Functional Defects From Endocrine Disease-Associated Mutations in HLXB9 and Its Interacting Partner, NONO.

The insulin-secreting pancreatic neuroendocrine tumors, insulinomas, characterized by increased pancreatic islet β-cell proliferation, express the phosphorylated isoform of the β-cell differentiation factor HLXB9 that interacts with NONO/p54NRB, a survival factor. Interestingly, two different homozygous germline mutations in HLXB9, p.F248L and p.F272L, were reported in neonatal diabetes, a condition with functional β-cell deficiency. Also, two somatic heterozygous NONO mutations were found in endocrine-related tumors, p.H146R (parathyroid) and p.R293H (small intestine neuroendocrine tumor). However, the biological consequence of the mutations, and the role of HLXB9-NONO interaction in normal or abnormal β cells, is not known. Expression, localization, and functional analysis of the clinically relevant HLXB9 and NONO mutants showed that HLXB9/p.F248L mutant localized in the nucleus but lacked phosphorylation, and NONO/p.R293H mutant was structurally impaired. The HLXB9 and NONO mutants retained the ability to interact, and overexpression of wild-type or mutant HXLB9 in MIN6 cells suppressed cell proliferation. To further understand the biological consequence of the HLXB9-NONO interaction, we mapped the NONO-interacting region in HLXB9. An 80-amino acid conserved region of HLXB9 could compete with full-length HLXB9 to interact with NONO; however, in functional assays, nuclear expression of this HLXB9-conserved region in MIN6 cells did not interfere with cell proliferation. Overall, our results highlight the importance of HLXB9 in conditions of β-cell excess (insulinomas) and in conditions of β-cell loss or dysfunction (diabetes). Our studies implicate therapeutic strategies for either reducing β-cell proliferation in insulinomas or alleviating normal β-cell deficiency in diabetes through the modulation of HLXB9 phosphorylation.

Pancreatic neuroendocrine tumors (PNETs; also known as PanNET or pNETs) arise from the cells of the islets of Langerhans in the pancreas. PNETs are broadly classified as clinically functioning when they hypersecrete the hormone characteristic of the cell of origin, or nonfunctioning when they do not. PNETs occur sporadically or as a part of inherited tumor syndromes such as multiple endocrine neoplasia type 1 (MEN1). MEN1 is characterized by germline heterozygous mutations in the MEN1 tumor suppressor gene (1, 2). Tissue-specific loss of the remaining normal copy of the MEN1 gene leads to tumors in multiple endocrine organs, mainly the parathyroids, pancreas, and pituitary (3). Inactivating somatic mutations in the MEN1 gene are also observed in 40% of nonfunctioning sporadic PNETs; however, mutations in the MEN1 gene are uncommon in functioning sporadic PNETs (which are mainly the insulin-secreting islet β-cell tumors known as insulinomas) (4–9). Interestingly, the PNETs that develop in the mouse model of Men1 loss (Men1+/−) are insulinomas, indicating that the MEN1 gene is critical for controlling β-cell proliferation (10–13). Therefore, it is possible that, although sporadic human insulinomas lack mutations in the MEN1 gene that encodes menin, alterations in the targets that are downstream of menin may be relevant. One such target is the β-cell differentiation factor, homeobox protein HLXB9 (also known as HB9, MNR2, or MNX1) (14).

HLXB9 is a transcription factor that is expressed during embryonic pancreas and β-cell development, and later in adult β cells. Mouse models have shown that this precise temporal regulation of HLXB9 expression is critical for pancreatic islet development and for maintaining β-cell fate. Genetically engineered mice with loss of Hlxb9 show dorsal pancreas lobe agenesis, abnormal islet structure, and reduced number of β cells in the ventral pancreas, but with normal exocrine function (15, 16). Mice with continuous HLXB9 expression in the whole pancreas under the control of the PDX1 promoter show impaired pancreas development with pancreatic cells that adopt intestinal fates (17). In a mouse model of β-cell–specific Hlxb9 loss, the inappropriate reexpression of HLXB9 in adult β cells has been shown to cause β-cell hyperplasia (18); therefore, Hlxb9 is critical for proper β-cell development, proliferation, and function. Mutations in β-cell differentiation factors have been reported in a subset of patients with diabetes, a condition characterized by inadequate or dysfunctional β cells (19). Also, two different homozygous germline mutations in HLXB9 have been identified in patients with permanent neonatal diabetes mellitus (PNDM), p.F248L and p.F272L (20, 21). However, the consequence on these mutations on HLXB9 expression or function is not known.

Loss of menin in mouse β cells correlates with increased HLXB9 messenger RNA (mRNA), and short hairpin RNA-mediated suppression of menin expression in mouse insulinoma cells has been shown to increase the level of HLXB9 protein (14, 22). HXLB9 protein is highly expressed and phosphorylated at serine-78 and serine-80 in both human sporadic insulinomas (which express menin) and mouse insulinomas (which lack menin, from a mouse model of Men1 loss) (14, 23, 24). This phospho isoform of HLXB9 was shown to interact with a survival factor non–POU domain–containing octamer-binding protein NONO (also known as p54NRB) that is capable of binding to long noncoding RNA to form subnuclear bodies called paraspeckles and of binding to DNA for regulating gene expression (24–27). The role of the HLXB9-NONO interaction in β-cell proliferation is not well understood. Interestingly, mutations in NONO have been reported in tumors of endocrine cells: p.H146R in a parathyroid adenoma and p.R293H in a small intestine neuroendocrine tumor (SI-NET) (28, 29). However, the consequence on these mutations on the expression of NONO or its function is not known.

To gain insights into the mechanisms of β-cell proliferation, we have determined the biological consequence of the clinically relevant HLXB9 and NONO mutations using expression, localization, and functional assays. One HLXB9 mutant (HLXB9/p.F248L) could localize in the nucleus but it lacked phosphorylation, and one NONO mutant (NONO/p.R293H) was structurally impaired. The HLXB9 and NONO mutant proteins were intact for interaction with normal NONO and normal HLXB9 protein, respectively. Similar to normal HLXB9, the two HXLB9 mutants could suppress the proliferation of MIN6 cells. We have mapped the NONO-interacting region in HLXB9 to an 80-amino acid (aa) conserved region that could compete with full-length HLXB9 to interact with NONO. However, in functional assays heterologous nuclear localization signal (NLS)–directed nuclear expression of this HLXB9-conserved region in MIN6 cells did not interfere with cell proliferation. These results from the analysis of defects in HLXB9 and NONO mutants highlight the importance of HLXB9 phosphorylation and, through the modulation of HLXB9 phosphorylation propose dual benefits, for the management of diseases associated with increased β-cell proliferation (insulinomas) or diseases associated with functional β-cell loss (diabetes).

Materials and Methods

Primers, plasmids, and antibodies

All of the antibodies, primers, and plasmids used in this study are listed in Table 1 and Supplemental Tables 1 and 2, respectively The mammalian expression vectors used were pcDNA3.1-myc-his (pcDNA3.1-mh) (Invitrogen), pCMV/myc/nuc (Invitrogen), and pFLAG-CMV-4 (Sigma). The HLXB9 mutant plasmids, pcDNA3.1-mh-HB9-F248L and pcDNA3.1-mh-HB9-F272L, were generated by site-directed mutagenesis (Stratagene) using pcDNA3.1-mh-HB9-wild-type (WT) as the template (14). The NONO mutant plasmids, pFlag-NONO-H146R and pFlag-NONO-R293H, were constructed by site directed mutagenesis using pFlag-NONO-WT as the template (Addgene) (24). Different regions of mouse HLXB9 ( 1 through 160, 1 through 240, 1 through 302, 160 through 273, 161 through 404, 241 through 404, and 301 through 404) were amplified by polymerase chain reaction (PCR) using Pfu polymerase (Stratagene) from the plasmid pcDNA3.1-mh-HB9-WT and cloned into pcDNA3.1-mh, in-frame with a C-terminal myc-his-tag. HLXB9 internal regions consisting of aa 160 through 240, 160 through 203, and 195 through 240 were PCR amplified from pcDNA3.1-mh-HB9-WT and cloned in-frame with the ATG of the Nco-I cloning site of pCMV/myc/nuc vector and in-frame with its C-terminal myc-tag and three copies of a nuclear localization signal (3XNLS). Regions of human NONO (aa 1 through 471, 1 through 236, and 228 through 471) were amplified by PCR from pFlag-NONO-WT and cloned into pFLAG-CMV-4, in-frame with an N-terminal Flag-tag.

Table 1.

List of Antibodies Used in This Study

Antibody Name	Manufacturer	Catalog No. or Reference	RRID
Rabbit anti-HB9-PO4	Bethyl (custom)	Desai et al. (23)	AB_2721068
Mouse anti-Myc-tag	Upstate/Millipore	05-724	AB_309938
Rabbit anti-Myc-tag	Upstate/Millipore	06-549	AB_310165
Mouse anti-Flag-tag	Sigma	F3165	AB_259529
Mouse anti-p54 (Nono)	Upstate/Millipore	05-950	AB_492627
Rabbit anti-cleaved-Caspase3	Cell Signaling Technology	9661	AB_2341188
Mouse anti-p84	GeneTex	GTX70220	AB_372637
Chicken anti-insulin	Abcam	ab14042	AB_300872
Mouse secondary antibody (HRP)	Santa Cruz	sc-2055	AB_631738
Rabbit secondary antibody (HRP)	Santa Cruz	sc-2054	AB_631748
Chicken secondary antibody (HRP)	Abcam	ab6753	AB_955464
Anti-rabbit Alexa Fluor 594 (red)	Invitrogen	A11037	AB_2534095
Anti-mouse Alexa Fluor 488 (green)	Invitrogen	A11029	AB_2534088

Abbreviations: HRP, horseradish peroxidase; RRID, Research Resource Identifier.

A bacterial expression vector (pGEX5X-1) (Pharmacia) was used to generate glutathione S-transferase (GST)-NONO expression constructs GST-NONO-WT, GST-NONO-(1-236), and GST-NONO-(227-471) by PCR amplification from pFlag-NONO-WT. GST-CDK2 and GST-HB9 expression constructs GST-HB9-WT, GST-HB9-(1-240), and GST-HB9-(241-404) were previously published (14, 30).

Cell culture, transfection, and protein extract

The MIN6-4N mouse insulinoma cell line was cultured in low-glucose Dulbecco’s modified Eagle medium supplemented with 15% fetal bovine serum and 1X antibiotic/antimycotic (Invitrogen) at 37°C with 5% CO2 (23, 31). Insulin production by this cell line was authenticated by reverse transcription PCR for insulin mRNA, and insulin enzyme-linked immunosorbent assay. Plasmids were transfected using Lipofectamine 2000 (Invitrogen) or by nucleofection (AMAXA/Lonza). Protein whole cell extracts (WCEs) were prepared 48 or 96 hours posttransfection in NETN buffer (20 mM Tris pH7.5, 100 mM NaCl, 0.5 mM EDTA, 0.5% NP40) containing a protease inhibitor cocktail (Roche). Protein quantitation was performed with the DC protein assay kit (Bio-Rad).

Western blot

Proteins from WCE or GST pull-down assay were subjected to sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE; 4% to 20%, 10%, or 8% gels) followed by western blot as per standard procedures (Invitrogen and Bio-Rad). Western blots were developed to detect proteins by enhanced chemiluminescence (Millipore). The blots were exposed to X-ray film (Kodak) or imaged on the G:BOX (Syngene).

Cell proliferation and apoptosis assay

MIN6-4N cells were transfected with the indicated plasmids by a reverse lipofectamine transfection protocol (Invitrogen) in 6-well plates to assess proteins by western blot of WCE and in 96-well plates to assess cell viability by MTT assay (Promega), respectively. Differences in cell proliferation were studied by using the MTT assay, for which 15,000 cells were plated in triplicate wells of 96-well plates containing lipofectamine-plasmid DNA mix. MTT assay was done 48 and 96 hours posttransfection, and the optical density was measured at 570 nm (SpectraMax190, Molecular Devices). To assess apoptosis, cleaved caspase-3 that is produced as a result of apoptosis was detected by western blot of WCEs prepared 96 hours posttransfection.

Immunofluorescence and microscopy

MIN6-4N cells transfected with various plasmids were cultured in chamber slides, formaldehyde-fixed, permeabilized, and processed for immunofluorescence (IF) (32). Microscopy and imaging were performed with the BZ-9000 microscope (KEYENCE).

GST-fusion proteins and GST pull-down assay

GST or GST-fusion proteins were expressed in the Escherichia coli strain BL21-PRIL (Stratagene) and purified using glutathione sepharose beads (GE Healthcare) (33). The purity of the GST-fusion protein preparations was analyzed by SDS-PAGE on 4% to 20% gels followed by Coomassie blue stain (Pierce). WCE were prepared from MIN6-4N cells untransfected or transfected with mammalian expression constructs in NETN buffer containing 0.2% bovine serum albumin and a protease inhibitor cocktail. WCE were precleared with glutathione sepharose beads for 30 minutes at 4°C. Equal amounts of GST or GST-fused proteins coupled to glutathione sepharose beads were incubated at 4°C overnight with the precleared WCE. The beads were washed five times with NETN buffer containing 0.1% bovine serum albumin. The bound proteins were detected by western blot with appropriate antibodies.

Immuno-detection of proteins in insulinomas

Formalin-fixed paraffin-embedded (FFPE) sections of human insulinoma were obtained after informed consent under an NIH institutional review board–approved protocol (NCT01005654) (23). Mouse work was conducted under an NIH/National Institute of Diabetes and Digestive and Kidney Diseases–approved animal study protocol. Men1+/− mice (FVB;129S-Men1tm1.1Ctre; also known as Men1ΔN3-8) were purchased from Jackson Laboratories (10). These mice develop insulin-secreting pancreatic islet tumors at age >15 to 18 months after the loss of the nontargeted Men1 allele resulting in complete loss of menin expression. Islet tumors are easily visible on the pancreas in 19-month-old mice. Such tumors were excised and processed for FFPE sections (Histoserv). FFPE tissue sections of human insulinoma and pancreatic islet tumors of Men1 mice were processed using immunohistochemistry (IHC) with anti-insulin, and dual-immunofluorescence with anti-HB9-PO4 and anti-NONO (24).

Statistical analysis

Averages and standard errors of the mean were plotted from at least three independent experiments. Student t test was used to compute significance (P < 0.05 was considered significant).

Results

Characterization of clinically relevant HLXB9 mutations associated with neonatal diabetes

Two different homozygous germline mutations in HLXB9, p.F248L and p.F272L, have been reported in patients with neonatal diabetes (20, 21). The effect of the two mutations was determined on protein expression, cell proliferation, and subcellular localization. The HLXB9 protein (HB9) contains two phosphorylation sites in the N-terminal region at aa residues serine-78 and serine-80, and two nuclear localization signals (NLS1: aa 240 through 273; NLS2: aa 293 through 300) in the homeodomain region (Fig. 1A). The two mutations were located in NLS1. The expression of HLXB9 mutants was analyzed by western blot analysis of WCE prepared from MIN6-4N cells transfected with myc-his-tagged HB9-WT, HB9-F248L, or HB9-F272L. On a western blot with anti-Myc-tag, normal HLXB9 (HB9-WT) and HB9-F272L were detected as double bands near 64 kDa (Fig. 1B). On a western blot with anti-HB9-PO4, normal HLXB9 (HB9-WT) and HB9-F272L were detected as a single band near 64 kDa, which corresponded to the top band of the doublet seen in the anti-Myc-tag western blot, indicating that the top band was for phosphorylated HLXB9 and the bottom band was for unphosphorylated HLXB9 (Fig. 1B). Interestingly, on western blots with anti-Myc-tag and anti-HB9-PO4, HB9-F248L lacked the top band of the double band near 64 kDa (Fig. 1B). These results indicate that the HB9-F248L mutant was defective in phosphorylation. Cell proliferation assay and the detection of cleaved caspase-3 to assess apoptosis showed that similar to HB9-WT both mutants were capable of reducing cell proliferation and inducing apoptosis (Fig. 1C and 1D). Therefore, lack of phosphorylation did not affect the ability of HB9-F272L to reduce cell proliferation or induce apoptosis of MIN6-4N cells. To determine the localization of the two HLXB9 mutants in vivo, transfected MIN6-4N cells were analyzed by dual IF with anti-Myc-tag and anti-HB9-PO4. HB9-WT and HB9-F272L showed nuclear staining with anti-Myc-tag and anti-HB9-PO4. However, HB9-F248L showed nuclear staining with anti-Myc-tag but not with anti-PO4-HB9 (Fig. 1E). These results further confirm the observation from western blot analysis that the HLXB9 mutation p.F248L blocks phosphorylation in the mutant protein.

Figure 1.

Characterization of HLXB9 mutations associated with neonatal diabetes. (A) Schematic diagram showing HLXB9 protein domains, two phosphorylation sites in the N-terminal region at serine-78 and serine-80, two NLSs (NLS1 and NLS2), and the two mutations in NLS1 associated with neonatal diabetes. HLXB9 is abbreviated as HB9 in all figures. (B) Western blot of WCE from MIN6-4N cells transfected with vector alone or with plasmids expressing myc-his-tagged HB9-WT, HB9-F248L, or HB9-F272L. WCE were run side by side on the same protein gel, and the blot was cut at the dotted line shown to probe with anti–Myc-tag to detect transfected HLXB9 and anti-PO4-HB9 to detect phosphorylated HLXB9. The blots were reprobed using anti-p84 (loading control). The top band of the doublet detected near 64 kDa in the lanes marked HB9-WT and HB9-F272L corresponds to phospho-HLXB9 (HB9-PO4), which is also detected with anti–HB9-PO4; bottom band corresponds to unphosphorylated HLXB9 (HB9-unPO4), which it is not detected with anti-HB9-PO4. (C) MIN6-4N cells transfected in panel (B) were analyzed by MTT assay to measure differences in cell proliferation, 48 and 96 hours posttransfection. *Significantly reduced cell proliferation (P < 0.05). (D) Western blot of WCE prepared from cells transfected in panel (B) 96 hours posttransfection to detect apoptosis by the presence of cleaved caspase-3. Anti-p84 was used as the loading control. (E) Cells transfected in panel (B) were cultured in chamber slides and analyzed by dual IR using anti–Myc-tag and anti-HB9-PO4. DAPI was used as a nuclear stain. Original magnification, ×1000. C-Term, C-terminal region; DAPI, 4′,6-diamidino-2-phenylindole; N-Term, N-terminal region.

Characterization of clinically relevant NONO mutations associated with endocrine tumors

Two somatic heterozygous NONO mutations, p.H146R and p.R293H, were identified via whole-exome sequencing in a parathyroid adenoma and an SI-NET, respectively (28, 29). The effect of the two mutations was determined on protein expression, cell proliferation, and subcellular localization. The NONO protein contains a Drosophila behavior/human splicing (DBHS) domain that is made up of two RNA recognition motifs [RRMs (RRM1 and RRM2)] followed by a NONA/paraspeckle (NOPS) domain and a coiled-coil domain (Fig. 2A). NONO-H146R is located between the two RRMs, and NONO-R293H is located in the coiled-coil domain. Western blot analysis of WCEs prepared from MIN6-4N cells expressing Flag-tagged WT or mutant NONO showed that the band for NONO-R293H was detected slightly higher than NONO-WT and NONO-H146R. Perturbed mobility of NONO-R293H in the SDS-PAGE gel could result from possible structural impairment of the coiled-coil domain (Fig. 2B). Cell proliferation was unaffected upon transfection of NONO-WT or the two mutants in MIN6-4N cells (Fig. 2C). IF analysis to detect transfected WT or mutant Flag-tagged NONO with anti-Flag-tag showed similar nuclear staining for NONO-WT and NONO-H146R in randomly distributed nuclear paraspeckles; however, nuclear staining of NONO-R293H was localized in brightly stained foci (Fig. 2D). These data indicate that the NONO mutation p.R293H from the SI-NET imparts structural aberrations to the mutant protein, which leads to subnuclear mislocalization.

Figure 2.

Characterization of endocrine disease–associated NONO mutations. (A) Schematic diagram showing NONO protein domains consisting of a DBHS domain that is made up of two RNA recognition motifs (RRM1 and RRM2) followed by a NOPS domain and a coiled-coil domain. Also shown is the location of the two mutations found in a parathyroid adenoma (p.H146R) and in a SI-NET (p.R293H). (B) Western blot of WCE of MIN6-4N cells transfected with vector alone or with plasmids expressing Flag-tagged NONO-WT, NONO-H146R, or NONO-R293H. Anti-Flag-tag was used to detect transfected Flag-NONO proteins. The blot was reprobed using anti-p84 (loading control). (C) Cells transfected in panel (B) were analyzed by MTT assay to assess cell proliferation, 48 and 96 hours posttransfection. (D) Cells transfected in panel (B) were cultured in chamber slides and analyzed by IF using anti-Flag-tag. DAPI was used as a nuclear stain. Original magnification, ×1000. DAPI, 4′,6-diamidino-2-phenylindole.

Interaction of HLXB9 and NONO mutants

The ability of the two HLXB9 and NONO mutants to interact with normal NONO and normal HLXB9 protein, respectively, was determined by GST pull-down assays. We have previously shown that cotransfection of plasmids expressing HLXB9 and NONO reduced the level of transfected NONO protein (24). This effect was also detected with the HLXB9 and NONO mutants (Supplemental Fig. 1A and 1B); therefore, interaction assays were not conducted by coimmunoprecipitation of cotransfected NONO and HLXB9 (WT or mutant). For GST pull-down assays, WCE of MIN6-4N cells overexpressing WT or mutant proteins were analyzed for interaction with GST-fusion proteins, GST-HLXB9 and GST-NONO, using GST-CDK2 as the negative control (Fig. 3A). Similar interaction was observed for WT and mutant HLXB9 with GST-NONO, and for WT and mutant NONO with GST-HLXB9 (Fig. 3B and Fig. 3C). These results indicate that the mutations did not disrupt HLXB9-NONO interaction.

Figure 3.

WT and mutant HLXB9 and NONO interaction. (A) Coomassie blue–stained gel (4% to 20%) showing the purity of the indicated GST-fusion proteins used in the GST pull-down assays. *GST or GST-fusion protein of the expected size. (B) WCE of MIN6-4N cells expressing myc-his-tagged HB9-WT, HB9-F248L, or HB9-F272L were incubated with GST-NONO-WT beads or negative control GST-CDK2 beads. Bound HLXB9 was detected by western blot with anti-Myc-tag. Inputs (1/10th amount of WCE used for the GST pull-down assays) were also analyzed on the same protein gel (4% to 20%). (C) WCE of MIN6-4N cells expressing Flag-tagged NONO-WT, NONO-H146R, or NONO-R293H were incubated with GST-HB9 beads or negative control GST-CDK2 beads. Bound NONO was detected by western blot with anti-Flag-tag. Inputs (1/10th amount of WCE used for the GST pull-down assays) were also analyzed on the same protein gel (4% to 20%).

HLXB9 and NONO protein in human and mouse pancreatic islet tumors

It has been shown that NONO is ubiquitously expressed, whereas HLXB9 expression is β-cell–specific in the pancreas, with higher level of phospho-HLXB9 in insulinomas (15, 23, 25). Germline Men1 heterozygous mice (Men1+/−) develop insulin-secreting pancreatic islet β-cell tumors (insulinomas) after the loss of the nontargeted allele (age >15 months) (10). We assessed the protein level of both phospho-HLXB9 and NONO in human insulinoma and Men1+/− mouse insulinoma. Both human and mouse insulinoma showed strong staining for insulin, phospho-HLXB9, and NONO (Fig. 4). These data show that both phospho-HLXB9 and NONO protein are abundantly expressed in insulinoma. Further analysis of the interaction between HLXB9 and NONO would be useful to determine the functional significance of their interaction in β-cells.

Figure 4.

HLXB9 and NONO in insulinomas. Representative images of IHC for insulin and dual IF for phosphorylated HLXB9 (HB9-PO4) and NONO performed on FFPE sections of a human sporadic insulinoma (left) and an islet tumor from a 19-month-old Men1+/− mouse (right). Hematoxylin or DAPI (blue color) was used as the counterstain for nuclei for IHC and IF, respectively. Original magnification, ×400. DAPI, 4′,6-diamidino-2-phenylindole.

Deletion analysis of HLXB9 and NONO to map interacting regions

The domain structure of HLXB9 protein consists of an N-terminal region (aa 1 through 160), a conserved region (aa 160 through 240), a homeodomain (aa 241 through 302) with two nuclear localization signals NLS1 and NLS2, and a C-terminal region (aa 303 through 404) (Fig. 5A). The domain structure of NONO protein consists of two RRMs at aa 74 through 141 and 148 through 229, a NOPS region (aa 230 through 267), and a coiled-coil region (aa 268 through 372) followed by the C-terminal region (Fig. 5B). We determined the regions of NONO and HLXB9 required for interaction by GST pull-down assays using GST-fusions of full-length or two halves of each protein (Fig. 5C and5D) and WCE of MIN6-4N cells expressing Flag-tagged NONO or myc-his-tagged HLXB9 proteins. Full-length NONO showed a strong interaction with full-length GST-HLXB9 but a weaker interaction with the N-terminal region of HLXB9 (Fig. 5E). In reciprocal interaction assays using GST-NONO-fusion proteins and HLXB9 expressed in MIN6-4N cells, HLXB9 interacted with full-length NONO (GST-NONO) but failed to interact with the two halves of NONO (Fig. 5F). Collectively, these data show that an intact domain structure of full-length NONO is required for interacting with HLXB9 at aa 1 through 240 in the N-terminal region.

Figure 5.

Mapping the regions of HLXB9 and NONO interaction. (A) Schematic diagram of full-length HLXB9 protein (aa 1 through 404), its domains, and the two regions (aa 1 through 240 and 241 through 404) used for GST-fusion protein constructs. (B) Schematic diagram of full-length NONO protein (aa 1 through 471), its domains, and the two regions (aa 1 through 236 and 227 through 471) used for Flag-tagged expression constructs or GST-fusion protein constructs. (C and D) Coomassie blue–stained gel (4% to 20%) showing the purity of the indicated GST-fusion proteins used in the GST pull-down assays. *GST or GST-fusion protein of the expected size. (E) WCE of MIN6-4N cells transfected with the indicated Flag-tagged NONO constructs were incubated with the indicated GST-HB9 beads. Bound NONO was detected by western blot with anti-Flag-tag. Inputs (1/10th amount of WCE used for the GST pull-down assays) were also analyzed by anti-Flag western blot on another protein gel. (F) WCE of MIN6-4N cells expressing myc-his-tagged HB9-WT was incubated with the indicated GST-NONO-fusion proteins on beads. Bound HLXB9 was detected by western blot with anti-Myc-tag. Input HLXB9 (1/10th amount of WCE used for the GST pull-down assays) was also analyzed on the same protein gel. CC, coiled-coil region; C-Term, C-terminal region; consv, conserved region; DAPI, 4′,6-diamidino-2-phenylindole; H.D., homeo-domain; N-Term, N-terminal region.

HLXB9-conserved region interacts with NONO

To determine the minimal region of HLXB9 required for interaction with NONO, plasmids containing various regions of HLXB9 were transfected in MIN6-4N cells (Fig. 6A). Western blot analysis revealed that two HLXB9 regions (aa 1 through 240 and 1 through 302) either showed a band of very low intensity or a band was not detected. It is possible that these portions of HLXB9 protein may be insoluble in the WCE buffer (Fig. 6B). GST pull-down assays showed that aa 161 through 404 was the only region that interacted with GST-NONO (Fig. 6C). Further deletion analysis of this region showed that the aa 241 through 404 region did not interact with GST-NONO (Fig. 6C). These two observations indicate that the aa 160 through 240 region of HLXB9 may contain the NONO interacting region. Interestingly, IF analysis of MIN6-4N cells transfected with the various HLXB9 constructs showed that only aa 161 through 404 localized to the nucleus (Supplemental Fig. 2A and 2B). The differential localization in the nucleus or cytoplasm was due to the presence or absence of the two nuclear localization signals in the HLXB9 constructs analyzed. However, the reason for the lack of nuclear staining for the aa 241 through 404 region that contains both nuclear localization signals is not known. Overall, the deletion analysis showed that the NONO interacting region in HLXB9 was located at aa 160 through 240; this region or parts of this region may serve as interaction competitors for assessing the importance of the HLXB9-NONO interaction in vivo.

Figure 6.

HLXB9 deletion mapping for NONO interacting region. (A) Schematic diagram of full-length HLXB9 protein (aa 1 through 404), its domains, and myc-his-tagged expression constructs containing the indicated amino acids. (B) Western blot of WCE from MIN6-4N cells transfected with myc-his-tagged full-length HLXB9 (1 through 404) or the indicated myc-his-tagged HLXB9 regions. Anti-Myc-tag was used to detect the transfected HLXB9 proteins. p84 was used as the loading control. (C) GST pull-down assay using GST-NONO beads and WCE from MIN6-4N cells transfected with the indicated myc-his-tagged HLXB9 regions (right). Note that for the HLXB9 aa 1 through 240 construct, a higher amount of WCE was used in the assay because of its low expression; the HLXB9 aa 1 through 302 construct was not used in the assay because it did not express well. Bound HLXB9 proteins were detected by western blot with anti-Myc-tag. Input HLXB9 proteins (1/10th amount of WCE used for the GST pull-down assays) were also analyzed by western blot with anti-Myc-tag on another protein gel (left). C-Term, C-terminal region; Consv, conserved region; DAPI, 4′,6-diamidino-2-phenylindole; H.D., homeo-domain; N-Term, N-terminal region.

Disruption of the HLXB9-NONO interaction and its biological consequence

The 80-aa region of HLXB9 from aa 160 through 240 has been designated as the “conserved region” because of high evolutionary conservation (Fig. 7A) (34). The functional contribution of this conserved region is not known. Our results from GST pull-down assays showed that this region of HLXB9 is important for protein–protein interaction; therefore, we intended to determine the ability of the aa 160 through 240 region to disrupt the endogenous HLXB9 interaction with NONO in vivo. This would help to examine the biological role of the HLXB9-NONO interaction. For the purpose of using the conserved region as a disruptor of the HLXB9-NONO interaction, the aa 160 through 240 region would have to localize to the nucleus. Because the aa 160 through 240 region lacks the HLXB9 nuclear localization signals NLS1 and NLS2 (aa 240 through 273 and aa 293 through 300), we made a plasmid construct of this conserved region so that it included NLS1 (aa 160 through 273) (Supplemental Figure 3A). However, expression of the aa 160 through 273 region was in the cytoplasm and not in the nucleus (Supplemental Figure 3B). Therefore, the aa 160 through 240 region and also its two halves were cloned in a vector that contained 3XNLS. WCE of MIN6-4N cells transfected with these plasmid constructs showed abundant expression of the conserved region and its two halves on western blot analysis (Fig. 7B). GST pull-down assays conducted with the WCE showed that aa *160 through 240 interacted with GST-NONO; breaking this region into two halves abrogated interaction with GST-NONO (*indicates 3XNLS) (Fig. 7B). Interestingly, even though all three regions (160 through 240, 160 through 203, and 195 through 240) were cloned in the same vector with 3XNLS, aa *160 through 240 localized to the nucleus, but the two halves were not localized to the nucleus (Fig. 7C).

Figure 7.

Interaction of HLXB9 conserved region with NONO and the consequence of disrupting the HLXB9–NONO interaction. (A) Schematic of HLXB9 protein showing the conserved region (aa 160 through 240) and its subparts used for cloning into pCMV/myc/nuc vector that contains a C-terminal myc-tag and 3XNLS. *Presence of the heterologous 3XNLS in the constructs. (B) WCE of MIN6-4N cells transfected with the indicated myc-tagged HLXB9 constructs were incubated with GST-NONO beads. Bound HLXB9 proteins were detected by western blot with anti-Myc-tag (right). Inputs (1/10th amount of WCE used for the GST pull-down assays) were also analyzed by western blot with anti-Myc-tag on another protein gel (left). (C) IF with anti-Myc-tag of MIN6-4N cells transfected with pCMV/myc/nuc vector containing the indicated HLXB9 regions. DAPI was used as a nuclear stain. Original magnification, ×1000. (D) Western blot of WCE of MIN6-4N cells cotransfected with vector or plasmids expressing myc-his-tagged HB9-WT alone or together with increasing amounts of myc-tagged HB9-(*160 through 240). Total DNA amount was maintained constant with vector DNA. HLXB9 proteins were detected by western blot with anti-Myc-tag, and p84 was used as the loading control. (E) GST pull-down assay using GST-NONO with WCE of MIN6-4N cells expressing myc-his-tagged full-length HLXB9 (HB9-WT) or myc-tagged HB9-(*160 through 240). The indicated ratios of HB9-WT WCE to HB9-(*160 through 240) WCE were incubated with GST-NONO beads. Input HLXB9 proteins (WCE used in the assay) (left) and GST-NONO bound HLXB9 proteins (right) were detected by western blot with anti-Myc-tag. Increasing amount of HB9-(*160 through 240) reduced the interaction of full-length HLXB9 (upper) with GST-NONO. (F) MIN6-4N cells transfected with the indicated plasmids were analyzed for protein expression by western blot with anti-Myc-tag. p84 was used as the loading control. (G and H) Transfected MIN6-4N cells analyzed in panel (F) were assessed by MTT assay to detect differences in cell proliferation 48 and 96 hours posttransfection. *Significantly reduced cell proliferation (P < 0.05). C-Term, C-terminal region; Consv, conserved region; DAPI, 4′,6-diamidino-2-phenylindole; H.D., homeo-domain; N-Term, N-terminal region.

These results showed that the nuclear localized conserved region of HLXB9 (aa 160 through 240) was the minimal region important for interaction with NONO; therefore, this region could be used in vivo to determine whether it could displace endogenous HLXB9 from NONO. We first conducted cotransfection of MIN6-4N cells to express HB9-WT with increasing amounts of HB9-(*160 through 240). Western blot analysis showed that increasing the amount of HB9-(*160 through 240) lowered the level of transfected HB9-WT protein (Fig. 7D); therefore, we could not use cotransfected HB9-WT and HB9-(*160 through 240) to determine whether HB9-(*160 through 240) could interfere with HLXB9-NONO interaction. Instead, interaction studies with GST-NONO were performed by mixing various ratios of WCE prepared from MIN6-4N cells that were transfected separately to express HB9-WT or HB9-(*160 through 240). Interaction of full-length HLXB9 with GST-NONO was significantly reduced by including HB9-(*160 through 240) in the GST pull-down assay (Fig. 7E). These data indicate that the aa 160 through 240 region of HLXB9 could block the interaction of HLXB9 with NONO as detected in in vitro assays. Next, we studied the consequence of disrupting the endogenous interaction between HLXB9 and NONO in vivo. In MIN6-4N cells, transfection of full-length HLXB9 reduced cell proliferation; however, transfection of HB9-(*160 through 240) did not reduce cell proliferation (Fig. 7F–7H). Therefore, disruption of the endogenous HLXB9-NONO interaction did not affect cell proliferation.

Discussion

Genetically engineered mice with null mutations have revealed the physiologic roles of several transcription factors that govern β-cell development and differentiation (19). These transcription factors regulate biological processes that control cell proliferation and hormone secretion to promote β-cell differentiation from progenitor cells and to maintain β-cell fate. Mutation or misregulation of the same transcription factors predict defects in β-cell proliferation and function leading to disease states of β-cell loss (diabetes) or β-cell excess (insulinoma).

Neonatal diabetes mellitus is a rare form of nonautoimmune diabetes occurring in transient or permanent states. In PNDM, symptoms of diabetes usually occur in infancy. Mutations in several β-cell differentiation factors (GATA4, GATA6, GLIS3, HNF1B, NEUROD1, NEUROG3, NKX2-2, PDX1, PTF1A, RFX6, and HLXB9/MNX1) have been reported in PNDM (19, 35). Although mutations in β-cell differentiation factors have not been reported in sequencing efforts of human insulinomas (5–7, 9), misregulation of mRNA or protein expression or phosphorylation have been observed in human or mouse insulinomas for a few β-cell differentiation factors (MafA, MafB, and Hlxb9/Mnx1) (14, 23, 36, 37). Because we previously observed misregulated increase of HLXB9 in insulinomas (14, 23), the subsequent reports of HLXB9 germline homozygous mutations in two different probands with PNDM (20, 21) prompted us to further examine the biological consequence of the mutations and the mechanistic role of HLXB9 in controlling β-cell proliferation. The proband with p.F248L mutation was reported deceased at age 10 months (age at diagnosis, 30 weeks); the proband with p.F272L mutation was 3 years of age in 2015 when the paper was published (age at diagnosis, 1 week) (21). We do not know any follow-up details for these patients. The two HLXB9 mutations were located in NLS1, which surprisingly did not affect nuclear localization. We found that the HLXB9/p.F148L mutant protein lacked phosphorylation in the N-terminal region. This phosphorylation of HLXB9 was present at a high level in insulinoma samples (23); therefore, this opposite effect on HLXB9 phosphorylation in diabetes vs insulinoma suggests functional consequence in β cells.

HLXB9 interacts with NONO (24), and two different somatic heterozygous mutations in NONO have been observed in a parathyroid adenoma (p.H146R) and a SI-NET (p.R293H) (28, 29). Because the functional significance of the HLXB9-NONO interaction is not known, further analysis of these NONO mutations and the critical interacting regions of HLXB9 and NONO was undertaken to gain insight into the importance of NONO in endocrine cells, including β cells. The p.R293H mutation, located in the C-terminal coiled-coil region of NONO, affected the structural integrity of NONO as indicated by the aberrant migration of the mutant protein on SDS-PAGE and the altered subnuclear location of the mutant protein as brightly stained foci compared with normal NONO. Similar aberrations in subnuclear localization were observed from truncation and mutagenesis analysis of the coiled-coil region of NONO (27). Therefore, the same abnormality from a naturally occurring mutation in a SI-NET proposes that proper subnuclear localization and function of NONO must be relevant in neuroendocrine cells.

HLXB9 has been shown to play a role in cell proliferation and apoptosis in Drosophila neuronal cells and in MIN6 insulinoma cells (14, 38). To determine whether the interaction of HLXB9 with NONO might regulate cell proliferation, we first mapped the minimal interacting region of HLXB9 that could displace HLXB9 from NONO in in vitro assays. However, when this HLXB9 minimal region (the aa 160 through 240 conserved region) was used as a NONO interacting competitor in vivo, cell proliferation was not affected. It is possible that the endogenous HLXB9-NONO interaction was not disrupted because of insufficient subnuclear colocalization of the competitor (HB9 aa 160 through 240) with NONO because the nuclear localization of the HLXB9 aa 160 through 240 region was supported by heterologous nuclear localization signals and not from the HLXB9 NLS. It is also possible that this interaction may not be associated with cell proliferation. Given that both HLXB9 and NONO are expressed abundantly in insulinoma, further work will be of interest to find out the functional role of their interaction in β cells.

In conclusion, the absence of HLXB9 phosphorylation from a mutation associated with β-cell loss (diabetes) and the abundant level of phospho-HLXB9 in a condition of excessive β-cell proliferation (insulinoma) suggests further exploration of the molecular regulation of β cells through this β-cell differentiation factor and its interacting partners. Future studies are warranted that include other β-cell differentiation factors known to be mutated in patients with diabetes and the potential misregulation of the same factors in insulinomas. Such studies will impact the development of therapeutic options for the treatment of β-cell tumors and β-cell mass expansion strategies for the treatment of diabetes.