Sulfatide Controls Insulin Secretion by Modulation of ATP-sensitive K⁺-Channel Activity and Ca²⁺-Dependent Exocytosis in Rat Pancreatic β-Cells

Male Lewis rats (250–300 g) were purchased from Møllegaard (Ll; Skensved, Denmark). The rats were anesthetized by pentobarbital (100 mg/kg i.p.), and the pancreases were removed quickly. Pancreatic islets were then isolated by collagenase digestion and dispersed into single cells using dispase (Dispase II; Boehringer Mannheim, Mannheim, Germany), and the β-cells were separated by fluorescence-activated cell sorting (8) using a FACSstar (Becton-Dickinson, Mountain View, CA). Window sortings were set to enclose the β-cell population, which in the combined high-scatter and high-fluorescence fraction includes 97 ± 3% β-cells (8).

Patch pipettes were pulled from borosilicate glass capillaries, coated with Sylgard near their tips and fire-polished before use. The pipette resistance (when filled with the pipette-filling solutions) was 2–4 MΩ. The zero-current potential was adjusted before establishment of the seal with the pipette in the bath. Whole-cell K⁺ currents were estimated by applying 10 mV hyper- and depolarizing voltage pulses (duration 200 ms; pulse interval 2 s) from a holding potential of −70 mV. The currents were recorded using an Axopatch 200B patch clamp amplifier (Axon Instruments, Foster City, CA), digitized, and stored in a computer using the Digidata AD-converter (Axon Instruments) and the software pClamp (Version 6.0; Axon Instruments). Whole-cell Ca²⁺ currents were evoked by 100-ms membrane depolarization (range −60 to 60 mV in 10-mV increments) from a holding potential of −70 mV using an EPC-9 patch clamp amplifier and the Pulse software (Version 8.31; HEKA Elektronik, Lamprecht/Pfalz, Germany). Exocytosis was monitored in single β-cells as increases in cell membrane capacitance (sample rate 2.5 Hz) using the EPC-9 amplifier and the Pulse software. The standard whole-cell configuration was used throughout this study, except when mentioned otherwise.

The extracellular medium consisted of (in mmol/l) 138 NaCl, 5.6 KCl, 2.6 CaCl₂, 1.2 MgCl₂, 5 HEPES (pH 7.4 with NaOH), and 5 d-glucose. The extracellular solution used for recordings of whole-cell Ca²⁺ currents were (in mmol/l) 118 NaCl, 20 tetraethylammonium, 5.6 KCl, 10 CaCl₂, 5 HEPES, and 5 glucose. The volume of the recording chamber was 0.4 ml, and the solution entering the bath (1.5–2 ml/min) was maintained at 33°C. For measurements of standard whole-cell K⁺-channel activity, the pipette solution contained (in mmol/l) 125 KCl, 30 KOH, 10 EGTA, 1.3 MgCl₂, 5 HEPES, 0.3 Mg-ATP, and 0.3 K-ADP (pH 7.15). For recordings of perforated-patch whole-cell K⁺-channel activity, the following pipette-filling solution was used (in mmol/l): 76 K₂SO₄, 10 NaCl, 10 KCl, 1 MgCl₂, and 5 HEPES (pH 7.35, KOH). Electrical contact was established by adding 0.24 mg/ml amphotericin B to the pipette solution. Perforation required a few minutes; the voltage clamp was considered satisfactory when the series conductance was constant and >35–40 ns. For recordings of whole-cell Ca²⁺ currents, the pipette solution was composed of (in mmol/l) 125 Cs-glutamate, 1 MgCl₂, 2 CaCl₂, 10 EGTA, 5 HEPES, and 3 Mg-ATP (pH 7.15 with CsOH). Exocytosis was elicited by infusion of the following Ca²⁺/EGTA buffers through the recording pipette (in mmol/l): 125 K-glutamate; 10 KCl; 10 NaCl; 1 MgCl₂; 5 HEPES; 3 Mg-ATP; 10 EGTA; and 0, 5, or 8 CaCl₂. The [Ca²⁺]_i of the resulting buffers were 10, 220, or 870 nmol/l using the binding constants of Martell and Smith (9).

The following glycolipids were used: sulfatide (3′sulfogalactosylceramide), gal-cer, and GM1. All three molecules are composed of a hydrophilic sugar chain and a hydrophobic part consisting of ceramide, i.e., sphingosine and fatty acid. Gal-cer is the precursor of sulfatide and lacks the negatively charged sulfate group. The GM1 has a negative charge like sulfatide but in this case expressed by sialic acid. All of the glycolipids were isolated from porcine brain, and their structures were analyzed by fast atom bombardment mass spectrometry (10,11). It is important to emphasize that the composition of sulfatide used in this study consists of molecules, which differ in length and degree of saturation of the fatty acid chain (12). Stock solutions of each glycolipid were made in chloroform/methanol/water (60:30:4.5 vol/vol/vol) in concentrations of 1 μmol/ml. The tubes were sealed and after 1 h at room temperature, stored at 4°C. The glycolipids were transferred to glass tubes before use and evaporated at room temperature overnight. The dried glycolipids were redissolved in phosphate-buffered saline (pH 7.1) and sonicated for 20 s using a Branson sonicator (Danbury, CT). The glycolipids were used within 2 h.

The antisulfatide monoclonal antibody Sulph I was raised after immunizing BALB/c mice with sulfatide coated on Salmonella minnesota bacterial membrane (13). The subclass of the antibody is IgG1, and it reacts specifically with sulfatide and the related structures lactosylceramide sulfate and seminolipid, of which only sulfatide is present in adult islets (2). For controls, the following two monoclonal antibodies (both IgG1) were used: antigalactosylceramide (14) and antiglucose oxidase (Dako, Copenhagen, Denmark).

Sera from 11 patients with newly developed type 1 diabetes were taken within 1 week of diagnosis. The patients were four women and seven men with a mean age of 27 ± 7 years (range 17–38). All fulfilled the following criteria at time of diagnosis: random blood glucose concentration >12 mmol/l and significant ketonuria and glucosuria but not overweight. Insulin treatment was initiated on the day of diagnosis. The study was performed in accordance with the principles of the Declaration of Helsinki. Five of the patients displayed ASA in titers between 1:800 and 1:1,600 as determined by thin-layer chromatography (7). Six patients were ASA-negative. As additional control, ASAs were removed from three ASA-positive patient sera by passing sera six times through an octylsepharose column coated with sulfatide. In addition, ASA-positive patient serum was run twice through a protein A column (Pharmacia, Uppsala, Sweden) and the samples with (eluate) or without IgG were used for capacitance measurements. The control group consisted of three women and two men with an average age of 28 ± 10 years (range 23–46). All subjects were healthy with no known allergies and with no family history of type 1 diabetes. Furthermore, sera from five patients (three women and two men) with Guillain-Barré syndrome were examined. These patients had no history of type 1 diabetes. Their average age was 57 ± 7 years (range 42–75), and they displayed ASA titers between 1:800 and 1:3,200.

The [Ca²⁺]_i measurements were made using an Axiovert 135 inverted microscope equipped with a Plan-Neofluar 100×/1.30 objective (Carl Zeiss, Oberkochen, Germany) and an Ionoptix fluorescence imaging system (Milton, MA) as described elsewhere (15). Before the experiments, the cells were loaded with 0.2 μmol/l fura-2/AM (Molecular Probes, Eugene, OR) for 20 min.

Insulin release was measured at 37°C after static incubation. Groups of 10 size-matched islets isolated from Lewis rats were cultured overnight in RPMI 1640 medium and preincubated for 20 min in modified RPMI-1640 medium supplemented with 1% penicillin/streptomycin and 3 mmol/l glucose in 96-well Durapore membrane plates (Millipore, Molsheim, France). The preincubation medium was aspirated using a vacuum control pump (Millipore) and discarded. The islets were then resuspended in 200 μl of RPMI-1640 with different glucose concentrations and in the absence or presence of test compounds. At the end of the test incubation, the medium was aspirated and kept at −20°C until assayed for insulin using the enzyme-linked immunosorbent assay technique. Each data point represents the average of three individual experiments from different islet preparations, and each experiment was performed in quadruplicate.

The effects of test substances on insulin release are expressed in percentage of control. The exocytotic rate is presented as the increase in cell capacitance occurring during the first 60 s after establishment of the whole-cell configuration excluding any rapid changes occurring during the initial ∼10 s required for equilibration of the pipette solution with cytosol. All voltage signals were filtered at 500 Hz and sampled at a rate of 1 kHz. Results are presented as mean ± SE for the indicated number of experiments. Statistical significance was evaluated using Student’s t test for unpaired observations.

Figure 1A shows the effects of sulfatide (20 nmol/ml for 24 h) on insulin secretion from rat islets at increasing glucose concentrations in the extracellular medium. Sulfatide inhibited insulin release at ≥11.2 mmol/l glucose by 40% (P < 0.01). Short-term incubation with sulfatide (20 nmol/ml for 1 h) did not affect insulin release (data not shown). Figure 1B suggests that the inhibitory effect of sulfatide on insulin secretion is dependent on the dose. The islets were incubated for 24 h in the presence of 16.8 mmol/l glucose. No significant effect was observed at 5 and 10 nmol/ml, whereas maximal inhibition occurred at 20 nmol/ml sulfatide. Sulfatide did not affect insulin release when Ca²⁺ was depleted from the extracellular medium (control 2.6 ± 0.5 vs. 2.8 ± 0.9 ng · islet⁻¹ · h⁻¹ in the presence of sulfatide; data not shown). Finally, insulin secretion was not affected by the related glycolipids gal-cer and GM1. Under these conditions, insulin secretion amounted to 16.6 ± 1.5 (control), 16.2 ± 2.1 (gal-cer), and 16.8 ± 1.9 ng · islet⁻¹ · h⁻¹ (GM1).

Previous experiments have demonstrated that the resting conductance in the β-cell in the absence of glucose is principally made up by the activity of the K_ATP channels (16,17). To study the effects of sulfatide on K_ATP-channel activity in single rat β-cells, we used the standard whole-cell configuration and intracellular dialysis with 0.3 mmol/l ATP and 0.3 mmol/l ADP to activate the channels (Fig. 2A). It is interesting that the maximal membrane conductance obtained after full replacement of the cytoplasm with the pipette solution (obtained 2–3 min after establishment of the whole-cell configuration) was increased by 41% in cells pretreated for 30 min with sulfatide (20 nmol/ml) compared with control. On average, the membrane current increased from 103 ± 4 pA/pF (n = 6) under control conditions to 145 ± 9 pA/pF (P < 0.01; n = 5) in sulfatide-treated cells. Figure 2B depicts the time course for the stimulatory action of sulfatide on the whole-cell K_ATP-channel activity. The K_ATP current was increased by 15% only 10 min after inclusion of sulfatide in the extracellular solution, whereas a maximal stimulatory action was observed after 30 min of exposure to the glycolipid. Subsequent application of tolbutamide (100 μmol/l) reduced the conductance by 95 ± 6% (P < 0.001; n = 30), suggesting that the currents principally reflect the activity of K_ATP channels (data not shown).

Activation of K_ATP channels by sulfatide was also observed using the perforated-patch configuration. Under these experimental conditions, the whole-cell K⁺ conductance amounted to 1.2 ± 0.2 ns in the presence of 5 mmol/l glucose. Application of sulfatide (20 nmol/ml for 30 min) increased the membrane conductance to 1.7 ± 0.1 ns (P < 0.01; n = 6). Tolbutamide (100 μmol/l) reduced the current by 93 ± 4% (P < 0.001; n = 6; data not shown).

Incubation of β-cells with sulfatide (20 nmol/ml for 30 min) did not affect the activity of the voltage-gated Ca²⁺ currents recorded during 100-ms voltage-clamp depolarization from −70 mV to membrane potentials between −60 and 60 mV with 10-mV increments (Fig. 3). Similar results were observed after incubation of the cells with 20 μg/ml sulfatide for 24 h (data not shown).

A similar lack of effect of sulfatide was observed on glucose-induced increases in [Ca²⁺]_i. A brief elevation in the extracellular glucose concentration from 5 to 20 mmol/l produced a transient increase in [Ca²⁺]_i, which did not differ between the control and the sulfatide-treated groups. On average, [Ca²⁺]_i increased by 258 ± 25 nmol/l (n = 13) in control cells and by 253 ± 11 nmol/l (n = 11) in cells exposed to sulfatide (20 nmol/ml). Furthermore, no difference was observed in basal Ca²⁺: control 104 ± 6 nmol/l (n = 24) and sulfatide-treated 111 ± 6 nmol/l (n = 20; data not shown).

Figure 4A shows that infusion of a single rat β-cell with a Ca²⁺/EGTA buffer with a free Ca²⁺ concentration of 220 nmol/l produced a small increase in cell capacitance (indicative of exocytosis) after establishment of the whole-cell configuration. In general, the rate of exocytosis reached a new steady-state level within 2–4 min. It is evident that preincubation of cells with sulfatide (20 nmol/ml for 30 min) exerted a strong stimulation (4.5-fold) of exocytosis (Fig. 4A). On average (Fig. 4B), exocytosis increased from 2.0 ± 0.6 (n = 5) to 8.7 ± 0.8 fF/s (n = 6; P < 0.01) in cells that were pretreated with sulfatide. Similar results were observed after incubation of cells for 1 h with sulfatide (4.2-fold stimulation) and after inclusion of sulfatide in the pipette-filling solution (4.3-fold enhancement). No stimulation of exocytosis was observed with gal-cer (2.2 ± 0.5 fF/s; n = 5) or GM1 (1.9 ± 0.6 fF/s; n = 5; cells incubated with 20 nmol/ml for 30 min; data not shown).

Figure 4B also shows that sulfatide (20 nmol/ml for 30 min) did not affect exocytosis in cells infused with a pipette-filling solution without added Ca²⁺ ([Ca²⁺]_i ∼10 nmol/l) but produced a significant stimulation in the presence of high [Ca²⁺]_i (870 nmol/l; P < 0.05; n = 5). The stimulatory action of sulfatide on exocytosis was also observed in islets exposed to 250 μmol/l diazoxide, 25 mmol/l KCl, and 16.8 mmol/l glucose (Fig. 4C). Under these experimental conditions, which reveal the K_ATP-channel independent stimulation of insulin secretion (18), sulfatide (20 nmol/ml for 24 h) enhanced insulin release by 44% (P < 0.05).

In the presence of 16.8 mmol/l glucose, the monoclonal sulfatide antibody Sulph I produced a dose-dependent inhibition of insulin secretion from rat islets (Fig. 5A). No significant inhibition was observed at 5 or 10 μg/ml, whereas 20 μg/ml Sulph I produced a maximal inhibition of insulin release by 42% (P < 0.01). Sulph I did not affect insulin release when Ca²⁺ was depleted from the extracellular medium (control 2.2 ± 0.7 vs. 2.6 ± 0.8 ng · islet⁻¹ · h⁻¹ in the presence of Sulph I; data not shown). Furthermore, insulin secretion was not affected by Sulph I boiled for 30 min or the related monoclonal antibodies against gal-cer or glucose oxidase (both 20 μg/ml; data not shown).

Next we explored whether the inhibitory action of Sulph I on insulin secretion could be mimicked by ASA-positive diabetes sera from subjects with newly diagnosed type 1 diabetes. Indeed, in rat islets treated for 24 h with five different ASA-positive diabetes sera (dilution 1:10), glucose-induced insulin secretion was reduced by 34 ± 7% (P < 0.05) compared with islets treated with either five different control sera from healthy individuals or from ASA-negative patients with type 1 diabetes (Fig. 5B).

Figure 6A depicts recordings of whole-cell K_ATP currents under control conditions and in cells exposed to either 20 μg/ml Sulph I or ASA-positive serum (diluted 1:10) for 24 h. The treatment did not affect the magnitude of the whole-cell K_ATP conductance (Fig. 6B). A similar lack of effect on K_ATP-channel activity was observed in cells pretreated for only 30 min with Sulph I (Fig. 6B).

Figure 7A shows that incubation of rat β-cells with Sulph I (20 μg/ml for 30 min) produced a strong inhibition of exocytosis. Secretion was evoked by intracellular dialysis with a pipette solution of 870 nmol/l [Ca²⁺]_i, which in itself produced a high rate of exocytosis. On average (Fig. 7B), the rate of capacitance increase was decreased from 9.0 ± 0.7 (n = 7) to 3.1 ± 0.9 fF/s (n = 7; P < 0.01) in cells pretreated with Sulph I. A similar reduction in the exocytosis rate (from 8.0 ± 0.3 to 3.2 ± 0.3 fF/s; P < 0.01; n = 8 for both conditions) was observed in mouse β-cell pretreatment with 20 μg/ml Sulph I for 30 min. No changes in the exocytotic capacity were observed in cells preincubated with boiled Sulph I or with monoclonal antibodies against gal-cer or glucose oxidase (all 20 μg/ml; data not shown). Figure 7B also shows that Sulph I (20 μg/ml for 30 min) did not affect exocytosis in cells infused with a pipette-filling solution without added Ca²⁺ ([Ca²⁺]_i ∼10 nmol/l) but reduced the rate of increase in cell capacitance in the presence of 220 nmol/l [Ca²⁺]_i.

It is interesting that ASA-positive sera (Fig. 7C) exemplified by serum P6 also reduced exocytosis elicited by infusing cells with 870 nmol/l [Ca²⁺]_i to less than half of that observed in the presence of either a control serum from a healthy individual (P1) or from an ASA-negative subject with type 1 diabetes (P11). On average (Fig. 7D), exocytosis measured in cells treated with five different ASA-positive sera was reduced by 59 ± 8% (P < 0.01) compared with the rate of exocytosis recorded in the presence of either five different control sera or five different ASA-negative sera. It is interesting that the mean capacitance increase was reduced to an extent comparable to that observed for Sulph I–treated cells (59% vs. 65% for Sulph I–treated cells; Fig. 7B). A similar reduction in exocytosis was observed in cells pretreated for 4–6 h with the ASA-positive sera (62 ± 11% inhibition; n = 5).

Figure 7E shows that when ASA-positive serum (P8) was passed through a protein A column to remove the IgG fraction (P8 pel; titer <1:100), no inhibition of exocytosis was observed. By contrast, the ASA protein A IgG–absorbed fraction (P8 prot A fraction; titer 1:800) decreased the rate of exocytosis by 35% (P < 0.01; n = 5; Fig. 7E). Furthermore, removal of ASA by passage over a sulfatide-coated Sephadex column (three different ASA-positive sera) abolished the inhibitory effect on exocytosis (Fig. 7F; −ASA). Under these experimental conditions, the average rate of increase in exocytosis (8.7 ± 0.5 fF/s; n = 26) was not different from that of cells treated with control sera (8.4 ± 0.4 fF/s; n = 25; Fig. 7D). For clarity, the corresponding increases in exocytosis for ASA-positive sera are depicted in Fig. 7F (+ASA) and amounted to 3.4 ± 0.4 fF/s (P < 0.01; n = 26).

Patients with the neurological disease Guillain-Barré syndrome (inflammatory demyelinating polyradiculoneuropathy) display antibodies against sulfatide, which is present in the myelin sheets around the peripheral nerves (19). It is interesting that exposure of rat β-cells for 30 min to five different ASA-positive sera from patients with Guillain-Barré syndrome (diluted 1:10) did not affect the average rates of increase in exocytosis (rates between 8.4 and 8.8 fF/s compared with 8.0 fF/s under control conditions). Correspondingly, increasing the exposure time to 4–6 h did not affect exocytosis (8.9 ± 0.4 vs. 8.8 ± 0.3 fF/s for control; n = 5). In parallel experiments, however, exposure of cells to ASA-positive type 1 diabetes sera decreased the rate of exocytosis by 59% (data not shown). In agreement with these data, glucose-induced insulin secretion (16.8 mmol/l) from rat islets was not affected by a 24-h treatment with ASA-positive sera from patients with Guillain-Barré syndrome: control serum 15.1 ± 1.1 ng · islet⁻¹ · h⁻¹ and ASA-positive serum 15.6 ± 0.9 ng · islet⁻¹ · h⁻¹ (data not shown). Again, ASA-positive serum inhibited insulin secretion evoked by 16.8 mmol/l glucose by 37% (P < 0.05).

The rate of insulin secretion is dependent on the external glucose concentration and is secondary to its uptake and metabolism by the β-cell. The resulting increase in the ATP-to-ADP ratio leads to closure of the K_ATP channels, membrane depolarization, and activation of the voltage-dependent Ca²⁺ channels. The ensuing rise in the [Ca²⁺]_i stimulates, through a series of less well-defined steps, exocytosis of the insulin-containing granules. Here we demonstrate that sulfatide plays an important role in controlling glucose-induced insulin secretion and that the glycolipid interacts with both proximal and distal regulatory steps in the β-cell stimulus-secretion coupling.

It is known that exogenously added glycosphingolipids are inserted into the plasma membrane (20,21). It is reasonable to assume, therefore, that an increase in the plasma membrane concentration of sulfatide is responsible for the present observations. The one control glycolipid used, GM1, will also be incorporated into the plasma membrane. It is more water-soluble than sulfatide but does form micelles. The other control molecule, gal-cer, has a low solubility, so much less is taken up by the membrane and micelles are more stable.

The turnover rate of plasma membrane–associated glycosphingolipids is not known, and theoretically there is a possibility that exogenously added sulfatide, GM1, and gal-cer are endocytosed and degradation products produced in the lysosomes being responsible for the effects. However, this seems unlikely because they all give rise to the same final degradation products.

In the current study, we demonstrated that sulfatide produced a 40% increase in the whole-cell K_ATP conductance and caused inhibition of glucose-induced insulin secretion. It was demonstrated previously that acute addition of long-chain acyl-CoA esters, the active intracellular form of free fatty acids, activate K_ATP channels in the β-cell plasma membrane (22,23), and it is tempting to speculate that sulfatide stimulated K_ATP-channel activity by a related mechanism. It is interesting that ASA (Sulph I or ASA-positive serum) did not affect K_ATP-channel activity. The reason for this is unknown, but we speculate that binding of the antibody to sulfatide in the plasma membrane does not affect the ability of sulfatide to interact with the channel protein. Alternatively, the membrane areas around the K_ATP channels, unlike the exocytose-fusing membrane areas, are not associated with appreciable amounts of sulfatide unless exogenous sulfatide is added.

Stimulation of insulin secretion by direct modulation of the secretory process distal to a rise in [Ca²⁺]_i is physiologically important. For example, it was demonstrated previously that the hypoglycemic sulfonylureas accelerate Ca²⁺-dependent exocytosis in mouse β-cells and that the direct effect of the sulfonylurea tolbutamide on exocytosis may account for up to 75% of the total stimulatory action (24). It is interestingly that the effect of sulfatide on exocytosis is of the same magnitude as that of tolbutamide. However, it is important to emphasize that these effects on exocytosis are secondary to closure of the K_ATP channels and the associated increase in [Ca²⁺]_i. This is consistent with the lack of effect of sulfatide on exocytosis in the absence of intracellular Ca²⁺ and that stimulation of K_ATP-channel activity by sulfatide leads to inhibition of insulin secretion. We recently reported that acute stimulation with long-chain acyl-CoA enhances exocytosis in β-cells (25). It is likewise tempting to speculate, in analogy to the effects on K_ATP channels, that sulfatide may stimulate exocytosis by a similar mechanism.

We demonstrate that both Sulph I and ASA-positive sera from subjects with newly diagnosed type 1 diabetes inhibit Ca²⁺-induced exocytosis. Evidence that ASA in ASA-positive sera mediate the inhibitory action on exocytosis comes from the observations that ASA-negative sera from patients with type 1 diabetes did not affect the rate of capacitance increase. Furthermore, ASA-positive sera passed through a protein A column was without inhibitory effect on exocytosis, whereas the ASA effluent from the column decreased the exocytotic capacity of the β-cells to a similar extent as the corresponding ASA-positive sera. Passage over a sulfatide-coated octylsepharose column also removed the inhibitory activity. Finally, monoclonal antibodies of the IgG1 subtype against gal-cer and glucose oxidase did not affect exocytosis. These data support the observation that sulfatide is important for the fusion of secretory granules with the plasma membrane and suggest that the ASAs inhibit secretion by reducing the effective concentration of this lipid in the plasma membrane. This could result from an inability of sulfatide to participate in the exocytosis process upon binding to the antibody.

It was demonstrated previously that serum (IgM fraction) from patients with type 1 diabetes increase L-type Ca²⁺-channel activity in rat β-cells (26). The subsequent increase in the [Ca²⁺]_i was associated with apoptosis, and it was suggested that an IgM-mediated increase in Ca²⁺ influx might be part of the autoimmune reaction associated with type 1 diabetes (26). In contrast, treatment of β-cells in the present study with either sulfatide or Sulph I was not associated with a significant change in the amplitude of the whole-cell Ca²⁺ current. Similar negative data were obtained after exposure of β-cells to ASA-positive diabetes sera (data not shown).

The association of type 1 diabetes with Stiff-man syndrome and Guillain-Barré syndrome (27) suggests common antigens between the pancreatic β-cells and neural tissue. One such antigen might be sulfatide, which is also a neural epitope to which 65% of sera from patients with Guillain-Barré syndrome react (19). However, incubation of β-cells with ASA-positive sera from patients with Guillain-Barré syndrome did not modify the capacity of these cells to secrete. This suggests that the ASAs associated with the development of Guillain-Barré syndrome do not recognize the same epitope(s) as the ASA-positive sera from patients with type 1 diabetes. This may correspond well with the finding that the fatty acid composition of sulfatide in the brain is different from that of sulfatide in the islets, which has a high proportion of short fatty acid C16:0 and a lack of hydroxy fatty acids (12,28).

Collectively, our data are consistent with a role for sulfatide as a prime signal molecule that interacts with both proximal and distal regulatory steps in the β-cell stimulus-secretion coupling. When released from the secretory granules, sulfatide may act as a physiological negative feedback mechanism on insulin secretion from the individual β-cells or islets, but only when the glucose concentration is at a certain stimulatory level (see Fig. 1A). Our findings suggest that a fine balance between the amount of sulfatide in the β-cell plasma membrane and the level of circulating ASAs may control insulin secretion. This observation may have implications for the understanding of the pathophysiology of type 1 diabetes because ASAs, which impair insulin secretion, have been found in the majority of serum samples from patients with newly diagnosed type 1 diabetes but also in prediabetic individuals. Depending on the time of occurrence of ASAs during the prediabetic phase, which is presently unknown, it is possible that ASAs may stress either the healthy β-cells or β-cells already affected by the insulitis process.

The study was supported in part by a grant from Swedish Medical Research Council (K2000-03X-09909-09A).

We thank chief physician Henrik Permin for providing the Guillain-Barré sera and Jens Peter Stenvang for excellent technical assistance.