is more rapidly reduced than the total number of insulin granules. It was noticed, however, that the mRNA level of the UDPgalactose:ceramide galactosyltransferase (CGT) (Stahl et al., 1994; Coetzee et al., 1996) was significantly increased in islets grown in high glucose medium. This finding indicated that, at least after 24 h, there was a need for increased synthesis of galactosylceramide, which is the immediate precursor of sulfatide. The most relevant enzyme mRNA to measure would have been 3 -phosphoadenylsulfate:galactosylceramide 3 -sulfotransferase. The only potential cDNA described is that found by Honke et al. (1997). Their report clearly showed that the gene encodes a lipid sulfotransferase but the acceptor(s) was not characterized. The exact specificity remains to be clarified. However, we could not detect any mRNA for this enzyme when using the probe kindly provided by Dr. Honke. One possible explanation is that this probe was made from a human renal cancer cDNA and there might be sequence differences between humans and rats. Another possibility might have been to measure the cell-free activity of the corresponding sulfotransferase but such analyses includes the use of detergent and other factors to optimize the milieu and might thus not reflect the cellular activity, which was the aim of this study. In conclusion, this study has provided evidence that sulfatide is produced in islets of Langerhans and that the fatty acid composition, and thereby the physicochemical properties, differ from those of sulfatide in myelin. The main pathway of sulfatide synthesis was recycling, involving partial degradation of sulfatide in the lysosome. The presence of both sulfatide and insulin, in the Golgi, in secretory granules and in the lysosomes, and the fact that most of the insulin is degraded without ever being released, indicates that insulin and sulfatide take the same intracellular route and supports a functional association of insulin and sulfatide. Based on these results, we have raised the hypothesis that sulfatide might play a role in the trafficking of in islets of Langerhans from rat pancreas secretory granules and possible also in the processing of insulin. Materials and methods Alumina backed high performance thin-layer chromatography plates (HPTLC-plates) and silica gel 60 were purchased from Merck AG, Darmstadt, Germany and plastic-backed silica gel thin-layer chromatography plates (TLC-plates) from MacheryNagel (Dren, Germany). The x-ray film (X-OMAT AR5) for autoradiography was obtained from Kodak and EnHance spray from Du Pont (Boston). Sulfatide and galactosylceramide used as standards, were purified from pig brain according to the procedure previously described (Rosengren et al., 1989). Other glycosphingolipid standards were prepared from human tissues (Svennerholm et al., 1979; Karlsson et al., 1990). The structures of the purified glycosphingolipid standards were verified by mass spectrometry (FAB-MS) (Mnsson et al., 1986ab). D-[6-3H]galactose (2040 m Ci/mmol), [U-14C]serine (150 mCi/mmol), and 35S-labeled sulfate (carrier free) were purchased from Amersham, Life Science. Inhibitors of sphingolipid synthesis used in the metabolic studies were Brefeldin A, Fumonisin B1, and chloroquine, all purchased from Sigma, St. Louis, MO. 14C-Labeled sulfatide and galactosylceramide, used as standards in autoradiography, were prepared in our laboratory. Briefly, sulfatide or galactosylceramide was deacetylated in alkaline methanol (Neuenhofer et al., 1985). The formed lysocompound was reacetylated with 14C-labeled palmitoylchloride in 50% sodium acetate (Karlsson et al., 1990). 3HLabeled sulfatide and galactosylceramide were prepared according to the procedure described by Schwarzmann (1978). The radiolabeled compounds were purified by preparative silica gel column chromatography and thin layer chromatography. The materials used for RNA isolation and Northern blots were the following: Genescreen Plus membrane from NEN (Boston); Elutip from Schleucher and Schuel, Germany; 32P (dCTP)from Amersham, Bucks, England; Sephadex G50 from Pharmacia, Uppsala, Sweden, and reflection film and intensifying screen from NEN. LR-white, used for electron microscopy, was obtained from Bio-Rad, Watford, UK. The phosphatase substrate 5 -bromo-4 -chloro-3 indolylphosphate was purchased from Sigma, St. Louis, MO. The production and characterization of the mouse monoclonal antibody Sulf I has been reported previously (Fredman et al., 1988). The glycosphingolipid antigens recognized have been shown to be sulfatide, lactosylceramide sulfate and seminolipid (Buschard et al., 1994), but not sulfated glucosylceramide (Iida et al., 1989). Alkaline-phosphatase-conjugated goat antimouse Ig antibody was purchased from Jackson laboratories (West Grove, PA), the rabbit anti-mouse Ig (F261) from Dako (Copenhagen, Denmark) and the anti-rabbit immunoglobulin conjugated with colloidal gold from Amersham. As control antibody to Sulf I the mouse anti-human CD8 (IgG1) antibody from Dako (M707) was used. The hybridoma-producing antigaclactosylceramide monoclonal antibody, O1 (Sommer and Schachner, 1981; Bansal et al., 1989) (mouse IgM) was kindly provided by Dr. Steve Pfeiffer, Department of Microbiology, University of Connecticut, Farmington, CT. Isolation of islet cells from rat pancreas Islets of Langerhans from 8-week-old male Lewis rats (purchased from Mllegrd, LI Stensved, Denmark) were isolated under sterile conditions by a collagenase digestion technique described previously (Buschard et al., 1990). Isolated islets were suspended in RPMI 1640 (Gibco), containing 11 mM glucose, 10% fetal calf serum, 1% penicillinstreptomycin (10,000 IU/ml-10,000 m g/ml, Gibco, Paisley, UK) and with the pH adjusted to 7.35. The islets were incubated for ~ 20 h at 37 C and thereafter used for the various experiments described below. Usually, about 2000 islets could be isolated on each occasion, using four to six rats. Since the precise number of cells per islet and the actual condition of the cells may vary from one isolation to another, a direct comparison can only be made on islets isolated on the same occasion. Growth of islets for preparation of sulfatide for determination of the ceramide composition Islets (~2000) were after 20 h in 11 mM glucose medium, divided into three equal portions and transferred to medium containing 2.8, 11, or 20 mM glucose for 24 h. Extraction and isolation of sulfatide from two experiments (~4000 islets) was performed as described below. The lipid extracts were analyzed with TLC-ELISA. Sulfatide fractions for mass spectrometry analysis were isolated by preparative HPTLC using C/M/W (65:25:4, by vol.) as the developing solvent. Bands with migration corresponding to reference pig brain sulfatide were scraped out and extracted from the gel with C/M/W (3.6:2, by vol.) The mass spectra were recorded in the negative mode on a double-focusing magnetic mass spectrometer (Autospec, Micromass, Manchester, UK) equipped with en electrospray ionization source (Ghardashkhani et al., 1995) Part of the isolated sulfatide fractions was also subjected to acid methanolysis and analyzed on borate-impregnated HPTLC plates as described below. In vitro biosynthesis of sulfatide in islets cells grown in high and low glucose medium The cells were isolated and grown in 11 mM glucose for 20 h as described above. After washing, the cells were incubated for another 24 h in RPMI 1640 (Gibco) with supplements as described above but with a reduced glucose content, from 11 to 2.8 mM. Thereafter, the cells were randomly divided into equal portions and grown in medium with 2.8 (low glucose) or 20 mM glucose (high glucose). To the media was added 1 m Ci/ml (final concentration) 14C-serine and/or 10 m Ci/ml 35S-sulfate. After 24 h the cells were washed twice in ice-cold phosphatebuffered saline, PBS, and frozen at 80 C until biochemically analyzed. Approximately 500 islets were used for each assay and two independent assay set-ups were analyzed. The sphingolipid synthesis was analyzed by autoradiography and in some experiments scintillation counting was performed on gel fractions scraped out from the TLC plate and corresponding to individual glycosphingolipids. Effect on the metabolism of sulfatide in islets cells in the presence of Brefeldin A Islets were isolated and grown for 20 h in 11 mM glucose medium as described above. The islets were then divided into equal portions (the number of islets in each experiment being ~ 500) and grown in the same medium supplemented with 14Cserine (1 m Ci/ml), 3H-galactose (2 m Ci/ml) or 35S-sulfate (10 m Ci/ml) and without or with Brefeldin A (0.5 or 5.0 m g/ml) for 6 h. (The reduced time of incubation as compared to the other experiments was due to the fact that the cells started to disintegrate after this time.) The islets were washed twice in ice-cold phosphate-buffered saline, PBS, and frozen at 80 C until biochemically analyzed. The experiment was performed on two occasions. The effect on sulfatide synthesis was analyzed by autoradiography and a scintillation count performed on gel fractions with individual glycosphingolipids. The experiments were performed as described for the Brefeldin A experiments, but with another incubation schedule. After the initial 20 h in 11 mM glucose medium fumonisin B1 (18.5 m g/ml) was added and the islets incubated for another 16 h. Thereafter, new medium (with 18.5 m g Fumonisin B1/ml) containing 14C-serine (1 m Ci/ml), 3H-galactose (2 m Ci/ml) or 35S-sulfate (10 m Ci/ml) was added and the cells grown for another 24 h. The experiment was performed on two occasions. To elucidate the short-term effect, one experiment, in duplicate, was performed where islets after the first 20 h in 11 mM glucose medium were transferred to medium (11 mM glucose) without or with Fumonisin B1 (18.5 m g/ml) incubated for 1 h, after which 35S-sulfate was added and incubation proceeded for 5 h. The effect on sulfatide synthesis was analyzed by autoradiography and scintillation counting performed on gel fractions with individual glycosphingolipids. Effect on the metabolism of sulfatide in islets cells in the presence of chloroquine The islets were isolated and grown in 11 mM glucose medium for 20 h as described above. The islets were then divided into four fractions (~ 500 islets per fraction), two of which were used as controls and two were grown in the presence of chloroquine (25 m g/ml). The experiment was performed on two occasions. The incubation medium used was, as described above, supplemented with 20 mM glucose and 14C-serine (1 m Ci/ml) or 35S-sulfate (10 m C/ml). The effect of chloroquine on the metabolic labeling during 6 h was analyzed by autoradiography and scintillation counting performed on gel fractions with individual glycosphingolipids. Effect on the metabolism of sulfatide in islets cells in the presence of Sulf I antibody Islet cells were isolated and grown in 11 mM glucose medium for 20 h as described above. Thereafter, the cells were incubated for 1 h in medium containing 20 mM glucose and the anti-sulfatide antibody Sulf I (40 m g/ml). Thereafter, without changing the media, 35S-sulfate (1 m Ci/ml) was added. The effect of exogenously added antibody to sulfatide was analyzed by autoradiography of formed sphingolipids and scintillation counting performed on gel fractions with individual glycosphingolipids. The extraction and separation of the lipid fractions were performed as previously described (Buschard et al., 1993bb). Briefly, the lipids were extracted from the islet cells by homogenization in C/M/W (4:8:3, by vol.). The pellet was reextracted and the two supernatants combined. This total lipid extract was evaporated to dryness and redissolved in 1 ml of C/M/W (65:25:4, by vol.). The lipids were then separated into two fractions by silica gel chromatography. Sulfatide, neutral monohexosylceramides, including galactosyl- and glucosylceramide, sulfated lactosylceramide and neutral glycosphingolipids with up to four sugar residues were eluted with 10 bed volumes of C/M/W (65:25:4, by vol.). This fraction, named the sulfatide fraction also contained ceramide and sphingomyelin and other lipids like fatty acids, cholesterol and phospholipids. To perform HPTLC- or TLC-separation for subsequent autoradiography and ELISA, respectively, the fraction had to be saponified (see below) to degrade phospholipids as some of these comigrated with the glycosphingolipid fractions and hampered the interpretation of the analyses. Saponification of the sulfatide fraction to degrade phospholipids was performed in the following way: an aliquot of the fraction was evaporated to dryness and redissolved in 200 m l of methanol/1 M KOH (1:1, v/v) and kept at room temperature over night. The samples were then neutralized with 0.5 M HCl and desalted on 2 0.5 g Sephadex G-25 (Wells and Dittmer, 1963). Autoradiography and scintillation counting Aliquots (corresponding to 50100 m g protein) of the lipid fractions were applied as 8 mm lanes to alumina-backed HPTLC plates and chromatographed in C/M/W (65:25:4, by vol.) The plates were air dried, sprayed with EnHance spray and exposed to x-ray film for 10 days. The autoradiogram was used to localize the individual glycosphingolipid fractions on the plate and the silica gel from these regions was scraped off and mixed with 10 ml scintillation liquid (Ultima Gold) and the samples counted in a Packard Tri-Carb 1500 Liquid Scintillation Analyzer. Separation of isomeric monohexosylceramides (galactosyland glucosylceramide) Isomeric monohexosylceramides comigrate on TLC and HPTLC plates in most organic solvents. Borate-impregnated plates on the other hand allow separation of isomeric hexaosylceramides. Briefly, HPTLC plates, were moistened by spraying them with 1% disodium-tetraborate-thiohydrate, Na2B4O7 10 H2O. The plates were air-dried for 4 h, and then dried further at 40 C in an oven for 16 h and kept in a desiccator until used. The chromatogram was developed with a solvent mixture of C/M/W (75:25:3, by vol.). The chromatography was repeated once in the same but newly prepared solvent. The plate was allowed to air-dry between runs. Autoradiography was performed as described above. Separation of ceramide and free fatty acids Ceramide and free fatty acids co-migrate in the solvents generally used in this study. To allow these molecules to be separated, the HPTLC plate was chromatographed in a solvent consisting of C/M/2.5M NH4 (90:10:1, by vol.). Autoradiography was then performed as described above. Determinations of endogenous content of sulfatide and galactosylceramide in the islets Sulfatide was quantified by TLC-ELISA using the sulfatidespecific SulfI monoclonal antibody (Fredman et al., 1988). Purified sulfatide standards and aliquots of the saponified sulfatide and galactosylceramide-containing lipid fractions from islets cells were applied as 5 mm lanes to plastic-backed HPTLC plates and chromatographed in C/M/W (65:25:4, by vol.). The plates were then incubated in sequential order with the Sulf I antibody, alkaline-phosphatase-conjugated antimouse antibody, and phosphatase substrate, 5 -bromo-4 chloro-3 indolylphosphate. Galactosylceramide was determined with the same method using the O1 antibody for detection and standards of galactosylceramide. Hydrolysis of the sulfate group from sulfatide This analysis was performed on 14C-serine or 35S-labeled fractions from individual experiments. An aliquot of the saponificated sulfatide/galactosylceramide-containing lipid fractions from islet cells was evaporated to dryness. Sulfate was removed by incubating these fractions dissolved in 200 m l 0.05 M HCl in methanol at room temperature over night. The fractions were directly applied to alumina-backed HPTLC plates for autoradiography. Formed products were identified by their migration in relation to labeled sulfatide and galactosylceramide standards on HPTLC-plates. The chromatographic solvent used was C/M/W (65:25:4,by vol.). To discriminate between galactosyl- and glucosylceramide, borate-impregnated plates were used as described above. Determination of UDP-galactose:ceramide galactosyltransferase mRNA in rat islets cells Islets were isolated and then incubated for 20 h in RPMI 1640 containing 11 mM glucose as described above. Thereafter, the cells were divided into two portions and transferred to medium with 2.8 mM or 20 mM glucose, respectively. After another 24 h, the cells were lysed in guanidinium thiocyanate-phenolchloroform (Chomczynski and Sacchi, 1987). An aliquot corresponding to 10 m g of total RNA was electrophoresed in 3[N-morpholino] propanesulfonic acid-buffered agarose gel containing 0.66 M formaldehyde. The RNA was then transferred to Genescreen Plus membrane (NEN Dupont, Boston, MA). Five independent experiments were performed. A fragment of the cDNA clone coding for UDP-galactose:ceramide galactosyltransferase containing the entire 2.4 kb insert was excised by BstxI digestion, isolated by agarose electrophoresis and Elutip-purified. The fragment was then 32P-labeled by random hexamer nucleotide priming (Amersham, UK) and purified by Sephadex G50 size exclusion chromatography (Pharmacia, Uppsala, Sweden). The blot was hybridized at 6 SSPS, 50% formamide at 48 C for 16 h and washed twice in 0.2 SSC, 0.05% sodium pyrophosphate, 1% SDS at 70 C and once in 0.2 SSC at room temperature. The filter was exposed for 20 h using Reflection film and intensifying screen (NEN Dupont). The autoradiographs were quantified using a laser scanning densitometer equipped with Image Quant software (Molecular Dynamics, Sunnyvale, CA). The results were corrected for differences in the amount of total RNA after hybridization to actin. Total brain RNA from adult and 10- and 27-week-old rats and total RNA from rat islets were isolated and blotted as described above. A 3 -phosphoadenylsulfate:galactosylceranide 3sulfotransferase probe was made using two internal PCR primers to amplify a 666 bp fragment from 1 ng pSV-hCST (Honke et al., 1997). The labeling reaction was performed in 25 m l using the buffer provided with the Taq polymerase: dATP, dGTP, and dTTp (5 m M each), 2.5m M 32P-dCTP (800 m Ci/m l, NEN, DuPont), primers 0.2 m M each, 1 U Taq polymerase (ICN Pharmaceuticals, Costa Mesa, CA). Purification and hybridization were performed as described above. The filters were hybridized to a GAPDH probe to correct for differences in total RNA amounts. Electron microscopy procedures Islets of Langerhans were isolated from Lewis rats as described above. The tissues were fixed for 1 h in a mixture of 2.5% paraformaldehyde and 0.1% glutaraldehyde. After washings in cacodylate buffer pH 7.3 and dehydration in 70% alcohol, specimens were embedded in LR-white (Bio-Rad, Watford, UK). To avoid unspecific staining, the ultrasections were blocked by 0.1% bovine serum albumin and 0.05% Tween 20 (Sigma, St. Louis, MO) in TrisHCl, pH 8.2. Sections were then incubated with Sulf I, diluted 1:75 in the same buffer at +4 C for 24 h. After incubation for 1 h at +20 C with rabbit anti-mouse immunoglobulin (F261, Dako) diluted 1:100 and absorbed with rat serum, sections were washed and treated for 1 h at +20 C with goat anti-rabbit immunoglobulin conjugated with colloidal-gold (10 nm, Bio Cell Research Laboratories, Cardiff, UK) diluted 1:100. After washing, specimens were postfixed in 2% glutaraldehyde for 5 min and stained with uranyl acetate/lead citrate before examination in a Philips EM208 electron microscope. Controls were treated similarly, except for incubation with the primary antibody, and showed no staining. The immunogold labeling of the Golgi apparatus was quantified in four experiments each with incubation of the islets in three different glucose concentrations (2.8, 11.0, and 20.0 mM). Ultramicrographs (5456) at a magnification of 30,000, showing an area of 5100 6900 nm of b cells, were taken from each of the incubations. Labeling (noted as present or absent) was registered for the Golgi profile on every micrograph. We are grateful to Dr. Brian Popko for kindly providing the CGT-probe (Stahl et al., 1994) and generous help. We are also grateful to Birgitta Dellheden, Susanne Srensen, and Elzbieta Christiansen for excellent technical assistance. The research project was supported by grants from the Swedish Medical Research Council (Proj. No. K99-03X-0990908C), Magnus Bergwalls foundation, ke Wiberg foundation, Trygg Hansa, the Danish Diabetes Association, and the Danish Medical Research Council


This is a preview of a remote PDF: https://glycob.oxfordjournals.org/content/10/1/39.full.pdf

Pam Fredman, Jan-Eric Månsson, Britt-Marie Rynmark, Knud Josefsen, Annette Ekblond, Linda Halldner, Thomas Osterbye, Thomas Horn, Karsten Buschard. The glycosphingolipid sulfatide in the islets of Langerhans in rat pancreas is processed through recycling: possible involvement in insulin trafficking, Glycobiology, 2000, 39-50, DOI: 10.1093/glycob/10.1.39