Parathyroid Hormone–Related Protein Induces Insulin Expression Through Activation of MAP Kinase–Specific Phosphatase-1 That Dephosphorylates c-Jun NH₂-Terminal Kinase in Pancreatic β-Cells

Early-passage MIN6 cells before passage 20 were used in this study because they lose their well-functioning β-cell features, including their high insulin content, with successive passages (4,5). The cells were cultured in Dulbecco’s modified Eagle’s medium (DMEM; Sigma Chemical Co., St. Louis, MO) with 15% fetal bovine serum (Gibco Laboratories, Grand Island, NY) at 37°C in 5% CO₂. Mouse pancreatic islets were isolated from C57BL/6J mice (CLEA Japan, Tokyo, Japan) by pancreatic ductal injection of 500 units/ml collagenase (type XI; Sigma Chemical Co.). After digestion with collagenase, the islets were separated using a discontinuous Ficoll 400 gradient. The islets were cultured in DMEM with 15% fetal bovine serum. All media contained 25 mmol/l glucose unless otherwise stated.

Cellular insulin content was assayed in acid-ethanol extracts from the cells. Immunoreactive insulin (IRI) was measured using a time-resolved immunofluorometric assay (4).

RT-PCR was used to detect MKP family RNAs in the MIN6 cells. Total RNAs from MIN6 cells were reverse-transcribed with an oligo(dT)17 primer at 42°C for 1 h. With this single-strand cDNA as a template, MKP family cDNAs were amplified by PCR using each of the MKP-specific primers: MKP-1, 5′-TCAACGTCTCAGCCAATTGTCCT-3′ and 5′-CGTCCAGCTTTACCCGGTTAGTC-3′ (237 bp); MKP-2, 5′-ACTGTCCCAATCACTTTGAGGGTC-3′ and 5′-CCTCACCCGCTTCTTCATCATC-3′ (215 bp); MKP-3, 5′-TAGATACGCTCAGACCCGTG-3′, and 5′-GTGTTCTCATTCCAGTCGCTG-3′ (322 bp); MKP-5, 5′-GCAGGATGCTCAGGACCTAGACA-3′ and 5′-CATCCGTGTGTGCTTCATCAAGT-3′ (285 bp); and VHR, 5′-CGTTCGAACTCTCGGTGCAAGA-3′ and 5′-CATTTTTATGGGCCAGCGCCT-3′ (342 bp). As a control for the β-cells, the presence of insulin gene one was confirmed with the following primers: 5′-GCTCTCTACCTGGTGTGTGG-3′ and 5′-GTTTTATTCATTGCAGAGGG-3′ (257 bp).

Total RNA was isolated from MIN6 cells using the reagent TRIZOL (Life Technologies, Gaithersburg, MD) and was then electrophoresed on a 1.0% agarose gel. Hybridization probes were made by PCR-amplified cDNAs with the above-described primers. Hybridization was performed with these cDNA probes for mouse MKP-1, MKP-2, MKP-3, MKP-5, VHR, rat insulin 1 gene (4), and glyceraldehyde-6-phosphate dehydrogenase (GAPDH) after labeling with [α-³²P] deoxy-CTP by the random priming procedure. The relative ratio of mRNA expression levels was obtained by densitometric measurement and calculated assuming an insulin or MKP mRNA band adjusted by control GAPDH mRNA band on a no stimulation culture as 100%.

PTHrP-stimulated insulin expression was assessed in the presence or absence of the following inhibitors: adenosine 3′,5′-cyclic monophosphorothioate, 8-bromo-Rp-isomer, sodium salt (Rp-8-br-cAMPS; Calbiochem-Novabiochem Co., San Diego, CA) as a cAMP antagonist for cAMP-dependent protein kinase A; PD98059 (Calbiochem-Novabiochem Co.) as a MAP kinase kinase (MEK) inhibitor; SB203580 (Calbiochem-Novabiochem Co.) as a SAPK inhibitor; SP600125 (Calbiochem-Novabiochem Co.) as a JNK inhibitor; and Ro-31-8220 (LC Laboratories, San Diego, CA) as a MKP-1 expression inhibitor (30,31).

MIN6 cells were solubilized in lysis buffer (25 mmol/l HEPES [pH 7.5], containing 0.3 mol/l NaCl, 1.5 mmol/l MgCl₂, 0.1% TritonX-100, 0.1 mmol/l vanadate, and a protease inhibitor cocktail). Cell lysates (50 μg protein/lane) were separated on a PAGE for nitrocellulose membrane blotting. The blotted membranes were treated with 5% nonfat dry milk in 10 mmol/l Tris-HCl, 0.05% Tween 20, and primary antibodies to ERK1/2, phospho-ERK1/2, p38, phospho-p38, JNK1/2, phospho-JNK1/2, c-jun, and phospho-c-jun. These primary antibodies were obtained from Cell Signaling Technologies (Beverly, MA), except ERK1/2 (Santa Cruz Biotechnology, Santa Cruz, CA), and was used at a dilution of 1:1,000. After the anti-rabbit IgG second antibody reaction, the immunoblots were visualized with an ECL kit (Amersham-Pharmacia Biotech, Buckinghamshire, U.K.).

The full-length MKP-1 cDNA was obtained from a MIN6 cDNA library and subcloned into a pCEP4 vector (InVitrogen, Carlsbad, CA). The JNK1 cDNA with the nonphosphorylatably mutated site Ala-Pro-Phe from Thr-Pro-Tyr (JNK-APF) was made as described previously (32, 33). The JNK1-APF cDNA was also subcloned into the pCEP4 vector. Each expression unit in the vector was transferred to the cassette cosmid pAdex vector. The cosmid pAd-MKP-1 or pAd-JNK1-APF was subjected to a homologous recombination with the Eco T22I-digested DNA-terminal protein complex of Ad5-dlX in human embryonal kidney 293 cells for generating the recombinant viruses Ad-MKP-1 and Ad-JNK1-APF. A virus titer was determined using the 293 cells and was expressed as a multiplicity of infection (MOI). Because the adenoviral expression of β-galactosidase was observed in >75% of MIN6 cells by MOI 5 and >95% by MOI 20, we performed recombinant adenovirus infection by MOI 20, unless otherwise stated.

We initially confirmed that the increase in cellular insulin content by PTHrP is mediated by its mRNA expression. The increase is obvious in both the content and mRNA level of insulin at 10⁻⁸ mol/l PTHrP (Fig. 1A). The increase was rapid during the first 24 h, then became gradual up to 72 h after the addition of PTHrP (Fig. 1B). With 10⁻⁸ mol/l PTHrP, the insulin content increased to 1.28 ± 0.06 nmol/10⁶ MIN6 cells for 24-h incubation and 1.44 ± 0.05 nmol/10⁶ cells for 72-h incubation from 0.69 ± 0.04 nmol/10⁶ cells at the start of incubation. Because the PTHrP effect is clearly observed at 10⁻⁸ mol/l by a 24-h incubation time, we used this dosage and incubation time for evaluating the effect of PTHrP on insulin expression.

The insulin secretion by high levels of glucose is enhanced by a cAMP-raising secretagogue such as glucagon-like polypeptide-1 (GLP-1; Fig. 1C). Because PTHrP is known to increase intracellular cAMP, we examined potentiation of glucose-stimulated insulin secretion by PTHrP. The original MIN6 cells secreted a 4.5-fold higher level of insulin when in a 25.0 mmol/l glucose solution than when in a 2.0 mmol/l glucose solution. GLP-1 significantly enhanced this glucose-stimulated insulin secretion by 6.5-fold (Fig. 1C). Although PTHrP apparently did increase the insulin secretion, the increase was not as significant as that of GLP-1. Thus, the increase in cellular insulin content by PTHrP was, at least, not due to stasis of insulin inside the cells.

We further confirmed that insulin expression by PTHrP is not limited to β-cell culture lines but was observed primarily in nontransformed β-cells. PTHrP increased both insulin content and its mRNA levels in mouse primary-cultured islets as much as in MIN6 cells (Fig. 1D).

The PTHrP signal for insulin induction is mediated by the cAMP pathway in MIN6 cells (4). We first confirmed that the PTHrP-induced effect on the content and mRNA level of insulin was suppressed to a control level by the inhibitory cAMP analogue Rp-8-br-cAMPS (10 μmol/l), which inhibits cAMP-dependent protein kinase A (Fig. 2). Because PTHrP is known to interact with the MAP kinase pathway (16, 17) and the cAMP pathway has been reported to interact with this pathway in an enhancing or suppressing manner depending on the cell types (12–15), we examined the involvement of the MAP kinases in insulin expression by PTHrP using MAP kinase pathway inhibitors. We initially used two inhibitors: PD98059 as a MEK inhibitor and SB203580 as a SAPK (p38 and JNK) inhibitor. PD98059 (50 μmol/l) did not affect the content or mRNA levels of insulin in either PTHrP-added or -nonadded culture (Fig. 2). We further used a wide range of PD98059 doses (0.5–50 μmol/l) but found no change in insulin expression, although the formation of phosphorylated ERK1/2 was markedly suppressed by a higher dosage (Fig. 3A).

In contrast, SB203580 (20 μmol/l) markedly increased the content and mRNA levels of insulin in both PTHrP-added and -nonadded cultures (Fig. 2). Although SB203580 is known to inhibit both p38 and JNK1/2, p38 is reportedly more sensitive to SB203580 than is JNK1/2 (34). Thus, we examined the dose-dependence of SB203580 on insulin induction from 0.5 to 50 μmol/l. SB203580-induced insulin expression was clearly evident at 20 μmol/l but not at <5 μmol/l. SB20380 consistently inhibited the phosphorylation of JNK1/2 at >20 μmol/l, but the SB203580-induced inhibition of p38 phosphorylation was even observed at <1 μmol/l, whereas its inhibition of JNK1/2 phosphorylation was ineffective (Fig. 3B). To confirm further the involvement of JNK1/2, we used the JNK-specific inhibitor SP600125 (35). SP600125 caused an increase in insulin content and inhibited JNK phosphorylation clearly at 5 μmol/l, a dosage at which the effect of SB203580 on insulin expression was not evident (Fig. 3C). This compound did not affect phosphorylation of ERK1/2 and p38, but it sensitively inhibited JNK1/2 phosphorylation.

We presumed that PTHrP may inhibit JNK1/2 activation by decreasing its phosphorylation, and this presumption was confirmed, as shown in Fig. 4. An obvious decrease in phosphorylated JNK1/2 was observed at 6 h after the PTHrP addition, and this continued for up to 24 h. To establish clearly the involvement of JNK1/2 in insulin expression, we made dominant-negative JNK1 (JNK1-APF), in which the specific phosphorylation site Thr-Pro-Tyr was mutated to the nonphosphorylatable site Ala-Pro-Phe by upstream MAPKK (32,33). When this JNK1-APF was adenovirally introduced to MIN6 cells, nonphosphorylated JNK1 blot became massive, and the phosphorylated form of JNK1/2 blots were unchanged between control and JNK1-APF–expressing MIN6 cells (Fig. 5A). However, c-jun phosphorylation was markedly inhibited in dominant-negative JNK1-APF–expressing MIN6 cells (Fig. 5A). As expected, JNK1-APF expression increased the content and mRNA level of insulin, indicating that a JNK/c-jun pathway is involved in insulin induction by PTHrP (Fig. 5B and C).

As an endogenous inhibitory regulator of MAP kinases, we investigated whether MKPs are involved in the insulin induction by PTHrP. MKPs are known to dephosphorylate MAP kinases for inactivation (28,29). However, we began by examining the expression of five MKPs in MIN6 cells using RT-PCR. We first confirmed the presence of the five MKPs: MKP-1, MKP-2, MKP-3, MKP-5, and VHR, by RT-PCR (data not shown) and cloned each cDNA fragment for a Northern blot probe. We next examined the PTHrP effect on the induction of MKPs by Northern blot. Among these five MKPs, only MKP-1 was markedly induced by PTHrP (Fig. 6A). The increase was clearly evident even 72 h after the PTHrP treatment, although MKP-1 is reportedly classified as an immediate early gene (36) (Fig. 6B). The increase in MKP-1 expression by PTHrP was also confirmed in primary-cultured mouse islets (Fig. 6C). Furthermore, this increase was suppressed by the inhibitory cAMP analogue Rp-8-br-cAMPS (Fig. 6D), suggesting that PTHrP-induced intracellular cAMP stimulates MKP-1 expression.

To confirm the involvement of MKP-1 in insulin induction, we used the inhibitor Ro-31-8220, which inhibits MKP-1 expression (30,31). Ro-31-8220 was initially identified as a protein kinase C (PKC) inhibitor but was later demonstrated to inhibit MKP-1 expression by cytokines such as tumor necrosis factor-α and interleukin 1β (30). We initially confirmed that Ro-31-8220 inhibited the PTHrP-induced MKP-1 expression and also decreased endogenous MKP-1 expression to a considerable extent (Fig. 7A). Ro-31-8220 clearly inhibited the PTHrP-induced increase in insulin content and its mRNA expression in MIN6 cells (Fig. 7B) and in primary-cultured mouse islets (Fig. 7C). Because we previously observed no involvement of PKC in PTHrP-induced insulin expression using PKC inhibitor H7 (4), it is evident that the effect of Ro-31-8220 is due to inhibition of MKP-1 expression (Fig. 7A).

PTHrP-induced insulin expression is suggested to be mediated through MKP-1 expression by PTHrP. To test this hypothesis, we made an adenoviral MKP-1 expression vector. As expected, cellular insulin content and its mRNA level increased two- to threefold over the control level of nontransformed cells with an increase in the viral MOI number from 20 to 50 (Fig. 8A and B). Next, we examined the effect of MKP-1 on the formation of phosphorylated JNKs and c-jun. Adenoviral expression of MKP-1 caused a decrease in phosphorylated JNK1/2 and a resultant reduction of a phosphorylated form of c-jun with an increase of the MOI number (Fig. 8C). We suggest that MKP-1 is involved in the induction of insulin expression through its dephosphorylation reaction of the JNK/c-jun pathway.

PTHrP caused the insulin content to increase approximately twofold in both MIN6 cells and primary-cultured mouse islets. Furthermore, PTHrP-inducible MKP-1 expression caused a two- to threefold increase in insulin content. The insulin content in rodent islets is ∼40–50 ng/islet, when the islet diameter is ∼150 μm (37). Islets can range in size from 50 to 500 μm, and each islet contains an average of ∼1,000 β-cells. Thus, insulin content can be calculated to be 8–10 nmol/10⁶ β-cells (38). Indeed, our data indicated ∼1.1 nmol insulin/100 mouse islets, corresponding to 11 nmol/10⁶ β-cells (Fig. 2). Among β-cell culture lines, MIN6 is thought to be one of the most highly functioning and contains 0.5–1.0 nmol insulin/10⁶ cells (Fig. 1). Although the magnitude of this value is one tenth lower than that of islet β-cells, other culture lines such as βTC3, HIT-15T, and RINm5F contain far less insulin (6). Thus, we think that early-passage MIN6 cells are an appropriate substitute for well-functioning islet β-cells.

For attaining a well-functioning β-cell mass, PTHrP is a very promising candidate among insulinotropic peptides including hepatocyte growth factor, GLP-1, and exendin-4 (3,39). Indeed, rat insulin gene–promoted overexpression of PTHrP in mouse islets presented a twofold increase in total islet number and total islet mass, as well as in insulin content (9). In contrast, transgenic mice expressing exendin-4, a long-acting GLP-1–like peptide, presented no remarkable change in total islet number and total islet mass (40). Thus, PTHrP seems to be potent in upholding a well-functioning β-cell mass. In the present study, PTHrP induced a twofold increase in insulin content and its mRNA in MIN6 cells and even in primary cultured mouse islets (Fig. 1). The insulin content increased maximally to ∼2 nmol/10⁶ MIN6 cells, which corresponds to a quarter of the value per 10⁶ nontransformed islet β-cells. To our knowledge, this high level of insulin content has not been reported in cultured β-cell lines.

PTHrP induced insulin expression by activating a cAMP/PKA pathway, because Rp-br-cAMPS, a cAMP antagonist for PKA, inhibited PTHrP-induced insulin expression (Fig. 2). Gaich et al. (41) studied a RIN1056-38 β-cell line and reported that PTHrP elicited an increase in [Ca²⁺]i. In contrast, in our previous report using MIN6 cells and primary-cultured mouse islets, we observed no such [Ca²⁺]_i influx, and the PKC inhibitor H7 did not interfere with insulin expression by PTHrP (4). Because MIN6 cells exhibit nontransformed, well-functioning β-cell features such as high insulin content, the signal transduction for insulin expression by PTHrP may differ from that of a rapidly growing RIN1056-38 cell line with far less insulin content (42). Another cAMP-raising insulinotropic peptide, GLP-1, is also known to exert potentiation of the insulin-secreting effect by glucose (Fig. 1C). However, GLP-1 did not increase insulin content in MIN6 culture, perhaps because of its much shorter half-life than that of PTHrP (data not shown). Here, we need to distinguish between short-acting and long-acting mechanisms of a cAMP-raising insulinotropic peptide. In insulin secretion from β-cells, a high intracellular cAMP level raised by a cAMP-raising peptide activates PKA, which results in enhancing glucose metabolism-dependent Ca²⁺ mobilization largely by activating the type 2 isoform of ryanodine receptor and L-type Ca²⁺ channels (43). This cAMP-dependent Ca²⁺ mobilization occurs within a minute order after the stimulation with glucose plus insulinotropic hormones. Indeed, PTHrP raises an intracellular cAMP level within 5 min in MIN6 cells (4). In this short cAMP-dependent reaction, MKP-1 may not be involved in the potentiating mechanism for insulin secretion because MKP-1 induction by PTHrP or cAMP is at least an hour-order event, as shown in Fig. 6B. In view of PTHrP production level of 20–40 fmol · ml⁻¹ · 24 h⁻¹ from MIN6 cells (4), PTHrP may not be an effective insulinotropic peptide. Instead, PTHrP may act as an autocrine/paracrine factor because of its long-half life by activating MKP-1 for β-cell functions in islets. In this study, PTHrP increased insulin content and its mRNA level for at least 3 days after its addition to MIN6 culture (Fig. 1).

During a MAP kinase inhibitor study on PTHrP-induced insulin expression, we noted a marked enhancing effect of the SAPK inhibitor SB203580 on the content and mRNA level of insulin. This effect was observed even without PTHrP in MIN6 culture. Another SAPK inhibitor, SP600125, also enhanced insulin expression. SB203580 is known to inhibit p38 and, to a lesser extent, JNK1/2, whereas SP600125 inhibits JNK1/2 specifically (34,35) (Fig. 3B and C). Thus, the involvement of JNK1/2 in insulin expression seems to be crucial. Indeed, PTHrP decreased a phosphorylation form of JNK1/2 in a time-dependent manner, at least up to 24 h after PTHrP addition to MIN6 culture (Fig. 4). Furthermore, the dominant-negative form of JNK, JNK-APF, repressed PTHrP-induced insulin expression by inhibiting phosphorylation of c-jun (Fig. 5). The phosphorylated form of c-jun reportedly interfered with the cAMP-activated human insulin gene promoter in HIT cells (24).

The influence of MAP kinases on insulin gene expression has been studied extensively in β-cell lines, including the INS-1 and MIN6 lines (18–23). Although glucose activated ERK1/2 for its subsequent translocation to the nucleus, the activation did not affect insulin secretion in INS-1 and MIN6 cells (18,19). Our data clearly showed that the MEK inhibitor PD98059 did not affect insulin expression in MIN6 cells, although it inhibited ERK phosphorylation specifically (Fig. 3A). Glucose is also known to activate p38 in β-cells, but an increase in insulin expression by p38 has not been reported previously (21,22). In fact, p38 has been reported to repress GLP-1–activated rat insulin 1 gene promoter activity (23). The JNKs play an important role in β-cell functions (44, 45). As described above, c-jun is known to inhibit cAMP-responsive element-binding protein–promoted insulin gene transcription (24). In β-cells, JNK interacts with JIP-1/JIP-2, which serves as a scaffold for inhibiting JNK phosphorylation activity, resulting in a decrease in c-jun activation. The trapping of JNK by JIP proteins seemed to further block phosphorylation of the apoptotic inhibitor Bcl-2 by JNK, resulting in the reduction of cytokine-induced apoptosis in β-cells (26). PTHrP induces the expression of Bcl-2 in growth plate chondrocytes, leading to delays in their maturation toward hypertrophy and apoptotic cell death (46). Thus, JNK inactivation by PTHrP seems to induce not only insulin activation but also prolongation of β-cell survival.

In addition to JIP proteins, we believe that MKPs are important for modulating MAP kinase functions in β-cells. We identified at least five MKPs in β-cells and found that MKP-1 is inducible by PTHrP. MKP-1 expression by PTHrP lasts up to 3 days after the addition of PTHrP to MIN6 culture (Fig. 6). The MKP-1 induction by PTHrP was inhibited by Rp-8-br-cAMPS, suggesting that a cAMP pathway is crucial for this induction by PTHrP. Indeed, MKP-1 was reportedly expressed by the cAMP-elevating agent forskolin in a rat pheochromocytoma PC12 cell line (47). Furthermore, the MKP-1 expression inhibitor Ro-31-8220 repressed the increase in the content and mRNA level of insulin induced by PTHrP (Fig. 7), indicating that the PTHrP-stimulated insulin induction is mediated by MKP-1 activity. Adenoviral expression of MKP-1 in MIN6 cells resulted in an increase in insulin expression by decreasing JNK phosphorylation and the resultant c-jun phosphorylation (Fig. 8).

MKP-1 is able to dephosphorylate ERK1/2 efficiently, with lower activity toward p38 and JNKs, although the binding capacity of MKP-1 is highest for JNKs (36,48). Among three major MAP kinases, we can exclude the involvement of ERK1/2 in insulin expression through a MKP-1–mediated pathway, as shown in Fig. 3A. We may also exclude the role of p38 in insulin expression: first, because a dominant-negative JNK-APF induced insulin expression by blocking c-jun phosphorylation (Fig. 5); and second, because p38 does not phosphorylate c-jun efficiently, although c-jun is a good substrate for JNK (44,49). Thus, we suggest that the dephosphorylation of JNKs by MKP-1 is essential for the expression of the insulin gene. Furthermore, the JNK/c-jun pathway is known to play an important role in cell apoptosis, perhaps by inducing caspase cascade activation and/or Bcl-2 inactivation by JNK phosphorylation (50,51). We think that MKP-1 expression leads to insulin gene expression extensively through JNK/c-jun–mediated pathways, together with a decrease in apoptotic cell death (31). Thus, the role of cAMP-raising peptides, including PTHrP and the cAMP-responsive mediator MKP-1, is essential for the control of β-cell functions. We believe that PTHrP is a promising therapeutic peptide for the improvement of dysfunctional β-cells.

We thank Mari Kosaki for secretarial assistance and Mari Hosoi for technical assistance.

This work is supported by Grants-in-Aid and the 21st Century COE Program from the Ministry of Education, Culture, Science, Sports and Technology.