GSK-3β Protein Phosphorylates and Stabilizes HLXB9 Protein in Insulinoma Cells to Form a Targetable Mechanism of Controlling Insulinoma Cell Proliferation*

Background: Germ line heterozygous loss of the MEN1 tumor suppressor gene causes tissue-specific tumors such as insulinomas. Results: GSK-3β and GSK-3β-mediated phosphorylation of HLXB9, a β-cell differentiation factor, is elevated in insulinomas, and GSK-3β inhibition blocks insulinoma cell proliferation. Conclusion: GSK-3β and phospho-HLXB9 form a targetable mechanism of insulinoma pathogenesis. Significance: Reactivation of a tissue-specific differentiation factor accounts for tumor tissue specificity. Insulinomas (pancreatic islet β cell tumors) are the most common type of functioning pancreatic neuroendocrine tumors that occur sporadically or as a part of the MEN1 syndrome that is caused by germ line mutations in MEN1. Tissue-specific tumor predisposition from germ line mutations in ubiquitously expressed genes such as MEN1 could occur because of functional consequences on tissue-specific factors. We previously reported the proapoptotic β cell differentiation factor HLXB9 as a downstream target of menin (encoded by MEN1). Here we show that GSK-3β inactivates the proapoptotic activity of HLXB9 by phosphorylating HLXB9 at Ser-78/Ser-80 (pHLXB9). Although HLXB9 is found in the nucleus and cytoplasm, pHLXB9 is predominantly nuclear. Both pHLXB9 and active GSK-3β are elevated in β cells with menin knockdown, in MEN1-associated β cell tumors (insulinomas), and also in human sporadic insulinomas. Pharmacologic inhibition of GSK-3β blocked cell proliferation in three different rodent insulinoma cell lines by arresting the cells in G2/M phase and caused apoptosis. Taken together, these data suggest that the combination of GSK-3β and pHLXB9 forms a therapeutically targetable mechanism of insulinoma pathogenesis. Our results reveal that GSK-3β and pHLXB9 can serve as novel targets for insulinoma treatment and have implications for understanding the pathways associated with β cell proliferation.

Insulinomas (pancreatic islet ␤ cell tumors) are the most common type of functioning pancreatic neuroendocrine tumors that occur sporadically or as a part of the MEN1 syndrome that is caused by germ line mutations in MEN1. Tissue-specific tumor predisposition from germ line mutations in ubiquitously expressed genes such as MEN1 could occur because of functional consequences on tissue-specific factors. We previously reported the proapoptotic ␤ cell differentiation factor HLXB9 as a downstream target of menin (encoded by MEN1). Here we show that GSK-3␤ inactivates the proapoptotic activity of HLXB9 by phosphorylating HLXB9 at Ser-78/Ser-80 (pHLXB9). Although HLXB9 is found in the nucleus and cytoplasm, pHLXB9 is predominantly nuclear. Both pHLXB9 and active GSK-3␤ are elevated in ␤ cells with menin knockdown, in MEN1-associated ␤ cell tumors (insulinomas), and also in human sporadic insulinomas. Pharmacologic inhibition of GSK-3␤ blocked cell proliferation in three different rodent insulinoma cell lines by arresting the cells in G 2 /M phase and caused apoptosis. Taken together, these data suggest that the combination of GSK-3␤ and pHLXB9 forms a therapeutically targetable mechanism of insulinoma pathogenesis. Our results reveal that GSK-3␤ and pHLXB9 can serve as novel targets for insulinoma treatment and have implications for understanding the pathways associated with ␤ cell proliferation.
Precise control of the expression and activity of differentiation factors is essential during embryonic development for the proper proliferation and differentiation of target cells and, later, for the maintenance of the differentiated state. It is important to understand how this control could go awry in differentiated cells, leading to re-entry into the cell cycle and increased cell proliferation. Tissue-specific tumor predisposition from germ line mutations in ubiquitously expressed genes could be attributed to the dysregulation of factors that specify the target tissues. Such tissue-specific endocrine tumors are manifested in the multiple endocrine neoplasia type 1 (MEN1) 3 syndrome and recapitulated in the mouse model of this syndrome (1,2). In both humans and in the mouse model, germ line heterozygous loss of the MEN1 tumor suppressor gene (encoding the protein menin) predisposes to simultaneous occurrence of tumors in three main endocrine organs: the parathyroids, the anterior pituitary, and the pancreas (3).
In mouse models, homozygous loss of Men1 in the insulinsecreting pancreatic islet ␤ cells or in the whole pancreas leads to only tumors of the ␤ cells (insulinoma) (4,5). Interestingly, Men1 loss in the islet ␣ cells causes insulinomas rather than glucagonomas because of trans-differentiation of menin-null ␣ cells into ␤ cells (6,7). Also, in a mouse model, Men1 loss in the liver did not cause tumors in the liver (8). These observations suggest the potential involvement of tissue-specific factors and differentiation factors in the pathogenesis of insulinomas. Furthermore, 40 -50% of sporadic pancreatic neuroendocrine tumors, including insulinomas, have somatic inactivation of at least one copy of MEN1 (9,10). Thus, the MEN1-encoded protein menin is critical for maintaining normal ␤ cell mass. Given that there are human tumors with MEN1 mutation and without 11q13 LOH (location of the gene), it is possible that menin could be haploinsufficient in certain tissues. For example, prior to the loss of the wild-type allele at 12 months, abnormal hyperplastic islets are observed in the conventional germ line Men1 heterozygous mouse model. Whether the effect on cell proliferation and function is due to menin haploinsufficiency together with other additional genetic or functional lesions is not known. Therefore, investigating downstream targets of menin could not only reveal the pathologic pathways associated with menin loss in MEN1 syndrome, but it could also provide insights into the cause of sporadic tumors that lack MEN1 mutations.
Kinases from the two major proliferation pathways, MAPK/ ERK and PI3K/AKT/mammalian target of rapamycin, have been investigated for targeted therapy of insulinomas (11). The serine/threonine kinase glycogen synthase kinase 3␤ (GSK-3␤) regulates a variety of physiological functions, including proliferation, differentiation, cell cycle progression, motility, and apoptosis (12). Interestingly, in ex vivo mouse model studies, GSK-3␤ inhibition suppressed the growth of medullary thyroid cancer, a type of neuroendocrine tumor (13). However, whether GSK-3␤ is important in insulinoma, a tumor of neuroendocrine cells of the pancreatic islet ␤ cells, has not been explored.
We have previously investigated a pancreatic ␤-cell differentiation factor, HLXB9 (HB9, MNX1, or MNR2) in the pathogenesis of insulinomas caused by menin loss (14,15). HLXB9 is a homeobox-containing transcription factor that acts early during embryonic ␤ cell development and differentiation and, later, in mature ␤ cells for the maintenance of the ␤ cell characteristic (16 -18). Also, it is involved in hematopoiesis and in the development of motor neurons (19,20). In the pancreas, HLXB9 is only expressed in ␤ cells (16).
We have shown that, similar to its function in motor neurons, HLXB9 overexpression caused apoptosis in ␤ cells (MIN6 cells). However, upon menin knockdown, HLXB9 could not cause apoptosis in ␤ cells (14). In this investigation, we found that HLXB9 was phosphorylated by GSK-3␤ and that this phosphorylation was increased upon menin knockdown, suggesting that the proapoptotic function of HLXB9 was inactivated by phosphorylation. Furthermore, both active GSK-3␤ and pHLXB9 were elevated under the following conditions: insulinoma cell line with menin knockdown, insulinomas from the mouse model of MEN1, and human sporadic insulinomas. Also, inhibition of GSK-3␤ in multiple insulinoma cell lines caused reduced cell viability, decreased proliferation, and induced apoptosis, implicating GSK-3␤ and pHLXB9 as potential targets to control cell proliferation in insulinoma.
Western Blot Analysis-Whole cell extracts (WCE) were processed for Western blot analysis with primary antibodies (supplemental Table 1), followed by HRP-conjugated secondary antibody and ECL (Millipore, Billerica, MA).
Cell Culture Treatments-All reagents were purchased from Sigma. For protein stability assays, 48 h post-transfection, cells were cultured with or without 20 g/ml cycloheximide for 1- Kinase Assay-GST, GST-HB9-wt, or GST-HB9-AA expressed in the Escherichia coli BL-21CodonPlus-RIL strain (Agilent, Santa Clara, CA) and purified on glutathione-Sepharose beads (14) was used for in vitro kinase assays with active recombinant GSK-3␤ or CDK5 (Signal-Chem, Richmond, Canada). Purified GST-HB9-wt and GST-HB9-AA showed a prominent band at 64 kDa for full-length protein together with multiple lower molecular weight bands that also contained GST-HB9, as assessed by Western blot analysis with anti-Hlxb9 (data not shown). GST-HB9 expression and purification from bacterial cells was prone to possible premature truncations or protein clipping.
Isolation of Mouse Islets-All mouse experiments were done under approved animal study protocols according to the guide-lines of the National Institutes of Health Animal Care and Use Committee. Mouse islets were isolated by intraductal collagenase perfusion following a standard protocol (25).
Mouse and Human Tissues and Immunohistochemistry (IHC)-Formalin-fixed, paraffin-embedded tissue sections from WT and Men1 ϩ/Ϫ mice (15 months old) and sporadic human insulinomas and normal human pancreas were obtained as follows. Mouse pancreas sections were cut from paraffin blocks provided by Dr. Peter Scacheri (Case Western Reserve University, Cleveland, OH). Human insulinomas (n ϭ 26) were obtained from patients under National Institutes of Health Institutional Review Board-approved protocols (NCT01005654) after written informed consent was obtained. The diagnosis of insulinoma was made on the basis of supervised fasting, with all patients having serum glucose levels of less than 45 mg/dl within 48 h of fasting, together with elevated plasma insulin, proinsulin, and C-peptide levels. All patients had their tumor localized and removed without any recurrence during followup. Biochemical testing for MEN1 was performed to confirm that patients had sporadic insulinoma. The MEN1 mutation status of these insulinomas is not known. Normal human pancreas (n ϭ 6) sections were obtained from different sources (Dr. Michael Emmert-Buck, NCI, National Institutes of Health; Origene, Rockville, MD; ProSci, Poway, CA; Abcam, Cambridge, MA; Zyagen, San Diego, CA; and US Biomax, Rockville, MD). IHC was performed using an EnVision Flex high pH kit (Dako, Carpinteria, CA). Microscopy and imaging was performed on a Keyence BZ900 microscope (BIOREVO series, Keyence, Itasca, IL) or Leica DMR microscope (Leica Microsystems, Buffalo Grove, IL).
GSK-3␤ Inhibition in Isolated Normal Mouse Islets-Handpicked islets seeded in 12-well plates (about 20 islets/well) were revived overnight in RPMI medium and treated with 1000parts per million camptothecin or 20 mM LiCl for 24 -48 h, followed by TUNEL staining (Promega). Immunofluorescence with anti-cleaved caspase 3 was also done, but the antibody was not very specific (data not shown).
Flow Cytometry and Cell Cycle Analysis-Cells seeded in 12-well plates (10 5 cells/well) were treated with LiCl (0 -20 mM) and incubated for 5 days (two cell divisions) with a change of LiCl-supplemented medium every other day. On day 5, cells were detached and incubated in Vindelov's propidium iodide buffer for 1 h on ice in darkness. Cell cycle histograms were generated by FACS (FACSCalibur, BD Biosciences). Each sample was analyzed in triplicate for 10,000 events each. The raw data were subjected to ModFit analysis to determine the percentage of cells in G 0 /G 1 , S, and G 2 /M phase.
Statistical Analysis-Data from at least three independent experiments were considered and presented as mean Ϯ S.E. Differences between groups were compared by Student's t tests. p Ͻ 0.05 (*) or p Ͻ 0.005 (**) were considered significant.

RESULTS
HLXB9 Phosphorylated at Ser-78/Ser-80 (pHLXB9) Is Predominantly Nuclear in ␤ Cells-Transfected myc-tagged-HLXB9 (mh-HB9-wt) in MIN6 cells was detected as a doublet band on Western blot analyses probed with an anti-Myc tag (Fig. 1A). This raised the possibility of secondary modifications of HLXB9 in the slower migrating upper band of the doublet. In the Uniprot database, two phosphorylation sites are reported in the mouse HLXB9 sequence at Ser-78/Ser-80 (conserved Ser-77/Ser-79 in human HLXB9) (Fig. 1B), on the basis of evidence from phosphopeptide sequencing of mitotic cells (27). Transfected myc-tagged-HLXB9 with alanine substitutions at Ser- 78/Ser-80 (mh-HB9-AA) showed a single band on Western blot analyses probed with an anti-Myc tag (corresponding to the lower band of the doublet) (Fig. 1A). Our Western blot analysis of mh-HB9-wt and mh-HB9-AA and the phospho-peptide sequencing data reported previously indicate that HLXB9 could be phosphorylated at Ser-78/Ser-80. Therefore, antibodies against a non-phosphorylated or phosphorylated peptide in the Ser-78/Ser-80 region of HLXB9 were generated (anti-HB9 and anti-HB9-PO 4 ) (Fig. 1B).
On Western blot analyses, anti-HB9-PO 4 detected no band for mh-HB9-AA, and a single band was detected for mh-HB9-wt that could be competed with the cognate phosphopeptide but not with the non-phosphorylated peptide (Fig. 1C). These observations demonstrate that the HB9-PO 4 antibody could specifically detect the phosphorylated isoform of HLXB9 (pHLXB9) and that the phosphorylation was located at Ser-78/Ser-80.
HLXB9 is found both in the nucleus and in the cytoplasm (14) Immunofluorescence analysis of transfected MIN6 cells with the Myc tag antibody detected mh-HB9-wt in the nucleus and cytoplasm (Fig. 1D, panel 2), whereas mh-HB9-AA was detected only in the cytoplasm (Fig. 1D, panels 6 and 7). However, with the HB9-PO 4 antibody, staining was observed predominantly in the nucleus of mh-HB9-wt-transfected cells (Fig.  1D, panel 3), whereas mh-HB9-AA-transfected cells did not show any staining (data not shown). Taken together, these results show that, in ␤ cells, HLXB9 could be phosphorylated at Ser-78/Ser-80 and that pHLXB9 was localized predominantly in the nucleus.
Decreased HLXB9 and pHLXB9 upon Treatment with Proteasomal or Lysosomal Inhibitors-During the characterization of HB9 antibodies, we observed that the phospho-defective form of HLXB9 (mh-HB9-AA) showed slightly reduced expression (Fig. 1A). RT-PCR analysis showed no significant differences at the mRNA level (data not shown), indicating that the decrease in expression was post-transcriptional, possibly because of lack of phosphorylation. A cycloheximide (Chx) treatment assay showed that the overall protein turnover was similar for both phospho-and phospho-defective HLXB9 ( Fig.  2A). Another possible reason for reduced expression of the mh-HB9-AA could be proteasome-or lysosome-mediated degradation. Unexpectedly, MG132 treatment (proteasome inhibitor) or NH 4 Cl treatment (lysosome inhibitor), rather than increasing the protein level, caused a decrease of both phosphoand phospho-defective HLXB9 (Fig. 2, B and C). These data indicate that inhibition of the proteasome or lysosome might stabilize another protein that could degrade HLXB9, irrespective of its phosphorylation state.
HLXB9 Is Phosphorylated at Ser-78/Ser-80 by GSK-3␤ in Vitro and in Vivo-To identify and characterize the kinase that could phosphorylate HLXB9, we first analyzed the kinases GSK-3␤ and CDK5 that were predicted by NetPhosK (Uniprot) for the consensus amino acid sequence near Ser-78/Ser-80. An in vitro kinase assay showed that GST-HB9-wt (Fig. 3A) was phosphorylated in the presence of GSK-3␤ (Fig. 3, B and C). The amount of phosphorylated GST-HB9 was low, possibly because of the absence of a priming kinase in the in vitro kinase assays. Usually, for GSK-3␤ to phosphorylate, a priming kinase phosphorylates a nearby amino acid (28). CDK5 could, in vitro, phosphorylate GST-HB9-wt and GST-HB9-AA almost equally. However, the overall phosphorylation was reduced in the presence of GSK-3␤ (Fig. 3, B and D), possibly because of competition between the two kinases.
To study the in vivo phosphorylation of HLXB9, MIN6 cells transfected with mh-HB9-wt were treated with various kinase inhibitors: bisindolylmaleimide (protein kinase C inhibitor), CR8 (CDK5 inhibitor), and GSK-3␤ inhibitors (kenpaullone, SB216763, and LiCl). Compared with the protein kinase C inhibitor or CDK5 inhibitor, the GSK-3␤ inhibitors decreased pHLXB9 (Fig. 3, E-G). Further analysis of MIN6 cells with Frat1 expression (known to degrade GSK-3␤) (29) showed decreased phosphorylation of mh-HB9-wt (Fig. 3H). The GSK-3␤ inhibitor LiCl efficiently reduced pHLXB9 (top band of the doublet). However, this did not cause an increase in the intensity of the lower band of the doublet (un-phosphorylated HLXB9), indicating that perhaps the unphosphorylated HLXB9 is not stable or is short-lived.
Together, these results demonstrate that GSK-3␤ could phosphorylate HLXB9 in vitro and in vivo. However, it is possible that other kinases, such as CDK5, could also phosphorylate HLXB9 at sites other than Ser-78/Ser-80.

pHLXB9 and GSK-3␤ Are Increased under Reduced Menin
Conditions-We have demonstrated previously that the protein level of HLXB9 increased upon menin knockdown in ␤ cells (14). Also, the proapoptotic effect of HLXB9 was reduced upon menin knockdown (14). Therefore, we investigated whether the phosphorylation state of HLXB9 was responsible for the reduction in HLXB9-induced apoptosis upon menin knockdown. In MIN6 cells, menin knockdown followed by overexpression of mh-HB9-wt showed that pHLXB9 was increased 24h post-transfection (Fig. 4, A and B). Similar results were obtained under another menin knockdown condition, islets from young Men1 ϩ/Ϫ mice (10 weeks) that retain one copy of Men1 (30). Elevated levels of pHLXB9 were seen in the Men1 ϩ/Ϫ islets compared with WT islets, and the staining for pHLXB9 was predominantly nuclear (Fig. 4C, panels 3 and 7). A Western blot analysis of MIN6 cells showed that reduced menin expression also led to increased GSK-3␤ and that menin overexpression led to decreased GSK-3␤ (Fig. 4D). However, GSK-3␤ mRNA levels were unaffected (Fig. 4, E and F). Analysis of posttranscriptional mechanisms of GSK-3␤ regulation revealed that overexpression or underexpression of menin did not affect the protein turnover or protein stability of GSK-3␤ (data not shown). Therefore, the precise mechanism by which menin regulates GSK-3␤ is not known. However, the above data show that, under reduced menin conditions, the protein level of GSK-3␤, HLXB9, and Ser-78/Ser-80 phosphorylation of HLXB9 was increased.
pHLXB9 and GSK-3␤ Are Elevated in Human Sporadic Insulinomas-Pathways affected in tumors of familial tumor syndromes (such as MEN1) could also be affected in the sporadic counterpart tumors. We explored this hypothesis in the context of human insulinomas. The levels of HLXB9, pHLXB9, and GSK-3␤ and its active and inactive forms were determined in normal pancreas sections (n ϭ 6) and in human sporadic insulinomas (n ϭ 26 total, n ϭ 19 for HLXB9 and n ϭ 22 for GSK-3␤; n ϭ 15 for both HLXB9 and GSK-3␤). Compared with the normal islets, in the insulinomas HLXB9, pHLXB9, and GSK-3␤ and its active form were elevated significantly (p Ͻ 0.05), and the inactive form of GSK-3␤ was absent or very low (Fig. 6, A-C).
In 20 mM LiCl-treated cell lysates of MIN6 cells, cleaved caspase 3 was detected, indicating that GSK-3␤ inhibition caused apoptosis (Fig. 8, A and B). We also examined whether GSK-3␤ inhibition in normal ␤ cells would cause apoptosis. Normal ␤ cell lines are not available. Therefore, the effect of 20 mM LiCl treatment was analyzed on normal mouse islets, which did not cause apoptosis, as assessed by TUNEL staining (Fig. 8C). Together, these results suggest that GSK-3␤ inhibition specifically reduces cell viability and proliferation in actively dividing insulinoma cell lines by inducing apoptosis.
We next investigated the mechanism by which GSK-3␤ inhibition reduces insulinoma cell proliferation. MIN6 cells treated with 20 mM LiCl showed a 16% increase in the number of cells in G 2 /M phase at the expense of cells in S phase of the cell cycle (Fig. 8, D and E). Thus, cells accumulated in G 2 /M phase might ultimately fail to enter the next cell cycle and die. These data indicate that GSK-3␤ modulates the cell cycle to control insulinoma cell proliferation.

DISCUSSION
Dysregulation of tissue differentiation factors could be potentially associated with aberrant cell proliferation and function in neoplasia of the target cell types. In this study, we investigated the modulation of a pancreatic ␤ cell differentiation factor, HLXB9, and its role in the insulin-secreting ␤ cell tumors (insulinomas) found in MEN1 syndrome, which are also the most common class of clinically significant and functional sporadic pancreatic neuroendocrine tumors. Identification of critical factors associated with insulinoma tumorigenesis is essential for the successful design of diagnostic and treatment modalities. Such investigations can also provide valuable insights into pathways that could be targeted for ␤ cell mass expansion during conditions of ␤ cell loss or increased insulin demand. We demonstrate that HLXB9 can be regulated by GSK-3␤-mediated phosphorylation at Ser-78/Ser-80, and we describe the importance of GSK-3␤ and pHLXB9 in MEN1associated and sporadic insulinomas and in the proliferation of insulinoma cell lines.
GSK-3␤-mediated Phosphorylation of HLXB9-The mouse and human HLXB9 protein consists of 404 and 401 amino acids, respectively, with a predicted molecular weight of 41 kDa.
However, on Western blot analyses, the endogenous and transfected myc-his-tagged HLXB9 migrate at a higher molecular weight, 55 and 64 kDa, respectively. The slow migration of HLXB9 predicts the presence of secondary modifications such as phosphorylation. The following evidence shows that HLXB9 undergoes GSK-3␤-mediated phosphorylation at Ser-78/Ser-80 in vitro and in vivo. Phosphorylation at Ser-78/Ser-80 in HLXB9 was observed by phosphopeptide sequencing (27). On Western blot analysis, mh-HLXB9-wt is detected as a doublet band, whereas alanine substitution at Ser-78/Ser-80 (mh-HB9-AA) abolishes the top band of the doublet. The pHLXB9-specific antibody reacts with the top band of the doublet but does not detect the lower band or mh-HB9-AA. Colocalization of pHLXB9 with active GSK-3␤ in the nucleus of ␤-cells, recombinant pure GSK-3␤ phosphorylated GST-HLXB9 by in vitro kinase assay, and GSK-3␤ inhibitor treatment or Frat-1 expression in MIN6 cells reduced the level of pHLXB9.
Consistent with our observation of reduced expression of phospho-defective HLXB9 in ␤-cells, in human fetal neural stem cells inhibition of PI3K/AKT or overexpression of active GSK-3␤ increased HLXB9 mRNA and protein expression (31), implicating that GSK-3␤-mediated phosphorylation could promote pHLXB9 stability. Also, decreased HLXB9 expression corelated with the presence of the inactive form of GSK-3␤ (31). Whether HLXB9 is phosphorylated on Ser-78/ Ser-80 in neural stem cells or in other cell types remains to be determined.
Phosphorylation at Ser-78/Ser-80 Inactivates the Proapoptotic Function of HLXB9-We have shown previously that, similar to the function of HLXB9 in motor neurons (20), overexpression of HLXB9 in ␤ cells (MIN6 cells) caused apoptosis (14). However, upon menin knockdown, overexpression of HLXB9 could not cause apoptosis (14). In this study, we found that, upon menin knockdown, the levels of pHLXB9 and GSK-3␤ were increased. These data indicate that phosphorylation of HLXB9 inactivates its proapoptotic function. Evasion of apoptosis is one of the hallmarks of cancer (32). The elevated levels of GSK-3␤ and pHLXB9 could drive the proliferation and accumulation of defective cells that evade apoptosis, leading to tumor formation.
Role of GSK-3␤ and pHLXB9 in MEN1-associated and Sporadic Insulinomas-In this study, we identified the elevated expression of GSK-3␤ and its active form and pHLXB9 as a common event in MEN1-associated mouse insulinomas and in sporadic human insulinomas. About 40 -50% of sporadic insulinomas show somatic mutation or deletion of at least one copy of the MEN1 gene (9,10,33). Therefore, we expected to find elevated pHLXB9 and GSK-3␤ in at least 50% of the insulinomas. However, our data show that the majority of the insulinomas examined had significantly elevated GSK-3␤ (73%), and staining is from hematoxylin counterstaining. B, pancreas sections of human sporadic insulinomas (n ϭ 22) and normal pancreas (n ϭ 6) were processed for IHC with antibodies against insulin, GSK-3␤, GSK-3␤-pTyr-216 (active form), or GSK-3␤-pSer-9 (inactive form). Bright field microscopy images from one normal section and one tumor section are shown. The islets stained with the indicated antibodies are marked with a dashed outline for clarity. Brown staining indicates positive staining. Blue staining is from hematoxylin counterstaining. The islet tumor part is marked with a red arrow, and the stromal/ exocrine part is marked with a green arrow. C, the intensities of IHC staining with the indicated antibodies of the pancreas sections processed in A and B were scored, and a heat map of the same was generated in Excel. The colors show the intensity of the staining, green being high and a gradual decrease to yellow shades being moderate to low to very low to absent. Cyto, cytoplasmic staining; Nuc, nuclear staining; N/D, not done because of the limited number of sections from the formalin-fixed, paraffin-embedded blocks. pHLXB9 (74%) levels, supporting a tumor promoter role of the combined expression of a tissue differentiation factor (HLXB9) and its kinase (GSK-3␤) in insulinomas. Thus, investigations that study familial tumor syndromes have the potential to unravel the mechanisms of tumorigenesis in their sporadic counterpart tumors.
The homeobox protein PDX1, which, like HLXB9, is essential in early pancreatic development, is stabilized by GSK-3␤mediated phosphorylation (34). The oncogenic activity of MafA, another ␤ cell differentiation factor, was enhanced by GSK-3␤-mediated phosphorylation (35). Therefore, in addition to HLXB9, GSK-3␤ targets such as PDX-1 and MafA may also activate oncogenic pathways during ␤ cell tumorigenesis. Identification of the affected pathways will provide insights into ␤ cell selective tumorigenesis in the pancreas with or without menin loss. Also, the elucidation of the precise mechanism by which menin regulates GSK-3␤ levels could provide insights into pathways that could regulate GSK-3␤ to understand menin loss-dependent and menin loss-independent tumorigenesis.
GSK-3␤ Inhibition Reduces Tumor Cell Proliferation-The tumor-promoter role of GSK-3␤ has been shown in various cancer cells: ovarian, colon, neuroblastoma, and non-endocrine pancreas and in neuroendocrine cell lines such as medullary thyroid cancer, BON1, gastrointestinal carcinoid, and H727 pulmonary carcinoid (36,37). Treatment of these cell types with the GSK-3␤ inhibitor LiCl caused apoptosis because of cell cycle arrest in G 2 /M, and decreased proliferation and tumor growth (13). GSK-3␤ inhibition studies in rodent pancreatic islet-tumor cell lines (InR1G9 and TGP-61) have been reported in the context of the role of menin in the Wnt signaling pathway, where LiCl reduced cell proliferation (38). We show that LiCl is toxic to 3 different rodent insulinoma cell lines (MIN6, ␤-TC3 and RINm5F), and in MIN6 cells LiCl reduced proliferation and increased apoptosis because of cell cycle arrest in G 2 /M. Therefore, GSK-3␤ inhibition can block cell proliferation via similar mechanisms in neuroendocrine tumor cells.
The importance of Wnt signaling has been investigated in the context of ␤ cell loss and diabetes where, in rodent cells, small molecule inhibitors of GSK-3␤ could improve ␤ cell function and ␤ cell mass (39). However, we found that, in actively dividing insulinoma cells, GSK-3␤ inhibition with LiCl could reduce cell proliferation and induce apoptosis, whereas LiCl did not cause apoptosis in normal mouse islets. It remains to be determined which signals up-regulate GSK-3␤ in ␤ cell tumors (other than menin loss) and how GSK-3␤ inhibition would affect the growth of usually slow-growing ␤ cell tumors.
Therapeutic Potential of Inhibiting GSK-3␤ and pHLXB9 in Insulinomas-Although we have not demonstrated a direct oncogenic effect of GSK-3␤ or pHLXB9, our data implicate a tumor-promoting role of GSK-3␤ and pHLXB9 in insulinomas that can be targeted by GSK-3␤ inhibitors such as LiCl to inhibit insulinoma cell proliferation. LiCl and other inhibitors of GSK-3␤ have been successfully used with no or minimal side effect in bipolar disorders for many years (40). The antimitotic effect of LiCl has been tested in mouse models (in vivo and ex vivo) of different cancers (13). Similarly, the mouse model of MEN1 could serve as an excellent preclinical model to test the efficacy of LiCl treatment in the context of insulinoma and other MEN1-associated endocrine tumors.
Our study establishes a targetable mechanism whereby the activity of a tissue differentiation factor (HLXB9) could be modulated by GSK-3␤-mediated phosphorylation to explain the tissue specificity of tumors in the ␤ cells of the pancreas. Whether similar mechanisms of tissue differentiation factor and kinase combination are active in other MEN1-associated tumors and their sporadic counterpart tumors remains to be determined.