Abstract
We demonstrated pancreatic reg gene overexpression in non-obese diabetic (NOD) mice during active diabetogenesis. The aim of this study was to determine in which part of the pancreas (endocrine and/or exocrine) the gene(s) and the protein(s) were expressed and if their localization changed with progression of the disease. In situ hybridization analysis and immunocytochemical studies were carried out on pancreas of female and male NOD mice. Both develop insulitis but diabetes develops only in females and in males only when treated by cyclophosphamide. Our results show that whatever the age, sex, and presence of insulitis and/or diabetes, the expression of reg mRNAs and of the corresponding protein(s) was restricted to exocrine tissue. Moreover, reg remains localized in acinar cells in the two opposite situations of (a) cyclophosphamide-treated males in a prediabetic stage presenting a high level of both insulin and reg mRNAs, and (b) the overtly diabetic females with no insulin but a high level of reg mRNA. These findings suggest that overexpression of the reg gene(s) might represent a defense of the acinar cell against pancreatic aggression.
Terazono et al., in 1988, isolated a clone from a cDNA library prepared from pancreatic islets isolated from 90% pancreatectomized nicotinamide-treated rats, a model of islet regeneration. The gene corresponding to the clone was named reg. This gene was found to be homologous to a previously described human exocrine pancreatic gene that encodes a 166 amino-acid protein (with an N-terminal signal peptide of 22 amino acids) whose proteolyzed form of 133 amino acids was designated “protein X”(Guy–Crotte et al. 1984), pancreatic thread protein (Gross et al. 1985), and “pancreatic stone protein” or lithostathine (De Caro et al. 1987; Sarles et al. 1990). The biological function of the reg protein in exocrine secretion remains elusive because its putative capacity to inhibit CaCO3 crystal growth and to prevent the formation of pancreatic stones has been recently ruled out (Bimmler et al. 1997; De Reggi et al. 1998).
Since then, several reg and reg-related genes have been isolated from human, rat, and mouse and have been shown to constitute a multigene family, the Reg family (for review see Okamoto 1999). In mouse, two reg genes, regI and regII, encoding proteins with 76% homology, were detected in the normal pancreas and hyperplastic islets of aurothioglucose-treated mice (but not in the normal islets) and might be involved in islet regeneration (Unno et al. 1993; Okamoto 1999). Because it was possible that defense phenomena and particularly cellular replication may occur during autoimmune aggression of pancreatic β-cells, we studied the expression of the reg gene in a spontaneous model of insulin-dependent diabete mellitus, the non-obese diabetic (NOD) mouse (Castano and Eisenbarth 1990). In this model, as in human patients, the disease has two distinct stages (Lampeter et al. 1989). An occult phase (insulitis) begins with invasion of the islets and may persist quite innocuously for a long period. The overt phase, diabetes, which in NOD mice is essentially confined in females, develops after most of the β-cells have been destroyed. We observed a pancreatic overexpression of the regII gene (and to a lesser degree of the regI gene) in female NOD mice, which develop diabetes early in life, whereas NOD males, which are relatively protected, have a low mRNA level similar to the level found in a control mouse strain (Baeza et al. 1996b, 1997). The specificity of this overexpression in the diabetogenic process was reinforced by the fact that it was also observed in NOD male mice after treatment with cyclophosphamide, an immunosuppressive drug known to accelerate diabetes, whereas this drug has no effect in control male mice. More importantly, we observed a high expression of the regII gene in NOD diabetic animals in which no insulin mRNA was detectable, leading us to hypothesize that the overexpression of reg genes does not depend on the presence of β-cells. It was therefore important to determine in which part of the pancreas (endocrine or exocrine) the reg gene was expressed and if this localization varied with the presence of insulitis and with the progression of the disease.
To resolve this issue, we determined in this present study the localization of the reg gene(s) in NOD mouse pancreatic tissues by an in situ hybridization (ISH) analysis with specific reg probes and an immunohistochemical evaluation of their corresponding protein(s).
Materials and Methods
Animals and Treatment
NOD mice bred from a parental stock were provided by Dr. C. Boitard (Department of Immunology, INSERM U25, Necker Hospital, Paris, France). In our colony, the prevalence of diabetes at Day 210 was 43% in female and <1% in male mice. Insulitis, characterized by a lymphocytic infiltration, was observed from the fourth week of age and was present to a similar degree in both sexes. Glycosuria was detected using labsticks (Ketodiastix; Bayer Diagnostics, Puteaux, France). To select situations with various degrees of activity of the diabetogenic process we studied non-diabetic NOD female and male mice (age range 12–245 days), diabetic NOD female, and cyclophosphamide-treated male NOD mice (aged 110 days). Cyclophosphamide (Endoxan; Lucien Laboratory, Colombes, France) in saline was injected SC at a dose of 300 mg/kg body weight 100 days after birth. Because in our colony the percentage of overt diabetes in NOD male mice was 16% after one injection and 83% after two injections of cyclophosphamide, the animals were sacrificed 10 days after the first injection while in a “prediabetic” period. In addition, female and male normal mice, age range 60–135 days (IOPS-OF1 supplied by Iffa Credo; L'Arbresle, France), were studied as controls.
Tissue Sections
The pancreas was rapidly collected and fixed in 4% paraformaldehyde in 0.1 M PBS for 12–48 hr at 4C. Tissues were treated in an automatic processor and embedded in paraffin. Paraffin sections of 3–4-mm thickness were cut for RNA ISH and immunohistochemistry and were mounted on Super-Frost Plus slides (Menzel-Glaser; Braunschweig, Germany) or Silane-prep slides (Sigma; L'Isle d'Abeau Chesnes, France).
Probes
The mouse regII probe was kindly provided by H. Okamoto (Sendai, Japan). It was full-length cDNA of 700 bp, sub-cloned into pBluescript vector at the XHoI site in the polycloning site. A 174-bp fragment of the mouse insulin cDNA was synthesized by RT-PCR as described by Baeza et al. (1997). The primers used were deduced from the nucleotide sequences of mouse preproinsulin gene I (Wentworth et al. 1986): sense primer 5′-CCCACCCAGGCTTTTGTCAA-3′ (nucleotides 889–908) and antisense primer 5′-CAAGGTCTGAAGGTCCCCGG-3′ (nucleotides 1043–1062). The 174-bp cDNA was further cloned into pLitmus 29 vector (New England Biolabs; Ozyme, France). For Northern hybridization and ISH, sense and antisense riboprobes were generated from linearized pBluescript plasmids containing regII cDNA and pLitmus plasmids containing insulin cDNA using T7 (for anti-sense reg II, antisense insulin, and sense insulin) or T3 (for sense reg) RNA polymerase (Promega; Charbonnieres, France) and digoxigenin-11-UTP (Boehringer Mannheim; Meylan, France). Optimal hybridization sensitivity was obtained for antisense and sense regII by reducing the RNA probe from its full size after transcription to an average of 200 nucleotides. The size reduction was carried out by controlled hydrolysis under alkaline conditions, as described by Cox et al. (1984). Briefly, the transcription products were incubated in the presence of 0.4 M NaHCO3 and 0.6 M Na2CO3, at 60C for a calculated period of time of 40 min and the reactions were stopped by adding Neutralizing Salt (Novagen; R&D Systems, Abingdon, UK). The mixure was incubated on ice for at least 30 min, and after RNA probe precipitation the pellets were resuspended in DEPC H2O and stored at −80C until use.
Northern Blotting Analysis
Total mouse pancreatic RNA and synthesized regI and regII RNAs were prepared as described previously (Baeza et al. 1997). Northern blotting analysis was performed to verify the hybridization specificity of antisense and sense regII RNA probes after hydrolysis. Hybridization and detection were carried out with the procedure described by the DIG system user's guide for filter hybridization of Boehringer Mannheim. In brief, Northern hybridization was performed overnight at 68C in a solution containing the DIG-labeled RNA probe (100–200 ng/ml), 50% formamide, 5 × SSC, 0.02% SDS, 0.1% N-lauroylsarcosine, 2% (w/v) blocking reagent (Boehringer Mannheim). After hybrization, the membranes were washed twice for 5 min in 2 × SSC containing 0.1% SDS at room temperature (RT) and twice for 15 min in 0.2 × SSC containing 0.1% SDS at 50C. Revelation of the membrane was performed with the colorimetric detection reagents NBT and X-phosphate (BCIP). The membranes were equilibrated in Buffer 1 (100 mM maleic acid, 150 mM NaCl, pH 7.5) for 5 min and blocked in Buffer 2 (1% Blocking Reagent dissolved in buffer 1) for 30 min. The membranes were incubated for 30 min with the alkaline phosphatase-conjugated anti-digoxigenin Fab fragment 1:5000 (Boehringer Mannheim) in Buffer 2 and washed twice for 15 min in Buffer 1. Results were visualized with the chromogenic substrate NBT/BCIP (Gibco BRL Life Technologies; Pontoise, France) in solution containing 100 mM Tris-HCl, pH 9.5, 100 mM NaCl, and 50 mM MgCl2.
In Situ RNA Hybridization
In situ hybridizations were performed essentially as described previously by Blouin et al. (1996). In brief, sections were deparaffinized with xylene and the tissues were rehydrated through graded ethanol solutions to PBS. Sections were treated in 0.2 M HCl at RT for 15 min and then washed in PBS before immersion in 0.3% Triton X-100 in PBS for 20 min. After rinsing with the same buffer, the sections were incubated for 8 min in ice-cold 20% acetic acid to reduce nonspecific background and then rinsed twice in PBS for 2 min. The slides were then digested with proteinase K (4 mg/ml) for 30 min at 37C. Digestion was stopped by a second fixation with paraformaldehyde (4%), followed by washing in 0.2% glycine in PBS. The sections were rinsed in PBS, acetylated with triethanolamine, and washed again in PBS. Sections were pre-hybridized in 5 × SSC, 5 × Denhardt's solution, 50% formamide, 250 mg/ml yeast tRNA, 250 mg/ml denatured salmon sperm DNA, and 4 mM EDTA at 42C for 3 hr. Tissues were then hybridized overnight at 42C in the same solution (but without salmon sperm DNA) with either antisense or sense RNA probe at a quantity of 100 ng/slide for insulin and 200 ng/slide for reg. After hybridization, the slides were washed twice in 2 × SSC for 15 min at RT, incubated once for 10 min with RNase A (1 mg/ml) at RT, rinsed twice in 2 × SSC at RT for 15 min, and finally washed twice in 1 × SSC at 37C for 15 min. The sections were blocked for 60 min at RT in 0.5% blocking reagent, followed by a 2-hr incubation with alkaline phosphatase-conjugated anti-digoxigenin antibody diluted 1:500 in 100 mM Tris-HCl, 150 mM NaCl (pH 7.5). After washing, the hybridized cells were visualized by a standard immunoal-kaline phosphatase reaction, using nitroblue tetrazolium chloride and 5-bromo-4-chloro-3-indolyl-phosphate as substrate (Boehringer Mannheim; Quebec, Canada). Photographs were taken under brightfield illumination using a Zeiss photomicro-scope and Kodak TMAX 100 print film.
Preparation of a Polyclonal Antibody Against Mouse reg Protein(s)
A polyclonal antibody was generated against mouse regI and regII protein using a peptide of 16 amino acids, VSLTSNT-GYKKWKDDN(A), corresponding to position 137–152 of regI and 145–160 of regII. The protein sequences were deduced from the mouse cDNA sequences (Unno et al. 1993). The peptide to be synthesized was determined by choosing among the protein amino acid sequences the most favorable regions for potential antigenic sites on the basis of acidic-basic residue alternates and hydrophilic-hydrophobic alternates.
The 16-mer reg peptide was synthesized by the solid-phase technique (Merrifield 1986). Stepwise elongation of the peptide was carried out on 0.35 mmol of HMP resin (0.96 mmol of hydroxyl sites) (Wang 1973) using an automated peptide synthesizer (Model 433A; Applied Biosystems, Foster City, CA). The crude peptide was purified by preparative reverse-phase HPLC (Perkin-Elmer, Norwalk, CT; ODS 20 μm, 100 × 10 mm) and the chemical identity of reg peptide was confirmed by amino acid analysis and mass spectrometry.
The peptide of 1927 kD was coupled to the carrier molecule at a peptide:carrier ratio of 1:50. Peptide (1 μmole) in 0.05 M phosphate buffer, pH 6.1, was coupled to 20 nanomoles of the carrier KLH (keyhole limpet hemocyanin; Pierce, Rockford, IL), using EDC (1-ethyl-(3-dimethylaminopropyl) carbodiimide (Pierce) at a final concentration of 10 mg/ml. The mixture was stirred for 120 min at RT and the coupling reaction was stopped by addition of 0.1 mM sodium acetate, pH 4.0.
At Day 0, two New Zealand rabbits were given 200 μg of KLH-coupled peptide emulsified in complete Freund's adjuvant, 200 μg at Days 21 and 42, then 400 μg at Days 80, 130, and 280. Blood was collected in heparinized plastic tubes 10 days after each boost. A preimmune serum was collected before starting the immunization protocol for each rabbit. Sera were then prepared by centrifugation, aliquotted, and kept frozen until use. The immunoreactivity of the various bleedings was determined by immunoblotting as described below. Positive signal occured after the third boost and increased to reach a maximum in the following bleedings.
Characterization of the Antibody on Pancreatic Homogenates by Immunoblotting
Pancreatic homogenates were prepared from frozen mouse pancreas washed with a 0.15 M NaCl solution, then homogenized in a 0.15 M Tris buffer, pH 8.5, containing 1 mM benzamidine (10 ml per g of pancreas). After ultracentrifugation at 100,000 × g for 1 hr at 4C, supernatants containing pancreatic proteins were collected and stored by aliquots at −80C. Proteins present in supernatants were submitted to SDS-PAGE in a 15% polyacrylamide gel according to the method of Laemmli (1970) in the presence of β-mercaptoethanol. Ten μg of protein per lane was loaded. After electrophoresis, the proteins were transferred to nitrocellulose according to Burnette (1981) for immunodetection. Proteins were probed with the first polyclonal antibody directed to the protein reg at different dilutions (1:100–1:500). After incubation with the horseradish peroxidase-coupled secondary antibody to rabbit IgG (1:5000), the immunoreactive bands were vizualized by enhanced chemiluminescence (ECL; Amersham, Poole, UK).
Purification of the Antibody for Immunocytochemistry
Mouse reg antibody was purified by adsorption to and elution from the protein immobilized on nitrocellulose according to Coudrier et al. (1983). A preparative electrophoresis through SDS-polyacrylamide gel was performed with 10 μg of pancreatic homogenate proteins per lane and transfered to nitrocellulose. After Ponceau S staining the colored band corresponding to reg protein was cut and washed with PBS buffer containing 5% milk for 1 hr, then with the antibody to be purified that was diluted five times in PBS buffer containing 5% milk and 0.3% Tween-20 for 2 hr. The band of nitrocellulose was washed four times with the PBS/milk/Tween-20 buffer, then in PBS alone, and cut in small pieces in an Eppendorf tube. The antibody was eluted by 700 μl of 0.2 M Gly-HCl buffer containing 0.2% gelatin for 5 min at 0C and the solution was immediately neutralized with 250 μl of 1 M Tris buffer, then stored until use.
Immunohistochemical Staining
The avidin-biotin-peroxidase complex method (Vectastain ABC kit; Vector, Burlingame, CA) was used for immunostaining. The paraffin sections were deparaffinized in xylene and hydrated through graded ethanols. They were immersed in preheated Dako target retrieval solution (Dako; Carpinteria, CA) in a water bath for 20 min to increase staining intensity with the primary antibodies and immersed in absolute methanol containing 0.5% hydrogen peroxide for 30 min to block endogenous peroxidase activity. Sections were incubated with non-immune goat serum for 20 min to prevent nonspecific binding, then with the primary antiserum (diluted 1:500 in 150 mM PBS, pH 7.3, before purification or concentrated after purification) for 1 hr, and finally with biotinylated secondary antibody diluted. Sections were incubated for 30 min with the ABC reagent and then with the peroxidase substrate 3-amino-9-ethylcarbazol (AEC substrate kit; Vector). Slides were counterstained with Mayer's hemalum. Several controls were routinely performed: PBS or non-immune rabbit serum was used in place of the primary antiserum.
Results
Specificity of the reg RNA Probes
Mouse RNA was analyzed by Northern hybridization using antisense and sense regII RNA probes after hydrolysis (reducing RNA probes from more than 700 nucleotides to an average of 200 nucleotides by controlled hydrolysis under alkaline conditions). As shown in Figure 1A, a single mRNA transcript of about 0.9 kb was detected in total mouse RNA with antisense regII RNA probes and no mRNA transcript was detected with sense regII RNA probes. Because of the homology that exists between regI and regII sequences, it was important to ensure the specificity of regII RNA probes after hydrolysis. We confirmed the cross-hybridization of the RNA probe (antisense) with synthetic RNAs regI and regII as shown in Figure 1B.
Comparative reg and Insulin Gene Expression in NOD Mouse Pancreas Evaluated by ISH
As shown in Figure 2, expression of reg mRNA was detected at high levels in the exocrine portion of the pancreas of a 12-day-old female mouse (Figure 2A) but not in the endocrine portion where, as expected, expression of insulin mRNA was clearly visible (Figure 2B). Negative control was obtained with the reg sense probe (Figure 2C). Similar results were obtained in female and male 30-day-old mice, in which reg mRNA expression was detected in the exocrine compartment, whereas cells of the islets of Langerhans appeared negative for this transcript (data not shown). Comparative data for insulin and reg mRNA expression in two adjacent sections of the pancreas of a 30-day-old female mouse are shown in Figures 2D and 2E. A strong hybridization signal was observed with the insulin antisense probe in the endocrine compartment (arrowheads) and in a few ductule cells (curved arrows) (Figure 2D), whereas expression of reg mRNA was detected at a high level in the exocrine portion of the pancreas but not in the endocrine portion (Figure 2E). As shown in Figure 3, a consistent exocrine localization of reg mRNA was observed in the presence of insulitis in the pancreas of a 95-day-old (Figure 3A) and a 140-day-old female NOD mouse (Figure 3B), as well as in the pancreas of a cyclophosphamide-treated male mouse (Figure 3C). Arrowheads in Figures 3A and 3B point to the site at which a lymphocytic infiltration was visible in pancreatic islets. The reg signal remained specifically detected in the pancreatic acini of a diabetic female NOD mouse (Figure 3D).
Specificity of regII RNA probes. Northern blotting analysis of mouse reg(I and II) mRNA. Northern blot containing total RNA (±5 μg/lane) was hybridized to an antisense or sense DIG-labeled regII RNA probe after hydrolysis (
Similar results were obtained on pancreas sections of a control mouse (OF1 strain) (data not shown). Expression of reg mRNA was detected in the exocrine portion of the pancreas but not in the islets of Langerhans, whereas insulin mRNAwas specifically detected in the endocrine compartment and in a few duct cells, as observed in the NOD mouse pancreas.
Comparative pancreatic expression of reg and insulin mRNAs in NOD mouse. (
Localization of reg Protein in NOD Mouse Pancreas by Immunocytochemistry
The specificity of mouse reg antibody was first determined by immunoblotting analysis performed on mouse pancreatic homogenates. Similar results were obtained with NOD or OF1 mouse pancreatic homogenates. As shown in Figure 4, the antibodies recognized a single band at 16.6 kD, consistent with the predicted molecular mass of the mouse reg protein. Immunocytochemical controls on NOD or OF1 mouse pancreas confirmed the specificity of the antibody. No immunoreactivity was observed (a) with preimmune serum, (b) after immunoabsorption of the antiserum with the corresponding free peptide before the immunocytochemical procedure, and (c) in appropriate controls in which the primary antibody was omitted.
Pancreatic localization of reg mRNA in NOD mice with insulitis. (
Immunohistochemical staining results presented in Figure 5 show that regardless of the age, sex, the severity of insulitis, and the presence of diabetes in NOD mouse, the results were identical. When pancreatic sections were treated with the purified antibody, acinar cells were intensely stained, whereas islet cells remained constantly negative. The presence of a mild (Figure 5B) or a severe (Figure 5E) insulitis did not modify the acinar staining. In a pancreatic section of a cyclophosphamide-treated male, the acinar staining appeared stronger in the cells surrounding the islets (Figure 5D) but the phenomenon was not constantly observed. Reg immunoreactivity was also observed in pancreatic ducts (Figures 5B, 5D, and 5G, arrowheads) demonstrating that, as expected, the reg protein was normally secreted. When the pancreatic sections were treated with the antibody against insulin, strong staining was observed only in islet cells and acinar cells were negative (Figure 5C).
Immunoblotting of mouse pancreatic homogenate with reg antiserum. Fifteen μg of protein was submitted to SDS-(15%) polyacrylamide gel electrophoresis. After transfer, the nitrocellulose was incubated with reg antiserum (dilution 1:250) and anti-rabbit IgG conjugated to horseradish peroxidase (dilution 1:5000) as described in Materials and Methods.
Discussion
The reg gene is normally expressed by the pancreatic acinar cells (Watanabe et al. 1990), and the protein encoded by this gene was found in a fairly high amount in pancreatic secretion in humans (Carrère et al. 1999) and is present in other species (for review see Okamoto 1999). The physiological role of the protein remains unknown, but overexpression of reg and reg-related genes has been described in different experimental models of pancreatic islet regeneration, such as aurothioglucose-treated mice (Terazono et al. 1988), removal of insulinoma tumor in the rat (Miyaura et al. 1991), “cellophane wrapping” of pancreas in the Syrian golden hamster (Rafaeloff et al. 1995), or in rat (Zenilman et al. 1996), supporting the hypothesis of an association between reg gene expression and islet cell proliferation (for review see Baeza et al. 1996a). However, in these models the site of pancreatic reg synthesis remains controversial. Terazono et al. (1990) reported the presence of reg protein both in rat acinar cells and within granules of β-cells several weeks after nicotinamide treatment and partial pancreatectomy, suggesting that the reg gene may be induced in proliferating β-cells. In contrast, Miyaura et al. (1991) localized by ISH an overexpression of the reg gene exclusively in acinar cells after implantation and resection of a solid insulinoma in rats. However, these last authors did not exclude the possibility that reg was synthesized in the acinar cells and accumulated preferentially in islets during periods of β-cell proliferation. In previous studies (Baeza et al. 1996b, 1997), we demonstrated an overexpression of the two mouse reg genes during active diabetogenesis in the NOD mouse pancreas. Therefore, it was important to study not only the site of synthesis of the gene(s) but also the localization of the protein(s) in this model.
We first prepared sense and antisense riboprobes from the mouse regII cDNA and verified after hydrolysis the hybridization specificity of the antisense probe on total mouse pancreatic RNA and synthetic regI and regII RNAs. Using these riboprobes, we followed by ISH pancreatic reg mRNA expression in NOD mice at various stages of the diabetogenic process and compared it with insulin mRNA expression. The present data show that, regardless of the age, sex, or presence of insulitis, expression of reg mRNA was restricted to the exocrine portion of the pancreas, whereas, as expected, insulin mRNA expression was clearly visible in the islets of Langerhans. In some cases, insulin expression was observed in a few ductule cells, in agreement with our previous observations showing the presence of isolated endocrine cells in human fetal and adult tissue (Sanchez et al. 1998). As expected, a similar exocrine localization of the reg gene was observed in the pancreatic sections of control mice, the OF1 strain, but the immunolocalization technique we used does not allow demonstration of lower expression of the reg mRNA levels that we previously observed in this control strain.
To complete our study, we performed immunocy-tochemical localization of the reg protein(s) in the NOD and the control mice. To identify the mouse protein(s), we raised antibodies to a synthetic peptide of 16 amino acids present in the two proteins and corresponding to the most favorable regions for potential antigenic sites. Immunoblotting analysis on mouse pancreatic homogenates with these polyclonal antibodies recognized a single band at 16.6 kD consistent with the predicted mass of the mouse regI protein. However, it is difficult to know if this single band also corresponds to the putative regII protein. Indeed, regI and regII share 76% homology and differ by the presence of a unique additional sequence of seven amino acids in the N-terminal region in regII (Unno et al. 1993). SDS gel electrophoresis could not discriminate between two proteins with such close molecular weights. When the antibodies were applied to pancreatic mouse sections, positive staining was distinctly shown in exocrine tissue of all the NOD mice studied, normal, prediabetic, or diabetic, and in the control strain, whereas it was never observed in the endocrine tissue. The reg immunoreactivity observed in pancreatic ducts demonstrates that the protein(s) is (are) normally secreted in the pancreatic juice. Those data confirm the ISH studies and exclude the possibility of a preferential accumulation of the protein(s) in islet cells. In some pancreatic sections of a cyclophosphamide-treated male, we observed a stronger acinar staining in the cells surrounding the islets. This phenomenon was not consistently observed but might be related to the observation of Bendayan and Gregoire (1987), in streptozotocin-induced diabetic rats, that a brighter staining for chymotrypsinogen was evident in acinar cells closely surrounding the islets of Langerhans vs those located at a distance from the islets. Therefore, in the present study, the changes in reg gene(s) expression observed in NOD mice during active diabetogenesis cannot be accounted for by changes that may be occurring in islets because the reg gene(s) and protein(s) are found to be expressed only in acinar cells.
Immunocytochemical localization of reg in the pancreas of NOD mice during diabetogenesis. Reg immunostaining is localized in the cytoplasm of the exocrine acinar cells in all cases, in normal 12-day-old female pancreas without insulitis (
It is important to note that reg mRNA and protein expression remains restricted to exocrine tissue, both in cyclophosphamide-treated males, which represent a prediabetic stage, and in overtly diabetic females. In these two situations, which correspond to a striking evolution in the organization of the lesion within the islet and in which reg mRNA levels were found to be very high, reg was never expressed in endocrine tissue. However, we previously observed that if, as expected, the diabetic NOD females did not express insulin mRNA, cyclophosphamide treated males exhibited a significantly higher level of insulin expression than that measured in non-treated males and in non-diabetic NOD females (Baeza et al. 1997). This finding, which might be related to hyperactivity of the β-cell and perhaps to a defense mechanism, is therefore not associated with reg expression within the islets.
Do these data imply that reg cannot be considered as a growth factor involved in regeneration and/or growth of pancreatic islets, as hypothesized by Okamoto and his group? To address this issue, it appears necessary to distinguish between β-cell regeneration and/or replication and proliferation of differentiated β-cells. From the data that we have gathered on NOD mice, the possibility that reg may have a direct proliferative effect on preexisting β-cells seems unlikely. On the other hand, an effect on β-cell replication was observed by Watanabe et al. (1994) using recombinant rat protein on rat islets isolated from 90% pancreatized rats treated by nicotinamide, which is a model for islet regeneration. Similarly, they observed an amelioration of the diabetes by IP injection of the same recombinant protein. In this experimental model, the possibility that reg may not act on the proliferation of normal β-cells but in the process of ductule epithelial proliferation followed by endocrine cell differentiation and islet neogenesis can be suspected. The hypothesis of a direct trophic effect of reg, either on ductule precursor cells or on β-cells within the forming islets, had been proposed by Zenilman et al. (1996) in a rat model of islet hyperplasia. But does this effect, which might be due to the lectin similarities of the reg protein (Patthy 1988), represent the physiological role of this protein? The high levels of reg mRNA normally present in humans and other animals in the absence of pancreatic regeneration are not in agreement with this idea. Our present findings showing that the overexpression of the reg gene observed in an experimental model of autoimmune diabetes, the NOD mouse, was restricted to exocrine tissue led us to suggest that this overexpression represents a defense not of the islet cell but of the acinar cell against the pancreatic endocrine aggression releasing inflammatory products. It is indeed well established that secretions from the endocrine pancreas influence the exocrine parenchyma via the islet–acinar portal system (Williams and Goldfine 1985). More generally, the hypothesis that the reg gene was produced by acinar cells to protect them might also explain why the gene was normally present and overexpressed in all the different models of pancreatic aggression studied thus far. Further investigations supporting this hypothesis by clarifying the mechanisms of this protective effect remain necessary.
Footnotes
Acknowledgements
DS is supported by a fellowship of AJD (Aide aux Jeunes Diabétiques), France.
We are grateful to H. Okamoto (Japan) for providing us with the mouse regII cDNA probe and to B. Vialettes (Laboratoire de Diabétologie; Marseille, France) for allowing free access to the NOD mice.