The sand rat (Psammomys obesus) is an animal model of nutritionally induced diabetes. We report here that several protein kinase C (PKC) isoforms (α, ε, and ζ, representing all three subclasses of PKC) are overexpressed in the skeletal muscle of diabetic animals of this species. This is most prominent for the ε isotype of PKC. Interestingly, increased expression of PKCε could already be detected in normoinsulinemic, normoglycemic (prediabetic) animals of the diabetes-prone (DP) line when compared with a diabetes-resistant (DR) line. In addition, plasma membrane (PM)–associated fractions of PKCα and PKCε were significantly increased in skeletal muscle of diabetic animals, suggesting chronic activation of these PKC isotypes in the diabetic state. The increased PM association of these PKC isotypes revealed a significant correlation with the diacylglycerol content in the muscle samples. Altered expression/activity of PKCε, in particular, may thus contribute to the development of diabetes in these animals; along with other PKC isotypes, it may be involved in the progression of the disease. This may possibly occur through inhibition of insulin receptor (IR) tyrosine kinase activity mediated by serine/threonine phosphorylation of the IR or insulin receptor substrate 1 (IRS-1). However, overexpression of PKCε also mediated downregulation of IR numbers in a cell culture model (HEK293), resulting in attenuation of insulin downstream signaling (reduced protein kinase B [PKB]/Akt activity). In accordance with this, we detected decreased 125I-labeled insulin binding, probably reflecting a downregulation of IR numbers, in skeletal muscle of Psammomys animals from the DP line. The number of IRs was inversely correlated to both the expression and PM-associated levels of PKCε. These data suggest that overexpression of PKCε may be causally related to the development of insulin resistance in these animals, possibly by increasing the degradation of IRs.

Psammomys obesus, often nicknamed “sand rat,” is a herbivorous desert gerbil living in the eastern Mediterranean and North Africa. The Jerusalem colony was established by domesticating the animals collected from the shores of the Dead Sea (1,2). In its native habitat, the Psammomys feeds on the halophilic plant Atriplex halimus and has never been found to be hyperglycemic or hyperinsulinemic. In captivity, it remains nondiabetic when fed a low energy (LE) diet containing 2.4 cal/g. However, when transferred to a relatively high energy (HE) diet of 3.0 cal/g, similar to the regular rodent diet, it gradually develops hyperinsulinemia and hyperglycemia. Four generally consecutive stages of progression to diabetes have been described (3). The basal stage A is followed by hyperinsulinemia without hyperglycemia (stage B), which precedes hyperinsulinemia with hyperglycemia (stage C), and lastly, stage D, marked by pancreatic insulin secretion collapse with apoptosis (4) and dependence on external insulin supply for survival.

Two Psammomys lines have been separated by selective breeding: the diabetes-resistant (DR) line, which remains normoglycemic and normoinsulinemic even on an HE diet, and the diabetes-prone (DP) line, which is susceptible to diabetes when exposed to the HE diet (4). The main difference between the two lines seems to be the cost of weight gain: the DP line uses 6.0 kcal/g for growth during 2 weeks after weaning, whereas the DR line requires 9.3 kcal/g (4). The DP animals exhibit insulin resistance even in the state of normoglycemia, as evidenced by the failure to induce hypoglycemia after external insulin administration and a minimal reduction of hepatic gluconeogenesis during the euglycemic-hyperinsulinemic clamp (5). Therefore, the insulin resistance in the Psammomys may be considered an innate characteristic of a desert animal according to the thrifty gene hypothesis (6).

Alterations in the expression level and/or activity of several protein kinase C (PKC) isoforms were found to be associated with insulin resistance in type 2 diabetic patients, animal models of diabetes, and different cell models (710). The serine/threonine kinase PKC consists of several isotypes, which according to structure and cofactor requirement for activation can be grouped into classic (α, βΙ, βΙΙ, γ), novel (δ, ε, θ, η, μ), and atypical (λ, ι, ζ) isoforms (1113). Evidence from in vitro studies indicated that several PKC isoforms can modulate insulin signaling at different levels of the signaling cascade. Several studies have suggested that PKC can inhibit insulin receptor kinase (IRK) activity through serine/threonine phosphorylation of the insulin receptor (IR) β-subunit (1417). Furthermore, classic PKC isoforms have been implicated in degradation of receptor tyrosine kinases, including the IR (18).

The aim of this study was to determine whether increased expression or chronic activation of PKC isoforms in insulin-responsive tissues (skeletal muscle and liver) are associated with the development of insulin resistance in an animal model of nutritionally induced diabetes, P. obesus. This animal system is extremely well characterized with respect to the development of diabetes and thus provides readily defined cohorts at various stages to determine whether alterations in PKC expression/activation are causally related to the development of the disease or a consequence of the diabetic milieu. Because in vitro studies have suggested that the IR could be a target for PKC-mediated downregulation, we used 125I-labeled insulin binding as an indirect measure of the number of IRs in skeletal muscle of these animals. Our data suggest that overexpression and chronic activation of PKCε might be causally related to the development and progression of diabetes in these animals and negatively regulate insulin signaling through degradation of the IR.

Materials.

Rabbit antipeptide antibodies against PKC isoforms α, βΙ, βΙΙ, δ, ε, θ, η, ι, and ζ and glutathione S-transferase (GST) were from Santa Cruz Biotechnology (Santa Cruz, CA). PKC l antibodies were from Transduction Laboratories (Lexington, KY). Phospho-Akt (Ser473) antibody and Akt antibody were from New England Biolabs (Beverly, MA). Polyclonal rabbit antisera against the IR were produced by immunization with synthetic peptides covering the last 15 COOH-terminal amino acids. Horseradish peroxidase–conjugated anti-mouse and anti-rabbit IgG were purchased from Amersham (Braunschweig, Germany).

Recombinant human insulin and [125I]TyrA14-labeled human insulin were obtained from Novo Nordisk (Bagsvaerd, Denmark).

The cDNAs for the type A IR (19), PKCε (Wang et al., unpublished data, Genbank accession number AF028009), and GST (kindly provided by R.A. Bradshaw) (20) were cloned into a cytomegalovirus promotor-enhancer-driven expression vector.

Experimental animals and dietary treatment.

Male and female Psammomys obesus from the DP and DR lines were bred at the Hebrew University-Hadassah Medical School animal farm. The animals were weaned at 3 weeks of age to an LE or HE diet (2.38 and 2.93 cal/g, respectively) manufactured by Koffolk, Petach (Tikva, Israel). Psammomys were housed in individual polypropylene cages; water and food were supplied ad libitum. All experimental procedures performed in this study were authorized by the Institutional Animal Care Committee.

Methods

Tissue and blood collection.

The animals were killed by decapitation in the fed state in the morning. Blood was collected for determination of serum glucose and triglyceride levels by enzymatic methods (Roche, Basel, Switzerland) and serum insulin levels by radioimmunoassay, using anti-human antibodies (Medgenix, Brussels, Belgium). The liver and gastrocnemius muscles from both legs were excised, immediately frozen in liquid nitrogen, and then stored at −80°C.

Preparation of lysates.

Muscles and livers were extracted in ice-cold solubilization buffer using a motor-driven Potter homogenizer (final concentrations: 20 mmol/l HEPES, 8 mmol/l EDTA, 0.2 mmol/l Na3VO4, 10 mmol/l Na4P207, 2.5 mmol/l phenylmethylsulfonyl fluoride, 1 mg/ml aprotinin, 2.5 mg/ml benzamidine, 2.5 μg/ml pepstatin, 2.5 μg/ml leupeptin, 160 mmol/l NaF, 2 mmol/l dichloroacetic acid, 1% Triton X-100, pH 7.4. After 20 min at 4°C, the samples were centrifuged at 20,000g for 60 min to remove insoluble material and the clarified extracts were stored at −80°C.

Subcellular fractionation.

In 2 ml of KH buffer (0.15 mol/l KCl, 20 mmol/l HEPES, pH 7.5, 1 mmol/l ATP), 0.2 g of powdered muscle was homogenized using a Polytron. Samples were maintained on ice during subsequent steps, and all centrifugations were performed at 4°C. KCl was added to a final concentration of 0.65 mol/l, followed by centrifugation at 700g for 10 min. The supernatant S1 was saved and the pellet was reextracted in 2 ml of KH buffer and centrifuged, yielding supernatant S2. S1 and S2 were combined in a total of 5 ml of KH buffer and centrifuged at 183,000g for 60 min. The supernatant was designated as the cytosolic fraction. The pellet was resuspended in 0.2 mol/l Tris-HCl, pH 7.4, 1 mmol/l EDTA, 1 mmol/l EGTA, 255 mmol/l sucrose, layered on a 38.3% sucrose gradient and centrifuged in a swing-out rotor at 60,000g for 40 min. The interphase was collected and centrifuged at 183,000g for 90 min. The pellet was resuspended in 5 ml of lysis buffer containing 20 mmol/l Tris-acetate, pH 7.0, 0.27 mol/l sucrose, 1 mmol/l EDTA, 1 mmol/l EGTA, 1 mmol/l 4-(2-aminoethyl)benzenesulfonyl fluoride, 1 mmol/l Na3VO4, 100 mmol/l NaF, 1% Triton X-100, 10 mmol/l Na4P2O7, 10 mmol/l C3H7O6PNa2, 1 mmol/l benzamidine, 1 mmol/l dithiothreitol, and 1 μmol/l leupeptin. The samples were centrifuged at 15,000g for 10 min to remove insoluble material and then stored at −80°C. Equal amounts (vol/vol) of cytosolic and membrane fractions were subjected to SDS-PAGE to determine the subcellular distribution of PKC isotypes.

Cell incubations.

Human embryonic kidney fibroblasts (HEK293; ATTC CRL 1573) were cultured in Dulbecco’s modified Eagle’s medium supplemented with 10% (vol/vol) fetal bovine serum (Life Technologies, Eggenstein, Germany), 10 mg/ml streptomycin, and 100 U/ml penicillin at 37°C in a 5% CO2-enriched, humidified atmosphere.

Cell transfection.

HEK293 cells were transiently transfected using CaCl2, as previously described by Chen and Okayama (21) and Gorman et al. (22). Twenty-four hours after transfection, the cells were starved overnight in medium containing 5 mmol/l glucose and 0.5% fetal bovine serum.

Cell lysis.

After stimulation, the cells were washed once with phosphate-buffered saline and homogenized in lysis buffer containing 20 mmol/l HEPES, pH 7.5, 150 mmol/l NaCl, 10% glycerol, 1.5 mmol/l MgCl2, 4 mmol/l EGTA, 1 mmol/l EDTA, 1% Triton X-100, 1 mmol/l AEBSF, 30 mmol/l Na4P2O4, 100 mmol/l Na3VO4, 10 mmol/l NaF, 5 μg/ml leupeptin, and 10 mmol/l benzamidine. After removal of cellular debris by centrifugation at 15,000g for 10 min at 4°C, the protein content in each sample was measured using Bio-Rad Protein Assay according to the manufacturer’s instructions (Bio-Rad, Richmond, CA).

Western blotting, enhanced chemiluminescence, and quantitation of bands.

Equal amounts of cell lysate were dissolved in 2× Laemmli buffer and subjected to SDS-PAGE. The proteins were transferred to nitrocellulose membranes (Schleicher & Schüll, BA85). Immunoreactive proteins were made visible using horseradish-peroxidase–coupled secondary antibodies and enhanced chemiluminescence reagents according to the manufacturer’s instructions (Amersham). The specific bands were quantitated using ImageQuant software (Molecular Dynamics, Sunnyvale, CA). For the Akt experiments, we used Fujifilm LAS1000 Luminescent Image Analyzer for detection of bands and Fujifilm Image Gauge V2.53 for quantification.

Determination of diacylglycerol (DAG) levels.

Powdered muscle tissue (100 mg) was extracted with 3 ml of chloroform/methanol (1:2, vol/vol) before addition of 0.8 ml of 1 mol/l NaCl. After monophase mixing, 1 ml of chloroform and 1 mol/l NaCl were added to break the phases. After centrifugation at 5,000g for 2 min, the chloroform phase was analyzed for DAG using the DAG kinase method (sn-1,2-DAG assay reagents system, Amersham). [32P] phosphatidic acid was separated by thin-layer chromatography, and the respective bands were scanned on a PhosphorImager and quantitated using ImageQuant software (Molecular Dynamics).

Insulin binding.

Insulin-binding activity was measured by the method described by Klein et al. (23), with a slight modification. Microwells coated with an anti-IR antibody were provided by Dr. H.H. Klein (Medical University of Lübeck, Lübeck, Germany). Lysate (60 μl) was added to each well and incubated for 16 h at 4°C. The lysates were then removed, the wells were washed five times with wash buffer (0.05% Triton X-100, 100 mmol/l NaCl, 2.5 mmol/l KCl, 1 mmol/l CaCl2, 0.1 mmol/l Na3VO4, 20 mmol/l HEPES, 10% glycerol, 0.05% NaN3, 0.5% BSA, pH 7.4), and 125I-insulin and unlabeled insulin (6.4 ng/ml and 121.6 ng/ml, respectively) were added. After 16 h at 4°C, the wells were washed again and bound 125I-insulin was subsequently collected by twice adding a solution of 2% SDS at room temperature for 30 min. The radioactivity in the collected SDS solution was measured by γ-counting.

Statistical analysis.

All data are presented as the mean ± SE. Group comparisons were made by unpaired Student’s t test. Correlations were assessed by Pearson’s least-squares method. P values less than 0.05 were considered significant.

Animal characteristics.

The general data on animals used in this study are summarized in Table 1. There was no significant difference in glucose and insulin values of animals in stage A, whether from the DR line maintained on an HE diet or DP animals maintained on an LE diet. As expected, the transfer of DP animals to the HE diet induced hyperinsulinemia and hyperglycemia. Well defined animal groups (n = 11–14 per group) were taken for PKC assays.

PKC isoforms α, ε, and ζ overexpressed in gastrocnemius muscles of Psammomys of the DP line.

We first aimed to determine the expression level of several PKC isoforms in skeletal muscle (gastrocnemius) tissues. Western blots of muscle lysates were probed with isozyme-specific antibodies for the classic PKC isoform α, βΙ, and βΙΙ, the novel PKC isoforms δ, ε, θ, and η, and the atypical PKC isoform ζ, ι, and λ. An example of such a Western blot is shown for a subset of the samples in Fig. 1. The quantitation and statistical analysis of the specific bands from all samples analyzed is summarized in Table 2. Expression levels of PKC isoforms α, ε, and ζ were significantly increased in the diabetic animals (DP/C) as compared with the DR line. Interestingly, there was also a significant increase in PKCε expression level in the normoinsulinemic, normoglycemic animals (which may therefore be termed the “prediabetic stage”) of the DP line as compared with the DR line. The values were calculated from 10–11 animals per group and are expressed as means ± SE. PKC isoforms βΙ, βΙΙ, θ, and η were not changed in the DP line, neither in the prediabetic nor in the diabetic state. The expression level of PKC isoforms δ, ι, and λ were undetectable by Western blotting or too low to allow quantitiation of the data.

Increased membrane association of PKCε in diabetic animals.

The Western blots had shown a significant increase in the expression level of several PKC isoforms, both in the prediabetic (PKCε) and the diabetic animals (PKCα, ε, ζ). To determine whether this was associated with increased activity of these isoforms, we performed subcellular fractionation of the muscle tissue. Translocation of PKCs from the cytosolic to the particulate fraction is a prerequisite of activation. Because of the lack of sensitive isoform–specific PKC activity assays, plasma membrane (PM) association of PKC is often used as a measurement of its activity. PM and cytosolic fractions of the muscle samples were run on Western blots and probed with PKC isoform–specific antibodies. The autoradiograms were scanned and quantitated using ImageQuant (Molecular Dynamics). The percentages of PM-associated PKCα, βΙ, βΙΙ, ε, η, and ζ in the three groups are shown in Table 3. Interestingly, the samples from diabetic animals revealed significant increases in the PKCα, βΙ, and ε content of the PM fraction, indicating a chronic activation of these PKC isotypes. Smaller increases in PKCα and ε PM association in the “prediabetic” animals were not significant within the study. PKCθ revealed the highest degree of association with the PM compared with other PKC isoforms; surprisingly, however, the percentage of PM-associated PKCθ decreased in diabetic animals. Despite significant increases in the expression level of PKCζ, we did not observe increased PM association of this isotype in the diabetic animals.

Increased PM association of PKC isoforms correlated with increased DAG levels.

Both classic and novel PKC isoforms require DAG for activation (12). To investigate whether the increased PM association of PKCα, θ, and ε could be caused by changes in DAG levels, we determined the DAG content in the muscle samples. Figure 2 shows linear regression analysis of PM association of PKC isoforms versus the DAG content in the individual samples. A significant correlation between these parameters was observed for α and ε isoforms of PKC. In contrast, we could not detect such a correlation for PKCζ. This was in agreement with the finding of unaltered PM content of PKCζ in the three different groups of animals (see Table 3). It was also expected because the activation of atypical PKC isoforms is independent of DAG (13).

Overexpression of PKCε preceding the onset of insulin resistance.

Interestingly, PKCε was already overexpressed in the prediabetic stage. To evaluate whether the increased PKCε expression level correlated with any of the parameters characteristic for insulin resistance/diabetes, we performed linear regression analysis of PKCε versus the serum levels of glucose, insulin, and triglycerides, respectively, in the individual samples. As demonstrated in Fig. 3, there was no significant correlation with any of these parameters.

PKCε expression in liver of diabetic Psammomys.

To determine whether the changes in PKCε expression levels were specific for skeletal muscle, we prepared liver extracts from the same animals to look at another insulin target tissue. No increase in PKCε expression was detectable in the group of prediabetic animals. The group of diabetic animals (DP/C) showed increased PKCε expression; however, this did not reach significance (DR, 7.5 ± 0.9; DP/A, 7.5 ± 0.9; DP/C, 10.1 ± 1.1, P = 0.08 vs. DR).

PKCε can mediate degradation of the IR and attenuates IR signaling.

A previous report suggested that PKCα was capable of degrading the IR (18). We were therefore interested to see whether a similar effect could be observed with PKCε. HEK293 fibroblasts were transfected either with the IR alone or with the IR and PKCε. As a control, GST was coexpressed in all samples. Cells were stimulated for 1 or 6 h with 1 μmol/l tetradecanoylphorbol acetate (TPA) to activate PKC and were subsequently lysed. Figure 4A shows a representative Western blot of lysates probed with an antibody specifically recognizing the IR β-subunit (upper panel), PKCε (middle panel), or GST (lower panel). Clearly, the endogenous background of PKCs has no effect on the expression level of the IR. However, when PKCε is coexpressed with the IR, there is a marked downregulation of IR expression, which is further enhanced after 6 h of TPA stimulation. PKCε itself is also partly downregulated after 6 h of TPA stimulation, in agreement with previous reports. No effect on the expression level of the GST could be detected, indicating that the changes in the expression level of the IR and PKCε are specific. A quantitation and statistical analysis of six experiments shown in Fig. 4B underlines the significance of this finding (P < 0.001). Similar effects could be detected for other classic and novel PKC isoforms, but not for atypical PKC isoforms (24).

Because PKCε overexpression and activation led to decreased IR expression, we would obviously expect an effect on IR downstream signaling. To test this hypothesis, we determined the activity of PKB/Akt, which upon insulin stimulation is activated through the IR via the phosphatidylinositol 3-kinase pathway. HEK293 cells were transfected with either the IR alone or a combination of IR and PKCε and stimulated with insulin or TPA plus insulin, respectively. Cells were then lysed; proteins were separated on SDS-PAGE and subjected to Western blotting. The respective bands of the IR and Akt were detected with specific antibodies and quantified. For Akt, we determined both the active (Akt-Ser) and total amount of protein. Figure 4C shows the result of a quantification of three independent experiments performed in duplicate (n = 6). Insulin-stimulated samples from cells expressing only the IR were set to 1, and all other samples were calculated accordingly. Coexpression and activation of PKCε led to a significant reduction of IR expression level (P < 0.001) and a concurrent inhibition of Akt activation (P < 0.01).

We therefore addressed the question of whether overexpression of PKCε could have a similar effect on the IR expression level in vivo. Because Psammomys have a very low number of IRs, we used a sensitive ELISA to indirectly determine the number of IRs, by measuring binding of 125I-insulin to the IR (23). There was a small but statistically significant reduction in 125I-insulin binding in lysates from the DP/C animals when compared with the DR animals (DR, 199.2 ± 9.2; DP/A, 188.8 ± 21.2; DP/C, 170.8 ± 8.5, P < 0.05 vs. DR). We therefore performed linear regression analysis of the IR expression level (125I-insulin binding) versus the PKCε expression levels. Figure 5A (2) shows that there is a significant inverse correlation between these two parameters (P < 0.05). This seemed to be specific for PKCε expression, because comparison of 125I-insulin binding to the expression levels of the other PKC isoforms (Fig. 5A) as well as to blood glucose and plasma insulin levels (data not shown) did not reveal any significant correlation. An even stronger correlation was detected between 125I-insulin binding and the PM-associated fraction of PKCε (Fig. 5B).

In this study, we provide evidence that several PKC isoforms (α, ε, ζ) are overexpressed in the skeletal muscle of diabetic P. obesus, taking the individuals from the DR line as a basal reference. This may indicate that this overexpression could be the primary reason for the insulin resistance in these animals. It is not a general phenomenon of all PKC isoenzymes, because, for example, PKC βΙ, βΙΙ, θ, and η expression levels are unchanged. Most notable is the expression of PKCε in even the prediabetic DP/A animals, which is consistent with the possibility that it could be causally related to the development of diabetes in these animals and may be an early event or mark an inherent susceptibility to the development of the disease. Overexpression of PKCε alone seems to have no consequences for the animals as long as they are restricted in their diet. Only after the introduction of a high-calorie diet is the increased expression of PKCε also reflected in higher PM association and thus presumably increased activity of the enzyme. The finding of increased plasma membrane association and therefore probable chronic activation of PKCε in diabetic DP/C, but not prediabetic DP/A animals, suggests that factors contributed by the diabetic milieu, i.e., hyperlipidemia, hyperglycemia, and/or hyperinsulinemia, are required to activate PKCε. Therefore, in light of a multifactorial origin for diabetes, the overexpression of PKCε might explain the genetic susceptibility of the DP line of P. obesus to the de-velopment of insulin resistance, which, together with environmental influences, may lead to development of the disease.

This PKC overexpression is particularly distinctive in the skeletal muscle of these animals but may also play a role in other insulin target tissues such as liver. In fact, when we determined the expression level of PKCε in the liver samples of the animals, we detected a trend of increased PKCε expression in the diabetic group; however, a larger number of samples would be needed to determine whether this was significant. Overexpression of PKC has previously been demonstrated in the liver of humans and rats with type 2 diabetes (8). In skeletal muscle of high-fat–fed rats (9), chronic activation of PKC isoforms ε and θ was observed. This was connected with decreased expression levels of PKCθ and no changes in PKCε expression level. In this study, the rats were fed a diet containing >30% fat, representing ∼60% of their caloric intake. They were insulin-resistant, as determined during an euglycemic-hyperinsulinemic clamp, but only mildly hyperinsulinemic or hyperglycemic. It was speculated that elevated DAG levels originating from lipolysis of two- to threefold increased triglyceride content in the muscles was responsible for increased PKC activation. This is different from the Psammomys that were fed a regular diet containing only ∼4% of fat, whose insulin resistance stemmed from a caloric load inappropriate to their metabolic capacity. Interestingly, however, we detected increased DAG levels in the DP line of P. obesus. The elevated DAG levels, possibly arising from increased synthesis and breakdown of glycerolipids, correlated well with membrane-associated fractions of PKCα and ε, suggesting that they may be causally related to the activation of these PKC isoforms. Increased DAG levels in aorta, heart, and retina of hypergalactosemic and diabetic dogs and rats have been suggested to be responsible for the activation of PKCβ (25,26). PKC enzyme activity and levels of immunoreactive PKCα, β, ε, and δ were shown to be increased in the membrane fractions of Goto-Kakizaki rats (10). Furthermore, an increased activity of PKC was linked to reduced IR phosphorylation in the liver of starved rats (27). These data, however, are in contrast to another study that found a decreased expression of PKCα, β, and ε in the soleus muscle of Zucker obese (fa/fa) rats (28). In most of these studies, the elevated expression/activation of PKC seemed to be a consequence of the diabetic milieu and therefore a late event in the development of the disease, correlating with muscle triglyceride levels (9) or hyperinsulinemia (10). The Psammomys have the advantage of being extremely well characterized with respect to the development of diabetes, and relatively homogeneous groups of animals representing the different stages (A, B, C, D) of the disease (4) can be studied. We can show that there is no correlation between PKC overexpression and hyperinsulinemia or serum triglyceride levels. In this animal model, overexpression of PKCε is clearly primary to the development of a diabetic milieu. Of course, it is possible that changes in the diabetic milieu add to the overexpression/activation of PKC in DP/C. Although it has been suggested that PKC activation arises from increased glucose incorporation into DAG (25,26), neither our own nor other studies (9,10) in diabetic animals show conclusive evidence that hyperglycemia could cause the observed changes. Hyperglycemia-induced changes of PKC activity (14) may, however, be an important factor in later stages of diabetes, in which PKCβ has been shown to be responsible for late complications of the disease (29).

Several studies in different cell lines as well as in vitro studies have suggested that PKC directly interferes with insulin signaling through serine/threonine phosphorylation of either the IR itself or one of its major substrates, the insulin receptor substrate 1 (IRS-1) (1417,30). There seems to be agreement that the ε isoform of PKC, in particular, is subject to regulation in diabetic individuals (710). Interestingly, an in vitro study provided evidence that PKCε may play an important role in mediating the inhibitory effect of tumor necrosis factor-α on insulin signaling (31,32), the major cause of obesity-linked insulin resistance. It is believed that serine/threonine phosphorylation of IRS-1, via a feedback loop, is able to inhibit the tyrosine kinase activity of the IR (33). In P. obesus, insulin resistance becomes considerably accentuated at stages B and C on the HE diet. This was shown to be associated with impaired IR function, which seems to be related to a decrease in the amount of phosphorylated tyrosine residues on the IR β-subunit (34). The extent of insulin-induced activation of the IRK was markedly reduced in stages B and C, using stage A as a reference. Hyperinsulinemia by itself was shown to be a detrimental factor for IR activation (34), because development of hyperinsulinemia, even without hyperglycemia, was associated with a decrease in insulin-induced IRK activation in liver and muscle. In the skeletal muscle, hyperinsulinemia was accompanied by a reduction in basal IRK activity. Diet restriction leading to the return of insulin level to normal restored the IRK activity (4,34). Therefore, the overexpression of PKC, particularly PKCε, may be an additional factor inducing the IR malfunction. In the present study, we provide evidence that, in addition to serine/threonine phosphorylation of the IR or IRS-1, a downregulation of IR numbers could be a possible mechanism for induction of insulin resistance. Skeletal muscle of diabetic Psammomys DP/C showed a reduction in 125I-insulin binding, most probably representing a decreased number of IRs, although we cannot completely rule out changes in the affinity for insulin. In fact, there was a clear correlation between PKCε expression levels or activity and 125I-insulin binding. This is in good agreement with previous reports showing that PKCα is capable of downregulating tyrosine kinase receptors (18) and that several PKC isoforms mediate degradation of the IR (24). The significant inverse correlation between insulin binding and PKCε expression further suggests that the PKC-induced downregulation of IR number is a causal factor in insulin resistance characterizing P. obesus. Insulin resistance is generally believed to have a major impact on skeletal muscle, liver, and adipose tissue. Surprisingly, mice with a targeted disruption of the IR gene in skeletal muscle displayed elevated levels of fat mass, serum triglycerides, and free fatty acids but never developed diabetes, suggesting that insulin sensitivity in muscle may not be the primary cause for the development of diabetes (35). However, the knockout mice reflect an isolated effect of IR expression in skeletal muscle, whereas in Psammomys, we study the progression of the disease in a much more complex situation of nutritionally induced diabetes. We describe only one aspect of several factors that may lead to insulin resistance in these animals. Other factors, such as the inhibition of IR tyrosine phosphorylation (34) and effects from other tissues, are contributing to the pathogenesis of diabetes. This is also a likely scenario in the context of a growing number of reports that suggest that type 2 diabetes is a multifactorial disease (36). In particular, this has been very elegantly demonstrated in heterozygotes combining several knockouts of genes of the insulin signaling pathway. It is also worthwhile to note that Psammomys has a very small number of IRs compared with the Albino rat (34). Therefore, even a minor change in IR number could have an influence on insulin signaling. In addition, the impact of decreased IR number could become more overt if, at the same time, IR tyrosine phosphorylation is inhibited (34).

Smaller changes in PKCε expression level and insulin binding (data not shown) were found in the liver of Psammomys. The hepatic insulin resistance is manifest mainly by nonsuppression of gluconeogenesis by insulin. This aspect of insulin resistance may not be exclusively related to the overexpression of PKCε.

In summary, we have demonstrated changes in the expression level and activity of PKC in gastrocnemius muscle of P. obesus. Our data suggest that, in particular, PKCε could be causally related to the development of insulin resistance in these animals. However, our correlative studies do not provide evidence of causality between PKCε activation and in vivo insulin resistance, although the proposed mechanism of its inhibitory action on insulin signaling through downregulation of IR numbers is consistent with such a model. Further work is required to establish a causative link between the observed changes in PKC expression/activity and the development of insulin resistance.

FIG. 1.

Expression of PKCα, ε, θ, and ζ in gastrocnemius muscle. Muscle lysates were subjected to SDS-PAGE followed by Western blotting and then probed with PKC isotype–specific antibodies. A representative panel for each isotype is shown.

FIG. 1.

Expression of PKCα, ε, θ, and ζ in gastrocnemius muscle. Muscle lysates were subjected to SDS-PAGE followed by Western blotting and then probed with PKC isotype–specific antibodies. A representative panel for each isotype is shown.

FIG. 2.

Correlation of plasma membrane–associated PKC to DAG levels in gastrocnemius muscle. DAG levels and PM-associated PKC levels were determined as described in research design and methods and the correlations in the individual Psammomys samples (n = 16) are plotted. Correlation coefficients (r) and P values are indicated.

FIG. 2.

Correlation of plasma membrane–associated PKC to DAG levels in gastrocnemius muscle. DAG levels and PM-associated PKC levels were determined as described in research design and methods and the correlations in the individual Psammomys samples (n = 16) are plotted. Correlation coefficients (r) and P values are indicated.

FIG. 3.

Correlation of PKCε expression level to serum glucose, insulin, and triglyceride levels. •, nondiabetic animals (DR + DP/A); ○, diabetic animals (DP/C). There was no significant correlation with any of the parameters in nondiabetic animals (•) nor in all the animals (• + ○).

FIG. 3.

Correlation of PKCε expression level to serum glucose, insulin, and triglyceride levels. •, nondiabetic animals (DR + DP/A); ○, diabetic animals (DP/C). There was no significant correlation with any of the parameters in nondiabetic animals (•) nor in all the animals (• + ○).

FIG. 4.

Effect of PKCε overexpression on IR expression level and Akt activation. A: HEK293 cells were transfected with IR expression plasmid alone or in combination with PKCε expression plasmid as indicated and starved overnight in medium containing 0.5% fetal calf serum. As a control, we cotransfected a GST construct in all samples. The cells were treated with 1 μmol/l TPA for the times indicated. Cell lysates were subject to SDS-PAGE, transferred to nitrocellulose, and incubated with anti-IR–specific antibodies (upper panel). Proteins were made visible using horseradish-peroxidase–coupled secondary antibodies and the enhanced chemiluminescence detection method. The middle panel shows PKCε expression levels, and the lower panel shows GST expression levels, as a control. B: Quantitation and statistical analysis of IR expression levels. Data are expressed as percentage of control (IR alone, no stimulation) and represent means ± SE of n = 6 independent experiments. P values were determined using Student’s t test. *P < 0.05 vs. IR alone, ***P < 0.001 vs. both IR alone and IR + PKCε unstimulated samples. C: HEK293 cells were transfected with IR expression plasmid alone or in combination with PKCε expression plasmid as indicated and starved overnight in medium containing 0.5% fetal calf serum. The cells were left untreated or stimulated with 107 mol/l insulin for 5 min or treated with 1 μmol/l TPA for 6 h before 107 mol/l insulin stimulation for 5 min, as indicated. Cell lysates were subjected to SDS-PAGE, transferred to nitrocellulose, and incubated with specific antibodies against the IR β-subunit, Akt, or phospho-Akt (Ser 473). Proteins were made visible using horseradish-peroxidase–coupled secondary antibodies and the enhanced chemiluminescence detection method on Fujifilm LAS1000 Luminescent Image Analyzer and were quantified using the program Fujifilm Image Gauge. The diagram shows the means ± SD of three independent experiments, performed in duplicate. The data are normalized to the expression levels in cells transfected with IR alone and stimulated with insulin.

FIG. 4.

Effect of PKCε overexpression on IR expression level and Akt activation. A: HEK293 cells were transfected with IR expression plasmid alone or in combination with PKCε expression plasmid as indicated and starved overnight in medium containing 0.5% fetal calf serum. As a control, we cotransfected a GST construct in all samples. The cells were treated with 1 μmol/l TPA for the times indicated. Cell lysates were subject to SDS-PAGE, transferred to nitrocellulose, and incubated with anti-IR–specific antibodies (upper panel). Proteins were made visible using horseradish-peroxidase–coupled secondary antibodies and the enhanced chemiluminescence detection method. The middle panel shows PKCε expression levels, and the lower panel shows GST expression levels, as a control. B: Quantitation and statistical analysis of IR expression levels. Data are expressed as percentage of control (IR alone, no stimulation) and represent means ± SE of n = 6 independent experiments. P values were determined using Student’s t test. *P < 0.05 vs. IR alone, ***P < 0.001 vs. both IR alone and IR + PKCε unstimulated samples. C: HEK293 cells were transfected with IR expression plasmid alone or in combination with PKCε expression plasmid as indicated and starved overnight in medium containing 0.5% fetal calf serum. The cells were left untreated or stimulated with 107 mol/l insulin for 5 min or treated with 1 μmol/l TPA for 6 h before 107 mol/l insulin stimulation for 5 min, as indicated. Cell lysates were subjected to SDS-PAGE, transferred to nitrocellulose, and incubated with specific antibodies against the IR β-subunit, Akt, or phospho-Akt (Ser 473). Proteins were made visible using horseradish-peroxidase–coupled secondary antibodies and the enhanced chemiluminescence detection method on Fujifilm LAS1000 Luminescent Image Analyzer and were quantified using the program Fujifilm Image Gauge. The diagram shows the means ± SD of three independent experiments, performed in duplicate. The data are normalized to the expression levels in cells transfected with IR alone and stimulated with insulin.

FIG. 5.

A: Correlation between PKC expression levels and 125I-insulin binding in muscle homogenates. PKC expression levels and 125I-insulin binding in the individual Psammomys samples (n = 22) are plotted. B: Linear regression analysis of PKCε associated to plasma membrane fraction versus 125I-insulin binding in muscle homogenates. PKCε associated to PM fraction and 125I-insulin binding in the individual Psammomys samples (n = 22) are plotted. Correlation coefficients (r) and P values are indicated.

FIG. 5.

A: Correlation between PKC expression levels and 125I-insulin binding in muscle homogenates. PKC expression levels and 125I-insulin binding in the individual Psammomys samples (n = 22) are plotted. B: Linear regression analysis of PKCε associated to plasma membrane fraction versus 125I-insulin binding in muscle homogenates. PKCε associated to PM fraction and 125I-insulin binding in the individual Psammomys samples (n = 22) are plotted. Correlation coefficients (r) and P values are indicated.

TABLE 1

Animal characteristics

Animal/stage (n)Weight (g)Maintenance dietSerum glucose (mg/dl)Insulin (mU/l)Triglyceride (mg/dl)
DR (13) 193 ± 9 HE 86 ± 5 104 ± 19 201 ± 42 
DP/A (11) 153 ± 6 LE 75 ± 4 75 ± 10 134 ± 14 
DP/C (14) 192 ± 5 HE 260 ± 13* 421 ± 40* 225 ± 35 
Animal/stage (n)Weight (g)Maintenance dietSerum glucose (mg/dl)Insulin (mU/l)Triglyceride (mg/dl)
DR (13) 193 ± 9 HE 86 ± 5 104 ± 19 201 ± 42 
DP/A (11) 153 ± 6 LE 75 ± 4 75 ± 10 134 ± 14 
DP/C (14) 192 ± 5 HE 260 ± 13* 421 ± 40* 225 ± 35 

Data are means ± SE.

*

P < 0.01 vs. DR line

P < 0.05

P < 0.01 vs. DP/A.

TABLE 2

Expression of PKC isotypes

Total PKC Expression (arbitrary units)
DRDP/ADP/C
PKCα 1.00 ± 0.20 1.02 ± 0.12 1.54 ± 0.12* 
PKCβI 1.00 ± 0.06 1.02 ± 0.08 1.00 ± 0.03 
PKCβII 1.00 ± 0.29 1.46 ± 0.36 1.09 ± 0.29 
PKCδ ND ND ND 
PKCε 1.00 ± 0.14 1.59 ± 0.16 1.82 ± 0.16 
PKCθ 1.00 ± 0.07 1.16 ± 0.11 1.28 ± 0.13 
PKCη 1.00 ± 0.13 1.17 ± 0.06 1.09 ± 0.10 
PKCζ 1.00 ± 0.17 1.67 ± 0.32 1.50 ± 0.14* 
PKCι ND ND ND 
PKCγ ND ND ND 
Total PKC Expression (arbitrary units)
DRDP/ADP/C
PKCα 1.00 ± 0.20 1.02 ± 0.12 1.54 ± 0.12* 
PKCβI 1.00 ± 0.06 1.02 ± 0.08 1.00 ± 0.03 
PKCβII 1.00 ± 0.29 1.46 ± 0.36 1.09 ± 0.29 
PKCδ ND ND ND 
PKCε 1.00 ± 0.14 1.59 ± 0.16 1.82 ± 0.16 
PKCθ 1.00 ± 0.07 1.16 ± 0.11 1.28 ± 0.13 
PKCη 1.00 ± 0.13 1.17 ± 0.06 1.09 ± 0.10 
PKCζ 1.00 ± 0.17 1.67 ± 0.32 1.50 ± 0.14* 
PKCι ND ND ND 
PKCγ ND ND ND 

Data are means ± SEM. P values were determined using Student’s t test.

*

P < 0.05

P < 0.01

P < 0.001. ND, not detectable.

TABLE 3

Subcellular distribution of PKC isotypes

Membrane/(Cytosolic + Membrane) (%)
DRDP/ADP/C
PKCα 26.1 ± 5.9 31.6 ± 4.1 39.6 ± 1.7* 
PKCβI 12.8 ± 2.0 17.7 ± 2.6 20.8 ± 1.3 
PKCβII 26.5 ± 3.7 32.1 ± 6.8 34.7 ± 3.1 
PKCε 26.7 ± 1.6 27.5 ± 1.5 33.5 ± 0.7§ 
PKCθ 72.9 ± 2.4 72.9 ± 4.0 65.5 ± 1.9* 
PKCη 16.4 ± 3.1 16.1 ± 4.9 18.8 ± 1.6 
PKCζ 35.8 ± 5.0 34.4 ± 6.0 39.6 ± 3.4 
Membrane/(Cytosolic + Membrane) (%)
DRDP/ADP/C
PKCα 26.1 ± 5.9 31.6 ± 4.1 39.6 ± 1.7* 
PKCβI 12.8 ± 2.0 17.7 ± 2.6 20.8 ± 1.3 
PKCβII 26.5 ± 3.7 32.1 ± 6.8 34.7 ± 3.1 
PKCε 26.7 ± 1.6 27.5 ± 1.5 33.5 ± 0.7§ 
PKCθ 72.9 ± 2.4 72.9 ± 4.0 65.5 ± 1.9* 
PKCη 16.4 ± 3.1 16.1 ± 4.9 18.8 ± 1.6 
PKCζ 35.8 ± 5.0 34.4 ± 6.0 39.6 ± 3.4 

Data are means ± SE. P values were determined using Student’s t test.

*

P < 0.05

P < 0.01

P < 0.001 vs. DR

§

P < 0.01 vs. DP/A.

We thank Patricia Holst for critical reading of the manuscript; K. Seedorf for valuable discussions throughout the study; A. Ullrich for IR cDNAs; H. Mischak, A. Ullrich, and T. Biden for PKC cDNAs; and H. Klein for ELISA plates coated with IR antibodies. The Hagedorn Research Institute is an independent basic research component of Novo Nordisk A/S.

1.
Ziv E, Shafrir E (Eds.): Psammomys obesus (sand rat): nutritionally induced NIDDM-like syndrome on a thrifty gene background. In
Lessons From Animal Diabetes
. Vol. 5. London, Smith-Gordon,
1995
, p.
285
–300
2.
Shafrir E: Development and consequences of insulin resistance: lessons from animals with hyperinsulinaemia.
Diabete Metab
22
:
122
–131,
1996
3.
Kalderon B, Gutman A, Shafrir E, Adler JH: Characterization of stages in the development of obesity-syndrome in sand rat (Psammomys obesus).
Diabetes
35
:
717
–724,
1986
4.
Bar-On H, Ben Sasson R, Ziv E, Shafrir E: Irreversibility of nutritionally induced NIDDM in Psammomys obesus is related to β-cell apoptosis.
Pancreas
18
:
259
–265,
1999
.
5.
Ziv E, Kalman R, Hershkop K, Barash V, Shafir E, Bar-On H: Insulin resistance in the NIDDM model Psammomys obesus in the normoglycemic, normoinsulinemic state.
Diabetologia
39
:
1269
–1275,
1996
6.
Neel JV: Diabetes mellitus: a “thrifty” genotype rendered detrimental by “progress.” Am J Hum Genet
14
:
353
–362,
1962
7.
Tang EY, Parker PJ, Beattie J, Houslay MD: Diabetes induces selective alterations in the expression of protein kinase C isoforms in hepatocytes.
FEBS Lett
326
:
117
–123,
1993
8.
Considine RV, Nyce MR, Allen LE, Morales LM, Triester S, Serrano J, Colberg J, Lanza-Jacoby S, Caro JF: Protein kinase C is increased in the liver of humans and rat with non-insulin-dependent diabetes mellitus: an alteration not due to hyperglycemia.
J Clin Invest
95
:
2938
–2944,
1995
9.
Schmitz-Peiffer C, Browne CL, Oakes ND, Watkinson A, Chisholm DJ, Kraegen EW, Biden TJ: Alterations in the expression and cellular localization of protein kinase C isozymes ε and θ are associated with insulin resistance in skeletal muscle of the high-fat–fed rat.
Diabetes
46
:
169
–178,
1997
10.
Avignon A, Yamada K, Zhou X, Spencer B, Cardona O, Saba-Siddique S, Galloway L, Standaert ML, Farese RV: Chronic activation of protein kinase C in soleus muscle and other tissues of insulin-resisant type II diabetic Goto-Kakizaki (GK), obese/aged, and obese/Zucker rats: a mechanism for inhibiting glycogen synthesis.
Diabetes
45
:
1396
–1404,
1996
11.
Nishizuka Y: The molecular heterogeneity of protein kinase C and its implications for cellular regulation.
Nature
344
:
661
–665,
1988
12.
Newton AC: Protein kinase C: structure, function and regulation.
J Biol Chem
270
:
28495
–28498,
1995
13.
Mellor H, Parker P: The extended protein kinase C superfamily.
Biochem J
332
:
281
–292,
1998
14.
Berti L, Musthaf L, Kroder G, Kellerer M, Tippmer S, Mushak J, Seffer E, Seedorf K, Haring H: Glucose-induced translocation of protein kinase C isoforms in rat-1 fibroblasts is paralleled by inhibition of the insulin receptor tyrosine kinase.
J Biol Chem
269
:
3381
–3386,
1994
15.
Bossenmaier B, Mosthaf L, Mischak H, Ullrich A, Häring HU: Protein kinase C isoforms β1 and β2 inhibit the tyrosine kinase activity of the insulin receptor.
Diabetologia
40
:
863
–866,
1997
16.
Danielsen AG, Liu F, Hosomi Y, Shii K, Roth RA: Activation of protein kinase Cα inhibits signaling by members of the insulin receptor family.
J Biol Chem
270
:
21600
–21605,
1995
17.
Takayama S, White MF, Lauris V, Kahn R: Phorbol esters modulate insulin receptor phosphorylation and insulin action in cultured hepatoma cells.
Proc Natl Acad Sci U S A
81
:
7797
–7801,
1984
18.
Seedorf K, Shearman M, Ullrich A: Rapid and long term effects of protein kinase C on receptor tyrosine kinase phosphorylation and degradation.
J Biol Chem
270
:
18953
–18960,
1995
19.
Ebina Y, Ellis L, Jarnagin K, Edery M, Graf L, Clauser E, Ou JH, Masiarz F, Kan YW, Goldfine ID, et al.: The human insulin receptor cDNA.
Cell
46
:
747
–758,
1985
20.
Thomas D, Bradshaw RA: Differential utilization of ShcA tyrosine residues and functional domains in the transduction of epidermal growth factor-induced mitogen-activated protein kinase activation in 293T cells and nerve growth factor-induced neurite outgrowth in PC12 cells.
J Biol Chem
272
:
22293
–22299,
1997
21.
Chen C, Okayama H: High-efficiency transformation of mammalian cells by plasmid DNA.
Mol Cell Biol
7
:
2745
–2752,
1987
22.
Gorman CM, Gies D, McCray G, Huang M: The human cytomegalovirus major immediate early promoter can be trans-activated by adenovirus early proteins.
Virology
171
:
377
–385,
1989
23.
Klein HH, Kowalewski B, Drenckhan M, Neugebauer S, Matthaei S, Kotzke G: A microtiter well assay system to measure insulin activation of insulin receptor kinase in intact mononuclear cells: decreased insulin effect in cells from patients with NIDDM.
Diabetes
42
:
883
–890,
1993
24.
Busch AK: Conventional and novel protein kinase C (PKC) isoforms interact with and mediate degradation of the insulin receptor. PhD Thesis. Copenhagen, University of Copenhagen,
1999
25.
Xia P, Inoguchi T, Kern TS, Engerman RL, Oates PJ, King GL: Characterization of the mechanism for the chronic activation of diacylglycerol-protein kinase C pathway in diabetes and hypergalactosemia.
Diabetes
43
:
1122
–1129,
1994
26.
Inoguchi T, Battan R, Handler E, Sportsman JR, Heath W, King GL: Preferential elevation of protein kinase C isoform beta II and diacylglycerol levels in the aorta and heart of diabetic rats: differential reversibility to glycemic control by islet cell transplantation.
Proc Natl Acad Sci U S A
89
:
11059
–11063,
1992
27.
Karasik A, Rothenberg PL, Yamada K, White MF, Kahn CR: Increased protein kinase C activity is linked to reduced insulin receptor autophosphorylation in liver of starved rats.
J Biol Chem
265
:
10226
–10231,
1990
28.
Cooper DR, Watson JE, Dao ML: Decreased expression of protein kinase C alpha, beta, and epsilon in soleus muscle of Zucker obese (fa/fa) rats.
Endocrinology
133
:
2241
–2247,
1993
29.
Ishii H, Jirousek MR, Koya D, Takagi C, Xia P, Clermont A, Bursell SE, Kern TS, Ballas LM, Heath WF, Stramm LE, Feener EP, King GL: Ameloriation of vascular dysfunctions in diabetic rats by an oral PKC beta inhibitor.
Science
272
:
728
–731,
1996
30.
Tanti JF, Gremeaux T, Van Obberghen E, Le Marchand Brustel Y: Serine/threonine phosphorylation of insulin receptor substrate 1 modulates insulin receptor signaling.
J Biol Chem
269
:
6051
–6057,
1994
31.
Kellerer M, Mushack J, Mischak H, Häring HU: Protein kinase C (PKC) epsilon enhances the inhibitory effect of TNF alpha on insulin signaling in HEK293 cells.
FEBS Lett
418
:
119
–122,
1997
32.
Hotamisligil GS, Murray D, Choy LN, Spiegelman BM: Tumor necrosis factor α inhibits signaling from the insulin receptor.
Proc Natl Acad Sci U S A
91
:
4854
–4858,
1994
33.
Hotamisligil GS, Budavari A, Murray DL, Spiegelman B: Reduced tyrosine kinase activity of the insulin receptor in obesity-diabetes.
J Clin Invest
94
:
1543
–1549,
1994
34.
Kanety H, Moshe S, Shafrir E, Lunenfeld B, Karasik A: Hyperinsulinemia induces a reversible impairment in receptor function leading to diabetes in the sand rat model of non-insulin-dependent diabetes mellitus.
Proc Natl Acad Sci U S A
91
:
1853
–1857,
1994
35.
Bruning JC, Michael MD, Winnay JN, Hayashi T, Horsch D, Accili D, Goodyear LJ, Kahn CR: A muscle-specific insulin receptor knockout exhibits features of the metabolic syndrome of NIDDM without altering glucose tolerance.
Mol Cell
2
:
559
–569,
1998
36.
Withers DJ, White M: Perspective: the insulin signaling system: a common link in the pathogenesis of type 2 diabetes.
Endocrinology
141
:
1917
–1921,
2000

Address correspondence and reprint requests to Luitgard Mosthaf-Seedorf, Lilly Research Laboratories, Essener Str. 93, 22419 Hamburg, Germany. E-mail: seedorf_luitgard@lilly.com.

Received for publication 18 March 1999 and accepted in revised form 27 October 2000.