Gene Therapy (2003) 10, 1712–1720 & 2003 Nature Publishing Group All rights reserved 0969-7128/03 $25.00 www.nature.com/gt RESEARCH ARTICLE Engineering physiologically regulated in non-b cells by expressing -like 1

LWu1, W Nicholson1, C-Y Wu1,MXu1, A McGaha1, M Shiota2 and AC Powers1,2,3 1Division of , , and , Department of Medicine, Vanderbilt University Medical Center, Nashville, TN, USA; 2Department of Molecular Physiology and Biophysics, Vanderbilt University Medical Center, Nashville, TN, USA; and 3VA Tennessee Valley Healthcare System, Nashville, TN, USA

Glucagon-like peptide 1 (GLP-1) is released from neuroen- . When both the GLP-1 receptor and insulin docrine cells in the intestine in the postprandial state and were introduced, GLP-1 stimulated cosecretion of human augments -stimulated insulin secretion from pan- insulin and endogenous pituitary hormones. GLP-1 was creatic b cells. To develop non-b cells that exhibit physiolo- similar in potency to the hypothalamic-releasing hormones gically regulated insulin secretion, we coexpressed the GLP- and stimulated secretion in a dose-dependent 1 receptor and human insulin in primary pituitary cells fashion. In contrast to pancreatic b cells, the hormone- using adenovirus-mediated transfer. The transduced releasing effect of GLP-1 on transduced pituitary cells was cells were analyzed in a perifusion system and after not dependent on the concentration of extracellular glucose. transplantation into mice. Normal pituitary cells do not After transplantation of pituitary cells coexpressing human express the GLP-1 receptor as shown by the absence of insulin and GLP-1 receptor into mice, enteral glucose GLP-1 receptor mRNA and the inability of GLP-1 to stimulate stimulated insulin secretion. These results demonstrate a pituitary hormone secretion. Following transduction with an new approach to engineer physiologically regulated insulin adenovirus carrying the GLP-1 receptor cDNA, the pituitary secretion by non-b cells. cells expressed functional GLP-1 receptors as reflected by Gene Therapy (2003) 10, 1712–1720. doi:10.1038/ the ability of GLP-1 to stimulate secretion of pituitary sj.gt.3302055

Keywords: insulin secretion; adenoviral vector; ; pituitary cells

Introduction glucose, and the insulin-releasing effect of GLP-1 on b cells requires the presence of elevated extracellular The fundamental characteristic of the pancreatic b cells is glucose.3,5,12–14 Although GLP-1 signaling interacts with their ability to secrete insulin in response to physiologic the signaling of glucose metabolism in b cells, the fact stimuli. Oral glucose stimulates a much greater insulin that GLP-1 stimulates insulin secretion from b cells in secretory response than intravenous glucose administra- suggests that it can also exert its effect tion because of factors () secreted by neuroendo- independent of the signaling of glucose metabolism.15–18 crine cells within the small and large intestine.1–4 Engineering regulated insulin secretion in both b and Glucagon-like peptide 1 (GLP-1) and glucose-dependent non-b cells has mostly focused on recreating the signal insulin-releasing polypeptide (GIP) are responsible for transduction pathway of glucose.19–23 Since the mechan- the vast majority of this ‘’ effect.3–5 GLP-1 isms in this pathway itself are not completely elucidated, augments glucose-stimulated insulin secretion (GSIS) engineering physiologically regulated insulin secretion has from b cells by binding to a cell-surface receptor, which been difficult. In this report, we used an alternative belongs to the incretin/ family that is approach to engineer regulated insulin secretion from non- coupled to heterotrimeric G . Binding of GLP-1 b cells by introducing insulin and the GLP-1 receptor into to its receptor stimulates cAMP formation and a rise in neuroendocrine pituitary cells. The rationale behind this intracellular Ca2+.6–10 GSIS from pancreatic b cells is the engineering strategy is that regulated hormone secretion is result of signals generated during glucose metabolism, a phenotype of a variety of endocrine and neuroendocrine which changes proportionally to elevations in the cells, including b cells and pituitary cells, and that these glucose level.11 The rise of incretins in the circulation cells utilize a conserved set of proteins and mechanisms. coincides with the postprandial increases in blood For example, these cell types contain secretory granules that not only serve as a reservoir for hormones, but also interact with pathways during the Correspondence: Dr AC Powers, Division of Diabetes, Endocrinology, and 24–27 Metabolism, 715 PRB, Vanderbilt University, Nashville, TN 37232, USA final steps of exocytosis. The distal events in insulin E-mail: [email protected] secretion utilize a set of proteins that are endocrine and Received 17 September 2002; accepted 17 February 2003 neuroendocrine cell specific, but not b-cell specific. To Regulated insulin secretion from non-b cells LWuet al 1713 investigate whether neuroendocrine cells are potentially suitable as a foundation cell to engineer physiologically regulated insulin secretion, both insulin and the GLP-1 receptor were introduced into primary rat pituitary cells through adenovirus-mediated gene transfer. Our results showed that insulin and GLP-1 receptor-expressing cells secreted human insulin in response to levels of GLP-1 that were similar to physiologic concentrations. When trans- planted into mice, these cells secreted human insulin in response to the endogenous GLP-1 released after an oral glucose load. These data open the possibility of engineer- ing non-b cells to secrete insulin in response to physiologic stimuli other than glucose.

Results Expression of insulin and GLP-1 receptor in rat pituitary cells We first examined if there was endogenous expression of insulin or GLP-1 receptor in the rat pituitary. Rat insulin was not detected in the culture supernatant and cell extracts of untransduced primary rat pituitary cells (data not shown). GLP-1 receptor transcripts were detected in rat pancreatic islets and hypothalami (both of which are known to express the GLP-1 receptor), but not in pituitaries (Figure 1a). A prior report indicated that GLP-1 released -stimulating hormone (TSH) from isolated rat pituitary cells in static culture.28 However, in our cell perifusion system, GLP-1 did not stimulate secretion of TSH or other pituitary hormones (2 Â 106 of isolated pituitary cells perifused with 10 nM of GLP-1; data not shown). Together, our results demonstrate that Figure 1 Expression of GLP-1 receptor and insulin in rat pituitary cells. (a) Total RNA was extracted from normal rat pancreatic islets, neither insulin nor the GLP-1 receptor is expressed in the hypothalami, and pituitaries. The GLP-1 receptor mRNA was detected normal rat pituitary. by RT-PCR that generated a band of 190 bp. Molecular weight standards Human and the human GLP-1 receptor are shown on the left. (b) Expression of human insulin in the transduced were expressed in primary rat pituitary cells using pituitary cells was examined by intracellular immunostaining, followed by adenovirus-mediated gene transfer. The recombinant flow cytometric analysis. This experiment was repeated three times and a adenovirus (AdINS/GLP-1R) carries a gene cassette representative graph is shown. containing both human proinsulin and human GLP-1 receptor cDNA and, thus, cells expressing insulin also express the GLP-1 receptor. Multiplicity of infection (MOI) was used to determine the amount of viral stock particular hormone in response to its respective hy- used and was selected by examining the insulin pothalamic-releasing hormone.29,30 Corticotrophs secrete expression and cell survival as a function of increasing adrenocorticotropin (ACTH) after stimulation with cor- MOI. After transduction with either AdINS/GLP-1R or ticotropin-releasing hormone (CRH); - the control adenovirus (AdLacZ), human insulin in the releasing hormone (GHRH) acts on somatotrophs and pituitary cells was detected by intracellular immuno- releases growth hormone (GH); gonadotrophs secrete using a human insulin-specific antibody, and /follicle-stimulating hormone (LH/ analyzed by flow cytometry. The results showed that the FSH) when they are stimulated by -releas- majority of AdINS/GLP-1R-treated cells expressed hu- ing hormone (GnRH); both thyrotrophs and lactotrophs man insulin when an MOI of 50 was used (Figure 1b). respond to thyrotropin-releasing hormone (TRH) and The adenoviral transduced cells were viable as examined secrete TSH and (PRL), respectively. It was by propidium iodide staining (data not shown). More demonstrated previously that recombinant adenovirus importantly, these cells remained metabolically active could mediate gene transfer into each of these endocrine and retained normal secretory capacity. Adenoviral- cell populations.31 We therefore examined if all pituitary transduced pituitary cells responded to hypothalamic- cell types could synthesize and secrete insulin when the releasing hormones or GLP-1 at levels that were insulin cDNA was introduced. AdINS/GLP-1R-treated comparable to wild-type pituitary cells in the perifusion pituitary cells were stimulated consecutively with the system (see below). An MOI of 50 for AdINS/GLP-1R hypothalamic-releasing hormones CRH, LHRH, GHRH, and AdLacZ was used for subsequent experiments. and TRH in the perifusion system; human insulin and pituitary hormones were measured in the perifusate. By Insulin secretion by transduced pituitary cells comparing the occurrence of insulin release with the The anterior rat pituitary is comprised of five different secretion of endogenous pituitary hormones, the results endocrine cell populations, each of which releases a showed that pituitary cell types cosecreted insulin with

Gene Therapy Regulated insulin secretion from non-b cells LWuet al 1714 ACTH, LH, GH, PRL, and TSH (Figure 2a) and the corresponding integrated pituitary hormone responses integrated insulin responses to stimulation with CRH, were 1.12 (ACTH), 0.52 (LH), 138 (GH), 24 (PRL), and GnRH, GHRH, and TRH were 0.34, 0.26, 0.42, and 0.21 (TSH) pmol. The order of the pituitary hormone 4.7 pmol, respectively (representative experiment). The response, GH4PRL4ACTH4LH4TSH, parallels the reported order of daily production rates in vivo as well as the prevalence of cell types in the pituitary. In the pituitary of an adult male rat, the most abundant cell types are somatotrophs and lactotrophs (constituting 50 and 15%, respectively, of the cells).29,30 The hormone release profiles presented in Figure 2a are consistent with these data, showing that GH and PRL secretion levels were the highest found in the perifusate. Insulin released by the transduced cells in response to TRH far exceeded the small, but consistent, amounts released in response to the other releasing factors. Attempts to identify the primary cell type responsible for insulin secretion was hindered by the fact that both lactotrophs and thyrotrophs respond to TRH stimulation and dopminergic inhibition and by the putative nature of the prolactin-releasing/inhibiting factors. Since recombi- nant adenoviruses do not preferentially transduce thyrotrophs and lactotrophs,31 these results suggest that thyrotrophs and/or lactotrophs can secrete insulin in response to GLP-1 more efficiently than other pituitary cell types when the insulin cDNA is introduced. The radioimmunoassay (RIA) for human insulin used in this study detects mature and selected intermediate- processed human insulin (together referred as processed insulin), with o0.2% cross-reaction with human proin- sulin. To further investigate the ability of pituitary cells to properly process proinsulin and secrete mature insulin, we measured human proinsulin in the perifusate by -linked immunosorbent assay (ELISA) and compared the results to the levels of processed insulin. After stimulation of AdINS/GLP-1R-treated cells with TRH and LHRH, the majority of immunoreactive insulin was processed insulin and only a small portion was proinsulin (Figure 2b). Together, these results demon- strate that all anterior pituitary cells, when transduced with the insulin adenovirus, can synthesize proinsulin, properly process proinsulin, and secrete mature insulin upon stimulation.

Expression of GLP-1 receptor in endocrine pituitary cells confers secretory response to GLP-1 To determine if pituitary cells could also express functional GLP-1 receptors and couple this to the signal transduction pathway for hormone secretion, we ana- lyzed GLP-1-stimulated hormone secretion from the transduced cells in the perifusion system. The results demonstrated that GLP-1 dose-dependently released human insulin from AdINS/GLP-1R-treated, but not AdLacZ-treated, pituitary cells (Figure 3a, upper panel). When assayed in parallel, the levels of GH, ACTH, and PRL in the perifusate showed similar patterns (Figure 3a, lower panel). These results indicate that pituitary cells can reconstitute the GLP-1 signaling pathway when the Figure 2 Secretion of insulin and endogenous pituitary hormones by GLP-1 receptor is introduced into these cells. transduced and perifused pituitary cells in response to releasing factors. In In pancreatic b cells, the insulin-releasing effect of this and subsequent perifusions, each stimulus was present for 6 min and GLP-1 is dependent on the concentration of extracellular the experiments were repeated three times and a representative graph is glucose. GLP-1 enhances insulin secretion from isolated shown. (a) Secretion of human insulin and pituitary hormones by the transduced and perifused pituitary cells in response to releasing hormones. islets only when the extracellular glucose is raised 12–14 (b) Secretion of processed human insulin and proinsulin by the transduced to 48mM. In addition, studies have suggested and perifused pituitary cells in response to TRH and LHRH. that GLP-1 participates in maintaining b-cell glucose

Gene Therapy Regulated insulin secretion from non-b cells LWuet al 1715

Figure 3 Secretion of human insulin and pituitary hormones in response to GLP-1. (a) Secretion of human insulin and pituitary hormone by perifused AdINS/GLP-1R- or AdLacZ-transduced cells in response to GLP-1. (b) Secretion of growth hormone by wild-type and AdINS/GLP-1R-transduced cells in response to increasing concentrations of extracellular glucose. (c) Effect of glucose concentration on GLP-1 receptor signaling in transduced and perifused pituitary cells.

responsiveness.14,32,33 We therefore examined if (1) ing increasing concentrations of glucose. Figure 3b shows expression of GLP-1 receptor in pituitary cells could that neither wild-type nor transduced pituitary cells render these cells glucose responsive and (2) the increase GH secretion in response to increased extra- hormone-releasing effect of GLP-1 on the transduced cellular glucose concentrations. The levels of ACTH and pituitary cells was dependent on the extracellular PRL secretion were unaltered by changes in extracellular glucose concentration. For the first set of studies, wild- glucose concentrations as well (data not shown). The GH type or AdINS/GLP-1R-treated pituitary cells were secretory responses to GHRH in wild-type and trans- analyzed in the perifusion system with medium contain- duced cells were comparable and demonstrate that

Gene Therapy Regulated insulin secretion from non-b cells LWuet al 1716 adenoviral transduction does not affect hormone secretory capacity or cell health. The second set of experiments was performed by perifusing wild-type or AdINS/GLP-1R-treated cells with GLP-1 in the presence of increasing concentrations of glucose. The results showed that the hormone-releasing effect of GLP-1 on AdINS/GLP-1R-treated pituitary cells was not depen- dent on the concentration of extracellular glucose. Perifusion medium containing 10 nM of GLP-1 and 2.5 mM glucose released the same amount of insulin as the perifusion medium containing 10 nM of GLP-1 and 20 mM of glucose (Figure 3c). In contrast to b cells, the insulin response to GLP-1 remained monophasic in the transduced cells, even when the exposure to GLP-1 in 20 mM glucose was extended from 6 to 24 min (data not shown). The insulin secretory profiles from the insulin and GLP-1 receptor-expressing pituitary cells were similar in the presence of both low and high glucose (Figure 3a (upper panel) and c). Together, these results suggest that, when expressed in pituitary cells, the GLP-1 receptor utilizes a signal transduction pathway that is independent of signals generated during glucose meta- bolism. This appears to be different from the interaction of the GLP-1 signal transduction and glucose metabolism pathways in b cells.

Insulin and GLP-1 receptor-expressing pituitary cells secrete insulin in response to physiological levels of GLP-1 To evaluate the potential use of this engineering strategy for generating a b-cell surrogate, we performed two sets of experiments. First, we examined if insulin and GLP-1 receptor-expressing pituitary cells could secrete insulin in response to physiologic levels of GLP-1 when analyzed in the perifusion system. GLP-1 could release human insulin from the transduced pituitary cells at a concentration as low as 10 pM (Figure 4a). Since the Figure 4 Insulin and GLP-1 receptor-expressing cells secrete insulin in plasma GLP-1 in normal reaches 10 pM after an response to physiological levels of GLP-1. (a) Secretion of insulin by intraduodenal glucose load,34 these data suggested that AdINS/GLP-1R-treated and perifused cells in response to increasing levels transduced pituitary cells could respond to physiological of GLP-1 (10 pM–30 nM). The experiment was repeated three times and a levels of GLP-1 in vivo. To test this hypothesis, we representative graph is shown. (b) Human insulin levels, during oral glucose tolerance testing, in the plasma of mice transplanted with AdINS/ measured human insulin secretion from the transduced GLP-1R-treated cells (day 1 after transplantation). The percentage increase cells after transplantation into the intraperitoneal cavity in plasma insulin over the fasting level after glucose load is plotted. The of nonobese diabetic/severe combined immunodeficient results are mean7s.e. of four separate experiments. (NOD/SCID) mice. Unlike NOD animals, NOD/SCID mice do not develop autoimmune diabetes and are a murine model that readily accepts xenografts. Experi- 6 Table 1 Levels of blood glucose and processed human insulin in mental animals received 3 Â 10 of AdINS/GLP-1R- mice 1 day after receiving a transplant of AdINS/GLP-1R-treated treated pituitary cells, and control mice were trans- cells planted with same number of AdLacZ-treated pituitary cells. At 1 day after transplantation, the mice were Time (min) Glucose (mg/dl) Processed human insulin (pg/ml) subjected to oral glucose tolerance testing (OGTT). The levels of blood glucose and human insulin are summar- 08376 241728 ized in Table 1. The fasting levels of plasma human 15 >400 274764 7 insulin (241728 pg/ml) significantly increased to 30 >400 330 45* 60 224778 268729 330745 pg/ml 30 min after the intragastric glucose loading (Po0.05). The levels of human insulin began to *Significantly different from the value at 0 min (Po0.05). increase 15-min after glucose administration and re- Values are mean7s.e. of four independent experiments. turned to fasting level 60 min after the glucose challenge (Figure 4b). In the mice that were transplanted with AdLacZ-treated cells, the levels of blood glucose fol- lowed a similar pattern, but human insulin was not was significantly reduced during OGTT on day 3, and detected in the plasma (data not shown). Human insulin was undetectable on day 5 after transplantation (data not in the animals that received AdINS/GLP-1R-treated cells shown).

Gene Therapy Regulated insulin secretion from non-b cells LWuet al 1717 Discussion insulin in mouse pituitary cells resulted in insulin secretion and reversal of diabetes in NOD mice, a model To generate genetically engineered cells that secrete of . However, these cells did not sense the insulin in a regulated fashion as a b-cell surrogate, past ambient glucose and adjust insulin secretion accord- efforts have focused mainly on recreating the signal ingly.36,37 Recently, several studies have attempted to transduction pathway of glucose. A variety of ap- engineer GSIS from by expressing insulin proaches have been explored using transformed insu- under the control of glucose responsive promoters, and lin-secreting b cells or non-b cells engineered to secrete have demonstrated that glucose regulated insulin secre- insulin. In the category of cells of b-cell origin, Newgard tion both in vitro and in diabetic animal models.22,23 and colleagues introduced important for GSIS into However, because the expression of insulin is controlled rodent b cell lines and showed improved GSIS by these at the transcriptional level, both the rise of insulin and cells.19,20 Efrat et al35 used a transgenic approach to the clearance of this hormone in the circulation in conditionally transform b cells with a SV-40 oncogene response to glucose load are delayed by hours.22 whose expression is reversible. When the oncogene was Alterations in insulin levels over a time course of hours inactivated, the b cells no longer grew, but had normal rather than minutes is likely to lead to and GSIS. This suggests that controlled oncogenesis might be nonphysiologic insulin secretion. used to amplify human b cells in vitro, and then the b The current report outlines an alternative experimental cells transplanted after oncogene inactivation. Hui et al33 approach to investigate whether non-b cells can secrete recently introduced GLP-1 into a mouse b cell line under insulin in response to other physiologic insulin secreto- the control of rat insulin II and showed gogues such as GLP-1. The studies presented here improved glucose responsiveness in these cells. The utilized neuroendocrine pituitary cells based on the advantage of using cells of b-cell origin is that these cells following considerations: (1) pituitary cells possess a already express many b-cell-specific genes. However, regulated secretory pathway and are closely related to b challenges and questions about such technologies as a cells in expression. This minimizes the number of treatment for human diabetes remain. For example, introduced genes required to endow physiologically although transformed b cells may be efficacious in regulated insulin secretion in these cells; (2) primary treating diabetes in animals, there are concerns about neuroendocrine cells are nonmalignant and nontrans- transplantation of islet cells with any malignant potential formed. This overcomes the disadvantage of using b cell as a treatment for human diabetes. Furthermore, geneti- lines, and (3) studies have shown that insulin-expressing cally engineered islet cells transplanted into pituitary cells are not destroyed by a recurrence of the with type 1 diabetes will be quickly destroyed by a autoimmune process in type 1 diabetes.36 Our long-term recurrence of the autoimmune process that destroys the objective is to develop an engineering strategy that can recipient’s b cells initially or by an allograft immune retrieve primary cells from a donor, engineer the cells in reaction. Thus, such a strategy will also require some vitro, and return the cells to the donor for insulin delivery form of immunosuppression of the recipient or immu- in diabetes. We coexpressed the GLP-1 receptor and nomodulation of the transplant. insulin in primary rat pituitary cells and demonstrated A strategy to utilize non-b cells as the foundation cell that such cells responded to physiologic levels of the into which proteins important for physiologically regu- GLP-1 in vitro and in vivo. This approach bypasses the lated insulin secretion are introduced overcomes the signaling pathway of glucose and utilizes signal trans- obstacle of a recurrence of the autoimmune process in duction pathways that are already present in pituitary type 1 diabetes.36,37 Despite the ability of many cell types cells. GLP-1 signaling is coupled to adenylate cyclase in b to express an introduced insulin gene, physiologic cells and to C in non-b cells expressing regulation of insulin secretion presents a number of GLP-1 receptor.43 Both the adenylate cyclase/protein challenges. Non-b cells can be engineered to express kinase A and phospholipase C/protein kinase C path- insulin, but most cells (eg hepatocytes, myocytes, ways are expressed in many cell types, including fibroblasts) lack the ability to store and properly process pituitary cells. Presumably, GLP-1 utilizes either or insulin because they do not contain secretory granules both of these pathways in the pituitary cells studied with their accompanying hormone-processing . here. Our approach was designed to utilize the native Furthermore, these cells secrete insulin via the constitu- physiology of GLP-1 in that it simulates insulin secretion tive secretory pathway and thus will secrete some insulin from b cells only when the blood glucose level is elevated regardless of the extracellular glucose concentrations.38–40 during the postprandial state. Although our data Insulin secretion independent of the glucose concentra- indicate that the hormone-releasing effect of GLP-1 in tion is a serious limitation of such approaches with non-b non-b cells is not dependent on an elevated glucose, the cells. Thus, most non-b cells must be engineered to not fact that GLP-1 is usually released only after food intake only sense the extracellular glucose concentration, but would allow insulin secretion from GLP-1 receptor- also to express a large number of proteins required for expressing non-b cells only during the postprandial regulated secretion via secretory vesicles. In the effort to period. search for the cells that already possess regulated A limitation of using pituitary cells is that the secretory pathways as foundation cells, insulin and/or endogenous pituitary hormones are cosecreted with the proteins important for GSIS in b cells (GLUT 2 insulin and this might create undesired physiologic and ) have been intro- effects (increased ACTH or GH). Other neuroendocrine duced into pituitary cells. Expression of insulin and cell types might better serve as the foundation cells for GLUT 2 in the mouse pituitary cell line AtT-20 conferred this engineering approach. Furthermore, a number of GSIS, but the maximal response to glucose occurred at other issues such as the long-term survival of trans- subphysiological levels.41,42 Transgenic expression of planted cells, the site of transplantation, and the amount

Gene Therapy Regulated insulin secretion from non-b cells LWuet al 1718 of insulin produced by engineered cells will require concentration of glucose in the culture medium during additional work. the transduction period was 25 mM except for the In conclusion, the outlined studies expand the horizon experiments evaluating glucose dependence, in which beyond engineering cells to simply possess GSIS by the cells were transduced in medium containing 5 mM developing insulin-secreting cells that are responsive to glucose. After transduction, the cells were harvested by other physiologic regulators of insulin secretion such as trypsinization and used for the subsequent assays. GLP-1 or GIP. These studies also demonstrate the ability to regulate insulin secretion via a cell-surface receptor Analysis of GLP-1 receptor expression and nonglucose insulin in non-b cells. Total RNA was extracted from normal rat pancreatic This opens the potential for other ligands, such as islets, hypothalami, and normal rat pituitaries. GLP-1 or other drugs, to act as in vivo regulators receptor transcripts were examined by reverse transcrip- of insulin secretion by insulin-expressing non-b cells. tase polymerase chain reaction (RT-PCR) using a set of primers that amplify a 190-bp fragment. Total RNA was used to synthesize cDNA using Maloney murine Materials and methods leukemia virus (MMLV) . The amount of cDNA equivalent to 10 ng of total RNA was Animals and reagents used in an RT-PCR reaction with 32P-dCTP and primers Male Sprague–Dawley rats were purchased from Harlan specific for the rat GLP-1 receptor (50 primer: ag- at 10 weeks of age (body weight 300–324 g). Male NOD/ tagtgtgctccaagggcat, 30 primer: aagaaagtgcgtaccccaccg). SCID mice were purchased from The Jackson Laboratory. One-third of the RT-PCR product was separated on an Animals were housed in the animal facility at Vanderbilt 8% nondenaturing polyacrylamide gel. The gel was University Medical Center on a 12-h light/dark cycle dried and exposed for autoradiography (18 h exposure). with free access to food and drinking water. Trypsin, soybean trypsin inhibitor, and DNAse were purchased Flow cytometric analysis of insulin expression from Worthington Biochemical Corporation. The hy- Expression of human insulin in AdINS/GLP-1R- or pothalamic-releasing hormones (CRH, LHRH, GHRH, AdLacZ-treated cells was examined by intracellular TRH), bovine serum albumin (BSA), saponin, and GLP-1 immunostaining of insulin, followed by flow cytometric (7-36) amide were obtained from Sigma. Bio-Gel P-2 resin analysis. The protocol was based on a published was from Bio-Rad. pCA14 and pJM17 as well procedure as described by Openshaw et al.47 Briefly, as 293 cells were purchased from Microbix Biosystems cells were washed with phosphate-buffered saline (PBS) Inc. Dulbecco’s modified eagle medium (DMEM) and and fixed in 2% formaldehyde in PBS at room tempera- fetal bovine serum (FBS) were purchased from Invitro- ture for 20 min. Subsequent washing and staining gen Life Technology. RNeasy mini for extraction of solutions contained 1% BSA and 0.5% saponin. All total RNA was from Qiagen. The analytical reagents and incubations and washing were carried out at room the analyzer for blood glucose were from HemoCue, Inc. temperature. After washing and a 10-min incubation in PBS/BSA/saponin, the cells were incubated with the Construction of recombinant adenoviruses following antibodies. Guinea anti-human insulin Recombinant adenovirus carrying both the human serum (Cat. # 1014, Linco Research, Inc.) was used at a proinsulin cDNA and human GLP-1 receptor cDNA44 1 : 10 000 dilution. Cy2-conjugated donkey anti-guinea (kindly provided by Dr Svetlana Mojsov at The Rock- pig IgG (the Jackson Immunoresearch Laboratories, Inc.) efeller University) was generated by homologous re- was used as second antibody at a 1 : 500 dilution. After combination as previously described.45 The human the final wash, the cells were resuspended in 0.5% BSA/ proinsulin cDNA or human GLP-1 receptor cDNA was PBS and analyzed on an FACScalibur flow cytometer first subcloned into pCA14 (pCA14INS or pCA14GLP- using Cellquest v3.1. software (Becton Dickinson). Cell 1R). To construct AdINS/GLP-1R, the fragment contain- debris was excluded from the analysis based on their ing human GLP-1 receptor cDNA, as well as the forward and sideway light-scattering properties. The upstream human cytomegalovirus immediate-early pro- fluorescence intensity from insulin staining was plotted moter/enhancer and the downstream SV40 polyadeny- on a logarithmic scale against cell number. lation signals, was retrieved from pCA14GLP-1R and inserted into pCA14INS, downstream of the SV40 Perifusion analysis polyadenylation signals. The resulting Cells were harvested by trypsinization after transduction (pCA14INS/GLP-1R) was then used to generate recom- and transferred onto pretreated support matrix Bio-Gel binant adenovirus AdINS/GLP-1R. The recombinant P2 resin.46 After overnight culture, the cell–resin mixture adenovirus carrying b-galactosidase gene (AdLacZ) has was loaded into the perifusion chambers.48 The perifu- been described.45 sion medium was DMEM containing 0.25% (w/v) BSA and, unless otherwise indicated, 25 mM glucose. After Isolation and transduction of rat pituitary cells initial equilibration period, the cells were perifused at a Pituitary cells were isolated from normal male Sprague– flow rate of 0.5 ml/min. The effluent was collected with Dawley rats as described previously.46 Isolated pituitary each fraction containing 3 ml/tube and stored at –701C cells were cultured in DMEM supplemented with 10% until assayed. FBS, 100 m/ml of penicillin, and 100 mg/ml of strepto- mycin. The cells were transduced with the recombinant Transplantation experiments adenoviruses by adding the viral stock directly to the cell Male NOD/SCID mice at 15–20 weeks of age were used suspension, followed by a 48-h incubation at 371Cina as recipients. A total of 3 Â 106 of either AdINS/GLP-1R-

humidified cell culture incubator with 5% CO2. The or AdLacZ-treated cells were injected into the intraper-

Gene Therapy Regulated insulin secretion from non-b cells LWuet al 1719 itoneal cavity of each mouse. The animals were subjected of Chicago (NIH DK20595) by Diane Ostrega and Dr to OGTT on days 1, 3, and 5 after transplantation. They Kenneth Polonsky. were fasted for 16 h, and then given an intragastric glucose load (2 mg/g body weight). Blood samples were collected from the retro-orbital sinus before and after oral glucose load (0, 15, 30, and 60 min). The levels of blood glucose were determined by a glucose oxidase method References using a Hemocue glucose analyzer. The plasma was separated and stored at –701C until assayed for insulin. 1 Perley MJ, Kipnis DM. Plasma insulin responses to oral and intravenous glucose: studies in normal and diabetic subjects. J Clin Invest 1967; 46: 1954–1962. Hormone assays 2 Creutzfeldt M. Candidate hormones of the gut. XV. Insulin- RIAs for the pituitary hormones ACTH, GH, LH, TSH, releasing factors of the gastrointestinal mucosa (Incretin). and PRL were performed using specific antisera obtained Gastroenterology 1974; 67: 748–750. from the National Hormone and Peptide Program 3 Holst JJ. Glucagon-like peptide 1 (GLP-1): an intestinal hormone, (National Institute of Diabetes and Digestive and signalling nutritional abundance, with an unusual therapeutic Disease) according to protocols recommended by the Potential. Trends Endocrinol Metab 1999; 10: 229–235. provider. The culture supernatant and cell extracts of 4 Drucker DJ. Minireview: the glucagon-like . Endocrinol- nontransduced rat pituitary cells were assayed for rat ogy 2001; 142: 521–527. insulin using an interspecies-specific and structurally 5 Thorens B. Glucagon-like peptide-1 and control of insulin non-specific insulin RIA kit, RI-13K, from Linco Research secretion. Diabetes Metab 1995; 21: 311–318. Inc. Human insulin in the perifusates of transduced rat 6 Thorens B. Expression cloning of the pancreatic pituitary cells was measured with a - and receptor for the gluco- incretin hormone glucagon-like peptide structurally specific human RIA kit, HI-14K, from Linco 1. Proc Natl Acad Sci USA 1992; 89: 8641–8645. Research Inc. Neither rat insulin, human proinsulin, des 7 Thorens B, Waeber G. Glucagon-like peptide-I and the control of 31, 32 human proinsulin nor human C-peptide is insulin secretion in the normal state and in NIDDM. Diabetes detected by this assay (manufacturer’s data). Human 1993; 42: 1219–1225. 8 Fehmann HC, Goke R, Goke B. Cell and molecular of the insulin in the plasma of mice bearing transduced rat incretin hormones glucagon-like peptide-I and glucose-depen- pituitary cell implants was measured with a modified dent insulin releasing polypeptide. Endocr Rev 1995; 16: 390–410. version of the above specific human insulin RIA, which 9 Sherwood NM, Krueckl SL, McRory JE. The origin and function permitted the determination of the insulin concentration of the pituitary adenylate cyclase-activating polypeptide (PA- in as little as 10 ml of plasma/assay tube. The sensitivity CAP)/glucagon superfamily. Endocr Rev 2000; 21: 619–670. of the assay was improved more than 10-fold by 10 Holz GG et al. cAMP-dependent mobilization of intracellular incubating the human insulin standard and mouse Ca2+ stores by activation of ryanodine receptors in pancreatic plasma samples with the primary antibody for 72 h at beta-cells. A Ca2+ signaling system stimulated by the insulino- 41C before the radioiodinated insulin was added. The glucagon-like peptide-1-(7-37). J Biol Chem 1999; incubation was then continued at 41C for an additional 274: 14147–14156. 24 h. Separation of antibody-bound radioactivity from 11 Meglasson MD, Matschinsky FM. Pancreatic islet glucose free radioactivity was then carried out as recommended metabolism and regulation of insulin secretion. Diabetes Metab by the manufacturer. Human proinsulin in the perifusate Rev 1986; 2: 163–214. was measured by ELISA according to the method of 12 Fehmann HC et al. The effects of glucagon-like peptide-I (GLP-I) Hartling et al49 and was performed by Diane Ostrega and on hormone secretion from isolated human pancreatic islets. Dr Kenneth Polonsky in the Diabetes Research and 1995; 11: 196–200. Training Center at the . This assay 13 Yamada S et al. Time-dependent stimulation of insulin exocytosis 0 0 does not measure human insulin and C-peptide.49 by 3 ,5 -cyclic in the rat islet beta-cell. Endocrinology 2002; 143: 4203–4209. Statistic analysis 14 Dachicourt N et al. Glucagon-like peptide-1(7-36)-amide confers 7 glucose sensitivity to previously glucose-incompetent beta-cells Data are presented as mean s.e. Statistical analysis was in diabetic rats: in vivo and in vitro studies. J Endocrinol 1997; 155: performed using Student’s t-test. Differences were 369–376. considered significant when Po0.05. 15 Gutniak M et al. Antidiabetogenic effect of glucagon-like peptide-1 (7-36)amide in normal subjects and patients with diabetes mellitus. N Engl J Med 1992; 326: 1316–1322. Acknowledgements 16 D’Alessio DA, Kahn SE, Leusner CR, Ensinck JW. Glucagon-like peptide 1 enhances glucose tolerance both by stimulation of This study was supported by a postdoctoral training insulin release and by increasing insulin-independent glucose grant from the Juvenile Diabetes Research Foundation disposal. J Clin Invest 1994; 93: 2263–2266. International, a Merit Review Award from the VA 17 Ahren B, Larsson H, Holst JJ. Effects of glucagon-like peptide-1 Research Service, a research grant from the National on islet function and insulin sensitivity in noninsulin-dependent Institutes of Health (DK55233), research grants from the diabetes mellitus. J Clin Endocrinol Metab 1997; 82: 473–478. American Diabetes Association and the Juvenile Diabetes 18 Nauck MA et al. Normalization of fasting hyperglycaemia by Research Foundation International, and the Vanderbilt exogenous glucagon-like peptide 1 (7-36 amide) in type 2 (non- Diabetes Research and Training Center (NIH DK20593). insulin-dependent) diabetic patients. Diabetologia 1993; 36: The GLP-1 receptor cDNA was kindly provided by Dr 741–744. Svetlana Mojsov at The . The 19 Clark SA et al. Novel cell lines produced by iterative proinsulin assay was graciously performed at the engineering of GLUT2, glucokinase, and human insulin expres- Diabetes Research and Training Center at the University sion. Diabetes 1997; 46: 958–967.

Gene Therapy Regulated insulin secretion from non-b cells LWuet al 1720 20 Hohmeier HE et al. Regulation of insulin secretion from novel 35 Efrat S et al. Conditional transformation of a pancreatic engineered insulinoma cell lines. Diabetes 1997; 46: 968–977. beta-cell line derived from transgenic mice expressing a 21 Ferber S et al. Pancreatic and duodenal homeobox gene 1 tetracycline-regulated oncogene. Proc Natl Acad Sci USA 1995; induces expression of insulin genes in and ameliorates 92: 3576–3580. -induced . Nat Med 2000; 6: 36 Lipes MA et al. Insulin-secreting non-islet cells are resistant to 568–572. autoimmune destruction. Proc Natl Acad Sci USA 1996; 93: 22 Lee HC et al. Remission in models of type 1 diabetes by gene 8595–8600. therapy using a single- chain insulin analogue. Nature 2000; 408: 37 Faradji RN et al. Glucose-induced toxicity in insulin-producing 483–488. pituitary cells that coexpress GLUT2 and glucokinase. Implica- 23 Chen R, Meseck M, McEvoy RC, Woo SL. Glucose-stimulated tions for metabolic engineering. J Biol Chem 2001; 276: and self-limiting insulin production by glucose 6-phosphatase 36695–36702. promoter driven insulin expression in hepatoma cells. Gene 38 Kolodka TM, Finegold M, Moss L, Woo SL. Gene therapy for Therapy 2000; 7: 1802–1809. diabetes mellitus in rats by hepatic expression of insulin. Proc 24 Tooze SA. Biogenesis of secretory granules. Implications arising Natl Acad Sci USA 1995; 92: 3293–3297. from the immature secretory in the regulated pathway 39 Gros L et al. Insulin production by engineered muscle cells. Hum of secretion. FEBS Lett 1991; 285: 220–224. Gene Ther 1999; 10: 1207–1217. 25 Tooze SA, Martens GJ, Huttner WB. Secretory granule biogen- 40 Falqui L et al. Reversal of diabetes in mice by implantation of esis: rafting to the SNARE. Trends Cell Biol 2001; 11: 116–122. human fibroblasts genetically engineered to release mature 26 Zhu X et al. Disruption of PC1/3 expression in mice causes human insulin. Hum Gene Ther 1999; 10: 1753–1762. dwarfism and multiple neuroendocrine peptide processing 41 Hughes SD, Johnson JH, Quaade C, Newgard CB. Engineering defects. Proc Natl Acad Sci USA 2002; 99: 10293–10298. of glucose-stimulated insulin secretion and in non- 27 Zhu X et al. Severe block in processing of proinsulin to insulin islet cells. Proc Natl Acad Sci USA 1992; 89: 688–692. accompanied by elevation of des-64,65 proinsulin intermediates 42 Hughes SD et al. Transfection of AtT-20ins cells with GLUT-2 but in islets of mice lacking prohormone convertase 1/3. Proc Natl not GLUT-1 confers glucose-stimulated insulin secretion. Acad Sci USA 2002; 99: 10299–10304. Relationship to glucose metabolism. J Biol Chem 1993; 268: 28 Beak SA et al. Glucagon-like peptide-1 (GLP-1) releases 15205–15012. thyrotropin (TSH): characterization of binding sites for GLP-1 43 Lu M, Wheeler MB, Leng XH, Boyd AEd. The role of the free on alpha-TSH cells. Endocrinology 1996; 137: 4130–4138. cytosolic calcium level in beta-cell signal transduction by gastric 29 Taniguchi Y, Yasutaka S, Kominami R, Shinohara H. Prolifera- inhibitory polypeptide and glucagon-like peptide I(7-37). tion and differentiation of rat anterior pituitary cells. Anat Endocrinology 1993; 132: 94–100. Embryol (Berl) 2002; 206: 1–11. 44 Wei Y, Mojsov S. Tissue-specific expression of the human 30 Kendall SK et al. Targeted disruption of the pituitary glycopro- receptor for glucagon-like peptide-I: brain, and pancreatic tein hormone alpha-subunit produces hypogonadal and hy- forms have the same deduced sequences. FEBS Lett pothyroid mice. Genes Dev 1995; 9: 2007–2019. 1995; 358: 219–224. 31 Castro MG et al. Expression of in normal 45 Wu L, Fritz JD, Powers AC. Different functional domains of and neoplastic anterior pituitary cells using recombinant GLUT2 glucose transporter are required for glucose affinity and adenoviruses: long term expression, cell cycle dependency, substrate specificity. Endocrinology 1998; 139: 4205–4212. and effects on hormone secretion. Endocrinology 1997; 138: 46 Watanabe T, Orth DN. Detailed kinetic analysis of adrenocorti- 2184–2194. cotropin secretion by dispersed rat anterior pituitary cells in a 32 Holz GGt, Kuhtreiber WM, Habener JF. Pancreatic beta-cells are microperifusion system: effects of ovine corticotropin-releasing rendered glucose-competent by the insulinotropic hormone factor and . Endocrinology 1987; 121: glucagon-like peptide-1(7-37). Nature 1993; 361: 362–365. 1133–1145. 33 Hui H, Yu R, Bousquet C, Perfetti R. Transfection of pancreatic- 47 Openshaw P et al. Heterogeneity of intracellular derived beta-cells with a minigene encoding for human synthesis at the single-cell level in polarized T helper 1 and T glucagon-like peptide-1 regulates glucose-dependent insulin helper 2 populations. J Exp Med 1995; 182: 1357–1367. synthesis and secretion. Endocrinology 2002; 143: 3529–3539. 48 Wang T et al. An encapsulation system for the immunoisolation 34 Kolligs F, Fehmann HC, Goke R, Goke B. Reduction of the of pancreatic islets. Nat Biotechnol 1997; 15: 358–362. incretin effect in rats by the glucagon-like peptide 1 receptor 49 Hartling SG et al. ELISA for human proinsulin. Clin Chim Acta antagonist exendin (9-39) amide. Diabetes 1995; 44: 16–19. 1986; 156: 289–297.

Gene Therapy