ORIGINAL RESEARCH published: 13 May 2020 doi: 10.3389/fendo.2020.00258 Repurposed Analog of GLP-1 Ameliorates Hyperglycemia in Type 1 Diabetic Mice Through Pancreatic Cell Reprogramming Adrian Villalba 1 , Silvia Rodriguez-Fernandez 1 , David Perna-Barrull 1 , Rosa-Maria Ampudia 1 , Laia Gomez-Muñoz 1 , Irma Pujol-Autonell 1 , Eva Aguilera 2 , Mireia Coma 3 , Mary Cano-Sarabia 4 , Federico Vázquez 2 , Joan Verdaguer 5,6 and Marta Vives-Pi 1,6*† 1 Immunology Section, Germans Trias i Pujol Research Institute, Autonomous University of Barcelona, Badalona, Spain, 2 Endocrinology Section, Germans Trias i Pujol Research Institute, Autonomous University of Barcelona, Badalona, Spain, 3 Anaxomics Biotech SL, Barcelona, Spain, 4 Catalan Institute of Nanoscience and Nanotechnology, CSIC and The Barcelona Edited by: Institute of Science and Technology, Bellaterra, Spain, 5 Immunology Unit, Department of Experimental Medicine, Faculty of Simona Chera, Medicine, IRBLleida, University of Lleida, Lleida, Spain, 6 CIBER of Diabetes and Associated Metabolic Disease (CIBERDEM), University of Bergen, Norway ISCIII, Madrid, Spain Reviewed by: Quan Zhang, Type 1 diabetes is an autoimmune disease caused by the destruction of the University of Oxford, United Kingdom Anabel Rojas, insulin-producing β-cells. An ideal immunotherapy should combine the blockade of the Andalusian Center of Molecular autoimmune response with the recovery of functional target cell mass. With the aim Biology and Regenerative Medicine (CABIMER), Spain to develop new therapies for type 1 diabetes that could contribute to β-cell mass *Correspondence: restoration, a drug repositioning analysis based on systems biology was performed Marta Vives-Pi to identify the β-cell regenerative potential of commercially available compounds. Drug mvives@igtp.cat repositioning is a strategy used for identifying new uses for approved drugs that are † ORCID: outside the scope of the medical indication. A list of 28 non-synonymous repurposed Marta Vives-Pi drug candidates was obtained, and 16 were selected as diabetes mellitus type 1 http://orcid.org/0000-0003-3735-0779 treatment candidates regarding pancreatic β-cell regeneration. Drugs with poor safety Specialty section: profile were further filtered out. Lastly, we selected liraglutide for its predictive efficacy This article was submitted to values for neogenesis, transdifferentiation of α-cells, and/or replication of pre-existing Cellular Endocrinology, a section of the journal β-cells. Liraglutide is an analog of glucagon-like peptide-1, a drug used in patients Frontiers in Endocrinology with type 2 diabetes. Liraglutide was tested in immunodeficient NOD-Scid IL2rg−/− Received: 13 January 2020 (NSG) mice with type 1 diabetes. Liraglutide significantly improved the blood glucose Accepted: 07 April 2020 Published: 13 May 2020 levels in diabetic NSG mice. During the treatment, a significant increase in β-cell mass Citation: was observed due to a boost in β-cell number. Both parameters were reduced after Villalba A, Rodriguez-Fernandez S, withdrawal. Interestingly, islet bihormonal glucagon+ insulin+ cells and insulin+ ductal Perna-Barrull D, Ampudia R-M, cells arose during treatment. In vitro experiments showed an increase of insulin and Gomez-Muñoz L, Pujol-Autonell I, Aguilera E, Coma M, Cano-Sarabia M, glucagon gene expression in islets cultured with liraglutide in normoglycemia conditions. Vázquez F, Verdaguer J and These results point to β-cell replacement, including transdifferentiation and neogenesis, Vives-Pi M (2020) Repurposed Analog of GLP-1 Ameliorates Hyperglycemia as aiding factors and support the role of liraglutide in β-cell mass restoration in type in Type 1 Diabetic Mice Through 1 diabetes. Understanding the mechanism of action of this drug could have potential Pancreatic Cell Reprogramming. clinical relevance in this autoimmune disease. Front. Endocrinol. 11:258. doi: 10.3389/fendo.2020.00258 Keywords: beta cell regeneration, neogenesis, transdifferentiation, liraglutide, drug repositioning Frontiers in Endocrinology | www.frontiersin.org 1 May 2020 | Volume 11 | Article 258 Villalba et al. Drug Repositioning and Type 1 Diabetes INTRODUCTION used for focusing the analysis toward β-cell regeneration in a human biological network context. An essential requirement to cure type 1 diabetes is the recovery The human biological network created incorporated the of β-cells lost after the autoimmune destruction. The pancreas available relationships (edges or links) between proteins (nodes) can restore β-cells from different sources, after injury (1) or from a regularly updated in-house database drawn from public drug administration (2, 3) in mice, and upon pathophysiological sources: KEGG (20, 21), REACTOME (22), INTACT (23), conditions in humans (4). With the aim to develop new therapies BIOGRID (24), HPRD (25), and TRRUST (26). All information for β-cell replacement, a repositioning analysis based on systems of the key proteins defined during the molecular and the biology was performed to identify the regenerative potential of biological characterization and stored in relevant databases (drug commercially available compounds. targets, other diseases effectors, biomarkers. . . ) was incorporated Drug repositioning is a strategy for identifying new uses into the biological networks (27). for approved drugs that are outside the scope of the medical Artificial neural networks (ANNs) are supervised algorithms indication. This approach uses a bioinformatics tool, based that identify relations between proteins (e.g., drug targets) and on networks of drugs, proteins, and diseases, which screens clinical elements of the network (18, 19) by inferring the approved compounds that can be repurposed for other diseases probability of the existence of a specific relationship between (5). This has resulted in successful drug discovery for diseases two or more protein sets. This is based on a validation of (6) such as cancer (7) and Alzheimer’s disease (8). Given that, the predictive capacity of the model toward the truth table, we aimed to look for drugs that can induce β-cell replacement a selected collection of known input (drug targets)–output through α-cell to β-cell transdifferentiation (9), neogenesis from (indications) relationships defined through specific scientific multipotent ductal progenitors (1), and/or replication of pre- literature search and hand-curated assignment of proteins to existing β-cells (10). This resulted in the identification of the conditions included in the biological effector database (17). liraglutide—an analog of glucagon-like peptide-1 (aGLP-1), a The learning methodology used consisted of an architecture of drug used in patients with type 2 diabetes (11), especially those stratified ensembles of neural networks as a model, trained with a with obesity (12). gradient descent algorithm to approximate the values of the given GLP-1 is produced in the gut after food intake and acts truth table. The neural network model used was a multilayer by increasing insulin release. Liraglutide ameliorates insulin perceptron (MLP) neural network classifier. The MLP gradient resistance in type 2 diabetes models (13, 14). Recently, GLP-1 descent training depends on randomization initialization and, to has been shown to promote transdifferentiation from α-cells to avoid random errors, 1,000 MLPs are trained with the training β-cells (15, 16). subset and the best 100 MLPs are used. In order to correctly We report here that liraglutide ameliorates hyperglycemia predict the effect of a drug independently of the number of in mice with type 1 diabetes by inducing insulin+ glucagon+ targets, different ensembles of neural networks are trained for cells and insulin-producing cells from the pancreatic ducts. different subsets of drugs according to their number of targets This is the first description of an approved drug—identified by (drugs with one target, two targets, three targets. . . ). Then, the a repositioning—that promotes the generation of insulin- predictions for a query drug are calculated by all the ensembles expressing cells from pancreatic ducts and ameliorates and pondered according to the number of targets of the query hyperglycemia in experimental type 1 diabetes. drug (the difference between the number of targets of the query and the number of targets of the drugs used to calculate each ensemble is used to ponder the result of each ensemble). A MATERIALS AND METHODS cross-validation with the truth table information showed that the accuracy of the described ANNs to reproduce the indications Systems Biology Analysis for Drug compiled in DrugBank (28, 29) is 81.23% for those drugs with Discovery and Repositioning Analysis all targets in the human biological network. ANNs were used to The therapeutic performance mapping system is a top-down screen the predicted relationship of 6,605 different drugs toward systems biology approach based on artificial intelligence and the β-cell regeneration molecular definition. The repurposed pattern recognition that integrates all available pharmacological drug candidates were sorted and selected by its relationship with knowledge to create mathematical models that simulate human β-cell regeneration mechanisms. pathophysiology in silico. The methodology employed has been previously described (17) and applied elsewhere (18, 19). A manually curated list of proteins known to be involved Mice and Diabetes Induction in the mechanisms of β-cell regeneration—α-cell to β-cell NOD mice, immunodeficient NOD.SCID-IL2Rγ−/− (NSG) transdifferentiation, neogenesis from ductal precursors, and mice, and C3HeB/FeJ mice were bred in our own facility. β-cell replication—was obtained (Supplementary Table 1) and The NOD and NSG mice were kept under specific pathogen- free conditions. The NSG model was selected for its lack of an adaptive immune system, resulting in the absence of Abbreviations: ANN, artificial neural network; AUC, area under the curve; aGLP- 1, analog glucagon-like peptide-1; NSG, immunodeficient NOD.SCID-IL2Rγ−/− autoimmunity, and allowing to determine the raw effect of the mice; IPGTT, intraperitoneal glucose tolerance test; ITT, insulin tolerance test; drug. Type 1 diabetes was induced in NSG mice at 8–14 weeks of STZ, streptozotocin. age by a single i.p. injection of streptozotocin (STZ, 150 mg/kg) Frontiers in Endocrinology | www.frontiersin.org 2 May 2020 | Volume 11 | Article 258 Villalba et al. Drug Repositioning and Type 1 Diabetes (Sigma-Aldrich). Type 1 diabetes was confirmed at 48–72 h post- total nuclei (34). The intensity of fluorescence was measured in STZ injection, after either two successive 2-h fasting blood arbitrary units using Fiji (32). glucose levels higher than 250 mg/dl or with one higher than To determine the insulin+ glucagon+ cells, pancreas from 300 mg/dl. The mice were euthanized by cervical dislocation. three mice from each group were analyzed (T1D, Lira 48 h, Lira, All animal studies were approved by the institutional animal post-Lira, and sham). Briefly, 12 non-overlapping pancreatic ethics committee. cryostat sections from each mouse were stained for insulin and glucagon. A minimum of 72 islets per mouse was considered Treatment With aGLP-1 and the percentage of islets that contained bihormonal cells was Immediately after type 1 diabetes diagnosis, the NSG mice (n = determined. To assess ductal insulin+ cells, pancreas from four 12 mice/group) were treated with liraglutide (Victoza R , Novo mice from each group were analyzed (T1D, sham, and Lira). Nordisk A/S), injected (s.c.) daily up to 30 days, following the Briefly, four non-overlapping pancreatic cryostat sections from dosage of 0.3 mg/kg at day 1, 0.6 mg/kg at day 2, and 1 mg/kg each mouse were stained for CK19 and insulin. A minimum of from day 3 onwards as described (30). After the withdrawal of the 23 ductal areas per section was considered and the percentage liraglutide treatment, the mice were maintained for 5 days. The of ducts that contained insulin+ cells was determined. To prove control group (n = 6 mice) received phosphate-buffered saline the colocalization of insulin and glucagon in islet cells and (PBS). Blood glucose was determined twice weekly, after 2 h of insulin and CK19 in ductal cells, confocal microscopy was fasting, throughout the study. performed using an Axiobserver Z1 (Zeiss) and by analyzing 1-µm sections. Intraperitoneal Glucose Tolerance Test and In vivo Tracking of Liraglutide Insulin Tolerance Test Liraglutide was conjugated to AlexaFluor750 (AF750, Invitrogen) Intraperitoneal glucose tolerance test (IPGTT) was performed in using a standard method (Thiol-Reactive Probes, Invitrogen). fasting conditions in the three groups: (1) diabetic NSG mice In vivo and ex vivo near-infrared fluorescence imaging was responding to liraglutide after 15 days of treatment (Lira, n = performed (Pearl Impulse imaging system, LI-COR) in NOD 3), (2) untreated diabetic and hyperglycemic NSG mice (T1D, mice anesthetized with ketamine (50 mg/kg) and xylazine (5 n = 3), and (3) healthy and normoglycemic NSG mice (sham, mg/kg) at 15 and 60 min after the s.c. administration of 1 n = 3). At point 0, basal glucose level was determined. The mg/kg of AF750-liraglutide in 50 µl of PBS. At the end of each mice were subsequently given an i.p. injection of 2 mg of glucose checkpoint, spleen, stomach, fat, heart, liver, pancreas, lungs, (Sigma-Aldrich) per gram of body weight and glycemia was kidney, salivary glands, thymus, and bladder were imaged ex vivo. measured after 15, 30, 60, 120, and 210 min. Insulin tolerance Fluorescent signal intensity was semi-quantitatively assessed: test (ITT) was performed in fasting conditions in 8-week-old the levels were normalized by subtracting the background and and normoglycemic NOD mice and C3HeB/FeJ mice injected s.c. represented as a relative index of fluorescence in each organ per with insulin (0.5 U/kg, n = 3) or liraglutide (1 mg/kg, n = 3). gram of tissue. Glycemia was determined after 15, 30, and 60 min. Islets and β-Cell Line Immunofluorescence Staining and Islets from young non-diabetic NOD mice (n = 7) were obtained Histometric Analysis from pancreas digested with collagenaseP (2.5 mg/ml in PBS; Immunofluorescence staining was performed to identify Sigma-Aldrich), after having been injected through the common pancreatic insulin-producing cells in a minimum of three bile duct, resected, and incubated at 37◦ C for 30 min. The mice per condition. Briefly, the pancreas were harvested and pancreas was dissociated and the islets were hand-picked. Groups snap-frozen in an isopentane/cold acetone bath. A minimum of 50 islets were cultured in a medium of high (10 mM, RPMI- of eight cryostat sections (5 µm) from every organ were 1640, Biowest) or normal glucose concentration (6.1 mM, Ham sequentially stained by indirect immunofluorescence with F-10, Gibco) as described (35), both in basal conditions—culture antibodies to insulin, glucagon, CK19 (Sigma-Aldrich), or Pdx1 media—and by adding liraglutide at 1,000 nM for 48 h. (Abcam) and FITC- or TRITC-labeled secondary antibodies The β-cell line NIT-1 (ATCC) (36) was cultured (35) with (Sigma-Aldrich) as described (31). The nuclei were stained liraglutide at 10, 100, and 1,000 nM for 12, 24, and 48 h. Viability with Hoechst (Invitrogen). The samples were observed in a was assessed with annexin V (AnnV) PE (Immunotools) and fluorescence microscope and analyzed (ImageJ Software) (32). 7-amino-actinomycin D (7aad) labeling (BD Pharmingen) and For histometric analysis, six mice per group were used. To analyzed with FACS Canto II (BD Biosciences). determine the β-cell counts, one section every 150 µm of tissue was sampled as described (33), resulting in 12–16 sections per Gene Expression Analysis pancreas. The β-cell mass was calculated by multiplying the To determine the effects of liraglutide on gene expression, relative insulin+ area per total pancreas weight, and the β-cell quantitative RT-PCR was performed. Briefly, RNA was number as well as the insulin+ aggregates were calculated by isolated from islets using RNeasy Micro Kit (Qiagen) and manually counting the nuclei within the insulin+ area and reverse-transcribed with a cDNA Reverse-Transcription Kit extrapolating to the whole organ as previously described (34). (ThermoFisher Scientific). cDNA synthesis was performed The β-cell size was assessed by dividing the insulin+ area per using random hexamers (0.5 mg/ml, BioTools) and reverse Frontiers in Endocrinology | www.frontiersin.org 3 May 2020 | Volume 11 | Article 258 Villalba et al. Drug Repositioning and Type 1 Diabetes transcriptase Moloney–murine–leukemia–virus (200 U/ml, recovered normoglycemia during liraglutide administration until Promega). Targeted cDNA was pre-amplified with TaqMan day 30 (Lira responders) when compared to the non-treated PreAmp MasterMix (ThermoFisher Scientific). Quantitative mice (T1D) (Figure 1A). After treatment withdrawal, all mice RT-PCR assays were performed with TaqMan universal assay became hyperglycemic. A statistical difference between the (ThermoFisher Scientific) on a LightCycler R 480 (Roche) treated and the control groups was observed when analyzing using the following TaqMan assays: Ins2 (Mm00731595_gH), the area under the curve (AUC) after 15 days of treatment Gcg (Mm00801714_m1), Ki67 (Mm01278617_m1), and Il17a (Figure 1B). The response of the NSG mice to liraglutide after (Mm00439618_m1). The expression for each gene of interest an acute increase in blood glucose levels showed that the treated was normalized to that of the housekeeping gene Gapdh diabetic mice (Lira) recovered normoglycemia at 210 min after (Mm99999915_g1), as described in the 2−1Ct method (37). glucose injection similarly to the non-diabetic group, whereas Values from islets were normalized using their respective basal the diabetic non-treated mice remained hyperglycemic (>400 controls and represented as a ratio (relative gene expression). mg/dl, T1D) (Figure 1C). Significant differences found between the treated and the non-treated mice demonstrate that insulin Effect of Liraglutide on NIT-1 Cell Line production and secretion improve by the effect of this aGLP- NIT-1 cells were stained with anti-CD44 BV786 (BD 1. The analysis of AUC showed an intermediate response to Biosciences), anti-class I major histocompatibility complex glucose stimulation in the liraglutide-treated group (Figure 1D). (MHC) eFluor-450 (eBioscience), anti-CD14 PE, and anti- To elucidate whether liraglutide acts in a similar manner as CD49b FITC (Immunotools). Viability was assessed with insulin administration, an ITT was performed in normoglycemic AnnV PE (Immunotools) and 7aad (BD Biosciences) as NOD mice. The group treated with insulin displayed a reduction detailed above. Median fluorescence intensity and viability in glycemia after 30 min that was maintained until 60 min. were determined using flow cytometry (LSR Fortessa, BD By contrast, the liraglutide-treated group displayed an increase Biosciences). Corresponding fluorescence minus one staining in glycemia until 30 min, which was normalized at 60 min was used as control. The data were analyzed using FlowJo similarly to the insulin-treated animals (Figure 1E). Despite that (Tree Star Inc). no significant differences were observed, the opposite tendency between both groups observed in the AUC (Figure 1F) revealed Statistical Analysis that liraglutide and insulin do not act similarly. To determine the Prism 7.0 (GraphPad Software Inc.) was used to perform the effect of the genetic background, the acute effect of liraglutide statistical analyses. For comparisons of unpaired data, a non- was also determined on the C3HeB/FeJ strain. The results parametric Mann–Whitney test was used. The statistical tests showed an insulinotropic effect of liraglutide in this strain of applied to each data set are specified in each figure legend. mice (Supplementary Figure 1) with a tendency to differ at the endpoint (60 min) in comparison to insulin. RESULTS Liraglutide Is a Repurposed Candidate for Liraglutide Transiently Increases the β-Cell β-Cell Regeneration Mass A total of 28 non-synonymous repurposed drug candidates The β-cell mass of the NSG mice after 7–15 days of showed a p < 0.05 (predicted value ≥76.075%), and 16 were treatment with liraglutide significantly increased when compared selected as type 1 diabetes treatment candidates in the context to that of the diabetic non-treated mice, although they of pancreatic β-cell regeneration. Drugs with poor safety profile did not reach normal levels (Figure 2A). This effect was were further filtered out. Lastly, we selected liraglutide (predicted lost 5 days after withdrawal. In this sense, an increase value of 96.88%) that fulfilled all the criteria and had a novel, in β-cell number was detected during the treatment in potentially beneficial mechanism toward β-cell regeneration. comparison to T1D and post-treatment groups (Figure 2B). These alterations were maintained even with normalized values Liraglutide Improves Hyperglycemia in to body weight or pancreatic tissue (Supplementary Table 2). Similarly, the percentage of islets emerging from ducts was Diabetic Mice increased both during treatment and after therapy removal We determined the effect of liraglutide in immunodeficient (Figure 2C). These alterations were not related to changes NSG mice because of their absence of autoimmunity and in either β-cell size (Figure 2D) or insulin fluorescence low phenotypic heterogeneity. Type 1 diabetes was induced intensity (33) (Figure 2E). in NSG mice by a single STZ injection. All mice showed disease symptoms at 48–72 h after administration. A total of 50% (six of 12) of the treated NSG mice were responders Liraglutide Induces Endocrine Cell to liraglutide according to the improvement of blood glucose Rearrangement in the Pancreas levels during treatment. The remaining six mice did not have The histological analysis revealed the presence of bihormonal decreased blood glucose levels in any way. No differences in cells (insulin+ glucagon+ ) in the islets of NSG mice at terms of initial fasting glycemia (mg/dl) were found between the the beginning of the treatment with liraglutide but not responder (346.8 ± 72.5, mean ± SD) and the non-responder after 7–15 days or withdrawal (Figure 3). A total of 48.14 (397.7 ± 57.71) mice. The NSG mice with diabetes partially ± 1.53% of islets contained bihormonal cells, and the Frontiers in Endocrinology | www.frontiersin.org 4 May 2020 | Volume 11 | Article 258 Villalba et al. Drug Repositioning and Type 1 Diabetes FIGURE 1 | Effect of liraglutide in diabetic immunodeficient NOD.SCID-IL2Rγ−/− (NSG) mice. (A) Two-hour fasting blood glucose levels (mg/dl) in mice rendered diabetic by a single dose of streptozotocin (150 mg/kg) and then treated with daily s.c. injections of liraglutide (Lira responders, circles) and phosphate-buffered saline (T1D, squares) from day 0 to 30 (dashed line). The filled area corresponds to normoglycemia levels in mice. (B) Area under the curve (AUC) of the graph in (A) at day 15, when all animals remain alive. Results are mean ± SD, and differences were found between groups (**p < 0.01, Mann–Whitney test). (C) Intraperitoneal glucose tolerance test in diabetic NSG mice treated with daily s.c. liraglutide for 15 days (Lira, dashed line), non-treated diabetic animals (T1D, continuous line), and normoglycemic mice (sham, dotted line). Statistical differences were found between Lira and T1D groups (+p < 0.05, ++p < 0.01, + + +p < 0.001, Mann–Whitney test) and in both groups when compared to sham (dotted line) (*p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001, Mann–Whitney test). The filled area corresponds to normoglycemia levels in mice. (D) AUC of the graph in (C). The results are mean ± SD, and differences were found between groups (**p < 0.01, ****p < 0.0001, Mann–Whitney test). (E) Insulin tolerance test performed in normoglycemic mice injected with Lira (1 mg/kg, dashed line) or insulin (0.5 U/kg, continuous line). (F) AUC of the graph in (E). The results are mean ± SD; no statistical differences were found between groups. Frontiers in Endocrinology | www.frontiersin.org 5 May 2020 | Volume 11 | Article 258 Villalba et al. Drug Repositioning and Type 1 Diabetes FIGURE 2 | Effect of liraglutide in the endocrine pancreas of diabetic immunodeficient NOD.SCID-IL2Rγ−/− (NSG) mice. (A) β-cell mass (mg), (B) β-cell number, (C) percentage of budding islets (arising from ducts), and (D) β-cell size (µm2 ) from NSG mice rendered diabetic by streptozotocin and treated with phosphate-buffered saline (T1D), with daily s.c. injections of liraglutide during 7–15 days (Lira), 5 days after the withdrawal of the liraglutide treatment (post-Lira) and normoglycemic mice (sham). The results are mean ± SD, and differences were found between groups (*p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001, Mann–Whitney test). (Continued) Frontiers in Endocrinology | www.frontiersin.org 6 May 2020 | Volume 11 | Article 258 Villalba et al. Drug Repositioning and Type 1 Diabetes FIGURE 2 | (E) Intensity of fluorescence of insulin staining in 5-µm cryostat pancreatic sections of NSG mice. The intensity was measured for all the islets analyzed per animal, and the mean of the intensities of all the islets is represented as a symbol. The results are mean ± SD, and no statistical differences were found between groups (Mann–Whitney test). percentage of insulin+ glucagon+ cells in relation to the total deleterious effect on the NIT-1 cell line at 100 and 1,000 nM number of beta cells in the islets was 11.38 ± 2.97%. during 24 h (Figure 6A). Interestingly, liraglutide increased No bihormonal cells were detected in the other groups. the expression of adhesion molecules such as CD49b, CD44 Confocal microscopy images prove the colocalization of (Figure 6B), and CD14, a receptor of the innate immunity, and insulin and glucagon in the same cell (bihormonal cells) reduced the expression of MHC class I. (Supplementary Figure 2). Moreover, the pancreatic sections of the NSG mice treated for 7–15 days and after withdrawal revealed the existence of insulin+ DISCUSSION bodies emerging from ducts and without resembling the classical Drug repositioning is an attractive strategy for identifying islet shape (Figure 4A). Between 23 and 47 ducts (CK19+) were new uses for approved drugs. The algorithm of this method analyzed in every pancreatic section. The percentage of ducts was fed with target proteins known to be involved in β-cell that contained insulin+ cells was 50.83 ± 7.31% in mice treated restoration processes—(i) α-cell to β-cell transdifferentiation (9), with liraglutide (Lira group), whereas no ducts with insulin+ (ii) neogenesis from multipotent ductal progenitors (1), and (iii) cells were detected in normoglycemic mice (sham group, 0%) self-replication of pre-existing β-cells (10)—and resulted in the or untreated T1D mice (T1D, 0%). The ducts that contained identification of liraglutide as a repurposed drug. As mentioned, insulin+ cells were analyzed and, in these ductal areas, 82.85 liraglutide is an aGLP-1 used to treat type 2 diabetes (11, 12). A ± 4.37% of doble-positive CK19+ insulin+ cells were found. recent clinical trial demonstrates that liraglutide reduced HbA1c Interestingly, the ductal areas positive for insulin in mice treated and insulin requirements in patients with long-standing type 1 with liraglutide were detected from 48 h to the end of the diabetes (38). Moreover, it has been described that liraglutide treatment and even 5 days after the withdrawal (Figure 4B, white improves β-cell function in alloxan-induced diabetic mice (39). arrows). Confocal microscopy images prove the colocalization Our data show that liraglutide improves hyperglycemia, of insulin and CK19 in the same cell (Supplementary Figure 3). even reaching normoglycemia, in NSG mice. This model was These insulin+ CK19+ cells were glucagon-negative (Figure 4C). selected for its lack of an adaptive immune system, resulting Moreover, we assessed the expression of the β-cell marker Pdx1 in immunodeficiency, and absence of autoimmunity. This in ductal cells, identifying a subpopulation of Pdx1+ cells in fact allows us to determine the effect of the drug without pancreatic ducts from treated animals (white arrow, Figure 4D). autoimmunity interferences and reduced heterogeneity. The To determine if liraglutide effects could be due to the amelioration of hyperglycemia was also observed upon glucose accumulation in the pancreas, its biodistribution was assessed. stimuli (IPGTT) in diabetic mice treated with liraglutide, whereas The in vivo tracking of AF750-liraglutide showed an affinity diabetic non-treated mice remained hyperglycemic. By contrast, for several organs including the pancreas that was higher at the administration of liraglutide to normoglycemic NOD mice 15 min (Figure 5A) than at 60 min (Figure 5B), revealing an resulted in a weak and transient increase of glycemia levels, as acute effect. opposed to the effects of insulin administration. Taken together, To further investigate the short-term effects of liraglutide, these results indicate that liraglutide ameliorates hyperglycemia the expression of specific genes was assessed in islets from both in fasting and fed conditions but acting differently to insulin. healthy NOD mice (three to four per group) cultured The acute effect of liraglutide on a different mouse strain was with or without liraglutide for 48 h in normal and high insulinotropic as expected. These differences between NOD and glucose concentrations. Previously, and in order to discard C3 mice could be due, at least in part, to genetic differences autoreactive insulitis in the islets, the expression of Il17a in the structure and the size of the endocrine pancreas in both was assessed and found negative (data not shown). After strains, specifically in the α- and β-cell mass (40). Another exposure to liraglutide, a trend to upregulate Gcg and to influencing factor should be the islet leukocyte infiltration, a downregulate Ki67 was observed both at doses of 6.1 mM feature of the NOD model, which promotes an inflammatory glucose (Figure 5C) and 10 mM glucose (Figure 5D). microenvironment, thus affecting insulin metabolism. The acute Finally, after exposure to liraglutide, the insulin gene (Ins2) effect of liraglutide on a different mouse strain—C3 mice appeared upregulated in normal glucose concentration were normoglycemic and free of insulitis—was insulinotropic conditions and slightly downregulated in high glucose as expected. The NOD mice are also normoglycemic, but concentration conditions. they display islet leukocyte infiltration despite no signs of overt diabetes having been observed at 8 weeks of age. The Liraglutide Alters Membrane Molecule observation of spontaneous insulitis is restricted to mice with Expression in NIT-1 Cell Line a diabetogenic genetic background, specifically NOD and NOR NIT-1 cell viability remained unaffected by liraglutide (10, 100, strains, and CD-1 outbred mice (31). The C3 mice display non- and 1,000 nM) at both 12 and 48 h (Figure 6A) but showed a diabetogenic genetic background, do not develop spontaneously Frontiers in Endocrinology | www.frontiersin.org 7 May 2020 | Volume 11 | Article 258 Villalba et al. Drug Repositioning and Type 1 Diabetes FIGURE 3 | Assessment of the effect of liraglutide in the induction of bihormonal insulin+ glucagon+ cells. Triple immunofluorescence staining of 5-µm cryostat pancreatic sections for α-cells (glucagon, green) and β-cells (insulin, red) in immunodeficient NOD.SCID-IL2Rγ−/− mice rendered diabetic by streptozotocin and treated with phosphate-buffered saline (T1D), or with liraglutide for 48 h (Lira 48 h), for 7–15 days (Lira) and after withdrawal of liraglutide at day 30 (post-Lira). Normoglycemic mice were included as control (sham). The white arrows in Lira 48 h staining depict bihormonal cells. The nuclei in all pictures were stained with Hoechst (blue). The scale bar indicates 5 µm. Frontiers in Endocrinology | www.frontiersin.org 8 May 2020 | Volume 11 | Article 258 Villalba et al. Drug Repositioning and Type 1 Diabetes FIGURE 4 | Assessment of the effect of liraglutide in the induction of insulin-expressing ducts. (A) Staining for β-cells in the pancreas of immunodeficient NOD.SCID-IL2Rγ−/− (NSG) mice rendered diabetic by streptozotocin and treated with phosphate-buffered saline (PBS; T1D), or with liraglutide for 48 h (Lira 48 h), for 7–15 days (Lira) and after the withdrawal of liraglutide at day 30 (post-Lira). The Lira and post-Lira groups show the presence of neo-islets emerging from the ducts. Normoglycemic mice were included as control (sham). (B) Staining for ductal cells (CK19, green) and β-cells (insulin, red) in the same groups than in (A). The white arrow depicts a CK19+ duct that is positive for insulin expression. (C) Staining for glucagon (green) and insulin (red) in ductal cells of animals treated with (Continued) Frontiers in Endocrinology | www.frontiersin.org 9 May 2020 | Volume 11 | Article 258 Villalba et al. Drug Repositioning and Type 1 Diabetes FIGURE 4 | liraglutide for 7–15 days (Lira). (D) Staining for Pdx1 (green) and insulin (red) in the ductal part of diabetic NSG mice treated with PBS (T1D) or liraglutide for 7–15 days (Lira). The white arrows depict positivity for both Pdx1 and insulin in the ducts. The nuclei in all pictures are stained with Hoechst (blue). The scale bar in all pictures indicates 50 µm. FIGURE 5 | Tracking of AF750-labeled liraglutide in prediabetic NOD mice and determination of gene expression in islets of Langerhans. (A) Histogram of ex vivo fluorescent signal relative to the grams of tissue of several organs from NOD mice at 15 min and (B) at 60 min after a s.c. injection of AF750-labeled liraglutide. The results are mean ± SEM of three independent experiments. (C,D) Ratio of relative gene expression of the selected genes in isolated islets of Langerhans after 48 h of 1,000-nM liraglutide treatment (n ≥ 3) in relation to basal conditions analyzed by qRT-PCR in conditions of normoglycemia and hyperglycemia. Gene expression was normalized to Gapdh. The bars show the mean ± SD of the log2 of fold change using basal transcription as standard value. autoimmune diabetes, and may have diabetes that can be induced speculate that the continuous presence of liraglutide is required by STZ treatment (41). In NOD mice, insulitis promotes an for the maintenance of islet β-cell mass by promoting the main inflammatory microenvironment, thus causing β-cell stress and mechanism responsible for the improvement of blood glucose metabolic abnormalities (42, 43). Liraglutide administered to levels in treated mice, at least during the first 30 days of treatment NOD mice may also impair insulin secretion at this stage because stages. Elucidating the mechanisms of action of liraglutide in of the inflammatory environment in this strain. diabetic mice with diabetogenic background will contribute to Intriguingly, the effect was transient as previously described the design of β-cell regenerative strategies. (30). This effect correlated with the transitory increase in the β- To further explore the regeneration mechanism, we then cell mass, confirming that insulin secretion induced by liraglutide searched for processes described in any of the pathways (11) is not the only event that contributes to the restoration of that liraglutide was predicted to act on. Suggesting normoglycemia. This increase is not due to β-cell hyperplasia but transdifferentiation, transient bihormonal cells were found by an increase in the number of insulin+ cells and the formation within the islets, a feature of α-cell to β-cell conversion detected of neo-islets emerging from ducts. Thus, it is reasonable to after β-cell loss (44). Because glycemia normalization occurs Frontiers in Endocrinology | www.frontiersin.org 10 May 2020 | Volume 11 | Article 258 Villalba et al. Drug Repositioning and Type 1 Diabetes FIGURE 6 | Effect of liraglutide in the NIT-1 β-cell line. (A) Percentage of viable NIT-1 cells (annexin V PE− , 7aad− ) after 12, 24, and 48 h of co-culture with increasing liraglutide concentrations (10, 100, and 1,000 nM) and control condition (*p < 0.05, Mann–Whitney test). (B) Median of fluorescence intensity for CD49b, CD44, CD14, and MHC class I surface expression (*p < 0.05, Mann–Whitney test). almost immediately after the first injections, newly formed β-cell, it is logical to speculate that these cells correspond to insulin+ cells with an α-cell origin could contribute to glucose emergent β-cells. In mice, α-cells can become insulin-expressing homeostasis. Our results fit well with the recent demonstration cells after β-cell ablation, thus promoting diabetes recovery. that exendin-4, another aGLP-1, also induces bihormonal cells Recent results confirm that human α-cells are able to secrete as a consequence of transdifferentiation (16). Other drugs that insulin and reverse diabetes, but surprisingly, these cells express induce this type of transdifferentiation also cause extreme islet α-cell markers (45). Nevertheless, to prove that liraglutide hyperplasia during the early stages of administration (3), but this induces a real conversion from glucagon-producing cells into fact has not been observed with liraglutide, thus suggesting that insulin-producing β-cells—as reported with GLP-1 peptide (16), liraglutide might act through different pathways. Bihormonal it would be of interest to perform lineage tracing experiments cells detected at 48 h after administration agreed with an increase with transgenic models. in both insulin and glucagon transcripts in islets treated in Furthermore, during the treatment and even after removal vitro with liraglutide. Although hormone colocalization per of liraglutide, insulin+ bodies were found to be emerging se is not a demonstration of a real conversion from α-cell to from CK19+ ductal cells that expressed insulin. Whether Frontiers in Endocrinology | www.frontiersin.org 11 May 2020 | Volume 11 | Article 258 Villalba et al. Drug Repositioning and Type 1 Diabetes FIGURE 7 | Proposed model of Lira-induced β-cell regeneration. The islets displaying their physiological conditions are depleted from most β-cells when treated with streptozotocin, while the ductal part of the pancreas remains unaffected. We propose that liraglutide can induce a regeneration process in two steps. The first one, triggered at least 48 h after the treatment, consists in the transdifferentiation from mature α-cells into new insulin-expressing β-like cells. During the second step, while β-cell mass seems to be partially recovered, neo-islets formed by insulin-expressing β-like cells from CK19+ cells in the ductal part of the pancreas arise from the ducts. these neo-islets emerged from specific multipotent progenitors despite the fact that the percentage of bihormonal cells at remains to be explored, but it has been proposed that early 48 h is almost 11% of total insulin+ cells, these data reflect transdifferentiation promotes later neogenesis from ducts a specific stage, and it is still unknown how many insulin+ (3, 46). Further experiments are required in order to confirm cells are arising from transdifferentiation throughout the this, given that the cell population identified in the ductal treatment. Indeed it is reasonable to speculate that ductal part are Neurogenin3+ progenitors that underwent epithelial- insulin+ cells are participating in insulin secretion, but it to-mesenchymal transition to differentiate into β-like cells remains to be investigated if they are glucose responsive. (9, 47). Neo-islets have also been reported in human pancreas Finally, both a decrease in the percentage of budding (48), including those from patients with type 1 diabetes (49), islets and the fact that liraglutide is an insulinotropic but this is the first demonstration of insulin+ cells with a agent may explain the transient effect in the glycemia ductal origin caused by liraglutide. Pdx1, a mature β-cell after withdrawal. and endocrine progenitor marker, also depicted a ductal Taking these observations together, we propose a two-step population. Because GLP-1 receptor is also expressed in ductal simultaneous mechanism of β-cell regeneration (Figure 7): first, cells (50), insulin expression could be induced by liraglutide an early, acute, and transient transdifferentiation mechanism in these cells. These results suggest that the role of this drug from α-cell to β-cell (3, 38) and, second, an early and in ameliorating hyperglycemia follows a mechanism that is permanent neogenic process of insulin+ cells from the ducts. non-exclusive to β-cells and has other players involved, such After withdrawal, the first mechanism seems to be suppressed— as islet insulin+ glucagon+ cells and ductal insulin-expressing probably after the loss of insulin+ cell identity—while the second cells. The herein reported findings go beyond the previously one persists. reported beneficial effect of liraglutide in β-cell function, In conclusion, liraglutide, a repurposed compound, both in vivo (39) and in vitro in terms of insulin secretion ameliorates hyperglycemia in experimental type 1 diabetes. (51). However, neogenesis from ducts was not reported in a Our results point to β-cell replacement, including previous study due to differences in dosage, disease induction, transdifferentiation and neogenesis, as aiding factor. Liraglutide and especially lineage tracing issues (39). It is difficult to could be a candidate to restore β-cell mass in combined assess which insulin-producing cell types are quantitatively therapies, together with an immunomodulatory strategy to contributing to the improvement of hyperglycemia. First, arrest autoimmunity. Frontiers in Endocrinology | www.frontiersin.org 12 May 2020 | Volume 11 | Article 258 Villalba et al. Drug Repositioning and Type 1 Diabetes DATA AVAILABILITY STATEMENT to the discussion. All the authors revised the manuscript and gave final approval of the version to be published. All datasets generated for this study are included in the article/ Supplementary Material. FUNDING This work has been funded by Fundació La Marató de TV3 ETHICS STATEMENT (project 201632_10). CIBER of Diabetes and Associated Metabolic Diseases (CIBERDEM) is an initiative from This study was carried out in strict accordance with the Instituto de Salud Carlos III, Spanish Ministry of Science recommendations in the Guide for the Care and Use of and Innovation. SR-F was supported by the Generalitat de Laboratory Animals of the Generalitat de Catalunya, Catalan Catalunya (AGAUR grant). Government. The protocol was approved by the Committee on the Ethics of Animal Experiments of the Germans Trias i Pujol Research Institute (Permit DAAM 9521) and has followed the ACKNOWLEDGMENTS principles outlined in the Declaration of Helsinki for animal We are grateful to Mr. M. Fernandez and Mr. G. Requena for experimental investigation. their support in flow cytometry. We acknowledge Dr. M. P. Armengol, from the Histopathology and Electron Microscopy AUTHOR CONTRIBUTIONS Facility, for her help. Special thanks to Ms. Deborah Culell- Young for English grammar assistance. AV, FV, IP-A, and MV-P designed the experiments. MC conducted the drug reposition analysis. AV, DP-B, SR-F, and SUPPLEMENTARY MATERIAL R-MA performed the experiments in mice. AV, DP-B, SR-F, and LG-M carried out the in vitro experiments. MC-S bound the drug The Supplementary Material for this article can be found to the fluorochrome. JV contributed to islet isolation. AV and online at: https://www.frontiersin.org/articles/10.3389/fendo. MV-P wrote the manuscript. SR-F, IP-A, EA, and JV contributed 2020.00258/full#supplementary-material REFERENCES 11. Jacobsen L V., Flint A, Olsen AK, Ingwersen SH. Liraglutide in type 2 diabetes mellitus: clinical pharmacokinetics and pharmacodynamics. 1. Xu X, D’Hoker J, Stangé G, Bonné S, De Leu N, Xiao X, et al. β cells can Clin Pharmacokinet. (2016) 55:657–72. doi: 10.1007/s40262-015- be generated from endogenous progenitors in injured adult mouse pancreas. 0343-6 Cell. (2008) 132:197–207. doi: 10.1016/j.cell.2007.12.015 12. Iepsen EW, Torekov SS, Holst JJ. Liraglutide for type 2 diabetes and 2. Li J, Casteels T, Frogne T, Ingvorsen c, Honoré C, et al. Artemisinins target obesity: a 2015 update. Expert Rev Cardiovasc Ther. (2015) 13:753–67. GABA A receptor signaling and impair α cell identity. Cell. (2017) 168:86–100. doi: 10.1586/14779072.2015.1054810 doi: 10.1016/j.cell.2016.11.010 13. Anholm C, Kumarathurai P, Pedersen LR, Nielsen OW, Kristiansen OP, 3. Ben-othman N, Vieira A, Courtney M, Record F, Gjernes E, et al. Long-term Fenger M, et al. Liraglutide effects on beta-cell, insulin sensitivity and GABA administration induces alpha cell-mediated βeta like cell neogenesis. glucose effectiveness in patients with stable coronary artery disease and Cell. (2017) 168:73–85. doi: 10.1016/j.cell.2016.11.002 newly diagnosed type 2 diabetes. DiabetesObes Metab. (2017) 19:850–7. 4. Dadheech N, Garrel D, Buteau J. Evidence of unrestrained beta- doi: 10.1111/dom.12891 cell proliferation and neogenesis in a patient with hyperinsulinemic 14. Jinnouchi H, Sugiyama S, Yoshida A, Hieshima K, Kurinami N, Suzuki T, et al. hypoglycaemia after gastric bypass surgery. Islets. (2018) 10:213–20. Liraglutide, a glucagon-like peptide-1 analog, increased insulin sensitivity doi: 10.1080/19382014.2018.1513748 assessed by hyperinsulinemic-euglycemic clamp examination in patients with 5. Lotfi Shahreza M, Ghadiri N, Mousavi SR, Varshosaz J, Green JR. A review uncontrolled type 2 diabetes mellitus. J Diabetes Res. (2015) 2015:706416. of network-based approaches to drug repositioning. Brief Bioinform. (2018) doi: 10.1155/2015/706416 19:878–92. doi: 10.1093/bib/bbx017 15. Zhang Z, Hu Y, Xu N, Zhou W, Yang L, Chen R, et al. A New Way for Beta 6. Pushpakom S, Iorio F, Eyers PA, Escott KJ, Hopper S, Wells A, et al. Cell Neogenesis: Transdifferentiation from Alpha Cells Induced by Glucagon- Drug repurposing: progress, challenges and recommendations. Nat Rev Drug Like Peptide 1. J Diabetes Res. (2019) 2019:2583047. doi: 10.1155/2019/25 Discov. (2018) 18:41–58. doi: 10.1038/nrd.2018.168 83047 7. Sleire L, Førde HE, Netland IA, Leiss L, Skeie BS, Enger PØ. Drug repurposing 16. Lee Y-S, Lee C, Choung J-S, Jung H-S, Jun H-S. Glucagon-like peptide 1 in cancer. Pharmacol Res. (2017) 124:74–91. doi: 10.1016/j.phrs.2017. increases β-cell regeneration by promoting α- to β-cell transdifferentiation. 07.013 Diabetes. (2018) 67:2601–14. doi: 10.2337/db18-0155 8. Kumar S, Chowdhury S, Kumar S. In silico repurposing of 17. Jorba G, Aguirre-Plans J, Junet V, Segú-Vergés C, Ruiz JL, Pujol A, et al. antipsychotic drugs for Alzheimer’s disease. BMC Neurosci. (2017) 18:76. In-silico simulated prototype-patients using TPMS technology to study a doi: 10.1186/s12868-017-0394-8 potential adverse effect of sacubitril and valsartan. PLoS ONE. (2020) 9. Al-Hasani K, Pfeifer A, Courtney M, Ben-Othman N, Gjernes E, Vieira 15:e0228926. doi: 10.1371/journal.pone.0228926 A, et al. Adult duct-lining cells can reprogram into β-like cells able to 18. Romeo-Guitart D, Forés J, Herrando-Grabulosa M, Valls R, Leiva-Rodríguez counter repeated cycles of toxin-induced diabetes. Dev Cell. (2013) 26:86–100. T, Galea E, et al. Neuroprotective drug for nerve trauma revealed using doi: 10.1016/j.devcel.2013.05.018 artificial intelligence. Sci Rep. (2018) 8:1879. doi: 10.1038/s41598-018- 10. Dor Y, Brown J, Martinez OI, Melton DA. Adult pancreatic beta-cells are 19767-3 formed by self-duplication rather than stem-cell differentiation. Nature. 19. Herrando-Grabulosa M, Mulet R, Pujol A, Mas JM, Navarro X, Aloy P, (2004) 429:41–6. doi: 10.1038/nature02520 et al. Novel neuroprotective multicomponent therapy for amyotrophic lateral Frontiers in Endocrinology | www.frontiersin.org 13 May 2020 | Volume 11 | Article 258 Villalba et al. Drug Repositioning and Type 1 Diabetes sclerosis designed by networked systems. PLoS ONE. (2016) 11:e0147626. with type 1 diabetes: the adjunct two randomized trial. Diabetes Care. (2016) doi: 10.1371/journal.pone.0147626 39:1693–701. doi: 10.2337/dc16-0690 20. Kanehisa M, Goto S. KEGG: kyoto encyclopedia of genes and genomes. 39. Tamura K, Minami K, Kudo M, Iemoto K, Takahashi H. Liraglutide improves Nucleic Acids Res. (2000) 28:27–30. doi: 10.1093/nar/28.1.27 pancreatic beta cell mass and function in alloxan-induced diabetic mice. PLoS 21. Kanehisa M, Furumichi M, Tanabe M, Sato Y, Morishima K. KEGG: new ONE. (2015) 1:e0126003. doi: 10.1371/journal.pone.0126003 perspectives on genomes, pathways, diseases and drugs. Nucleic Acids Res. 40. Bock T, Pakkenberg B, Buschard K. Genetic background determines the (2017) 45:D353–61. doi: 10.1093/nar/gkw1092 size and structure of the endocrine pancreas. Diabetes. (2005) 54:133–7. 22. Croft D, Mundo AF, Haw R, Milacic M, Weiser J, Wu G, et al. The doi: 10.2337/diabetes.54.1.133 reactome pathway knowledgebase. Nucleic Acids Res. (2014) 42:D472–7. 41. Leiter EH Schile A. Genetic and pharmacologic models for doi: 10.1093/nar/gkt1102 type 1 diabetes. Curr Protoc Mouse Biol. (2013) 3:9–19. 23. Orchard S, Ammari M, Aranda B, et al. The MIntAct project–IntAct as a doi: 10.1002/9780470942390.mo120154 common curation platform for 11 molecular interaction databases. Nucleic 42. Strandell E, Sandler S, Boitard C Eizirik DL. Role of infiltrating T cells for Acids Res. (2014) 42:D358–63. doi: 10.1093/nar/gkt1115 impaired glucose metabolism in pancreatic islets isolated from non-obese 24. Salwinski L, Miller CS, Smith AJ, Pettit FK, Bowie JU, Eisenberg D. The diabetic mice. Diabetologia. (1992) 35:924–31. doi: 10.1007/BF00401420 database of interacting proteins: 2004 update. Nucleic Acids Res. (2004) 43. Eizirik DL, Colli ML Ortis F. The role of inflammation in insulitis and 32:D449–51. doi: 10.1093/nar/gkh086 beta-cell loss in type 1 diabetes. Nat Rev Endocrinol. (2009) 5:219–26. 25. Keshava Prasad TS, Goel R, Kandasamy K, Keerthikumar S, Kumar S, doi: 10.1038/nrendo.2009.21 Mathivanan S, et al. Human protein reference database - 2009 update. Nucleic 44. Thorel F, Népote V, Avril I, Kohno K, Desgraz R, Chera S, et al. Conversion Acids Res. (2009) 37(Suppl. 1):767–72. doi: 10.1093/nar/gkn892 of adult pancreatic α-cells to β-cells after extreme β-cell loss. Nature. (2010) 26. Han H, Shim H, Shin D, Shim JE, Ko Y, Shin J, et al. TRRUST: a reference 464:1149–54. doi: 10.1038/nature08894 database of human transcriptional regulatory interactions. Sci Rep. (2015) 45. Furuyama K, Chera S, van Gurp L, Oropeza D, Ghila L, Damond N, et al. 5:11432. doi: 10.1038/srep11432 Diabetes relief in mice by glucose-sensing insulin-secreting human α-cells. 27. Pache RA, Aloy P. Incorporating high-throughput proteomics experiments Nature. (2019) 567:43–8. doi: 10.1038/s41586-019-0942-8 into structural biology pipelines: identification of the low-hanging fruits. 46. Druelle N, Vieira A, Shabro A, Courtney M, Mondin M, Rekima S, Proteomics. (2008) 8:1959–64. doi: 10.1002/pmic.200700966 et al. Ectopic expression of Pax4 in pancreatic δ cells results in β - 28. Wishart DS, Knox C, Guo AC, Cheng D, Shrivastava S, Tzur D, et al. like cell neogenesis. J Cell Biol. (2017) 216:449–462. doi: 10.1083/jcb.2017 DrugBank: a knowledgebase for drugs, drug actions and drug targets. Nucleic 04044 Acids Res. (2008) 36:D901–6. doi: 10.1093/nar/gkm958 47. Bru-Tari E, Cobo-Vuilleumier N, Alonso-Magdalena P, Dos Santos RS, 29. Wishart DS, Feunang YD, Guo AC, Lo EJ, Marcu A, Grant JR, et al. DrugBank Marroqui L, Nadal A, et al. Pancreatic alpha-cell mass in the early-onset and 5.0: a major update to the DrugBank database for 2018. Nucleic Acids Res. advanced stage of a mouse model of experimental autoimmune diabetes. Sci (2018) 46:D1074–82. doi: 10.1093/nar/gkx1037 Rep. (2019) 9:9515. doi: 10.1038/s41598-019-45853-1 30. Rydén AK, Perdue NR, Pagni PP, Gibson CB, Ratliff SS, Kirk RK, et al. Anti- 48. Bogdani M, Lefebvre V, Buelens N, Bock T, Veld PI, Pipeleers D. IL-21 monoclonal antibody combined with liraglutide effectively reverses Formation of insulin-positive cells in implants of human pancreatic established hyperglycaemia in mouse models of type 1 diabetes. J Autoimmun. duct cell preparations from young donors. Diabetologia. (2003) 46:830–8. (2017) 84:65–74. doi: 10.1016/j.jaut.2017.07.006 doi: 10.1007/s00125-003-1118-4 31. Alba A, Puertas MC, Carrillo J, Planas R, Ampudia R, Pastor X, et al. IFN beta 49. Somoza N, Vargas F, Roura-Mir C, Vives-Pi M, Fernández-Figueras MT, accelerates autoimmune type 1 diabetes in nonobese diabetic mice and breaks Ariza A, et al. Pancreas in recent onset insulin-dependent diabetes mellitus. the tolerance to beta cells in nondiabetes-prone mice. J Immunol. (2004) J Immunol. (1994) 153:1360–77. 173:6667–75. doi: 10.4049/jimmunol.173.11.6667 50. Xu G, Stoffers DA, Habener JF, Bonner-Weir S. Exendin-4 stimulates both 32. Schindelin J, Arganda-Carreras I, Frise E, Kaynig V, Longair M, Pietzsch T, beta-cell replication and neogenesis, resulting in increased beta-cell mass et al. Fiji: an open-source platform for biological-image analysis. Nat Methods. and improved glucose tolerance in diabetic rats. Diabetes. (1999) 48:2270–6. (2012) 9:676–82. doi: 10.1038/nmeth.2019 doi: 10.2337/diabetes.48.12.2270 33. Chintinne M, Stangé G, Denys B, In ’T Veld P, Hellemans K, Pipeleers- 51. Xu X, Chen J, Hu L, Liang M, Wang X, Feng S, et al. Liraglutide regulates Marichal M, et al. Contribution of postnatally formed small beta cell the viability of pancreatic α-cells and pancreatic β-cells through cAMP-PKA aggregates to functional beta cell mass in adult rat pancreas. Diabetologia. signal pathway. Life Sci. (2018) 195:87–94. doi: 10.1016/j.lfs.2017.12.012 (2010) 53:2380–8. doi: 10.1007/s00125-010-1851-4 34. Chintinne M, Stange G, Denys B, Ling Z, In P, Pipeleers D. Beta cell count Conflict of Interest: MC-S and MV-P are co-founders of Ahead Therapeutics SL, instead of beta cell mass to assess and localize growth in beta cell population which aims at the clinical translation of immunotherapies for the treatment of following pancreatic duct ligation in mice. PLoS ONE. (2012) 7:e43959. autoimmune diseases. MC was employed by the company Anaxomics Biotech SL. doi: 10.1371/journal.pone.0043959 35. Perna-Barrull D, Rodriguez-Fernandez S, Pujol-Autonell I, Gieras A, The remaining authors declare that the research was conducted in the absence of Ampudia-Carrasco RM, Villalba A, et al. Prenatal betamethasone interferes any commercial or financial relationships that could be construed as a potential with immune system development and alters target cells in autoimmune conflict of interest. diabetes. Sci Rep. (2019) 9:1235. doi: 10.1038/s41598-018-37878-9 36. Hamaguchi K, Gaskins HR, Leiter EH. NIT-1, a pancreatic beta-cell line Copyright © 2020 Villalba, Rodriguez-Fernandez, Perna-Barrull, Ampudia, Gomez- established from a transgenic NOD/Lt mouse. Diabetes. (1991) 40:842–9. Muñoz, Pujol-Autonell, Aguilera, Coma, Cano-Sarabia, Vázquez, Verdaguer and doi: 10.2337/diabetes.40.7.842 Vives-Pi. This is an open-access article distributed under the terms of the Creative 37. Livak KJ, Schmittgen TD. Analysis of relative gene expression data using real- Commons Attribution License (CC BY). The use, distribution or reproduction in time quantitative PCR and the 2[-Delta Delta C(T)]. MethodMethods. (2001) other forums is permitted, provided the original author(s) and the copyright owner(s) 25:402–8. doi: 10.1006/meth.2001.1262 are credited and that the original publication in this journal is cited, in accordance 38. Ahren B, Hirsch IB, Pieber TR, Mathieu C, Gomez-Peralta F, Hansen TK, et al. with accepted academic practice. No use, distribution or reproduction is permitted Efficacy and safety of liraglutide added to capped insulin treatment in subjects which does not comply with these terms. Frontiers in Endocrinology | www.frontiersin.org 14 May 2020 | Volume 11 | Article 258