Autoimmune Type 1 diabetes mellitus (T1DM) is caused by specific destruction of the insulin-producing beta cells in the islets of Langerhans in the pancreas [1]. During this process the islets are infiltrated with macrophages and lymphocytes, and these cells release a mixture of cytokines, such as interleukin-1β (IL-1β), tumour necrosis factor-α (TNF-α) and interferon-γ (IFN-γ) to which beta cells are particularly sensitive [2]. Cytokines have been shown to induce different free radicals [3] including nitric oxide (NO), catalyzed by inducible nitric oxide synthase (iNOS), as described in the Copenhagen Model of beta-cell destruction in T1DM [4].

During development of the pancreas, all four endocrine cell-types (α, β, δ and pp cells) are believed to arise from the same stem cell [5] and this specialization is dependent upon activation of different transcription factors [6, 7, 8, 9]. Since sensitivity to cytokines, free radicals and toxic chemicals is exquisite for the beta cell, this could reflect an acquired trait during the maturation of stem cells into mature insulin producing beta cells. We therefore hypothesize, that PiPP leading to susceptibility to IL-1β-damage is a consequence of beta-cell maturation. Here we address this hypothesis by using a glucagon producing pre-beta-cell cell line (NHI-glu), which following in vivo passage as a subcutaneous tumour in a rat, was re-established as an insulin producing beta-cell line (NHI-ins) [10, 11]. Our previous analyses of these two phenotypes showed that this maturation process was accompanied by an acquired sensitivity to the toxic effects of IL-1β [12]. Hence, the aim of our study was to characterize by proteome analysis [13, 14, 15] changes in protein expression accompanying the final maturation from the glucagon positive and IL-1β resistant pre-beta-cell phenotype to the insulin positive and IL-1β sensitive beta-cell phenotype and to relate these data to our previous analysis of PiPP in IL-1β exposed rat islets. [14, 16].

Materials and methods

Cell culture

The NHI-cell system [12] is based on the subclone NHI-glu derived from the glucagon-producing MSL-G2 culture [17]. Following in vivo passage by transplantation into syngeneic NEDH rats, the NHI-glu cells mature into insulinomas [10, 18]. The insulinomas re-established in vitro have an insulin-producing phenotype (NHI-ins) for prolonged periods of time [19], closely resembling beta cells with respect to the mRNA expression profile [20].

The two NHI-phenotypes were cultured in RPMI 1640 Glutamax (GibcoBRL Life Technologies, Paisley, Scotland, UK) supplemented with 10% FCS (GibcoBRL Life Technologies) and 1% penicillin/streptomycin (GibcoBRL Life Technologies) at 37°C in a 5% CO2 atmosphere in four independent experiments. For metabolic labelling 2×105 cells/well were set up in 24-well plate (Costar, Cambridge, Mass., USA) containing 3 ml media. To prepare a large amount of protein for preparative gels and MS, cells were cultured in 50 ml tissue-culture flasks (107 cells/bottle) for 4 days.

Cell labelling

Cells used for 2D-GE analysis were cultured for 44 h and then washed twice in HBSS and labelled for 4 h in 250 µl/well methionine-free Dulbecco’s modified Eagle’s medium (DMEM) [21] with 10% dialyzed NHS and 18.5 MBq/ml [35S]-methionine (SJ 204 Amersham International, Amersham, UK). After labelling the cells were washed twice in HBSS and lysed with 100 µl lysis buffer (8.5 mol/l urea, 2% nonidet P-40, 5% 2-mercaptoethanol and 2% carrier ampholytes, pH range 7–9) and stored at −80°C.

Determination of [35S]-methionine incorporation

[35S]-methionine incorporation was quantitated in duplicate by adding 10 µl BSA (0.2 µg/ml in H2O) as a protein-carrier to 5 µl of each sample diluted 1:10 with lysis buffer, followed by 0.5 ml of 10% TCA. This was left to precipitate for 30 min at 4°C before being filtered through 0.25 µm HAWP filters (Millipore) presoaked in BSA-saturated 10% TCA. The filters were washed twice with 1 ml BSA-saturated 10% TCA, once with ethanol, dried and placed into scintillation liquid for counting.

2D-gel electrophoresis

The procedure was as previously described [22]. Briefly, first-dimensional gels contained 4% acrylamide, 0.25% bisacrylamide and carrier ampholytes. An equal number of counts (106 cpm) of each sample were applied to the gels. The samples were analyzed on both isoelectric focusing (IEF; pH 3.5–7.0) and non-equilibrium pH-gradient electrophoresis (NEPHGE; pH 6.5–10.5) gels. Second-dimension gels contained 12.5% acrylamide and 0.063% bisacrylamide run overnight at 20°C. After electrophoresis, gels were fixed with Amplify (Amersham) before being dried. The gels were exposed to a PhosphoImage screen and the protein-pattern was scanned on a PhosphorImager (ADC 70 AGFA, Belgium). The resulting approximately 8 MB images were converted to 16-bit tiff images for image analysis.

Computer-assisted analysis

Computer analysis was carried out using the BioImage program 2D-Analyzer (version 6.1; Genomic Solutions, Ann Arbor, Mich., USA). The computer program assists in the matching of the spots among the eight independent gels, four from each phenotype, into a composite image followed by manual editing (Fig. 1). All protein spots that were marked were then re-evaluated to ensure that the boundaries assigned to the protein spots were correct and that no matching errors between the gels had been made. Significant changes in percent integrated optical density (IOD) between individual spots in the NHI-glu compared with the NHI-ins phenotype were determined using a double-sided non-paired Student’s t-test comparing four independent experiments from each phenotype. Changes with a p value less than 0.01 were accepted as being statistically significant.

Fig. 1
figure 1

Sections of the NEPHGE and the IEF 2D-gels of the NHI-glu phenotype covering the spots of changed expression level. The spot number refers to the match number assigned by the Bio Image computer program and used throughout the manuscript

Full size image

Protein characterization

For preparative purpose, a pool of 2×107 cells were prepared and separated on 2D-gels as described above. For localization of the spots, a sample of 10% of the cells was radioactively labelled to be used as a tracer as previously described. We ran six gels of each cell type, and loaded each with 200 µg of non-labelled protein and 4×106 cpm/gel of [35S]-methionine labelled protein per gel.

Protein identification using matrix assisted laser desorption/ionization mass spectrometric (MALDI-MS) peptide mass mapping

Protein spots of interest were obtained by cutting them out of the dried gel using a clean scalpel. All 135 selected spots could technically be cut out of the preparative-gels for protein identification. The proteins were enzymatically digested in the gel as described [23] with minor modifications [24]. In short, the excised gel plugs were washed in 50 mmol/l NH4HCO3/acetonitrile (60/40) and dried by vacuum centrifugation. Digestion buffer (50 mmol/l NH4HCO3) containing endoproteinase lys-C (8 ng/µl, Calbiochem, San Diego, Calif., USA) was added to the dry gel pieces and incubated on ice for 1 h for re-swelling. After removing the supernatant, an additional 20–40 µl digestion buffer was added, and the digestion was continued at 37°C for 4–18 h. The peptides were desalted and concentrated prior to MALDI-MS as previously described [15]. The peptides were deposited directly onto the MALDI target by elution with matrix solution [α-cyano-4-hydroxy cinnamic acid in 70% acetonitrile/29.9% water/0.1% TFA (20 g/l)]. MALDI-MS analysis was carried out using a Voyager STR instrument (PerSeptive BioSystems, Framingham, Mass., USA) equipped with delayed extraction. Spectra were obtained in reflector positive mode using an accelerating voltage of 20 kV. Protein identification was done by searching the peptide mass maps in a comprehensive non-redundant protein sequence database (nrdb, European Bioinformatics Institute, Hinxton, Cambridge, UK) using the ProFound software (http://129.85.19.192/profound_bin/WebProFound.exe) with a mass accuracy of below 50 ppm [25]. Both first and second pass searches were carried out and the validation of the individual protein identifications was carried out [25]. In short, the Z-score described in the ProFound software was used as the primary evaluation of a positive identification, supported by information on sequence coverage, induced oxidation of methionine residues and terminal sequence tags, For identifications with a Z-score between 1.2 and 2.0 a careful evaluation including information on protein molecular weight together with the above-mentioned parameters were taken into account.

Determination of Mw and pI

Determination of the Mw and pI of all the spots were done by cutting 17 spots with high density form the NEPHGE 2D-gels and 24 spots of the IEF 2D-gels. Identification of the spots was carried out by MS and the Mw and pI values, from the identified proteins, were used as markers in the Bioimage Program to calculate the Mw and pI values for all the spots in the gel. Theoretical pI and Mw were calculated for all the identified proteins using the ‘Compute pI/Mw tool’ at the ExPASy Molecular Biology Server (http://www.expasy.ch/tools/pi_tool.html) taking into account any post-translational cleavage known.

Results

Expressions of 135 (53 spots from the NEPHGE 2D-gels and 82 spots from the IEF 2D-gels) out of 2239 reproducibly detectable spots were significantly (p<0.01) changed when samples of NHI-glu and NHI-ins were compared by 2D-GE. A total of 97 different proteins were unambiguously identified in 94 out of the 135 spots by mass spectrometry (a success rate of 70%, Table 1). The remaining 41 spots (Table 2) yielded either good mass spectra but no matches in the databases (17 spots), weak spectrum but no protein identification (19 spots) or no spectra at all (5 spots).

Table 1 Known and putative functions of identified proteins during maturation from the pre-beta to the beta-cell phenotype. Spots on the NEPHGE side of the 2D-gels have prefix “N” and spots on the IEF side have prefix “I”. Each protein is listed according to its major known function. The percent IOD of the spots from the NHI-glu phenotype is given, and the IOD ratios are given relative to the expression level in the NHI-glu phenotype. Thus, values below 1 represent proteins that are down-regulated and values above 1 represent proteins that are up-regulated after beta-cell maturation. Furthermore proteins that are suppressed or undetectable after beta-cell maturation are indicated by “supp”. Theoretical pI and Mw values were calculated using the ‘Compute pI/Mw tool’ at the ExPASy Molecular Biology Server, and the observed pI and Mw values are obtained directly from the gel. The expression data are based on the matching and comparisons of gels from four independent experiments for each phenotype
Full size table
Table 2 Unidentified protein spots with statistically significant changed expression level after beta-cell maturation are shown. The percent IOD of the spots in the NHI-glu phenotype are given and the IOD ratios are given relative to the expression level in the NHI-glu phenotype. Furthermore, proteins that are suppressed or undetectable after beta-cell maturation are indicated by “supp”. One protein only expressed in the beta-cell phenotype is indicated by “de novo”. Observed pI and Mw values obtained directly from the gel are given
Full size table

Our data show that at least 37% of the proteins identified might be post-translationally modified, since they were either found in more than one spot on the 2D-gel (22 proteins) or their observed and calculated theoretical Mw and/or pI were in discordance (19 proteins, see below). A total of five proteins were overlapping between these two groups. We found that 15 proteins were present in two spots each (ATP synthase α chain, γ-enolase, pyruvate kinase M1 isozyme, flag structure-specific endonuclease, ezrin, T-complex protein 1 ε and γ, protein disulfide isomerase, Hsc70-ps1, alpha-2-macroglobulin receptor-associated protein, dihydropyrimidase-related protein 2, voltage-dependent anion-selective channel protein 1, sorting nexin 6, septin-like protein and Erp 44). Furthermore six proteins were present in three spots each (fructose-bisphosphate aldolase A, T-complex protein 1 ζ, probable protein disulfide isomerase ER-60, P60 protein, voltage-dependent anion-selective channel protein 2 and 26S protease regulatory subunit). Finally, one protein was present in four spots (78 KDa glucose regulated protein). Major inconsistencies were found between observed and calculated theoretical Mw and/or pI, for 19 protein-spots (20%) (NEPHGE 59, 124, 332, 335, 357, 377, 454, 582, 4410, 19991, 20140, 45036 and IEF 330, 376, 846, 935, 922, 1014, 1451). These inconsistencies could reflect post-translational modifications.

Some spots (33 spots) contained two identified proteins (*Table 1): NEPHGE 59, 77, 230, 335, 357, 377, 454, 4234, 4410, 20127 and IEF 85, 109, 166, 193, 255, 330, 383, 469, 728, 775, 846, 871, 881, 973, 1014, 1039, 1219, 1451, 1463, 1472, 1473, 1500 and 12315. The identified proteins have been assigned in seven groups according to their major known or putative functions allowing easy comparison with previous publications on PiPP of IL-1β exposed rat islets [14, 16]: (i) energy transduction and redox potential, (ii) glycolytic enzymes, (iii) protein synthesis, chaperones and protein folding, (iv) signal transduction, regulation, differentiation and apoptosis, (v) cellular defence, (vi) cellular transport, (vii) other functions. The function and possible importance for beta-cell maturation and T1DM pathogenesis are discussed below for selected proteins, including the 22 proteins in common with our previous analysis of PiPP in IL-1β exposed rat islets [14, 16].

Discussion

Substantial research has been carried out to characterize the molecular mechanisms involved in the development of the pancreatic beta cells. It is generally believed that the endocrine, exocrine and ductal cell types are derived from endodermal cells [5]. Several studies have shown that the endocrine stem cell require several specific transcription factors to mature into single-hormone-expressing cells [6, 7, 8, 9]. The pancreatic duodenum homeobox 1 (PDX-1) has been shown to be important both for pancreas development [26] and subsequent maturation of the endocrine cells [27], whereas later in the mature islets PDX-1 expression is restricted to the insulin-producing beta cells [8]. Another transcription factor, Nkx-6.1, is important for maturation of the insulin-producing beta cells [8], whereas Brain-4 is important for maturation of the glucagon-producing α-cells [7]. Previous analysis of mRNA gene expression profiles by semi-quantitative multiplex RT-PCR of the two NHI-phenotypes used in this study, grown under the same conditions, showed that several genes associated with late stages of beta-cell maturation, such as PDX-1 were expressed in both phenotypes, whereas Nkx-6.1 was restricted to the beta-cell phenotype [20]. We showed that maturation into the beta-cell phenotype was accompanied by insulin gene expression and an acquired susceptibility to toxic effects of IL-1β, not seen in the pre-beta-cell NHI-glu phenotype [12]. Proteome analyses of the two phenotypes could therefore identify proteins and pathways important for beta-cell maturation and the acquired IL-1β sensitivity.

High resolution 2D-gel technology can efficiently separate thousands of proteins. Statistical analysis of the expression pattern of the NHI-glu and NHI-ins phenotypes allowed us to separate a small group of 135 protein spots, which were differentially expressed between these phenotypes. Subsequent MALDI-MS identified the majority of the proteins (70%) in these protein spots using available protein or translated nucleotide sequence databases. Compared to transcriptome analyses of mRNA expression, proteome analysis offers the possibility to quantitate changes in protein expression and to identify post-translational protein modifications, such as phosphorylation [15]. Post-translational modifications are often required for the functional activation of a protein or a protein complex [28], and indeed the importance of these modifications in the beta-cell maturation process is suggested by our finding that 37% of the identified proteins are present either in more than one spot, or exhibited modified mobility. Several of the other proteins might also be modified in more subtle ways. The nature of the post-translational modifications observed was not studied as part of this study. Thus, currently our data provide information on changes in protein expression profiles during beta-cell maturation. Future analyses will allow identification of the precise mechanisms as well as the functional relevance responsible for these changes.

Due to the high sensitivity of the MALDI-MS technique, 33 spots were found to contain two proteins (34%). One study has described in detail how one spot is able to contain more than one protein [29]. In short, this could be due to overlapping protein bands or co-migration in polyacryamide gel bands due to SDS-PAGE limited resolving power or contamination with cytokeratins. This is important to note, since it is therefore presently not possible to say which protein in the protein spot in question actually changed expression level during the beta-cell maturation process. Future analysis using zoom gels covering narrow pH ranges with considerably higher resolution could resolve this [30]. Another limitation in proteome analysis is that only the proteins that are metabolically labelled with methionine under the present experimental conditions and are within the detectable pI and Mw ranges could be identified. Therefore proteins (e.g., pro-insulin and insulin) below a Mw of 12 kDa are not detected.

The protein spots that changed expression levels after maturation into the beta-cell phenotype are grouped in Table 1 according to their known or putative functions. Similar group assignments were recently used in the description of protein changes in rat islets exposed to IL-1β [14, 16]. It is beyond the scope of our study to discuss in detail the possible role of all identified proteins. Hence, only proteins previously shown to be influenced by IL-1β [14, 16], and/or relevant for beta-cell maturation or beta-cell destruction are discussed.

We have previously shown that IL-1β influences PiPP in islets of Langerhans [13, 14, 16]. PiPP could reflect up- or down-regulation of protein expression levels and/or gel-position shifts resulting from post-translatory modifications. Consequently, when in this paper the expression level of a protein is described as up- or down-regulated, this could reflect changes in the expression level of the native protein (including alterations in protein synthesis and degradation) or expression changes in a modified form of that protein. Therefore protein expression changes in common between beta-cell maturation and IL-1β exposed WF rat islets [14] and BB rat islet [16] (Table 1, ¤ and #, respectively) could reflect both native or modified forms of the protein. Of the 97 proteins of changed PiPP reflecting maturation into the beta-cell phenotype identified in this study, the expression pattern of 26 of these was previously shown to be influenced by IL-1β in rat islets. Of these, 16 were identified in IL-1β exposed WF rat islets [14] and 10 in IL-1β exposed BB rat islets [16], representing five of the common pathways (Table 1). Of these 26 proteins, only four (ATP synthase alpa chain, glyceraldehyde-3 phosphate-dehydrogenase, probable protein disulfide isomerase ER-60 and 78 Kda glucose-regulated protein) were common in all three studies.

Due to the nature of the in vitro and in vivo metastasing cell-lines used in the present experiments, it cannot be excluded that proteins involved in cell-proliferation and cell-transformation might also be included in Table 1.

Energy transduction and redox potentials

Protein spots involved in the Citric acid cycle (such as isocitrate dehydrogenase, citrate synthase and fumarate hydratase) were down-regulated or suppressed after beta-cell maturation, suggesting decreased ATP production. In contrast ATP synthase α chain and a different form of isocitrate dehydrogenase were highly up-regulated. ATP synthase α chain, creatine kinase B and isocitrate dehydrogenase were also changed after IL-β exposure of rat islets [14, 16]. IL-1β has an inhibitory effect on the mitochondrial energy production in islets [31], and IL-1β exposure of rat islets showed down-regulation of several proteins involved in energy generation [14, 16].

Glycolytic enzymes

During beta-cell maturation six glycolytic enzymes changed expression level. Of these, five protein spots were also changed in rat islets after IL-1β exposure (¤, # Table 1). The functional relevance of the glycolytic pathway in insulin production still needs to be elucidated, since it has previously been shown that IL-1β does not affect the activity of any glycolytic enzyme [32]. However, it has been reported that insulin and cellular ATP levels regulate both the activity and synthesis of several glycolytic enzymes [32].

Protein synthesis, chaperones and protein folding

Proteins involved in protein synthesis were generally down-regulated after maturation of the beta-cell phenotype (the first six proteins in this group, Table 1). IL-1β has been shown to inhibit gene transcription and protein synthesis [13, 33] and to decrease preproinsulin biosynthesis in particular [34]. Therefore, a global decrease in protein synthesis could reduce the capacity of a beta cell to survive IL-1β exposure.

Citrullinaemia is an autosomal disorder of the urea metabolism characterized by high circulating levels of citrulline as a result of a deficiency in the activity of the urea cycle enzyme argininosuccinate synthetase (ASS) [35]. In addition, mutations in ASS cause high levels of citrulline. The expression level of ASS was significantly (p<0.0001) lower in the beta-cell phenotype compared to the pre-beta-cell phenotype. This could result in increased levels of citrulline in beta cells. Beta-cell sensitivity to IL-1β is in part mediated through iNOS expression. iNOS converts arginine and oxygen to citrulline and NO [36]. Arginine might be provided by uptake from the extracellular fluid, by intracellular protein degradation or by synthesis from citrulline [37]. Thus, it is conceivable that the down-regulation of ASS expression results in increased levels of citrulline in beta cells that in turn could induce higher intra-cellular levels of arginine. The increased pool of arginine could then serve as substrate for iNOS and result in increased production of NO [37]. However, our previous analyses have shown that despite the different sensitivity to IL-1β, no significant difference in IL-1β induced NO production could be detected between the two phenotypes [12]. Nevertheless, it has been shown that IL-1β induces the citrulline-NO cycle in beta cells, and that extracellular arginine or citrulline is required for NO production [38]. Furthermore, the arginine level has been shown to be increased in IL-1β exposed beta cells [38]. Thus, it is conceivable that the beta-cell phenotype could contain increased concentrations of arginine due to low ASS expression as an acquired trait occurring during beta-cell maturation.

During beta-cell maturation seven protein spots involved in the formation and maintenance of the actin/myosin cytoskeleton changed expression level. One example is caldesmon also shown to be down-regulated in IL-1β exposed rat islets [14]. Whether this influences beta-cell stability and IL-1β induced toxicity is yet unknown.

The remaining 15 proteins of this group are all involved in chaperone functions and/or protein folding. The ζ and γ subunits of the T-complex protein-1 were also changed in WF rat islets after IL-1β exposure (¤; Table 1) [14]. This complex is involved in molecular chaperoning and in the production and folding of native actin and tubulin [39]. We identified two proteins found in more than one spot (probable protein disulfide isomerase ER-60 and protein disulfide isomerase) specifically involved in the re-arrangement of both intra-chain and inter-chain disulfide bonds in proteins to form native protein structures. Interestingly, spots containing these two proteins were also previously shown to change expression levels in rat islets exposed to IL-1β [14, 16].

Of the heat shock protein spots (HSPs) eight changed their level of expression after beta-cell maturation and several of these also in response to IL-1β (Table 1; ¤, #). HSPs function as molecular chaperones assisting in protein folding, transport, translocation and degradation [40], and could have protective functions against exposure to several damaging stimuli, such as heat [41], cytokines [42] and NO [43]. Furthermore, lower expression levels of HSP have been shown in streptozotocin induced diabetic rats [44].

Signal transduction, regulation, differentiation and apoptosis

A protein spot containing PKC-interacting cousin of thioredoxin (PICOT), a Jun N-terminal kinase (JNK) inhibitor [45], was down-regulated during beta-cell maturation. JNK is activated in response to different stress factors such as cytokines, heat and oxidative compounds. Dependent upon the type of stimulation, signalling through JNK activates programmed cell death (apoptosis), differentiation/proliferation or tumour development [46], and has been shown to be critically involved in IL-1β mediated beta-cell destruction [47].

Apoptosis plays an important role during cell maturation [48]. Apoptosis is characterized by a set of cellular events including cell shrinkage, chromatin condensation and DNA fragmentation. Many factors have been shown to be involved in this process [49]. Toxic levels of cytokines have been shown to induce apoptosis [50, 51]. In line with this, a cleavage product of lamin A was up-regulated after IL-1β exposure of rat islets in vitro [14]. After maturation of the beta-cell phenotype protein spots containing lamin B1 and C2 were up-regulated. Proteolysis of lamins, the major structural proteins of the nuclear envelope, are observed in cells undergoing apoptosis [52]. The relevance of this in relation to beta-cell sensitivity to IL-1β is not clear. However, it has been shown that inhibitors of lamin cleavage prevent apoptosis [53]. Voltage-dependent anion channel (VDAC) 1 was down-regulated while VDAC 2 was up-regulated after beta-cell maturation. VDAC 1 was also changed after IL-1β exposure of rat islets [14]. VDAC proteins form pores in the outer mitochondrial membrane serving as the major permeability pathway for metabolite flux between the cytoplasm and the mitochondria [54]. Pro-apoptotic proteins Bax and Bak bind to VDAC channels and induce apoptosis by releasing cytochrome C, whereas the anti-apoptotic protein bcl-xL closes the VDAC channels and inhibits apoptosis [55].

Cellular defence

Gluthatione-S-tranferase (GST) is involved in the glutathione pathway where gluthathione peroxidase and GST work together with glutathione in the decomposition of hydrogen peroxide (H2O2) or other organic hydroperoxides [56]. In mammalians, GSTs are divided into six classes (α, µ, κ, θ, π and σ). Protein spots containing GST (class π and µ) were both down-regulated during beta-cell maturation, suggesting that the beta-cell phenotype has impaired ability to reduce H2O2 compared to the pre-beta-cell phenotype. This reduction in cellular defence was also seen in the IL-1β exposed BB rat islet [16], since glutathione synthase was down-regulated. This is in line with previous analysis showing an impaired ability of islet beta cells to reduce H2O2 [57]. Glutathione has been shown to protect a human insulinoma cell-line against the toxic effects of tumour necrosis factor α (TNF-α) [58], and together with catalase, glutathione peroxidase protects RIN cells against H2O2,reactiveoxygen species and cytokines [59].

Cellular transport

Nuclear or cellular transport is important for the survival and function of specialized cells and 14 proteins with such functions changed their levels of expression during beta-cell maturation. The Coatomer delta subunit was also up-regulated in IL-1β exposed rat islets [14]. Maturation into the beta-cell phenotype was also associated with an increased expression level of clathrin light chain, suggestive of increased capacity for exocytosis in the beta cells. This correlates well with maturation of the beta-cell phenotype, and it could be associated with increased insulin-production secreted through exocytosis in clathrin-coated vesicles [60] in the stimulatory culture condition (11.5 mmol/l glucose) used in these experiments.

We previously hypothesized that cytokines induce a race between protective and deleterious mechanisms in beta cells, and when the deleterious mechanisms prevail, beta-cell destruction and T1DM develop [4]. In addition we suggested that IL-1β sensitivity is an acquired trait during beta-cell maturation. This is supported by our previous studies of the two phenotypes [12]. Of the 97 proteins of changed expression patterns following beta-cell maturation some could be crucial for the required sensitivity to cytokines and resulting PiPP in response to IL-1β [14, 16]. Thus, the beta cell may “pay a price” influencing several common cellular pathways to become a highly specialized insulin-producing beta cell.

Based upon the Copenhagen Model of T1DM we attempted simple mathematical modelling of the earliest phase of the pathogenesis of T1DM [61]. Resulting from this, one could view the unperturbed beta cell as being in a parameter space characterized by dynamic stability separated by a virtual parameter surface from a parameter space characterized by dynamic instability. The closer to the parameter surface in its still dynamic instability space a beta cell is, the easier it is “pushed” into the space of dynamic instability or ongoing destruction. Maturation of a pre-beta cell (NHI-glu) into a beta-cell phenotype (NHI-ins) might position the beta cell closer to the parameter surface. When metabolically active the beta cell could, although still in the dynamic stability space, be even closer to penetrate the surface, since its sensitivity to cytokine toxicity is increased. Exposure to IL-1β or a mixture of cytokines might be what “pushes” beta cells through the parameter space surface and into dynamic instability and destruction. Dependent on the stimulus this could involve the interaction of several different proteins and pathways and the outcome might in part be genetically determined.

In conclusion, differentiation and maturation from a stem cell to a beta cell is a complex series of events. Future functional analyses are needed to further elucidate the role of the proteins identified in this study for the last maturation stage towards insulin production and accompanying acquired sensitivity to IL-1β and other cytokines. We anticipate that such knowledge could provide useful information for the design of novel and rational preventive and/or curative strategies in T1DM, both in relation to beta cell growth and differentiation as well as in inhibition of cytokine-mediated toxicity.