MIN6 murine pancreatic $\beta$-cell line were cultured in Dulbecco’s modified Eagle’s medium (DMEM) containing 11 mM glucose, 10% fetal bovine serum (FBS), 2 mM L-glutamine, and 5 µM $\beta$-mercaptoethanol. 100 U/mL penicillin and 100 µg/mL streptomycin were added to non-transfected MIN6 cells, while 400 µg/mL geneticin (G418, InvivoGen, Toulouse, France) was supplemented to stably transfected MIN6 cells. Cells were grown at 37 °C in a 5% CO₂ humidified atmosphere. MIN6 cells were used within 20 passages and were routinely tested for mycoplasma contamination. No additional authentication was performed, as the identity of this $\beta$-cell line was confirmed by its characteristic morphology and glucose-stimulated insulin secretion response.

pEGFP-c1 plasmid (Addgene, https://www.addgene.org/) harboring G418 resistance was used for creating stable MIN6 cell lines that overexpress Otub2 (mouse isoform 2) with a Flag tag (instead of EGFP). The pEGFP-c1 vector, in which the Flag tag was cloned instead of the EGFP sequence (pFlag), served as a control. pEGFP-c1 plasmids which contained the original EGFP sequence (p-EGFP-c1) and EGFP-Otub2 sequence (pEGFP-Otub2) were used for transient transfection of MIN6 cells. To create these constructs, the Otub2 cDNA was amplified by PCR using the 5^′ primer (5^′-CAGTCCGGAGACACTATGAGTGAAACATCTTTCAACC-3^′); and the 3^′ primer (5^′-CGCGGATCCGCGGTAGTCAGTGTTTCTCGGCTGC-3^′). The PCR product was digested using BspE1 and BamHI restriction enzymes and ligated into the pEGFP-c1 plasmid. An additional construct was generated using the following primers containing Flag tag: 5^′ primer (5^′-CTAGCATGGACTACAAAGACGATGACGACAAGT-3^′) and 3^′ primer (5^′- CCGGACTTGTCGTCATCGTCTTTGTAGTCCATG-3^′). The primers were hybridized and ligated into pEGFP-c1 digested with NheI and BspEI restriction enzymes. The EGFP cDNA was then excised from the plasmid and replaced by a Flag tag, creating the pFlag-Otub2 construct. pFlag plasmid was generated after digestion of pFlag-Otub2 construct with BspE1 and BamHI restriction enzymes (excising Otub2 out of the construct) and used as a control.

Polyclonal anti-Flag and anti-GAPDH antibodies were from Sigma (St. Louis, MO, USA). Polyclonal anti-Otub2 antibodies were from Novus Biologicals (Centennial, CO, USA). Polyclonal anti-GFP antibodies were from Abcam (Waltham, MA, USA). Alexa488 conjugated goat anti-mouse secondary antibodies were from Life Technologies (Waltham, MA, USA).

The cytokine mixture referred to as ‘1x-cytomix’ consisted of 3 nM TNF-$\alpha$ (Prospec-Tany Technogene, Rehovot, Israel), 3 nM IFN-$\gamma$ (Prospec-Tany Technogene, Rehovot, Israel), and 1.5 nM IL-1$\beta$ (MD Biosciences, Ness Ziona, Israel). Their biological activities were 10 units/ng (TNF-$\alpha$, IFN-$\gamma$) and 200 units/ng (IL-1$\beta$).

Isolated human islets (~80% purity confirmed by dithizone staining) were provided by the European Consortium for Islets Transplantation (Islet for Basic Research program) through a Juvenile Diabetes Research Foundation Award 31-2008-413. Isolated human islets were grown in CMRL 1066 medium supplemented with 10% (v/v) FBS, 2 mM L-glutamine, 100 units/mL penicillin, 100 µg/mL streptomycin, 0.25 µg/mL amphotericin, and 40 µg/mL gentamycin, and were cultured at 37 °C in a 5% CO₂ humidified atmosphere. Islets media was replaced every other day. Intact human islets were dispersed by incubation at 37 °C (4 minutes) with 1 mg/mL Trypsin/EDTA by pipetting the cells, and by passing them twice through a 21G needle. Trypsinized islets were washed and resuspended with CMRL 1066 medium supplemented with 10% FBS. Cells were used within 48 hours following dispersion. Human islets studies received Ethics approval from the Bioethics and Embryonic Stem Cell Research Oversight Committee (ESCRO) of the Weizmann Institute of Science.

MIN6 cells or human islets were seeded in 96 well plates (30,000 cells/well or 1000 dispersed human islets cells/well) in 100 µL medium (DMEM for MIN6 cells and CMRL for human islets), and immediately transfected with sequences of non-targeting siRNA or Otub2 siRNA (to a final concentration of 25 nM) using DarmaFECT-4 transfection reagent (Dharmacon, Waterbeach, UK) for MIN6 cells and Darma-FECT-1 transfection reagent (Dharmacon) for human islets, according to manufacturer’s instructions.

MIN6 cells or human islets were seeded (500,000 cells/well in 6-well plates for MIN6 cells, and 1000 islets/well in 96-well plates for human islets) and 24 hours post-seeding, the cells or human islets were transfected with pFlag or pFlag-Otub2 vectors (3 µg/well for a 6-well plate and 0.2 µg/well for a 96-well plate) using the transfection reagent Lipofectamine 2000 (Thermo-Fisher Scientific, Waltham, MA, USA), according to the manufacturer’s instructions.

MIN6 cells were seeded in 12-well plates and 24 hours post-seeding, cells were transfected using jetPEI transfection reagent (Polyplus Transfection, Illkirch, France), with pEGFP-c1 vectors (1 µg/well) harboring G418 resistance to calibrate for the cells’ G418 resistance. pEGFP-c1 vector in which Flag tag was cloned instead of EGFP sequence (pFlag) served as control; and pEGFP-c1 with Flag-Otub2 cDNA clone without EGFP sequence (pFlag-Otub2) was utilized to overexpress Otub2. Non-transfected cells served as a control and 48 hours after transfection media were replaced with media containing elevated concentrations of G418. After the calibration process was complete and the proper concentration of G418 required for selection was established, MIN6 cells were seeded in 6-well plates (150,000 cells/well) to create stable cell lines, 24 hours post-seeding, cells were transfected with pEGFP-c1 constructs (3 µg/well), and 48 hours thereafter, the media were replaced with media containing 400 µg/mL G418 for the selection process. After completion of selection, the media were replaced with media containing 200 µg/mL G418 for maintenance.

MIN6 cells were seeded in 96-well plates (30,000 cells/well) and transfected with siRNA and 48 hours later, cells were treated with a cytokine mixture, referred to as 1x-cytomix. For stably transfected cell lines, cells were treated with 1x-cytomix 24 hours post-seeding and 24 hours later caspase 3/7 activity was examined using Sensolyte homogeneous RH110 caspase 3/7 assay kit (AnaSpec, Fremont, CA, USA). Human islets were dispersed and seeded in 96-well plates (1000 islets/well). Islets were transfected with p-Flag or pFlag-Otub2 constructs that stably express Flag-Otub2. The rest of the experiment was similar to the assay in MIN6 cells as described above.

NF-$\kappa$B activity was measured using the Ready-To-Glow secreted luciferase assay kit (Clontech, San Jose, CA, USA). Stably transfected MIN6 cells were seeded in 96-well plates (30,000 cells/well) and 24 hours post-seeding, cells were transfected with a secreted luciferase plasmid (250 ng/well), coupled to the NF-$\kappa$B enhancer elements. Constitutively secreted luciferase plasmid, transfected into MIN6 cells, served as a control. After 24 hours, cells were treated with 1x-cytomix for 4 hours to induce NF-$\kappa$B activity and 50 µL of the media of the cells was harvested and transferred to black 96-well plates containing a luciferase assay mix (5 µL/well). Luminescence was read using an Infinite 2000 PRO reader (Tecan, Männedorf, Switzerland).

Stably transfected MIN6 cells were seeded in 96-well plates (30,000 cells/well). After 48 hours, cells were incubated for 1 hour in Krebs-Ringer bicarbonate HEPES buffer (KRBH), containing 124 mM NaCl, 5.6 mM KCl, 2.5 mM CaCl₂, and 20 mM HEPES, pH 7.4, at 37 °C, followed by incubation for 1 hour in KRBH with 0 mM or 20 mM glucose. Insulin concentration in the culture medium was determined using insulin detecting HTRF kit (Cisbio, Codolet, France) according to the manufacturer’s instructions.

Following treatment in 24-well plates, cells were harvested, and total RNA was extracted using the PerfectPure RNA kit (5-prime, Montreal, Canada). RNA was quantified using nano-drop. cDNA was generated by the cDNA Reverse Transcription kit (Applied Biosystems, Waltham, MA, USA), following the manufacturer’s instructions. qRT-PCR was carried out using an ABI-Prism 7300 instrument (Applied Biosystems, Waltham, MA, USA), utilizing SYBR Green PCR mix (Invitrogen, Carlsbad, CA, USA) and specific primers (100 nM final concentration). Expression levels of Actin or Hypoxanthine Phosphoribosyltransferase 1 (HPRT) were used to normalize mRNA levels. Primers used are given herein (see Table 1).

RNA was extracted from the pancreas and liver of mice using the PerfectPure RNA kit (5-prime, Montreal, Canada) following the manufacturer’s instructions. An Agilent 4200 TapeStation System (Agilent Technologies, Santa Clara, CA, USA) was used to assess RNA quality. RNA-seq libraries were generated by the MARS-seq protocol [16]. Libraries were sequenced by the Novaseq 6000 (Illumina, San Diego, CA, USA), using SP mode 100 cycles kit (Illumina, San Diego, CA, USA). Alignment of sequences to the genome and count matrix determination were performed by the UTAP pipeline (Weizmann Institute, Rehovot, Israel). Normalization of libraries, low count genes filtration, and calculation of differentially expressed genes were performed using the edgeR and Limma packages in R (https://www.r-project.org/). Gene Ontology (GO) and MsigDB pathways enrichment analysis were performed using the Camera method from the Limma package in R. The enrichment results are given in p-values, with –log10(p-value) $>$1.3 considered to be significant. A functional protein association network was built using STRING (https://string-db.org/).

Protein A-agarose beads were washed with 0.1 M Tris-HCl (pH 8.0, 4 °C), followed by incubation with Flag or GFP antibodies in 0.1 M Tris-HCl (pH 8.0) for 4 h at 4 °C. Supernatants (centrifuged at 20,000 $\times$g 15 min 4 °C) of cell extracts in extraction buffer (~0.8–1.2 mg of protein), were incubated at 4 °C overnight with the immobilized antibodies. Immunocomplexes were washed three times with extraction buffer and then incubated for 30 min at 4 °C with PBS containing Flag or GFP peptide in order to elute Otub2 and the proteins that formed a complex with it. Immunocomplexes were mixed with Laemmli sample buffer and resolved by 10% SDS-PAGE. The complexes were subjected to mass spectrometry analysis as previously described [17].

Cells were seeded and treated as indicated. Treated cells were washed three times with PBS and were harvested in lysis buffer (25 mM Tris-HCl [pH 7.4], 10 mM sodium orthovanadate, 10 mM pyrophosphate, 100 mM sodium fluoride, 10 mM EDTA, 10 mM EGTA, and 1 mM phenylmethylsulphonyl fluoride). Lysates were centrifuged (20,000 $\times$g, 15 min, 4 °C). Samples (40–150 µg) were mixed with 5$\times$ Laemmli buffer, boiled, and resolved by 10–12% SDS-PAGE. Proteins were transferred to a nitrocellulose membrane for Western blotting with the indicated antibodies: Polyclonal anti-Flag (Sigma, F7425, diluted 1:1000), Polyclonal anti-GAPDH (Sigma, ABS16, diluted 1:1000).

All in vivo experiments were carried out in accordance with relevant guidelines and regulations. Experiments were approved by the Animal Care and Use Committee of the Weizmann Institute of Science (application number 23980116-3). Mice were housed under standard light/dark conditions and were given access to food and water ad libitum. The study is reported in accordance with the ARRIVE guidelines.

The mouse strain C57BL/6NTac-Otub2^{tm1a(EUCOMM)Wtsi}/WtsiH was ordered from the Wellcome Trust Sanger Institute, London, UK, as part of the EUCOMM Mutant Mouse Project. One heterozygous male and two heterozygous female mice on a C57BL/6NTac background were ordered from the European Mouse Mutant Archive (EMMA) mouse repository. EUCOMM vector contains 5^′ and 3^′ homology arms, facilitating homologous recombination, and a targeting cassette that disrupts gene function, flanked by flipase recognition target (FRT) recombination sites to allow removal by Flp recombinase. Further, exons 3/4 of Otub2 were flanked by a pair of loxP recombination sites, inducing a frame shift upon removal of the exons and leading to complete gene inactivation. For genotyping, mouse tail genomic DNA preparations were extracted using REDExract-NAmp tissue PCR kit (Sigma), followed by amplification reactions performed with oligonucleotide pairs (see Table 2) specific for the foreign or wild-type sequences, to amplify ~200–900 bp fragments.

Otub2 knock-out and wild-type (WT) mice were sacrificed by cervical dislocation. An incision was made from the upper abdomen downwards to expose the liver and intestines. The duodenum was exposed by shifting the intestines to the right, and the pancreas was isolated by pulling the intestines carefully from the duodenum downwards. Spleen was then removed, and the pancreas was detached from the large intestine to complete its isolation.

Adeno-associated viral vectors (AAVs) that can stably express Flag or Flag-Otub2 were generated by inserting the Flag or Flag-Otub2 inserts into the pAAV-RIP-VEFG-WPRE vector, after eliminating the VEGF coding sequence, as described [18, 19]. The pAAV-RIP-Flag (for control purposes) and pAAV-RIP-Flag-Otub2 vectors were used for production of the AAV vectors. To ensure selective expression of Otub2 in pancreatic $\beta$-cells, the rat insulin promoter (RIP) was used to drive expression of Otub2. WPRE (WHP Posttranscriptional Regulatory Element) was used to enhance expression. Following vector generation, transgene expression was evaluated first in vitro in cultured cells. Then the DNA was amplified, and single-stranded AAV-Otub2 vectors were produced. Single-stranded AAV8 was used throughout. Administration of the AAV-8 viral vectors into the pancreatic duct of live mice was carried out as described [20]. For vector administration, mice were anesthetized with ketamine (100 mg/kg body weight, working concentration 25 mg/mL) and xylazine (10 mg/kg body weight, working concentration 2.5 mg/mL), administered intraperitoneally prior to surgery.

Mice were fasted overnight with water access ad libitum. Mice were intraperitoneally injected with glucose (1.8 g/kg body weight). Blood samples were taken at timed intervals (0–120 minutes) from a tail vein, and Glucose levels were monitored using MediSense Optium Blood Glucose test strips (Abbott Laboratories, Abbott Park, IL, USA). Glucose levels were plotted versus the time points, and the area under the curve (AUC) was calculated using the computeAUC function from the PharmacoGx package in R.

For immunofluorescence, pancreata were fixed for 24 h in 4% paraformaldehyde and then transferred to 70% ethanol until embedding in paraffin by a standard protocol (by using automated tissue processing) as follows: pancreata were first dehydrated by immersing them sequentially for 45 minutes each in a series of ethanol-water mixtures (70%; 95% ($\times$3) and 100% ($\times$2; 30 minutes)) followed by immersion with ethanol-xylene mixture (1:1; for 45 minutes) and xylene ($\times$2; 1 hour). The tissues were then embedded in paraffin at 60 °C (3 times for 1 hour). Cross sections of paraffin blocks were incubated in xylene ($\times$2) and ethanol (95–70%). Antigen retrieval was performed by boiling in 0.01 M sodium citrate (pH 6.0; 12.5 minutes in microwave). Sections were then blocked (with 0.5% Triton) and incubated overnight at 4 °C with anti-Otub2 antibodies (Novus Biologicals, OTI11B3, diluted 1:1000). Anti-Otub2 was detected with fluorescent-tagged secondary antibodies (conjugated to Alexa488 dye). Sections were later washed with PBS and DAPI for nucleus staining.

Statistical analysis was performed in R. Differences between experimental conditions were determined by a two-tailed Student’s t test, unless otherwise mentioned in the figure legends.

To characterize the effects of Otub2 on $\beta$-cells, a MIN6 cell line that stably overexpresses Flag-tagged Otub2 was generated. Over-expression of Otub2 was validated by qRT-PCR (for mRNA levels, Fig. 1A) and by Western blotting (for protein levels, Supplementary Fig. 1A,B). Fluorescence microscopy indicated that there are no morphological changes between control Flag-expressing cells and Flag-Otub2 expressing cells (Supplementary Fig. 1C).

Fig. 1.

Effects of OTUB2 overexpression on NF-$\kappa$B activity in MIN6 cells. (A) MIN6 cells were transfected with pEGFP-c1 plasmid constructs: pFlag and pFlag-Otub2 vectors. The stable cell line was subjected to qRT-PCR analysis following treatment with 1x-cytomix for 4 h. Non-treated cells served as a control. mRNA levels were normalized to Actin. (B) MIN6 cells overexpressing Flag (control) or Flag-OTUB2 were transfected with pNF-$\kappa$B-MetLuc2 or with pMetLuc2 (control). 24 h following transfection, cells were treated with or without 1x-cytomix for 4 h. NF-$\kappa$B activity was determined. Data presented were normalized to control vector transfection values. (C) MIN6 Flag or Flag-OTUB2 over-expressing cells were treated with 1x-cytomix for 4 h and then harvested. Total mRNA was extracted, and qRT-PCR was conducted. mRNA levels of NF-$\kappa$B target genes were normalized to Actin. Data are means $\pm$ SEM of four replicates in each of four experiments (B) and of duplicates in each of three experiments (C). Asterisks indicate the p-values as follows: *p 0.05, ***p 0.001, ****p 0.0001 vs. control. Flag stable cells.

To examine the effects of Otub2 on NF-$\kappa$B activity, a luciferase-reporting system was employed. Min6 cells were treated with ‘cytomix’, a mixture of cytokines (TNF-$\alpha$, IL-1$\beta$, IFN-$\gamma$) that significantly increases the expression of NF-$\kappa$B target genes [21] and promotes $\beta$-cell apoptosis [21, 22]. Indeed, cytomix treatment significantly increased the level of NF-$\kappa$B promoter activity by ~1.8-fold, while stable overexpression of Otub2 resulted in ~78% decrease in NF-$\kappa$B promoter activity, both in control and cytomix-treated cells (Fig. 1B). Overexpression of Otub2 in cytokine-treated cells also reduced by 60% and 80%, respectively, the mRNA levels of the NF-$\kappa$B target genes MCP1 and iNOS (Fig. 1C). These results complement our previous findings that Otub2 knock-down increased NF-$\kappa$B activation that significantly increased the expression of NF-$\kappa$B target genes [21].

The reduced NF-$\kappa$B activity in cells overexpressing Otub2 may induce anti-apoptotic effects. To test this possibility, we examined Caspase 3/7 activity in Otub2 over-expressing or Otub2-depleted MIN6 cells. As expected, basal Caspase 3/7 activity was significantly decreased (by ~40%) in Otub2-overexpressing cells compared to the control Flag-overexpressing cells (Fig. 2A). In addition, the ~5-fold increase in caspase activity observed upon cytokine treatment in control (Flag over-expressing) cells, was also significantly reduced (~65%) in Flag-Otub2 overexpressing cells (Fig. 2A). Conversely, and consistent with our previous observations [9], silencing of Otub2 in MIN6 cells using siRNAs, significantly increased caspase 3/7 activity, under basal conditions as well as following treatment with cytokines (Fig. 2B).

Fig. 2.

Effects of Otub2 on caspase 3/7 activity. (A) Control, Flag over-expressing MIN6 cells or Flag-Otub2 over-expressing cells were treated with 1x-cytomix for 24 h (or remained untreated). Apoptosis was assayed by caspase-3/7 activity measurements. (B) MIN6 cells, transfected with the indicated siRNAs, remained untreated or were treated with 1x-cytomix for 24 h. Apoptosis was assayed by caspase-3/7 activity measurements. Control-siRNA transfected cells served as control. (C,D) Dispersed human islets transiently transfected for 48 h with pFlag (control) or pFlag-Otub2 constructs, remained untreated or were treated with IFN-$\gamma$ (C) or transfected with the indicated siRNAs (D), were assayed for apoptosis by caspase-3/7 activity. Data represent means $\pm$ SEM of five replicates in each of three experiments (A), 4–5 replicas (B), or four replicas of human islets (C,D). In all panels, asterisks indicate the p-values as follows: *p 0.05, **p 0.01, ****p 0.0001 vs. control cells.

Even stronger effects were observed in dispersed human islets. An approximate 5-fold decrease in caspase 3/7 activity was observed in human islets that were transfected with the Flag-Otub2 construct, when compared with Flag-transfected islets (Fig. 2C). Similar results were observed upon treatment of dispersed human islets with IFN-$\gamma$ (Fig. 2C). Conversely, opposite effects were observed in dispersed human islets in which Otub2 has been silenced by siRNA. These islets demonstrated ~2.4-fold increase in caspase 3/7 activity compared to control, mock-transfected cells (Fig. 2D). These results indicate that overexpression of Otub2 decreases caspase 3/7 activity and, therefore, confers a pro-survival effect on human pancreatic $\beta$-cells.

To determine whether overexpression of Otub2 affects $\beta$-cell function, glucose-stimulated insulin secretion (GSIS) was studied in MIN6 cells. Basal insulin secretion was ~2.5 higher in Flag-Otub2 expressing cells, when compared to controls (Fig. 3A). Glucose challenge (20 mM) increased insulin secretion 2.5-fold in control cells and 2.2-fold in Otub2 expressing cells, resulting in an approximately 2.2-fold higher GSIS in Otub2 expressing cells. These results favor our hypothesis that Otub2 promotes $\beta$-cell function, in addition to or independent of its inhibitory effects on NF-$\kappa$B activity. Next, we assessed the influence of Otub2 on glucose transporter 2 (Glut2), the glucose transporter mediating glucose uptake in $\beta$-cells [23]. As shown in Fig. 3B, overexpression of Otub2 dramatically increased the mRNA levels of Glut2 by approximately 20-fold. A similar increase was found in cells treated with cytokines, which by themselves drastically inhibited Glut2 expression, as previously described [24].

Fig. 3.

Effects of OTUB2 stable over-expression on $\beta$-cell function. (A) MIN6 cells over-expressing Flag or Flag-Otub2 were incubated for 1 h in Krebs-Ringer bicarbonate HEPES buffer (KRBH) at 37 °C, followed by an additional 1 h incubation in 0 mM or 20 mM glucose. Insulin secretion was then determined using insulin detecting Homogeneous Time-Resolved Fluorescence (HTRF) kit according to the manufacturer’s instructions. (B,C) MIN6 cells overexpressing Flag or Flag-Otub2 were treated with 1x-cytomix for 4 h and harvested. Non-treated cells served as the control. Total mRNA was extracted. Quantitative detection of GLUT2 (B) and NKX6.1 (C) was carried out by qRT-PCR. mRNA levels were normalized to Actin. Data are means $\pm$ SEM of four replicates in each of four experiments (A), or two replicates in each of three experiments (B,C). In all panels, asterisks indicate the p-values as follows: *p 0.05, **p 0.01, and ***p 0.001.

Another gene of interest is NK6 homeobox 1 (Nkx6.1), a known transcription factor required for $\beta$-cell development, as well as glucose sensing and insulin secretion [25]. As shown in Fig. 3C, overexpression of Otub2 significantly increased the mRNA levels of Nkx6.1 by approximately 2- and 2.8-fold under basal or cytomix treatments, respectively. Collectively, these findings suggest that overexpression of Otub2 is involved in preserving $\beta$-cells functionality, mainly in terms of insulin secretion.

To identify proteins that interact with Otub2, immunoprecipitations (IPs) with Otub2 antibodies were carried out using MIN6 cells transiently overexpressing Otub2 (and appropriate controls), followed by mass spectrometry analysis. Several proteins were enriched at least 100-fold in immunoprecipitates derived from Flag-Otub2 overexpressing cells (Table 3, Supplementary Table 1) compared to controls. A similar trend was observed when GFP-Otub-2 Min6 cells were used. Top hits were Peg3 and Camk2d, which enhance the NF-$\kappa$B pathway and $\beta$-cell death [26, 27]. We assume that Otub2 deubiquitinates protein complexes containing Peg3 and Camk2d, and thereby might inhibit propagation of the NF-$\kappa$B cascade that promotes $\beta$-cell apoptosis. The third protein, kv9.3, is a potassium channel responsible for repolarization of $\beta$-cell membranes after insulin release [28]. Potentially, Otub2 can deubiquitinate and inhibit this channel and thereby prolong insulin release and improve $\beta$-cell function.

The effects of Otub2 on $\beta$-cell functionality in vivo were evaluated next. To this end, we employed a mouse model that harbors either heterozygous (Otub2^+⁣/- (or homozygous) Otub2^-⁣/-(whole-body deletion of the Otub2 gene. Pancreata were isolated from wild-type (WT, Otub2^+⁣/+) and heterozygous Otub2 knockout mice (het, Otub2^+⁣/-), and the expression of several NF-$\kappa$B target genes was evaluated. As shown in Fig. 4A, the expression levels of Otub2 itself were decreased by approximately 2-fold in Otub2^+⁣/- mice compared to Otub2^+⁣/+ mice, confirming the expected partial down-regulation of the Otub2 gene expression in the Otub2^+⁣/- animals. Conversely, there was a marked increase in the expression of several NF-$\kappa$B targets in Otub2^+⁣/- mice, including IP-10, MCP-1 and IL-1$\beta$, whose expression levels increased by approximately 5-, 6- and 2-fold, respectively (Fig. 4A). These results suggest an increased inflammatory state of the pancreas upon partial Otub2 depletion, supporting the physiological role of Otub2 as a negative regulator of NF-$\kappa$B activity in pancreatic islets in vivo. Immunofluorescence staining of pancreatic sections from those mice confirmed the expected decrease in Otub2 protein levels in the Otub2^+⁣/- and Otub2^-⁣/- mice (Fig. 4B).

Fig. 4.

Effects of OTUB2 knockdown on pancreatic NF-$\kappa$B activity in vivo. Pancreata from Otub2^+⁣/+ (wild-type (WT)) and Otub2^+⁣/- (het) female and male mice were excised. (A) Total mRNA was extracted, and qRT-PCR was conducted. mRNA levels of NF-$\kappa$B target genes were normalized to Actin. Data is Otub2^+⁣/+ n = 5; Otub2^+⁣/- n = 5 (2 males and 3 females per group). Data represent means $\pm$ SEM of three experiments. Asterisks indicate the p-values: *p 0.05, **p 0.01 and ***p 0.001 vs. Otub2^+⁣/+ mice. (B) Pancreas sections were stained for Otub2 and DAPI. Scale bar = 100 µm. (C,D) Otub2 knock-out mice are glucose intolerant. Male and female mice at 8 weeks of age were subjected to an intraperitoneal (i.p.) glucose tolerance test (GTT; 1.8 g D-glucose per kg body weight) after overnight fasting. Blood samples were taken at the indicated time points (0’–120’) and glucose levels were determined by a glucometer (C). A graph representing the area under the curve of the GTT is indicated as (D). Data represent means $\pm$ SEM of six experiments. Mice numbers: Males: Otub2^+⁣/+ n = 7; Otub2^+⁣/- n = 12; Otub2^-⁣/- n = 8. Females: Otub2^+⁣/+ n = 8; Otub2^+⁣/- n = 10; Otub2^-⁣/- n = 8. Asterisks indicate the p-values: *p 0.05, **p 0.01.

Next, glucose tolerance tests (GTT) were performed in Otub2^+⁣/+, Otub2^+⁣/-, and Otub2^-⁣/- mice. As shown in Fig. 4C, basal glucose levels of Otub2^+⁣/+, Otub2^+⁣/-, and Otub2^-⁣/- male and female mice were approximately similar. However, following glucose injection, blood glucose levels, at all-time points, were significantly higher in Otub2^+⁣/- and Otub2^-⁣/- male mice, when compared to Otub2^+⁣/+ animals (Fig. 4C). Accordingly, the area under the curve (AUC) was ~40% greater in Otub2^+⁣/- and Otub2^-⁣/- male mice compared to WT controls (Fig. 4D). The effects of Otub2 deletion on the female mice were a bit more complex. There were no differences in GTT between Otub2^+⁣/+ and Otub2^-⁣/- female mice, but the Otub2^+⁣/- animals exhibited a slight, yet significant higher GTT response (Fig. 4C) that was also evident by a modest, yet significant higher (approximately 20%) AUC (Fig. 4D). These results suggest a positive role for Otub2 in the improvement of islet functionality under physiological conditions in an in vivo setting, with possible compensatory pathways in KO female mice.

To complement and validate the above findings that were based upon Otub2 KO mouse models, we used a second, independent model in which Otub2 was selectively overexpressed in the pancreas of C57BL wild-type mice. To introduce the Otub2 gene into the pancreas, we made use of adeno-associated viral constructs (AAVs) that are considered a favorable delivery system to express exogenous proteins due to their low immunogenicity and excellent safety profile [29]. C57BL wild-type mice were subjected to pancreatic intraductal administration of AAV8 constructs containing Flag (control) or Flag-Otub2 sequences (Supplementary Fig. 2A). As shown in Supplementary Fig. 2B, islets derived from mice, subjected to intraductal injection of AAV8-Flag-Otub2, showed a strong fluorescent signal, while a much weaker signal was observed in mice injected with AAV8-Flag viral control vector. We further validated these findings by extracting RNA from the pancreases and performing qRT-PCR (Supplementary Fig. 2C). These results verified the efficacy of increasing Flag-Otub2 expression in pancreases, by intra-ductal AAV8 injection.

Glucose-tolerance tests in these mice revealed that basal glucose levels in AAV8-Flag-Otub2 mice were almost identical to the AAV8-Flag control animals, however, blood glucose levels of Flag-Otub2 mice were significantly lower (approximately 30%) at 15–45 minutes post glucose injection (Fig. 5A). Accordingly, the AUC decreased by 20% in AAV8-Flag-Otub2 injected mice compared with control animals (Fig. 5B). These findings further support our conclusion that Otub2 exerts favorable functional effects on $\beta$-cells in vivo.

Fig. 5.

Effects of intra-ductal injection of AAV8-Flag-Otub2 vector on glucose tolerance. (A) C57BL WT, 10 weeks old, male mice, were anesthetized and subjected to retrograde pancreatic intraductal AAV administration. After two weeks of recovery and overnight fasting, mice were subjected to an intraperitoneal (i.p.) glucose tolerance test (GTT; 1.8 g D-glucose per kg body weight). Blood samples were taken at the indicated time points (0–120 minutes), and glucose levels were determined by a Glucometer. (B) Area under the curve of the GTT. Data represent means $\pm$ SEM of five experiments. T-test was used to measure the p-values: *p-value 0.05, **p-value 0.01 of AAV8-Flag mice compared to AAV8-Flag-Otub2 mice. Mice numbers: n = 5 AAV8-Flag mice; n = 7 AAV8-Flag-Otub2 mice.

To reveal the genes whose expression is affected upon Otub2 deletion, RNA was extracted from the pancreases of Otub2^-⁣/-, Otub2^-⁣/+, and Otub2^+⁣/+ (WT) mice, and RNAseq analysis was performed. The expression levels of Otub2 were indeed below detection in most Otub2^-⁣/- and Otub2^+⁣/- pancreases (Supplementary Fig. 3A). Both heterozygous and homozygous pancreases showed significantly different patterns of gene expression compared to WT pancreases, as shown by the volcano plots (Supplementary Fig. 3B). Of note, more than 20 genes were significantly up- or down regulated, both in heterozygous and homozygous mice when compared to wild-type animals (Supplemental Table 2). Enrichment analysis of curated signatures employing the MsigDB database and the Camera method [30] revealed several gene-sets that were significantly upregulated (Supplementary Fig. 3C). Most significant were the MOOTHA_VOXPHOS gene families [31], involved in oxidative phosphorylation. Other relevant families were WANG_NFKB_TARGETS, which includes NF-$\kappa$B target genes [32]; as well as REACTOME_DIABETES_TARGETS [33], and ALCALA_APOPTOSIS [34]; all gene-sets expected to be upregulated upon depletion of Otub2.

To gain a deeper insight into the gene networks affected by Otub-2 depletion, pathway enrichment analysis was carried out using the GO database. Such analysis revealed that potassium ion transport was significantly affected in Otub-2^-⁣/- and Otub-2^-⁣/+ mice (Fig. 6A,B), with several genes related to this pathway being down-regulated in OTUB2 KO animals (Fig. 6B). One is Ank2 that encodes the ankyrin-2 (AnkB) protein, which is essential for localization and membrane stabilization of potassium channels. Other relevant genes were (potassium voltage-gated channel subfamily C member 3) KCNC3, encoding a voltage-dependent potassium channel involved in insulin secretion [35], and Cacna1a, encoding the $\alpha$1A pore-forming subunit of the neuronal channel P/Q, which is a member of the voltage-gated calcium channels superfamily.

Fig. 6.

RNA-seq analysis of pancreas from OTUB2 KO mice. Pancreases and livers were extracted from OTUB2^+⁣/+ (WT, n = 3), OTUB2^+⁣/- (Heterozygotes, n = 5) and OTUB2^-⁣/- (Homozygote, n = 3) mice. RNA was extracted, and RNA-seq was performed to determine differential gene expression between the homozygotes/heterozygotes and the WT mice. (A) Geneset enrichment of gene ontology (GO) terms, performed by the CAMERA method. Arrows indicate interesting GO terms. Analysis is based on gene expression in the pancreas. (B,C) Heatmaps representing the fold change (log2) in gene expression between heterozygotes and WT mice in the pancreases and livers. Genes related to potassium ion transport (B) and oxidative phosphorylation (C) are taken from the respective GO terms. (D) STRING analysis of genes significantly upregulated in Otub2^+⁣/- vs. Otub2^+⁣/+ pancreases. The black circle depicts a cluster of genes associated with oxidative phosphorylation.

Conversely, genes related to oxidative phosphorylation were mostly upregulated in heterozygous and homozygous Otub2 KO mice (Fig. 6A,C), in accordance with the increased enrichment of the MOOTHA_VOXPHOS geneset (Supplementary Fig. 3C,D). Independently, STRING analysis revealed a cluster of oxidative phosphorylation-related genes among those genes that were significantly upregulated in pancreata of Otub-2^-⁣/+ mice (Fig. 6D).

Given that $\beta$-cells are highly susceptible to oxidative stress [36, 37], the increase in oxidative phosphorylation might be associated with $\beta$-cells dysfunction and apoptosis observed upon Otub2 depletion. Of note, the ATP-dependent potassium channels, discussed above, are specific targets of oxidative damage [38]. Indeed, the upregulated genes in Otub2 KO pancreases, related to oxidative phosphorylation, were almost exclusively encoding for components of the electron transport chain (ETC) complexes in the mitochondria (Fig. 6C). Conversely, the highest upregulated gene in this pathway, Cox4i1, was previously shown to be upregulated in type 2 diabetic mice [39]. This suggests that oxidative stress is a potential trigger of the impaired $\beta$-cells function observed in the Otub2 KO mice.

The alterations in the transcriptomic landscape of pancreata depleted of Otub2 were quite unique, as largely different alterations in transcriptomic landscapes were observed in livers of heterozygous and homozygous Otub2 KO mice when compared to livers of WT animals (Supplementary Fig. 4A,B). Still, pathway enrichment analysis revealed that oxidative phosphorylation genes, as well as components of the ETC complexes in the mitochondria, were enriched in livers as well (Supplementary Fig. 4C). Although these hepatic genes were not upregulated to the same extent as they did in the pancreas (Fig. 6C, Supplementary Fig. 3D), our results suggest that knockdown of Otub2 ubiquitously upregulates genes involved in oxidative phosphorylation and mitochondrial ETC.