AU7179994A - Vectors for genetically engineered cells that produce insulin in response to glucose - Google Patents

Description

DESCRIPTION

VECTORS FOR GENETICALLY ENGINEERED CELLS THAT

PRODUCE INSULIN IN RESPONSE TO GLUCOSE

BACKGROUND OF THE INVENTION

The present application is a continuation-in-part U.S. Serial Number 08/084,742, filed June 28, 1993 which is relied upon for an earlier filing date.

The government owns certain rights in the present invention pursuant to NIH grants R29-DK40735 and P01-DK42582.

1. Field of the Invention

The present invention relates generally to the preparation, culture and use of engineered cells having the ability to secrete insulin in response to glucose, to gene transfer protocols, to methods for the detection of diabetes-associated antigens, and to methods employing engineered cells in the production of human insulin. In particular aspects, the present invention relates to recombinant adenoviruses containing genes involved in glucose sensing, such as GLUT-2, the gene encoding glucokinase, and antisense versions of the hexokinase gene, and to the use of adenovirus-mediated gene therapy to direct the expression of recombinant genes in pancreatic islet cells, islet-derived cells or other neuroendocrine cell lines capable of insulin secretion.

2. Description of the Related Art

Insulin-dependent diabetes mellitus (IDDM, also known as Juvenile-onset, or Type I diabetes) represents approximately 15% of all human diabetes. IDDM is distinct from non-insulin dependent diabetes (NIDDM) in that only IDDM involves specific destruction of the insulin producing β cells of the islets of Langerhans in the pancreas. The destruction of β cells in IDDM appears to be a result of specific autoimmune attack, in which the patient's own immune system recognizes and destroys the β cells, but not the surrounding α (glucagon producing) or δ (somatostatin producing) cells that comprise the islet.

The precise events involved in β cell recognition and destruction in IDDM are currently unknown, but involve both the "cellular" and "humoral" components of the immune system. In IDDM, islet β cell destruction is ultimately the result of cellular mechanisms, in which "killer T-cells" destroy β cells which are erroneously perceived as foreign or harmful. The humoral component of the immune system, comprised of the antibody-producing β cells, is also inappropriately active in IDDM patients, who have serum antibodies against various β cell proteins. Antibodies directed against intracellular proteins probably arise as a consequence of β cell damage which releases proteins previously "unseen" by the immune system. However, the appearance of antibodies against several cell surface epitopes such as insulin, proinsulin, the "38kD protein", immunoglobulins, the 65kD heat shock protein and the 64kD and 67kD forms of glutamic acid decarboxylase (GABA) are believed to be linked to the onset of IDDM (Lernmark, 1982) . Antibodies in diabetic sera may also interact with the islet GLUT-2 glucose transporter (Johnson et al . , 1990c; Inman et al . , 1993) .

A progressive loss of β cell function is observed in the early stages of NIDDM and IDDM, even prior to the autoimmune β cell destruction in IDDM. The specific function of glucose-stimulated insulin release is lost in islets of diabetic patients, despite the fact that such islets continue to respond to non-glucose secretagogues such as amino acids and isoproterenol (Srikanta et al . , 1983) .

The participation of the pancreatic islets of Langerhans in fuel homeostasis is mediated in large part by their ability to respond to changes in circulating levels of key metabolic fuels by secreting peptide hormones. Accordingly, insulin secretion from islet β cells is stimulated by amino acids, three-carbon sugars such as glyceraldehyde, and most prominently, by glucose. The capacity of normal islet β cells to "sense" a rise in blood glucose concentration, and to respond to elevated levels of glucose (as occurs following ingestion of a carbohydrate containing meal) by secreting insulin is critical to control of blood glucose levels. Increased insulin secretion in response to a glucose load prevents chronic hyperglycemia in normal individuals by stimulating glucose uptake into peripheral tissues, particularly muscle and adipose tissue.

Mature insulin consists of two polypeptide chains, A and B, joined in a specific manner. However, the initial protein product of the insulin gene in β cells is not insulin, but preproinsulin. This precursor differs from mature insulin in two ways. Firstly, it has a so-called N-terminal "signal" or "pre" sequence which directs the polypeptide to the rough endoplasmic reticulu , where it is proteolytically processed. The product, proinsulin, still contains an additional connecting peptide between the A and B chains, known as the C-peptide, which permits correct folding of the whole molecule. Proinsulin is then transported to the Golgi apparatus, where enzymatic removal of the C-peptide begins. The processing is completed in the so-called secretory granules, which bud off from the Golgi, travel to, and fuse with, the plasma membrane thus releasing the mature hormone.

Glucose stimulates de novo insulin biosynthesis by increasing transcription, mRNA stability, translation, and protein processing. Glucose also rapidly stimulates the release of pre-stored insulin. While glucose and non-glucose secretagogues may ultimately work through a final common pathway involving alterations in K⁺ and Ca⁺⁺ channel activity and increases in intracellular Ca⁺⁺ (Prentki et al . , 1987; Turk et al . , 1987), the biochemical events leading from changes in the levels of a particular fuel to insulin secretion are initially diverse. In the case of glucose, transport into the β cell and metabolism of this sugar are absolute requirements for secretion, leading to the hypothesis that its specific stimulatory effect is mediated by, and proportional to, its flux rate through glycolysis and related pathways (Ashcroft, 1980; Hedeskov, 1980; Meglasson and Matchinsky, 1986; Prentki et al . , 1987;

Turk et al. 1987; Malaisse et al . , 1990). Strong support for this view comes from the finding that non- metabolizable analogues of glucose such as 3-O-methyl or 2-deoxy glucose fail to stimulate insulin release (Ashcroft, 1980; Meglasson and Matchinsky, 1986).

A substantial body of evidence has accumulated implicating a specific facilitated-diffusion type glucose transporter known as GLUT-2, and the glucose phosphorylating enzyme, glucokinase, in the control of glucose metabolism in islet β cells. Both proteins are members of gene families; GLUT-2 is unique among the five-member family of glucose transporter proteins in that it has a distinctly higher Km and V ax for glucose. Glucokinase is the high Km and high Vmax counterpart of GLUT-2 among the family of hexokinases (Weinhouse, 1976) . Importantly, both proteins have affinities for glucose that allow dramatic changes in their activities over the physiological range of glucose. This has led to the hypothesis that these proteins work in concert as the "glucose-sensing apparatus" that modulates insulin secretion in response to changes in circulating glucose concentrations by regulating glycolytic flux (Newgard et al . , 1990; Johnson et al . , 1990a).

In normal β cells, glucose transport capacity is in excess relative to glycolytic flux. Thus, the GLUT-2 transporter likely plays a largely permissive role in the control of glucose metabolism, while glucokinase represents the true rate-limiting step (Meglasson and Matchinsky, 1986; Newgard et al . , 1990). Implicit in this formulation, however, is the prediction that severe underexpression of GLUT-2 will result in loss of glucose- stimulated insulin secretion in islets, an idea that has recently received strong experimental support from studies with spontaneous (Johnson et al . , 1990b; Orci et al . , 1990) as well as experimentally induced (Chen et al . , 1990; Thorens et al. , 1990b) animal models of β cell dysfunction, which have clear similarities to the β cell impairment observed in human NIDDM. Furthermore, RINm5F clonal insulinoma cells derived from islet β cells express GLUT-1, a transporter with a substantially lower Km and Vmax for glucose, as their predominant glucose transporter instead of GLUT-2. This may explain the finding that the clonal cells fail to respond to glucose as an insulin secretagogue (Thorens et al . , 1988).

Currently, there are significant deficiencies in the diagnosis and treatment of both IDDM and NIDDM. In diagnostic tests, for example, the most common clinical test, the oral glucose tolerance test (OGTT) suffers from severe drawbacks, such as subjective interpretation and the ability to only identify individuals with advanced disease. The serological test for cytoplasmic islet cell antibodies (ICA-cyt) (Bright, 1987; Gleichmann et al . , 1987) is a diagnostic procedure for detecting the onset of diabetes, which involves binding of patients' antibodies to cryostat sections of fresh human or primate pancreas. One evident disadvantage of this is the requirement for fresh human or primate tissue. Further difficulties are: false negatives (40%) ; subjective interpretation; poor reproducibility; and the inability to detect cell surface-directed antibodies which are known to specifically damage β cells (Doberson et al . , 1980) .

Even less progress has been made in developing new therapeutic strategies for diabetics. Significant effort has been devoted to the strategy of islet or pancreas fragment transplantation as a means for permanent insulin replacement (Lacy et al . , 1986). However, this approach has been severely hampered by the difficulties associated with obtaining tissue, as well as the finding that transplanted islets are recognized and destroyed by the same autoimmune mechanism responsible for destruction of the patients original islet β cells.

Treatment for diabetes is still centered around self-injection of insulin once or twice daily. Both recombinant and non-recombinant methods are currently employed for the industrial production of human insulin for therapeutic use. Recombinant methods generally include the expression of recombinant proinsulin in bacteria or yeast, followed by chemical treatment of the proinsulin to ensure correct disulfide bond linkages between the A and B chains of the mature insulin molecule. The proinsulin produced by microorganisms is processed to insulin by the addition of proteolytic enzymes. Thereafter, the mature insulin peptide must be purified away from the bacterial or yeast proteins, as well from the added proteases. The bacterial procedure involves 40 distinct steps. The non-recombinant methods typically include the purification of pig insulin from freshly isolated porcine pancreas or pancreatic islets. Each of the above methods suffer from the drawbacks of being technically difficult and laborious. The latter method is further complicated by the fact that the pancreas is a complex proteinaceous tissue with high levels of active proteases that can degrade insulin and render it inactive as a hormone.

Accordingly, it is evident that improvements are needed both in the treatment and diagnosis of diabetes and in the methods of insulin production for current therapeutic application.

SUMMARY OF THE INVENTION

The present invention is intended to address such disadvantages present in the prior art. In general, the invention is based on the discovery that recombinant DNA technology and cell culture methods may be employed to engineer an "artificial β cell" that secretes insulin in response to glucose. The present invention provides a means of preparing artificial β cells that it is proposed can be employed in a variety of applications, such as, e.g., in the detection of diabetes-associated antigens, in the clinical treatment of IDDM and even in the large- scale production of correctly-folded insulin. In further aspects, the current invention provides methods for growing artificial β cells in liquid culture on gelatin beads and for the increased production of human insulin by perfusion of such recombinant cells with glucose- containing buffers.

Turning first to embodiments directed to the recombinant engineering of cells secreting insulin in response to glucose, it should be pointed out that this aspect of the invention relates generally to an engineered cell that incudes a gene, preferably a recombinant gene, encoding a functional glucose transporter protein, wherein the engineered cells secrete insulin in response to glucose. This aspect of the invention is based generally on the finding that where a cell is competent to secrete insulin generally, it may be converted to a glucose-responsive cell through the introduction of a gene encoding a functional glucose transporter protein, such as a GLUT gene. For most purposes leading up to the ultimate treatment of the diabetic condition, one will desire to employ GLUT-2 as the recombinant glucose transporter gene. This is because the GLUT-2 gene corresponds to that found and normally expressed in β cells, and it is believed that this gene will ultimately provide a more physiological response than other types of glucose transporters.

As used herein, the term "engineered" or

"recombinant" cell or even "recombinant host cell" is intended to refer to a cell into which a recombinant gene, such as a gene encoding a functional glucose transporter protein, has been introduced. Therefore, engineered cells are distinguishable from naturally occurring cells which do not contain a recombinantly introduced gene. Engineered cells are thus cells having a gene or genes introduced through the hand of man. Recombinantly introduced genes will either be in the form of a cDNA gene (i.e., they will not contain introns) , a copy of a genomic gene, a gene or genes positioned within a recombinant adenovirus, genes produced by synthetic means, and/or genes positioned adjacent to a promoter not naturally associated with the particular introduced gene.

Generally speaking, it will be more convenient to employ as the recombinant gene a cDNA version of the gene. It is believed that the use of a cDNA version will provide advantages in that the size of the gene will generally be much smaller and more readily employed to transfect the targeted cell than will a genomic gene, which will typically be up to an order of magnitude larger than the cDNA gene. However, the invention does not exclude the possibility of employing a genomic version of a particular gene in any of the embodiments disclosed herein.

Recombinant host cells of the present invention will generally be derived from a cell line comprised of cells capable of forming secretory granules. Secretory granules are generally confined to mammalian cells whose main function is the synthesis and secretion of peptides. Generally speaking, secretory granules are found in endocrine cells. Secretory granules are formed by budding of intracellular membranous structures known as the Golgi apparatus. Polypeptide hormones are usually synthesized as prohormones and undergo proteolytic processing to yield the shorter, mature version of the hormone.

Thus, for example, the initial protein product of the insulin gene in β cells is preproinsulin. This precursor differs from mature insulin in that it has a so-called "signal sequence" at its N-terminus, consisting of a stretch of hydrophobic amino acids that guide the polypeptide to the rough endoplasmic reticulum. It also has a connecting peptide between the A and B chains that comprise the mature insulin molecule; this connector is known as the "C-peptide". The preproinsulin molecule enters the lumen of the endoplasmic reticulum, in the process having its hydrophobic N-terminal "pre" region proteolytically removed. The processed, correctly folded proinsulin molecule (still containing the C-peptide) is then transported to the Golgi apparatus. As the precursor is transported through the Golgi apparatus, enzymatic removal of the C-peptide connector begins.

Secretory granules are derived from Golgi membranes by a process of budding off and eventual separation. The resulting granule envelopes the mixture of unprocessed proinsulin and the small amount of mature insulin. Most of the processing of proinsulin to insulin occurs shortly after formation of the secretory granules by virtue of the fact that the enzymes responsible for this processing are found at highest concentration within the granules. The granules are transported to the plasma membrane surface of the cell in response to secretory stimuli such as glucose; whereupon they fuse with the plasma membrane and release their stores of the mature hormone. The important and unique features of this system are 1) the secretory granules allow a supply of a particular hormone to be built up and stored for release at the time when it is needed to perform its function and 2) the presence of processing enzymes in the granules allow efficient conversion of the precursor forms of hormones to the mature forms. Cells that lack secretory granules will thus likely not be useful for the purposes of this aspect of the invention.

Therefore, cells used in this aspect will preferably be derived from an endocrine cell, such as a pituitary or thyroid cell. Particularly preferred endocrine cells will be AtT-20 cells, which are derived from ACTH secreting cells of the anterior pituitary gland, GH1 or the closely related GH3 cells, which are derived from growth hormone producing cells of the anterior pituitary, or other cell lines derived from this gland. AtT-20 cells are preferred for the following reasons. First, these cells have been modified for insulin gene expression by stable transfection with a viral promoter/human proinsulin cDNA construct (this derivation of the AtT-20 cell line is known as AtT-20_ins; both the parental AtT-20 cell line and the insulin expressing AtT- 20^ cell line are available from American Type Culture Collection, 12301 Parklawn Drive, Rockville, MD 20852) . Second, AtT-20_jns cells are able to process the preproinsulin mRNA and protein to yield the correctly processed insulin polypeptide. Third, their insulin secretory response to analogues of cAMP compares favorably with the well-differentiated hamster insulino a (HIT) cell line which is derived from hamster islet β cells. Finally, studies from one of the inventors' laboratories have recently shown that AtT-20_ins cells contain significant amounts of the islet isoform of glucokinase, making this the only tissue other than liver or islets in which glucokinase gene expression has been reported (Hughes et al . , 1991).

GH1 and GH3 cells were originally derived from the same batch of cells isolated from a rat pituitary gland tumor. GH3 cells differ from GH1 cells in that they secrete more growth hormone and also secrete prolactin (both lines are available from the American Type Culture Collection) . These cells are believed to be preferred for the practice of the invention because it has been shown that introduction of a recombinant preprosomatostatin gene into these cells results in secretion of the mature somatostatin peptide (Stoller et al . , 1989). Processing of the endogenous preprosomatostatin gene also occurs in 5-cells of the islets of Langerhans. The finding that an islet hormone precursor can be correctly processed in growth hormone secreting cells of the anterior pituitary suggests that proinsulin processing will also occur in these cells, perhaps even more efficiently than in AtT-20_ins cells.

A number of cell lines derived from β cells, commonly known as insulinoma cells, are also preferred for the practice of this invention and are readily available, particularly as concerns the therapeutic aspects of the work. For example, hamster insulinoma (HIT-T15) cells are well studied and are readily available from the American Type Tissue Collection. A number of rat insulinoma cell lines are also available. The RINm5F and RINrl046-38 cell lines were derived from a radiation induced tumor of the islet β cells (Gazdar et al . , 1980; Clark et al . , 1990). MSL-G2 cells were derived from a liver metastasis of an islet cell tumor.

These cells require periodic passage in an animal host in order to maintain expression of their endogenous insulin gene (Madsen, et al . , 1988). Finally, the 0-TC insulinoma cell line has been recently derived from transgenic mice injected with a T-antigen gene driven by an insulin promoter, resulting in specific expression of T-antigen in islet β cells and consequent immortalization of these cells (Efrat et al. , 1988).

RIN 1046-38 cells have been shown in one of the inventors' laboratory to express both GLUT-2 and glucokinase (Hughes et al . , 1991), and have been shown by Clark et al. (1990) to be responsive to glucose. Glucose stimulation of insulin release from these cells is maximal at 0.5 mM glucose, however, a level far below the stimulatory concentration of glucose required for insulin release from normal β cells. Recent studies in the inventor's laboratory have shown that this hypersensitivity to glucose in RIN 1046-38 cells may be due to high levels of hexokinase activity. Hexokinase performs the same function as glucokinase (glucose phosphorylation) but does so at much lower glucose concentrations (hexokinase has a Km for glucose of approximately 0.05 mM versus 8 mM for glucokinase). It is proposed that lowering of hexokinase activity by methods of recombinant DNA technology described below might make RIN cells useful for the practice of this invention.

Of course, the type of engineering that will be required in order to achieve a cell that secretes insulin in response to glucose will depend on the property of the starting cell. In general, the invention proposes that in addition to the ability to form secretory granules, the ability to functionally express certain genes is important. The functional genes that are required include an insulin gene, a glucose transporter gene and a glucokinase gene. In the practice of the invention, one or more of these genes will be a recombinant gene. Thus, if the starting cell has a functional insulin gene and a functional glucokinase gene, and these genes are expressed at levels similar to their expression in β cells, but the cell does not have a functional glucose transporter gene, introduction of a recombinant glucose transporter gene will be required. Conversely, if the starting cell expresses none of the aforementioned genes in a functional fashion, or at physiologic levels, it will be necessary to introduce all three. Since recombinant versions of all three categories of genes are available to the art, and the specific technology for introducing such genes into cells is generally known, the construction of such cells will be well within the skill of the art in light of the specific disclosure herein.

As stated above, particularly preferred endocrine cells for use in accordance with the present invention are AtT-20_ins cells, which have been stably transfected to allow the production of correctly processed human insulin. Also as stated, it is generally preferred to employ the GLUT-2 glucose transporter isoform to provide recombinant cells with a functional glucose transporter. Engineered cells that combine both of these features have been created, and one form of cell expressing high levels of GLUT-2 mRNA, termed CTG-6 cells, are envisioned to be of particular use in aspects of the present invention.

Where the introduction of a recombinant version of one or more of the foregoing genes is required, it will be important to introduce the gene such that it is under the control of a promoter that effectively directs the expression of the gene in the cell type chosen for engineering. In general, one will desire to employ a promoter that allows constitutive (constant) expression of the gene of interest. Commonly used constitutive promoters are generally viral in origin, and include the cytomegalovirus (CMV) promoter, the Rous sarcoma long- terminal repeat (LTR) sequence, and the SV4O early gene promoter. The use of these constitutive promoters will ensure a high, constant level of expression of the introduced genes. The inventors have noticed that the level of expression from the introduced gene(s) of interest can vary in different clones, probably as a function of the site of insertion of the recombinant gene in the chromosomal DNA. Thus, the level of expression of a particular recombinant gene can be chosen by evaluating different clones derived from each transfection event. Once that line is chosen, the constitutive promoter ensures that the desired level of expression is permanently maintained. It may also be possible to use promoters that are specific for a specific cell type, such as the insulin promoter in insulinoma cell lines, or the prolactin, proopiomelanocortin (POMC) or growth hormone promoters in anterior pituitary cell lines.

Certain particular embodiments of the invention are directed to engineering cells with reduced hexokinase activity relative to the parent cell line. There are four known isoforms of hexokinase in mammals.

Hexokinases I, II, and III have very low K s (high affinities) for glucose, on the order of 0.05 mM. Hexokinase IV is glucokinase, which has a high Km for glucose of around 8-10 mM. In the islet β cell, glucokinase is the predominant glucose phosphorylating enzyme, while in most clonal cell lines grown in culture, the low Km hexokinase I isoform predominates. The inventor proposes that expression of hexokinases other than glucokinase at high levels in clonal cells used for engineering will tend to make the cell glucose-responsive in terms of insulin release at lower concentrations of glucose than is desirable. Thus, it is proposed that the lower the hexokinase/glucokinase ratio, the more physiologic the insulin response.

Various approaches may be taken to reduce the hexokinase activity in engineered cells. One approach involves the introduction of an antisense RNA molecule. Antisense RNA technology is now fairly well established, and involves the juxtaposition of the targeted gene in a reverse orientation behind a suitable promoter, such that an "antisense" RNA molecule is produced. More specifically, a segment of DNA encoding the protein to be inhibited is oriented in the antisense orientation relative to its controlling promoter and then expressed from that promoter. This results in the production of mRNA which is complementary to the mRNA that encodes the active protein. This "antisense" construct is then introduced into the engineered cell and, upon its expression, produces an mRNA molecule that will hybridize with, and prevent the processing/translation of mRNA produced by the targeted gene, in this case the hexokinase gene.

An alternative approach to the reduction of hexokinase action is through a technique known as positive/negative selection. This technique involves selection for homologous recombination of a hexokinase gene segment that renders the endogenous hexokinase gene nonfunctional.

In certain embodiments, the present invention is directed to a method of providing glucose-responsive insulin secreting capability to a cell. This method comprises obtaining an insulin producing cell and expressing a GLUT-2 and/or a glucokinase enzyme in the cell. A preferred cell is an islet β cell. In order to obtain a more physiological response to glucose, the method may further comprise inhibiting the hexokinase activity in the cell. A preferred method of expressing the GLUT-2 and/or glucokinase activity is by transfecting the cell with an adenovirus vector expressing the protein(s) and a preferred method of inhibiting the hexokinase activity is by expressing an antisense copy of the hexokinase gene or a fragment thereof.

As is well understood in the art of antisense technology, an antisense fragment of less than the full length gene will inhibit expression of a gene. Therefore, as used herein a "fragment thereof" is taken to mean any contiguous stretch of bases from about 20 bases up to the full length encoding region and even including the upstream promoter region of the gene.

Particularly preferred fragments would include as a part of the sequence, the ATG start site of the coding region of the gene.

In other embodiments, the present invention is directed to a method of providing a glucose-responsive insulin-secreting capability to a mammal in need of such capability. The method includes generally implanting engineered cells which secrete insulin in response to glucose as described in the previous paragraph into such a mammal. It is proposed that techniques presently in use for the implantation of islets will be applicable to implantation of cells engineered in accordance with the present invention. One method involves the encapsulation of engineered cells in a biocompatible coating. In this approach, cells are entrapped in a capsular coating that protects the encapsulated cells from immunological responses, and also serves to prevent uncontrolled proliferation of clonal engineered cells. A preferred encapsulation technique involves encapsulation with alginate-polylysine-alginate. Capsules made employing this technique generally contain several hundred cells and have a diameter of approximately 1 mm.

An alternative approach is to seed Amicon fibers with engineered cells. The cells become enmeshed in the fibers, which are semipermeable, and are thus protected in a manner similar to the micro encapsulates (Altman et al . , 1986). After successful encapsulation or fiber seeding, the cells, generally approximately 1,000-10,000, may be implanted intraperitoneally, usually by injection into the peritoneal cavity through a large gauge needle (23 gauge) .

A variety of other encapsulation technologies have been developed that are proposed will be applicable to the practice of the present invention (Lacy et al . , 1991; Sullivan et al . , 1991; WO 9110470; WO 9110425; WO 9015637; WO 9002580; US 5,011,472; US 4,892,538; WO 8901967, each of the foregoing being incorporated by reference) . The company Cytotherapeutics has developed encapsulation technologies that are now commercially available that will likely be of use in the application of the present invention. A vascular device has also been developed by Biohybrid, of Shrewsbury, Mass., that may have application to the technology of the present invention. In regard to implantation methods which may be employed to provide a glucose-responsive insulin- secreting capability to a mammal, it is contemplated that particular advantages may be found in the methods recently described by Lacy (Lacy et al . , 1991) and Sullivan (Sullivan et al . , 1991), each incorporated herein by reference. These concern, firstly, the subcutaneous xenograft of encapsulated islets, and secondly, the long-term implantation of islet tissue in an "artificial pancreas" which may be connected to the vascular system as an arteriovenous shunt. These implantation methods may be advantageously adapted for use with the present invention by employing engineered cells, as disclosed herein, in the place of the "islet tissue" of the prior art methods.

Further important embodiments of the present invention concern compositions and methods related to alternative strategies by which to provide glucose- responsive insulin-secreting capability to animals or humans with non-insulin dependent diabetes (NIDDM) or other NIDDM-like syndromes. These aspects of the invention are directed to restoring β cell function in such animals or patients by the application of adenovirus-mediated gene transfer, thus allowing physiological control of glucose homeostasis.

These aspects of the invention are based upon the use of recombinant vectors, and particularly, on the use of recombinant adenoviruses, to effect high level expression of specific proteins in islet cells. The invention generally concerns adenovirus vector constructs which comprise a recombinant insert including an expression region encoding at least one protein involved in the predisposition to diabetes, and particularly, those proteins involved in glucose sensing, such as GLUT- 2 and glucokinase, which vectors are capable of expressing these proteins in cells infected with a recombinant adenovirus. In this method, the construct would be administered to the mammal in a pharmacologically acceptable form as discussed below.

As used herein, a protein which is "involved in the predisposition to diabetes" is defined as a protein which is involved in insulin responsiveness, or resistance, in peripheral tissues or a protein which is involved in glucose-sensing in the β cell. Defects, mutations or alterations in such proteins will generally lead to disturbances in insulin action and glucose metabolism and homeostasis, and may lead to disease states such as various forms of diabetes and/or Maturity-Onset Diabetes of the Young (MODY) .

Proteins which are candidates for involvement in insulin responsiveness in peripheral tissues, and in which mutations thus lead to insulin resistance, include the insulin receptor itself and the GLUT-4 glucose transporter. In addition, other candidate proteins include hexokinase II, glycogen synthase and glycogen phosphorylase; various kinases and phosphatases which act upon the former enzymes; and various components of the insulin signal transduction pathways, such as G proteins, tyrosine kinases, the insulin-regulated substrate-1 (IRS- 1) and the like.

The present invention is concerned with proteins which are involved in glucose-sensing in the β cell. As such, particular aspects of the invention embody adenovirus vector constructs which comprise a recombinant insert including an expression region under the control of a promoter and including a coding region that encodes at least one glucose transport protein, glucose phosphorylating protein or a fragment of a glucose transport protein or a glucose phosphorylating protein. The expression region in the vector may comprise a genomic sequence, but for simplicity, it is contemplated that one will generally prefer to employ a cDNA sequence. The recombinant insert of the vector will also generally comprise a promoter region and a polyadenylation signal, such as an SV40 or protamine gene polyadenylation signal. The promoter may be a constitutive promoter, or it may be a jS-cell preferential promoter. For example, see Shelton et al . 1992, and Selden et al . , 1987 for a discussion of /3-cell preferential control elements such as the promoters from GLUT-2, insulin and glucokinase promoters and methods of use of these promoters for jS-cell preferential expression.

Preferred proteins for use in such an adenoviral vector include the glucose transporter GLUT-2, the enzyme glucokinase, and the antisense versions of the hexokinase I and GLUT-1 cDNAs or fragments of the full length proteins. GLUT-2 glucose transporters of the invention may be of any type including e.g. the HepG2 form as described in Mueckler et al . , 1985, and the more preferred islet cell GLUT-2 glucose transporter described in Permutt et al . , 1989. The single glucokinase gene is also known to be alternatively regulated and processed in liver and islets (Iynedjian et al . , 1989; Magnuson and Shelton, 1989; Newgard et al . , 1990; Hughes et al . , 1991) , resulting in distinct transcripts that predict proteins with unique N-termini. Both isoforms may be used in the practice of the present invention with the islet glucokinase being particularly preferred. AtT-20_jns cells express the islet isoform of glucokinase (Hughes et al . , 1991).

β cell dysfunction in NIDDM is believed to involve a deficit in the expression of GLUT-2, glucokinase, or both such proteins, or possibly overexpression of GLUT-1 or hexokinase I. Thus, adenoviral vectors capable of expressing glucose transporters and/or glucose phosphorylating enzymes are particularly preferred, with the islet cell isoforms of each protein being even more preferred where applicable. In certain embodiments, particularly in those designed for human treatment, it is envisioned that one would wish to employ expression units encoding normal forms of one or both of the GLUT-2 and glucokinase enzymes. The term "normal form" is used herein to refer to a form of a protein, such as an enzyme or glucose transporter, which functions essentially as that form of the protein expressed in individuals which do not have diabetes or a diabetes-associated disorder.

In certain embodiments, however, the use of adenoviral vectors capable of expressing proteins other than normally-functional GLUT-2 and glucokinase is also contemplated. For example, adenoviral vectors expressing the glucose transporter GLUT-1 are contemplated to have utility as control' vectors, particularly for use as controls in in vitro or in vivo studies concerning GLUT-2 expression by adenoviral vectors. Adenoviral vectors capable of expressing mutant or altered proteins are also encompassed by the present invention. Such vectors will have utility in that they will allow the mutant form, such as the glucokinase enzyme associated with maturity onset diabetes of the young (MODY) , a form of NIDDM, to be specifically expressed in a given cell type so that the effects of the mutation may be directly assessed at the molecular level. Finally, individuals or rodent models with NIDDM and related syndromes tend to exhibit markedly elevated insulin secretion at glucose concentrations that are substimulatory for normal islets. This dysfunction may be due to overexpression of a high affinity (low Km) glucose metabolizing protein such as GLUT-1 or hexokinase I. It is proposed that this abnormality may be corrected by expression of antisense versions of hexokinase or GLUT-1. In preferred embodiments, it is contemplated that one will desire to position the recombinant insert cDNA or gene under the control of a promoter that will restrict expression of the gene to /3-cells. An example of such a promoter would be the insulin promoter, which is specifically active in _/3-cells. Alternatively, expression of the gene of interest in jS-cells and a small subset of other cell types (preferential 3-cell expression) is also preferred. Examples of the latter class of promoters include the GLUT-2 promoter, which directs expression in islet /3-cells, liver, and certain cells of the intestine and kidney (Takeda et al . , 1993; Bell et al . _f 1990; Thorens et al . , 1990a), and the β-cell glucokinase promoter, which causes expression in 0-cells and other neuroendocrine cells of the pituitary (Hughes et al . _r 1991; Shelton et al . , 1992).

Specific or preferential targeting of the recombinant genes to islet jS-cells by the indicated techniques is preferred because it should prevent wholesale expression of the genes in all tissues. This is advantageous in that inappropriate expression of GLUT- 2 or glucokinase in other tissues such as muscle or fat, for example, tissues which normally express the insulin- regulatable transporter GLUT-4 and the glucose phosphorylating enzyme hexokinase II, might alter the glucose metabolizing properties of these tissues and affect whole-body glucose homeostasis. Recent studies suggest, however, that systemic injection of recombinant adenoviruses results in preferential expression in liver, probably because the blood supply of this tissue is in direct contact with the cells of the organ, as opposed to most other tissues, in which the blood supply is separated from the tissue mass by the endothelial lining of blood vessels. Interestingly, the islets of

Langerhans are a highly vascularized tissue which also have some direct contact with the blood supply, suggesting that they may also be a preferred target for adenovirus. Thus, it is proposed that expression of the gene of interest could be driven by any of a host of strong constitutive promoters such as CMV, viral LTR, or SV-40, or alternatively by promoters associated with genes that are expressed at high levels in all cells such as translation elongation factor-1 or actin. Recent studies conducted by the inventors indeed demonstrate that infusion of recombinant adenovirus containing the reporter gene /3-galactosidase demonstrates successful gene transfer into islets in vivo without significant transfer of the gene in the surrounding exocrine tissue.

The adenoviral vectors of the present¹ invention will also have utility in embodiments other than those connected directly with gene therapy. Alternative uses include, for example, in vitro analyses and mutagenesis studies of various candidate genes, and the recombinant production of proteins for use, for example, in antibody generation or other embodiments. An important use is envisioned to be the introduction of GLUT-2 cDNA into islets isolated from ZDF rats, or other models of β cell dysfunction, allowing the issue of whether GLUT-2 underexpression is sufficient to explain ablated glucose sensing in these cells to be addressed (Johnson et al . , 1990b) .

Furthermore, as concerns engineering of insulin producing cell lines in vitro, the adenovirus system will allow rapid introduction of genes into such cells. Thus, for example, current cell lines gain glucose-stimulated insulin secretion upon transfection with GLUT-2, but respond to glucose at subphysiological concentrations of the sugar. As described earlier, a likely explanation for the hypersensitive response to glucose is that hexokinase is the predominant glucose phosphorylating enzyme in the cell lines under study. Hexokinase is maximally active at subphysiological glucose concentrations, in contrast to glucokinase, which only becomes active at glucose concentrations within the physiological range. Adenovirus-mediated introduction of glucokinase and/or antisense hexokinase cDNAs into cells already engineered for GLUT-2 expression can allow rapid testing of this concept. The advantage of adenovirus- mediated gene transfer for the foregoing purpose is that it obviates the need to select clonal cell lines, a laborious and time-consuming procedure. The effect of genetic maneuvers on glucose-stimulated insulin secretion can be studied within 24 hours of viral infection, as opposed to cloning of stably transfected lines, which generally requires a month of work. In support of this contention, we have recently demonstrated potent induction of GLUT-2, glucokinase, and hexokinase I expression by adenovirus-mediated gene transfer into CV-1 cells, RIN cells, and normal primary islet cells (Ferber et al . 1994; Becker et al . , 1994b; Becker et al . , 1994c). Two specific examples of the use of adenovirus for concept testing in engineered cells are as follows. First, stable transfection of RIN cells of intermediate passage number with GLUT-2 confers glucose-stimulated insulin secretion and causes a spontaneous 4-fold increase in glucokinase enzymatic activity (Ferber et al . , 1994). The relationship between the presence of GLUT-2 and the increase in glucokinase activity has now been confirmed by adenovirus-mediated GLUT-2 expression in RIN cells, which also leads to a spontaneous increase in glucokinase activity (Ferber et al . , 1994). A second example concerns a virus containing the cDNA for hexokinase I in antisense orientation (the virus is designated AdCMV-HKIrev) , which was used to treat GLUT-2 transfected, glucose responsive RIN 1046-38 cells. A 70% reduction in immunodetectable hexokinase protein was observed in cells treated with AdCMV-HKIrev relative to untreated cells or other control cells treated with the virus expressing /3-galactosidase (BeltrandelRio et al . , 1994) . Importantly, cells treated in this manner shift their glucose-stimulated insulin response by two orders of magnitude such that maximal responsiveness is achieved at 5mM glucose in the AdCMV-HKIrev treated cells, compared to 50μM in control cells. These studies therefore validate the concept that down-regulation of hexokinase activity in RIN cells is an effective strategy for "normalizing" the glucose concentration dependence of the insulin secretion response.

Adenovirus-mediated gene transfer may also have utility in the context of cells destined for implantation into animals or humans with IDDM. Current information suggests that adenovirus DNA integrates into chromosomal DNA with poor efficiency, meaning that most of the adenovirus genome will reside in the nucleus as extrachromosomal DNA (Van Doren and Gluzman, 1984) . This means that adenovirus-mediated gene expression will be transient, as opposed to permanent. Current experience suggests that adenovirus-directed gene expression persists at high levels for 3-4 weeks in vivo when the gene expression is directed by the CMV promoter. However, it may be possible to augment gene expression by multiple injections of the relevant recombinant adenoviruses directly into the implanted tissue mass. Thus, in this embodiment, cells would be engineered for the correct levels of expression of GLUT-2, glucokinase, hexokinase and insulin by one or a combination of the methods described herein, and subsequently, gene expression could be supplemented by direct injection of relevant viruses into the implant. Since the implant is envisioned to be contained in a permselective device or capsule capable of excluding molecules of the immune system, the adenovirus will not exit the device, and gene therapy will therefore be restricted to the engineered cells within the device. In this way, relatively constant levels of expression of the genes of interest might be maintained, and therefore, relatively constant insulin secretion parameters. Accordingly, in such embodiments, the human cyto egalovirus (CMV) immediate early gene promoter (Thomsen et al . , 1984) may be used to achieve constitutive, high-level expression, as may any other viral or mammalian cellular promoters known to those of skill in the art (Sambrook et al. , 1989).

The adenovirus vectors of the present invention have been rendered replication defective through deletion of the viral early region 1 (ElA) region such that the virus is competent to replicate only in cells, such as human 293 cells, which express adenovirus early region 1 genes from their cellular genome. The recombinant virus will therefore not kill normal cells which do not express adenoviral early gene products and will thus be suitable for use in human gene therapy regimens. Techniques for preparing replication defective adenoviruses are well known in the art (Ghosh-Choudhury and Graham, 1987; McGrory et al., 1988; Gluzman et al . , 1982; Gerard et al . , USSN 07/823,747, filed 1/22/92).

In that the vectors of the present invention are replication defective, they will typically not have an adenovirus El region. Thus, it will be most convenient to introduce the region encoding the glucose-sensing protein at the position from which the El coding sequences have been removed. However, the position at which the coding region is inserted is not critical to the invention, and it may thus also be inserted in lieu of the deleted E3 region in E3 replacement vectors as described previously by Karlsson et al . (Karlsson et al ., 1986) .

Other than the requirement that the adenovirus vector be replication defective, the nature of the adenovirus vector is not believed to be crucial to the successful practice of the invention. The adenovirus may be of any of the 42 different known serotypes or subgroups A-F. Adenovirus type 5 of subgroup C is the preferred starting material in order to obtain the conditional replication-defective adenovirus vector for use in the method of the present invention. This is because Adenovirus type 5 is a human adenovirus about which there is significant amount of biochemical and genetic information known, and which has historically been used for most constructions employing adenovirus as a vector. In certain embodiments the present invention is also an adenoviral virion or adenovirus particle containing a vector construct expressing the proteins or mRNA as described above.

Further related aspects of the invention concern recombinant, or engineered, host cells which incorporate an adenoviral vector prepared in accordance herewith. In preferred embodiments, the recombinant adenovirus- containing host cell will be a eukaryotic or mammalian recombinant host cell, with pancreatic islet cells, such as β cells, being particularly preferred. These engineered cells are distinguishable from naturally occurring cells in that one or more cDNAs, genes or other nucleic acid expression units, have been introduced through the hand of man. Recombinant host cells may be obtained from an animal or human host subsequent to the systemic in vivo administration of an adenoviral construct prepared in accordance herewith. The cells may be obtained from the animal, for example, by biopsy, and employed for further detailed analyses, such as to determine the degree or longevity of recombinant gene expression in vivo . Alternatively, the recombinant adenovirus may be introduced into a recombinant host cell directly in vitro , for example, in an analysis of β cell gene function regarding the introduction of normal or mutant proteins, such as transporters, enzymes or isoenzymes, or in the over-production of certain proteins.

Further embodiments concern compositions comprising a vector construct which encodes a protein involved in glucose-sensing, such as GLUT-2, glucokinase and/or antisense hexokinase dispersed in a pharmacologically acceptable solution or buffer. As discussed above, in preferred embodiments, it is contemplated that one will desire to use a vector in which the expression region of the recombinant insert is positioned under the control of a β cell-specific promoter, such as the insulin promoter, so that the resultant construct may be employed in human gene therapy, directing the expression of recombinant genes in β cells only.

Preferred pharmacologically acceptable solutions include neutral saline solutions buffered with phosphate, lactate, Tris, and the like. Of course, one will desire to purify the vector sufficiently to render it essentially free of undesirable contaminants, such as defective interfering adenovirus particles or endotoxins and other pyrogens such that it will not cause any untoward reactions in the animal or individual receiving the vector construct. A preferred means of purifying the vector involves the use of buoyant density gradients, such as cesium chloride gradient centrifugation.

In still further embodiments, the invention relates to a method for providing glucose-responsive insulin secreting capability to a mammal with NIDDM or an NIDDM- like syndrome. In effect, this method provides a means of restoring normal β cell function in such animals or human patients. To achieve this, one would administer to the animal or human subjects a physiologically effective amount of a pharmaceutical composition comprising an adenovirus vector which encodes a protein involved in glucose-sensing, and preferably, a vector which encodes GLUT-2 and/or glucokinase, either alone or in combination with a virus containing the hexokinase cDNA in antisense orientation.

In that adenovirus is a virus that infects humans, there may be certain individuals that have developed antibodies to certain adenovirus proteins. In these circumstances, it is possible that such individuals might develop an immunological reaction to the virus. Thus, where an immunological reaction is believed to be a possibility, one may desire to first test the subject to determine the existence of antibodies. Such a test could be performed in a variety of accepted manners, for example, through a simple skin test or through a test of the circulating blood levels of adenovirus-neutralizing antibodies. In fact, under such circumstances, one may desire to introduce a test dose of on the order of 1 x 10^ to 1 x 10" or so virus particles. Then, if no untoward reaction is seen, the dose may be elevated over a period of time until the desired dosage is reached, such as through the administration of incremental dosages of approximately an order of magnitude.

It should also be pointed out that because the adenovirus vector employed is replication defective, it will not be capable of replicating in the cells that are ultimately infected. Moreover, it has been found that the genomic integration frequency of adenovirus is usually fairly low, typically on the order of about 1%. Thus, where continued treatment in certain individuals is required it may be necessary to reintroduce the virus every 6 months to a year. In these circumstances, it may therefore be necessary to conduct long term therapy, where the individual's plasma glucose and insulin levels are monitored at selected intervals. The particular cell line used to propagate the recombinant adenoviruses of the present invention is not critical to the present invention. The recombinant adenovirus vectors can be propagated on, e .g. , human 293 cells, or in other cell lines that are permissive for conditional replication-defective adenovirus infection, e.g., those which express adenovirus ElA gene products "in trans" so as to complement the defect in a conditional replication-defective vector. Further, the cells can be propagated either on plastic dishes or in suspension culture, in order to obtain virus stocks thereof.

In a further group of embodiments, the present invention is directed to methods of detecting the presence of diabetes-associated, or islet-cell directed, antibodies in a sample as a means of assessing the occurrence or risk of diabetes onset. For uses in connection with diagnostic or antibody detection aspects of the present invention, it is contemplated that numerous additional types of engineered cells will prove to be important, particularly those which exhibit an epitope of a selected antigen on their cell surface. Exemplary antigens include particularly GLUT-2, and also glutamic acid decarboxylase (the 64KD islet antigen and the less antigenic 67kD form) , insulin, proinsulin, islet 38KD protein, 65 kDa heat shock protein, selected immunoglobulins, insulin receptors or other types of islet cell antigens, whether cytoplasmic or surface. However, it may be desirable to employ cells that do not secrete insulin, in that antibody reactivity with insulin has been associated with false positive reactions.

Generally speaking, the cells are prepared by introducing genes expressing relevant epitopes into cultured cell lines that can be grown in unlimited quantity. However, in the context of immunologically- based detection methods there is no requirement that the cells be glucose-response or have insulin-secreting capability. All that is required is that the these cells express on their surface an epitope associated either with the onset of diabetes or, more generally, an islet cell epitope. Furthermore, there is no requirement that the cell actually express the entire protein, in that all that is ultimately required is that the cell express an epitope that is recognized by the antibody that is sought to be detected. Therefore, the invention contemplates that subfragments which comprise antigenic epitopes may be employed in place of the complete antigenic protein.

The first step of the detection methods of the invention will generally include obtaining a biological sample suspected of containing diabetes-associated or islet cell-directed antibodies. Generally speaking, the biological sample will comprise serum, plasma, blood, or immunoglobulins isolated from such samples. However, the method will be applicable to any sample containing antibodies, regardless of its source or derivation.

Next, the sample is contacted with an engineered cell expressing a diabetes-associated or islet cell- expressed epitope, under conditions effective to allow the formation of an immunocomplex between the expressed epitope and antibodies that may be present in the sample. This aspect is not believed to be particularly critical to the successful practice of the invention in that any incubation technique or conditions that favor immunocomplex formation may be employed. Preferred conditions include incubation of the cells with serum in isotonic media such as phosphate buffered saline or Hanks balanced salt solution.

Lastly, the method is completed by testing for the formation of an immunocomplex between the diabetes- associated or islet cell epitopes expressed by the cell and antibodies present in the sample, wherein a positive immunoreaction indicates the presence of the respective antibody in the sample. The testing method is not believed to be crucial to the overall success of the invention. Many types of testing procedures for detecting immunocomplex formation are known in the art and are applicable, including RIA, EIA, ELISA, indirect immunofluorescence, and the like. In general, all that is required is a testing/detection procedure that allows one to identify an interaction of immunoglobulins present in the sample and epitopes expressed on the surface of the engineered cell.

Certain approaches to the foregoing method will provide particular advantages. One such approach involves contacting the immunocomplexed cell with a molecule having binding affinity for the immunocomplexed antibody. The binding molecule is, generally speaking, any molecule that is capable of binding the immunocomplexed antibody, and that is detectable. Exemplary binding ligands include protein A, anti- immunoglobulin antibodies, protein G, or even complement. Preferably, the binding ligand includes an associated label that allows for the convenient detection of immunocomplexed antibodies. Typical labels include radioactive materials, fluorescent labels, and enzymes. Often, one may achieve advantages through the use of an enzyme such as alkaline phosphatase, peroxidase, urease, β-galactosidase or others that can be detected through use of a colorimetric substrate.

Other specific embodiments may include the use of associating ligands such as biotin, which can complex with avidin or streptavidin and thereby bring the enzyme or other label into association with the antibody or binding ligand. The detection of immunocomplexed cells through the use of a label may be further improved, and even automated, through the application of cell sorter technology that can identify or quantify cells having associated immunocomplexed antibodies. Particularly preferred is the use of a fluorescent label in conjunction with sorting of cells on a fluorescence- activated cell sorter. It has been found that such a system can screen 40-50 sera per hour using a single fluorescence-activated cell sorter (Inman et al . , 1993).

In other embodiments, one may simply employ a microscope slide test wherein cells are grown on polylysine coated slides, exposed to a test sample and then treated with an appropriate reagent capable of detecting immunocomplex formation. The presence of complexes can then be determined by direct viewing in a microscope, especially when the detecting reagent is an antibody that is labeled with a fluorescent marker.

An extension of such embodiments concerns the delineation of the specific epitope (or epitopes) within an antigenic protein, for example GLUT-2, that is recognized by antibodies in the sera of patients with diabetes. It is proposed that mutant or chimeric protein molecules can be constructed and expressed in recombinant AtT-20 cells, and used to investigate the binding of patients' antibodies, as described above. The failure of antibodies to bind to a mutant molecule after a specific deletion, or likewise, the ability of antibodies to bind to a chimeric molecule after a specific insertion, would allow the identification of the diabetes-specific epitope. Candidate epitopes include multiple extracellular "loop" regions of the GLUT-2 molecule. Once such an epitope is identified, synthetic peptides corresponding to the specific region of the protein sequence can be produced and used to develop simpler diagnostic procedures, for example, utilizing ELISAs or RIAs to detect the formation of an antibody/peptide complex.

It is further believed that the foregoing method may be employed as a technique for selection of engineered clonal cells that express epitopes recognized by autoantibodies. That is, one may prepare a series of clones which comprise, for example, cDNA prepared to islet cell mRNA, express these DNAs in a recombinant cell and screen the resultant recombinant cells with a known antibody composition to identify diabetes associated antigens in addition to those specific antigens discussed above.

In still further embodiments, the invention concerns a method for detecting the presence of diabetes- associated antibodies in a biological sample, such as a sample of serum, plasma, blood, or in immunoglobulins isolated therefrom. This method comprises contacting the sample suspected of containing diabetes-associated antibodies with intact GLUT-2-expressing cells under conditions effective to allow the interaction of any antibodies which may be present with GLUT-2, and then determining the degree of glucose uptake by the cells. Inhibition of glucose uptake indicates the presence of diabetes-associated antibodies in the sample.

Preferred cells for use in such embodiments are GLUT-2-expressing engineered cells, and particularly,

GLUT-2-expressing AtT20_ins cells. Suitable conditions for assays of this kind include incubating the cells with an IgG sample and determining the degree of glucose uptake using 3-O-methyl-β-D-glucose.

Further important embodiments concern methods of using the engineered cells of the present invention in the production of insulin, and particularly, in the production of human insulin which can be used in the treatment of IDDM. In certain aspects, the engineered artificial β cells are grown in culture and then contacted with a buffer containing glucose, thus stimulating the cells to produce and secrete insulin which can be collected and purified from the surrounding media. For use in connection with this aspect of the present invention, CGT-6 engineered cells are contemplated to be of particular use, but any cell prepared to secrete insulin in response to glucose may be employed.

It has discovered that a particularly useful approach to the production of human insulin in the above manner is the glucose-stimulation of artificial β cells grown in liquid culture. As such, the recombinant cells are contained within a column and subjected to perfusion with a buffer at a physiological pH, such as Krebs Ringer salt (KRS) solution, pH 7.4. To stimulate the production and secretion of insulin, the column of cells is perfused with a glucose-containing buffer, such as KRS, 5mM glucose. At this stage, the insulin-containing eluent from the column is collected, which provides ideal starting material for the purification of increased amounts of high-quality insulin for human use.

An alternative strategy for the isolation and purification of human insulin for use in IDDM therapy is to purify insulin directly from CGT-6 cells or other GLUT-2 transfected AtT-20 cell lines. This is now a viable possibility as the present inventors have demonstrated that GLUT-2 transfection causes an increase in intracellular insulin of approximately 5-fold in CGT-6 cells. These recombinant cells thus contain sufficient insulin to enable the large scale production of human insulin from CGT-6-like cells possible. BRIEF DESCRIPTION OF THE DRAWINGS

FIGURE 1. Northern blot demonstrating the presence of GLUT-2 mRNA in tissues and AtT-20_ins cell lines. Each lane contains 6 μg of total RNA. Samples were prepared from liver, anterior pituitary and islet tissue samples, as well as from untransfected (AtT-20_ins) and GLUT-2 transfected (AtT-20_ins CGT-5 and CGT-6) AtT-20_ins cell lines. The blot was probed with radiolabeled antisense GLUT-2 cRNA, and as a control for gel loading, with an antisense oligonucleotide probe for 18S rRNA (Chen et al . , 1990).

FIGURE 2. Immunoblot of GLUT-2 in tissues, untransfected cells (AtT-20_ins) and cells transfected with the CMV/GLUT-2 construct (AtT-20_ins CGT-5, -6) .

FIGURE 3A-C. Glucose transport into AtT-20i:_nns_<! cells.

FIGURE 3A: Measurements of 3-O-CH3 glucose uptake as a function of glucose concentration for untransfected AtT-20_ins cells (parental) and GLUT-2 transfected lines CGT-5 and CGT-6. The symbol legend is shown in the upper left corner of this panel.

FIGURE 3B: Reciprocal plot of glucose uptake versus 3-O-CH3 glucose concentration for GLUT-2 transfected lines CGT-5 and CGT-6. The calculated Km and Vmax values for glucose transport and the symbol legends are given in the upper left corner of the panel.

FIGURE 3C: Reciprocal plot of glucose uptake versus 3-0-CH3~glucose concentration for untransfected AtT-20_ins cells (parental cell line) . The calculated Km and Vmax values for glucose transport are indicated. Note the difference in the scales between figures 3B and 3C.

FIGURE 4A-B. Insulin release for AtT-20_ins cells in response to glucose, and glucose potentiation of forskolin induced secretion.

FIGURE 4A: Insulin release was measured from untransfected (AtT-20_ins) and GLUT-2 transfected (CGT-6) AtT-20ins lines incubated with varying glucose concentrations over the range of 0-20mM, or with 0.5tM forskolin (F) or 0.5μM forskolin + 2.5mM glucose (F + G) for a period of three hours. Data are normalized to the total cellular protein present in each secretion well and represent the mean ± SEM for 3-9 independent secretions per well condition. *, p < 0.001 compared to secretion at OmM glucose; #, p = 0.002 compared to secretion at OmM glucose.

FIGURE 4B: Insulin release was measured from untransfected (AtT-20ins) and GLUT-2 transfected (CGT-6) AtT-20ins lines incubated with 0.5μM forskolin (Fors) and 2.5mM glucose (Glc) in combinations indicated by the legend. Data are normalized to total cellular DNA in each secretion well and are expressed as the mean ± SEM for 3-9 independent measurements. Statistically significant increases in secretion relative to the -Glc, -Fors control are indicated by the symbol * (p < 0.001) .

FIGURE 5A-D: β-galactosidase expression in isolated rat islets treated with AdCMV-/?GAL recombinant adenovirus.

FIGURE 5A: Light microscopic view of a representative intact islet four days after transduction with AdCMV-jSGAL and treatment of the islets with chromogenic substrate, viewed at 20 x magnification.

FIGURE 5B: An islet 21 days after viral transductions treated similarly to that in Figure 5A, viewed at 20 x magnification.

FIGURE 5C: A multicell aggregate that was prepared by trypsin-mediated dispersal of intact islets prior to incubation with the chromogenic substrate, viewed at 40 x magnification.

FIGURE 5D: A control islet treated with β- galactosidase substrate after 4 days in culture without exposure to AdCMV-βGAL, viewed at 20 x magnification.

FIGURE 6A-B: Adenovirus-mediated expression of variant glucokinases in CV-1 cells.

FIGURE 6A: Glucokinase expression was assayed by western blot hybridization analysis using 100 μg islet protein and an antibody that detects rat islet glucokinase (Antibody U343, Quaade et al . , 1991). Lanes contain the following samples: Lane 1, Islet glucokinase expressed in bacteria (Quaade et al . , 1991). Lane 2, Uninfected islets; Lane 3, Islets infected with recombinant adenovirus containing the _/3-galactosidase cDNA; Lane 4, Islets infected with recombinant adenovirus containing cDNA encoding a glucokinase variant containing an amber mutation (premature stop codon) in place of glutamic acid 279 that is not synthesized as a stable protein; Lane 5, Islets infected with recombinant adenovirus containing cDNA encoding a glucokinase variant with a point substitution that results in replacement of amino acid glycine 261 with arginine; Lane 6, Islets infected with recombinant adenovirus containing cDNA encoding a glucokinase variant with a point substitution that results in replacement of amino acid glutamic acid 279 with glutamine.

FIGURE 6B: The glucokinase enzymatic activity measured in crude homogenates of the CV-1 cells described in the western blot.

FIGURE 7: Adenovirus-mediated expression of variant glucokinases in primary rat islets. Lanes contain the following samples: Lane 1, Uninfected islets; Lane 2, Islets infected with recombinant adenovirus containing the jS-galactosidase cDNA; Lane 3, Islets infected with recombinant adenovirus containing cDNA encoding a glucokinase variant containing an amber mutation (premature stop codon) in place of glutamic acid 279 that is not synthesized as a stable protein; Lane 4, Islets infected with recombinant adenovirus containing cDNA encoding a glucokinase variant with a point substitution that results in replacement of amino acid glycine 261 with arginine; Lane 5, Islets infected with recombinant adenovirus containing cDNA encoding a glucokinase variant with a point substitution that results in replacement of amino acid glutamic acid 279 with glutamine; Lane 6, empty; Lane 7, Islet glucokinase expressed in bacteria (Quaade et al . , 1991).

FIGURE 8A-F: Delivery of recombinant adenovirus to the islets of Langerhans in vivo .

FIGURE 8A: Five days after termination of the infusion, fresh-frozen 15μm pancreatic sections were prepared, incubated with the jβ-galactosidase substrate solution.

FIGURE 8B: Five days after termination of the infusion, fresh-frozen 15μm pancreatic sections were prepared, incubated with the jS-galactosidase substrate solution, and then treated with an anti-insulin antibody.

FIGURE 8C: Five days after termination of the infusion, fresh-frozen 15μm pancreatic sections were prepared, incubated with the jS-galactosidase substrate solution.

FIGURE 8D: Five days after termination of the infusion, fresh-frozen 15μm pancreatic sections were prepared, incubated with the 3-galactosidase substrate solution, and then treated with an anti-insulin antibody.

FIGURE 8E. A representative islet isolated from a virally infused animal that was treated with β- galactosidase substrate and then embedded and sectioned into 5 μm sections. Multiple blue nuclei are evident.

FIGURE 8F: In contrast to Figure 8E, islets isolated from a control animal are completely devoid of blue color.

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

ENGINEERING OF "ARTIFICIAL" β CELLS

Insulin dependent diabetes mellitus (IDDM) is caused by autoimmune destruction of insulin producing jS-cells. Islet transplantation has been extensively investigated as a strategy for curing IDDM, but suffers from the difficulties associated with procuring enough tissue. The present invention is based in part on the recognition that the problem of islet supply could potentially be circumvented if a non-islet cell type could be engineered to secrete insulin in response to metabolic signals, since such cells could be grown in unlimited quantity in vitro . Such cells could ultimately replace daily insulin injections as therapy for Type I diabetes.

The participation of the pancreatic islets of Langerhans in fuel homeostasis is mediated in large part by their ability to respond to changes in circulating levels of key metabolic fuels by secreting peptide hormones. Accordingly, insulin secretion from islet β- cells is stimulated by amino acids, three-carbon sugars such as glyceraldehyde, and most prominently, by glucose. While these diverse secretagogues may ultimately work through a final common pathway involving alterations in K+ and Ca++ channel activity and increases in intracellular Ca++ (Prentki et al . , 1987; Turk et al . , 1987) , the biochemical events leading from changes in the levels of a particular fuel to insulin secretion are initially diverse. In the case of glucose, transport in the jS-cell and metabolism of this sugar are absolute requirements for secretion, leading to the hypothesis that its specific stimulatory effect is mediated by and proportional to its flux rate through glycolysis and related pathways (Ashcroft, 1980; Hedeskov, 1980; Meglasson and Matchinsky, 1986; Prentki et al . , 1987; Turk et al . , 1987; Malaisse et al . , 1990). Strong support for this view comes form the finding that non- metabolizable analogues of glucose such as 3-O-methyl or 2-deoxy glucose fail to stimulate insulin release (Ashcroft, 1980; Meglasson et al . , 1986).

IDDM has traditionally been treated by insulin replacement, either classically, by external administration, or experimentally, by transplantation of islets or pancreas fragments. The latter strategy is not likely to be broadly applicable because of the difficulty and expense associated with the isolation of large numbers of islets. The present invention is directed to an alternative approach, that of using molecular techniques to engineer an "artificial /3-cell", i.e., a non-islet cell capable of performing glucose-stimulated insulin secretion, which can be grown in unlimited quantity in vitro.

The anterior pituitary cell line AtT-20_iπs is preferred because of important similarities to 3-cells. First, these cells have been modified for insulin gene expression by stable transfection with a viral promoter/proinsulin cDNA construct (Moore et al . , 1983). Second, AtT-20_ins cells are able to process the preproinsulin mRNA and preprotein to yield the correctly processed insulin polypeptide. Third, their secretory response to analogues of cAMP compares favorably with the well differentiated hamster insulinoma (HIT) cell line (Moore et al . , 1983). Finally, AtT-20_ins cells contain significant amounts of the islet isoform of glucokinase (Hughes et al . , 1991), making this the only tissue other than liver or islets in which glucokinase gene expression has been reported.

On the other hand, AtT-20_ins cells differ from islet jβ-cells in two important ways. First, they do not secrete insulin in response to glucose, and second, they express the low Km GLUT-1 glucose transporter mRNA and not GLUT-2 (Hughes et al. , 1991). The inventors hypothesized that the lack of glucose responsiveness in Atτ-20_ins cells could be explained either by deficient capacity or altered affinity of glucose uptake relative to normal islets. To test this hypothesis, AtT-20_ins cells were stably transfected with GLUT-2 cDNA. Surprisingly, it has been found that cells engineered in this way gained glucose-stimulated insulin secretion and glucose potentiation of non-glucose secretagogue stimulation, albeit with a dose-response curve that is different from normal islets (Hughes et al . , 1992). Engineering of the AtT-20_ins cells generally involved construction of a suitable GLUT-2 expression vector, transfection of AtT-20_ins cells with the vector, and selection of stable transfectants. To accomplish this, rat islet GLUT-2 cDNA (Johnson et al . , 1990a) was cloned into the vector pCB-7, a derivative of vector pCMV4 (Anderson et al . , 1989), immediately downstream of its cytomegalovirus (CMV) promoter. pCB-7 was constructed by Drs. Michael Roth and Colleen Brewer of the Biochemistry Department, University of Texas Southwestern Medical Center. It differs from pCMV-4 in that it contains a hygromycin resistance gene; thus, cells transfected with the pCB7/GLUT-2 construct can be selected for stable integration of the vector DNA into the cell's genome by treatment with hygromycin. AtT-20_ins cells were transfected with this construct using electroporation, and stable transfectants were selected with hygromycin.

PHYSIOLOGICAL RESPONSE TO GLUCOSE

Although the engineering of an AtT-20_ins cell line with glucose-stimulated insulin secretion has been accomplished, maximal insulin secretion from these cells occurs at a much lower glucose concentration than observed for normal islets, which do not respond at levels less than the fasting glucose concentration of approximately 4-5 mM, and which have not reached maximum secretion at the upper range of physiological glucose (10 mM) . The potentiating effect of glucose on forskolin, dibutyl cAMP, or IBMX induced insulin secretion from AtT- ²⁰ins ^ceH^s ^^{s a}l^so maximal at low glucose. The heightened sensitivity of GLUT-2 transfected AtT-20_ins cells to both the direct and potentiating effects of glucose is reminiscent of a number of cell lines derived from insulinoma (β cell) tumors (Praz et al . , 1983;

Halban et al . , 1983; Giroix et al . , 1985; Meglasson et al . , 1987; Clark et al . , 1990). For example, the rat insulinoma cell line RIN 1046-38 is responsive to glucose when studied after short periods of time in cell culture (between passages 6-17) , albeit with a maximal response at sub-physiological glucose levels, as in transfected AtT-20_ins cells. With longer time in culture (passage number greater than 50) , all glucose-stimulated insulin secretion is lost (Clark et al . , 1990). Low passage RIN 1046-38 cells contain both glucokinase and GLUT-2, but lose expression of these genes when studied at higher passages.

The fact that both transfected AtT-20_ins cells and RIN1046-38 cells of low passage number respond to subphysiological levels of glucose, despite expression of glucokinase and GLUT-2, suggests that these cells share metabolic determinants that can override the regulatory function of the high Km components. Given that the glucose transport kinetics of normal islets are recapitulated in GLUT-2 transfected AtT-20_ins cells, the increased sensitivity of the clonal cells to glucose might alternatively be explained by alteration in regulation of glucose phosphorylation. While hexokinase activity is readily measured in islet cell extracts, this enzyme is thought to be potently inhibited (by as much as 95%) inside the intact islet cell (Trus, et al . , 1981; Giroix et al . , 1984). Thus, in the presence of stimulatory concentration of glucose, normal islets have both sufficient glucokinase activity and inhibited hexokinase (the levels of glucose-6-phosphate, an inhibitor of hexokinase, increase during glucose stimulation) to allow the control of glucose metabolism to be tied directly to glucokinase activity (Km of - 10 mM in islets) (Meglasson and Matchinsky, 1986) .

Recent studies have shown that restoration of GLUT-2 expression by transfection of intermediate passage RIN 1046-38 cells with the same GLUT-2 cDNA containing plasmid used for engineering of AtT-20_ins cells confers glucose-stimulated insulin release (Ferber et al . , 1993; Ferber et al . , 1994). This result validates the concept that GLUT-2 transfection of multiple endocrine cell types confers the capacity for glucose response. In addition, GLUT-2 transfection of RIN 1046-38 cells caused a 4-fold increase in glucokinase enzymatic activity, an effect not seen in GLUT-2 transfected AtT-20_ins cells. These results suggest that GLUT-2 and glucokinase expression are somehow linked, and that RIN cells may not require a separate glucokinase gene transfer step for physiologically relevant glucose sensing. The relationship between the presence of GLUT-2 and the increase in glucokinase activity has been confirmed by adenovirus-mediated GLUT-2 expression in RIN cells, which also leads to a spontaneous increase in glucokinase activity (Ferber et al . , 1994).

In support of this point, the inventors have recently performed the following study. GLUT-2 transfected RIN 1046-38 cells were preincubated with 50 mM 2-deoxyglucose for 30 minutes. 2-deoxglucose is readily transported into mammalian cells and phosphorylated by glucokinase and hexokinase, but once phosphorylated, this glucose analog is not metabolized further. Thus, administration of 2-deoxyglucose to cells will result in accumulation of 2-deoxyglucose-6- phosphate, a potent inhibitor of hexokinase, but not glucokinase. When cells were preincubated with 2- deoxyglucose and subsequently examined for their glucose- stimulated insulin secretion response, it was found that these cells secreted insulin in response to glucose concentrations of 5 mM or greater. This is the expected glucose concentration threshold for normal β-cells. In sum, this study validates the concept that inhibition of hexokinase activity in insulin producing clonal cells in the face of enhanced glucokinase activity allows expression of a physiologically relevant glucose- stimulated insulin secretion response.

In addition to the chemical approach to reduction of hexokinase activity with 2-deoxyglucose, the inventors have also demonstrated the importance of control of the glucokinase/hexokinase ratio by molecular approaches. First, a recombinant adenovirus containing hexokinase I in sense orientation (AdCMV-HKI) was used to overexpress the enzyme in normal islets. Such islets secreted significantly more insulin at nonstimulatory glucose concentrations (3mM glucose) than control islets (Becker, et al . , 1994b) . The inventors have also constructed a virus containing the cDNA for hexokinase I^■in antisense orientation (the virus is designated AdCMV-HKIrev) and used it to treat GLUT-2 transfected, glucose responsive RIN 1046-38 cells. A 70% reduction in immunodetectable hexokinase protein was observed in cells treated with AdCMV-HKIrev relative to untreated cells or other control cells treated with the virus expressing jS-galactosidase

(BelTrandelRio et al . , 1994). Importantly, cells treated in this manner shift their glucose-stimulated insulin response by two orders of magnitude such that maximal responsiveness is achieved at 5 mM glucose in the AdCMV- HKIrev treated cells, compared to 50μM in control cells. These studies therefore validate the concept that down- regulation of hexokinase activity in RIN cells is an effective strategy for "normalizing" the glucose concentration dependence of the insulin secretion response.

AtT-20_ins cells have glucokinase activity, but it represents only 9% of total glucose phosphorylation in these cells, and only 32% of the activity measured in normal islets (Table 1 in Example I below) ; the proportions of glucose phosphorylating enzymes in RIN1046-38 cells are similar to those found in AtT-20_jns cells. Hexokinase I, the isoform that is expressed in most clonal cell lines (Arora et al . , 1990) is found bound to mitochondria and in a free cytosolic form (Lynch et al . , 1991). In the former state, the enzyme is less sensitive to glucose-6-phosphate inhibition (Wilson,

1984) . Thus, in addition to the fact that AtT-20^ and RIN cells have reduced glucokinase activity, they may also have altered regulation of hexokinase such that it becomes the predominant glucose phosphorylating enzyme at any concentration of glucose studied.

The increased sensitivity of GLUT-2 expressing AtT-20ins or RIN cells can be explained as follows. Expression of the GLUT-2 transporter not only increases the Km for transport, but also the transport capacity at all glucose concentrations studied. Our data show that at 2.5 mM glucose, for example, there is an approximately 10-fold increase in glucose uptake in the GLUT-2 transfected cells compared to the parental line (see Figure 3A) . This means that even at glucose concentrations that would be sub-stimulatory for islets, transport into GLUT-2 transfected AtT-20_ins cells will be rapid and hexokinase activity (Km for glucose of -0.01 mM) will be maximal, and the generation of glucose- related secretory signals will be maximized at low glucose as a consequence.

As discussed above, an imbalance in the hexokinase/glucokinase ratio may at times result in maximal insulin secretory response at subphysiological glucose concentrations. It is proposed that a more physiologic glucose response may be achieved by "knocking out" hexokinase activity in engineered cells of the present invention. One approach is to co-transfect these cells with antisense hexokinase constructs. This can be achieved, for example, using the CMV vector system described for GLUT-2 transfection, with the exception that the plasmid will contain an alternate resistance gene, such as puromycin or histidinol, since the AtT-20_ins cell line is resistant to both neo ycin (due to stable integration of the SV40-insulin-neo chimeric construct) and hygromycin (due to stable integration of the CMV- GLUT-2-hygromycin chimeric construct) . Alternatively, recombinant adenoviruses can be used for introduction of multiple genes at one time. Recently, the hexokinase isozyme expressed in mouse hepatoma cells has been cloned and characterized (Arora et al . , 1990) and shown to be approximately 92% identical to the hexokinase I sequences derived from rat brain (Schwab et al . , 1989) and human kidney (Nishi et al . , 1988).

In order to generate antisense probes with exact sequence identity to the homologue of hexokinase I being expressed by the engineered cell, the hexokinase variant present in the cell was converted to cDNA by reverse transcribing the mRNA and amplification of the DNA product, a procedure recently employed in one of the inventors' laboratory for amplification of glucokinase mRNA from islets, RIN cells, AtT-20_ins cells, and primary anterior pituitary cells (Hughes et al., 1991). The oligonucleotides used for amplification were based on the published sequence of the mouse hepatoma hexokinase I (Arora et al . , 1990). The oligonucleotides included restriction enzyme recognition sequences at their 5' ends to facilitate directional cloning of the amplified cDNA into the selected vector in an antisense orientation.

It is proposed that engineered lines may be transfected with antisense constructs by electroporation. After appropriate selection to obtain colonies that have stably integrated the antisense hexokinase construct into their genome, expression of the antisense mRNA can be evaluated by hybridization to labeled sense RNA, e.g., prepared with the pGEM vector system (Promega) . Alternatively, antisense constructs can be introduced via adenovirus. Blot hybridization analysis may be carried out not only with the probe corresponding to the antisense construct, but to regions outside as well, since cellular factors known to unwind RNA:RNA duplexes result in modification of that RNA, thus interfering with its detection on Northern blots (Walder, 1988) . One may also assess whether the presence of antisense mRNA is capable of affecting the level of hexokinase protein(s) through the use of antibodies against relevant hexokinase sequence(s) .

Should the foregoing general antisense approaches fail to provide adequate hexokinase suppression in the particular system selected, modified antisense oligonucleotides may be employed. For example, an antisense oligonucleotide may be prepared to sequences surrounding and/or containing the ATG initiation codon, for example, and introduced into cells by simply incubating the cells in media containing the oligonucleotide at high concentration. This approach bypasses uncertainties about the stability of longer antisense hexokinase transcripts synthesized from the construct and should provide suppression of hexokinase activity for a period of time sufficient to assess the functional consequences. On the negative side, the oligonucleotide antisense procedure can only cause a transient reduction in endogenous expression, and is thus not applicable to the engineering of a stable "artificial" β cell.

Should complete normalization of glucose responsiveness require more than 70% suppression of hexokinase activity in engineered cells, this may be achieved by the "ribozyme" modification of the antisense approach. In this strategy, the ribozyme catalytic domain, which is a piece of RNA with the capacity to cleave other RNA molecules (Forster and Symons, 1987; Hasellof and Gerlach, 1988) is inserted between fragments of antisense RNA complementary to the gene to be targeted (hexokinase I, for example). The RNA degrading activity is thereby targeted to the gene of interest, theoretically providing more efficient reduction in the level of expression of the targeted RNA. The system has recently been used to achieve a 45% reduction in glucokinase mRNA in islets of transgenic animals (Efrat et al . , 1994b) . It remains to be determined whether this strategy will provide better than 70% reduction in hexokinase I expression.

A second alternative that bypasses the issue of effectiveness of antisense strategies altogether would be to knock out the endogenous hexokinase gene of interest in the cells using a positive/negative selection protocol (Mansour et al . , 1988; Capecchi, 1989; Zheng et al . , 1990) to select for homologous recombination of a hexokinase gene segment that renders the endogenous hexokinase gene nonfunctional. This approach involves cloning of at least a segment of the hexokinase gene(s) expressed in the engineered cells either by library screening or PCR amplification, and construction of a vector that contains a genomic fragment, preferably containing exons that encode the putative ATP or glucose binding sites (Arora et al . , 1990; Schwab et al . , 1989; Nishi et al . , 1988; Andreone et al . 1989). These also are then interrupted by insertion of an antibiotic resistance gene (e.g., puromycin) and cloned into a targeting vector adjacent to a copy of, e.g., the herpes simplex virus (HSV) thymidine kinase gene.

The plasmid is then introduced into cells by electroporation and homologous recombination events are selected for by incubation of the cells in puromycin and FIAU, a recently described thymidine kinase substrate (Capecchi, 1989; available from Dr. Richard White, Bristol Myers/Squibb, Walingford, CT) . The action of FIAU is exerted as follows. If recombination occurs at a nonhomologous site, the viral thymidine kinase gene is retained in the genome and expressed, rendering cells extremely sensitive to FIAU. If the disrupted gene is inserted at its homologous site (the endogenous hexokinase gene) , in contrast, the viral thymidine kinase gene is lost, and the cells are tolerant of the drug. While homologous recombination in mammalian cells is a relatively rare event, the selection strategy is sound, and has recently been applied to mammalian tissue culture cells (Zheng et al . , 1990).

Although glucokinase activity is present in

AtT-20ins cells, the activity of 0.7 U/g protein is only about 25% of the activity in normal islet cells, which contain approximately 3.1 U/g protein. One may therefore desire to increase the glucokinase activity of engineered cells, using a cDNA clone for the islet isoform of glucokinase (Newgard, 1990; Hughes et al . , 1991) and the strategies and vector systems described above. If the assumption about hexokinase overexpression is correct, it may be necessary to increase glucokinase expression in engineered cells having a reduced hexokinase activity in order to observe any effects on glucose responsiveness. Note that creation of a GLUT-2⁺, insulin⁺, glucokinase- overexpressing, hexokinase^" cell line will require cotransfection of some of the relevant constructs, since limited numbers of resistance-gene containing plasmids are available. Efficient cotransfection can be expected when using either electroporation or CaP0₄ precipitation transfection strategies (Sambrook et al. , 1989). Again, recombinant adenovirus vectors can also be used for cases in which the introduction of multiple gene products is desired. As described above, the complication of increasing glucokinase expression may not be relevant to GLUT-2 transfected RIN cells, which undergo a 2-3 fold increase in glucokinase activity upon transfection with GLUT-2.

USE OF ENGINEERED CELLS IN HUMAN TREATMENT

A: TREATMENT OF NON-INSULIN DEPENDENT DIABETES MELLITUS (NIDDM)

It is proposed that the engineered cells of the present invention will be particularly advantageous in the treatment of insulin dependent diabetes following their introduction into such diabetic animals or human patients. However, these cells are also contemplated to be of use in the treatment of non-insulin dependent diabetes mellitus (NIDDM; type 2 diabetes) , particularly for the reasons set forth below.

The inventors have demonstrated in rat and human models of type 2 diabetes that the ratio of amylin to insulin is extremely high. Amylin has been shown to cause insulin resistance in muscle and to increase the production of glucose by the liver, effects that lead to hyperglycemia. Amylin is normally co-secreted by the β cells with insulin (Ogawa et al . , 1990). In nondiabetics the ratio of amylin to insulin is always below 2%. In animals or humans with type 2 diabetes the ratio of amylin to insulin exceeds 2%. In obese rats in which the β cells, the source of insulin and amylin, have been destroyed the inventors determined that replacement of insulin and amylin so as to maintain an amylin-insulin ratio below 2% is not associated with hyperglycemia, but when the ratio exceeds 2% hyperglycemia invariably occurs. Based on these studies they propose that an excess of amylin relative to insulin is a cause of type 2 diabetes. The inventors have previously shown that insulin inhibits the secretory activity of β cells in obese humans (Elahi et al . , 1982). Although amylin was not known to exist at the time that these studies were performed, it is proposed that insulin will be a powerful suppressant of amylin secretion. By implanting a source of insulin secretion into type 2 diabetic individuals, i.e., the engineered insulin-secreting cells of the present invention, the β cells will become hypoactive and the insulin levels will be derived not from β cells that co-secrete amylin but rather from the engineered cells that secrete only insulin. This should greatly enhance the effectiveness of the insulin by eliminating amylin- mediated insulin resistance.

B: TREATMENT OF OBESITY

Studies of the inventors, and other workers, have demonstrated that amylin levels are high in obese individuals and low in thin persons. The present inventors propose that amylin directs the anabolic actions of insulin towards an increase in the formation of fat from dietary glucose. Based on this hypothesis, they predict that suppression of the relative concentration of amylin to insulin will reduce the flow of glucose into the formation of fat. By putting the β cells at rest through implantation of cells that secrete insulin without co-secretion of amylin, such as those described herein, one would greatly reduce the amylin-insulin ratio. This would reduce the flow of glucose carbon into biosynthesis of fat and thereby facilitate the dietary control of obesity. C: TREATMENT OF INSULIN DEPENDENT DIABETES MELLITUS (IDDM)

It is proposed that the engineered cells of the present invention may be particularly advantageously employed in the treatment of animals or human patients with insulin dependent diabetes in that such cells can sense glucose and respond by secreting insulin. Although ideally cells are engineered to achieve glucose dose responsiveness more closely resembling that of islets, it is believed that implantation of the CGT-5 or CGT-6 GLUT-2 expressing cells will also achieve advantages in accordance with the invention. It should be pointed out that the studies of Madsen and coworkers have shown that implantation of poorly differentiated rat insulinoma cells into animals results in a return to a more differentiated state, marked by enhanced insulin secretion in response to metabolic fuels (Madsen et al . , 1988) . These studies suggest that exposure of engineered cell lines to the in vivo milieu enhances some of their response(s) to secretagogues.

Engineered cells may be implanted using the alginate-polylysine encapsulation technique of O'Shea and Sun (O'Shea and Sun,1986), with modifications as recently described by Fritschy (Fritschy et al . , 1991). The engineered cells are suspended in 1.3% sodium alginate and encapsulated by extrusion of drops of the cell/alginate suspension through a syringe into CaC^. After several washing steps, the droplets are suspended in polylysine and rewashed. The alginate within the capsules is then reliquified by suspension in 1 mM EGTA and then rewashed with Krebs balanced salt buffer. Each capsule should contain several hundred cells and have a diameter of approximately l mm. Implantation of encapsulated islets into animal models of diabetes by the above method has been shown to significantly increase the period of normal glycemic control, by prolonging xenograft survival compared to unencapsulated islets (O'Shea and Sun, 1986; Fritschy et al . , 1991). Also, encapsulation will prevent uncontrolled proliferation of clonal cells. Capsules containing cells are implanted (approximately 1,000- 10,000/animal) intraperitoneally and blood samples taken daily for monitoring of blood glucose and insulin.

Recently, further methods for implanting islet tissue into mammals have been described (Lacy et al . , 1991; Sullivan et al . , 1991; each incorporated herein by reference) . Lacy and colleagues encapsulated rat islets in hollow acrylic fibers and immobilized these in alginate hydrogel. Following intraperitoneal transplantation of the encapsulated islets into diabetic mice, normoglycemia was reportedly restored. Similar results were also obtained using subcutaneous implants that had an appropriately constructed outer surface on the fibers. It is therefore contemplated that engineered cells of the present invention may also be straightforwardly "transplanted" into a mammal by similar subcutaneous injection.

The development of a biohybrid perfused "artificial pancreas", which encapsulates islet tissue in a selectively permeable membrane, has also been reported (Sullivan et al . , 1991). In these studies, a tubular semi-permeable membrane was coiled inside a protective housing to provide a compartment for the islet cells. Each end of the membrane was then connected to an arterial polytetrafluoroethylene (PTFE) graft that extended beyond the housing and joined the device to the vascular system as an arteriovenous shunt. The implantation of such a device containing islet allografts into pancreatectomized dogs was reported to result in the control of fasting glucose levels in 6/10 animals. Grafts of this type encapsulating engineered cells could also be used in accordance with the present invention.

An alternate approach to encapsulation is to simply inject glucose sensing cells into the scapular region or peritoneal cavity of diabetic mice or rats, where these cells are reported to form tumors (Sato et al . , 1962). Implantation by this approach may circumvent problems with viability or function, at least for the short term, that may be encountered with the encapsulation strategy. This approach will allow testing of the function of the cells in animals but will require further safety testing before it is used as a strategy for treating human diabetes.

With what is learned from engineering of clonal cell lines, it may ultimately be possible to engineer primary cells isolated from patients. Dr. Richard Mulligan and his colleagues at the Massachusetts Institute of Technology have pioneered the use of retrovirus vectors for the purposes of introducing foreign genes into bone marrow cells (Cone et al . , 1984; Danos et al . , 1988). The cells of the bone marrow are derived from a common progenitor, known as pluripotent stem cells, which give rise to a variety of blood borne cells including erythrocytes, platelets, lymphocytes, macrophages, and granulocytes. Interestingly, some of these cells, particularly the macrophages, are capable of secreting peptides such as tumor necrosis factor and interleukin 1 in response to specific stimuli. There is also evidence that these cells contain granules similar in structure to the secretory granules of β cells, although there is no clear evidence that such granules are collected and stored inside macrophages as they are in β cells (Stossel, 1987). Nevertheless, it may ultimately be possible to use the recombinant DNA for glucose transporters and glucose phosphorylating enzymes in combination with the recombinant insulin gene in a manner described for clonal cells to engineer primary cells that perform glucose- stimulated insulin secretion. This approach would completely circumvent the need for encapsulation of cells, since the patient's own bone marrow cells would be used for the engineering and then re-implanted. These cells would then develop into their differentiated form (i.e., the macrophage) and circulate in the blood where they would be able to respond to changes in circulating glucose by secreting insulin. A particularly advantageous strategy for such an approach' involves the use of recombinant adenovirus which the inventors have shown allows the introduction of genes into primary cells with very high efficiency.

USE OF ADENOVIRUS CONSTRUCTS IN ISLET CELL GENE THERAPY

Adenovirus gene transfer systems are based upon recombinant, engineered adenovirus which is rendered replication-incompetent by deletion of a portion of its genome, and yet still retains its competency for infection (Sen et al . , 1988). Adenovirus-mediated gene transfer has recently been investigated as a means of mediating gene transfer into eukaryotic cells and into whole animals. For example, in treating mice with the rare recessive genetic disorder ornithine transcarbamylase (OTC) deficiency, it was found that adenoviral constructs could direct the expression of normal levels of the OTC enzyme in 4 out of 17 instances (Stratford-Perricaudet et al . , 1990). Similarly, the gene for human a^-antitrypsin has been introduced into the liver of normal rats by intraportal injection, where it was expressed and resulted in the secretion of the introduced human protein into the plasma of these rats (Jaffe et al . , 1992) .

Recent studies by the present inventors have shown that adenovirus vectors can transfer a cDNA encoding rabbit muscle glycogen phosphorylase into primary hepatocytes with very high efficiency (Gomez-Foiz et al . , 1992) such that 86% of the primary hepatocyte expressed the muscle phosphorylase gene as determined by in situ hybridization. The remarkable efficiency of phosphorylase gene transfer into primary, non-dividing cells suggested to the inventors that this system had great potential for use in gene therapy protocols for the treatment of diabetes. This has been substantiated in the more recent work of the inventors in which a recombinant adenovirus containing the bacterial β- galactosidase gene has been used to demonstrate a 75-80% efficiency of gene transfer into isolated islets of Langerhans (Becker et al . , 1993), see figures 5-8. In addition, infusion of this virus into animals results in _/3-galactosidase expression in islets but not in surrounding exocrine tissue. The inventors further determined that luciferase, a reporter gene, is actively expressed in multiple tissues when a recombinant adenovirus is injected into the systemic circulation. It is proposed that the various properties of adenovirus may be adapted for use in therapeutic strategies directed to the treatment of diabetes, particularly in the treatment of NIDDM, but also in IDDM.

A: ADENOVIRUS-MEDIATED GENE THERAPY FOR NIDDM

NIDDM, in contrast to IDDM, is a disease with two coexisting but distinct derangements: (l) Insulin resistance, or the inability of insulin to exert its normal metabolic effect on fat, muscle, and liver, its primary target tissues, and (2) β cell failure, in which the insulin secretory response to glucose is lost, resulting in the inability to correct hyperglycemia. The challenges for therapy in NIDDM are thus complex, in that the genetic and physiological bases for insulin resistance and β cell failure are still incompletely understood.

There is a long-standing, and still unresolved debate about whether either the insulin resistance or β cell dysfunction are "primary", and whether one syndrome causes the other. On the one hand, many NIDDM patients are hyperinsulinemic, especially in the early phases of the disease, suggesting that aberrant glucoregulation is a function of reduced insulin effectiveness. On the other hand, first degree relatives of NIDDM probands with normal glucose tolerance are found to exhibit an exaggerated insulin secretory response to a glucose load, suggesting that the β cell defect may be in place prior to frank insulin resistance and hyperglycemia, and that insulin resistance is a consequence of rather than a cause of hyperinsulinemia. In the absence of any clear resolution of the controversy, workers in the field are pursuing identification of the gene(s) predisposing to both components of the disease, given the clear clustering of disease incidence in families.

Thus far, the favored approach for human studies has been to search for mutations in "candidate genes" that are involved in insulin responsiveness in peripheral tissues or in glucose-sensing in the β cell. However, other candidate genes have emerged from more traditional studies of animal models of β cell dysfunction and NIDDM- like syndromes. Naturally, the identification of particular genes from the possible candidate genes that may be involved in either arm of the NIDDM disease processes would open up new avenues of therapeutic intervention in the form of gene therapy.

Gene Therapy for Insulin-Resistance in NIDDM

Investigation of the relationship between two logical candidate genes, the insulin receptor itself and the GLUT-4 glucose transporter, to the insulin-resistance aspects of NIDDM has been undertaken. A variety of insulin receptor mutations have been detected in rare severe insulin resistance syndromes such as type A insulin resistance and leprechaunism. In NIDDM, however, only mild, diet-reversible defects in insulin receptor function can be documented, and in general; the insulin receptor gene and its expression are found to be normal in such patients (Bell, 1991) . Similarly, no obvious association of GLUT-4 gene mutations with the disease have yet been found.

There are also a number of other genes that could potentially be involved in insulin resistance. For example, the reduced levels of muscle glucose-6-phosphate in patients with NIDDM (Rothman et al . , 1992) may be due to a reduction in glucose transporter activity, or to altered hexokinase II activity. In addition to hexokinase, glycogen synthase, glycogen phosphorylase and their attendant regulatory kinases and phosphatases have been considered as candidate genes, due in large part to demonstrated impairment in non-oxidative glucose metabolism (i.e., glycogen synthesis) in muscle tissues of NIDDM patients. Finally, candidate genes are likely to emerge from further studies on the mechanism of insulin-receptor mediated signal transduction. Possible players could include G proteins, which have been implicated in insulin action, or proteins with tyrosine kinase activity and their substrates that may participate in an insulin-activated "phosphorylation cascade", leading to altered target enzyme functions.

Following the identification of genes causing a predisposition to insulin resistance in muscle, gene therapy strategies for this aspect of NIDDM will then be possible. To date, transfer of genes into muscle cells has been demonstrated by two methods, but both have their drawbacks. First, direct injection of plasmid DNA into skeletal muscle in vivo has been shown to result in expression of the transferred gene for periods of up to two months (Wolff et al . , 1990), however, this is a relatively inefficient mode of gene delivery. The second protocol involves engineering of myoblasts in vitro, followed by injection of the altered cells into muscle, whereupon they fuse with existing tissue and form multi- nucleated myotubes (Barr and Leiden, 1991; Dhawan et al . , 1991) . The expression of introduced genes, directed by a retrovirals long-terminal repeat (LTR) , can persist for up to three months in syngeneic mice. This technique is limited by the use of either muscle satellite cells obtained by biopsy, or the permanent mouse myoblast cell line, C2C12, which can form tumors in recipient animals.

Broad application of this technology would therefore require development of strategies for gene transfer into primary myoblasts, either derived directly from the patient, thus avoiding immunological complications, or involving allografts that must be protected by immunosuppression. It remains to be determined whether primary myoblast lines will fuse with existing muscle tissue and whether transgene expression will be maintained in such cells. However, adenovirus is known to infect mouse skeletal muscle cells in vivo (Stratford- Perricaudet et al . , 1992; Ragot et al . , 1993; Herz and Gerard, 1993) following either intravenous or intramuscular injection. Given these limitations, the use of adenovirus-mediated gene transfer in supplying copies of normal genes involved in insulin responsiveness is therefore very attractive.

Gene Therapy for β Cell Dysfunction in NIDDM

Two of the genes discussed above in the context of engineering glucose-sensitive insulin secretion have also been implicated in the development of β cell dysfunction in NIDDM-like syndromes, namely GLUT-2 and glucokinase. Remarkably, underexpression of GLUT-2 has been found in every rodent model studied to date in which glucose- stimulated insulin secretion is compromised (Unger et al . , 1991). This suggests that loss of GLUT-2 expression in islet β cells may be a primary event in the pathogenesis of NIDDM, although this has yet to be unambiguously established. A further gene of glucose metabolism encoding the glucose-phosphorylating enzyme glucokinase has recently been reported to be altered in patients from French families afflicted with a disorder known as Maturity Onset Diabetes of the Young (MODY; Frougel et al., 1992; Vionnet et al . , 1992). As such, it is believed that β cell dysfunction in NIDDM may involve a deficit in expression of GLUT-2, glucokinase, or both.

Another hallmark of _/3-cell function in the early stages of NIDDM is that insulin secretion is elevated in the presence of basal, normally non-stimulatory concentrations of glucose. Thus, /3-cell failure procedes through a phase of insulin hypersecretion at fasting glucose levels and continues to an exacerbated state in which stimulatory concentrations of glucose no longer cause increased insulin release. The first hypersensitive phase of /3-cell dysfunction may be a consequence of overexpression of a low Km glucose metabolizing protein such as GLUT-1 or hexokinase. If so, this condition could be treated by gene therapy with adenovirus-mediated introduction of antisense hexokinase or GLUT-1 cDNAs as discussed above.

The inventors envision that gene therapy regimens directed to restoring normal GLUT-2 and glucokinase functions would be particularly appropriate for use in treating NIDDM. However, certain drawbacks in gene therapy techniques currently exist which limit their utility in this regard. For example, despite improvements in physical gene transfer techniques such as Ca2P0₄ co-precipitation, lipofection, and electroporation and their demonstrated utility for studies on islet gene expression (German and Rutter, 1991; Welsh et al . , 1990), such procedures are not yet efficient enough to ensure introduction of the gene of interest into more than 20- 30% of the cells. Retroviral vectors allow more persistent expression and can be used to infect a wide range of cell types, but they have limited effectiveness for infection of non-proliferating cells, since integration of the retroviral genome into chromosomal DNA is required for expression (Miller, 1992) . Retroviral infection of adult hepatocytes, for example, generally results in expression of the heterologous gene in 20-30% of the cells, even when the culture media is supplemented with growth factors (Ledley, 1990) .

In utilizing recombinant adenovirus as a therapeutic tool in the treatment of NIDDM, it is contemplated that one would prepare an adenovirus containing β cell- specific promoter/enhancer elements, such as the insulin promoter, followed immediately by a suitable gene (or genes) involved in glucose sensing, particularly either the GLUT-2 and/or glucokinase gene, and then a suitable polyadenylation signal, such as the mouse protamine polyadenylation signal. Administration of such a recombinant virus via injection into animals, and after proper testing, into humans, will result in infection of multiple tissues. Alternatively, although the recent work of the inventors suggests that tissues with direct blood contact such as liver and the islets of Langerhans may represent preferred sites, the virus may be delivered directly into the pancreatic circulation via a catheter, providing for preferential expression in the highly vascularized islet cells. Despite this, the recombinant glucose-sensing gene(s), such as GLUT-2 or a suitable glucokinase gene, would be preferably expressed only in islet β cells, since the expression of the transgene(s) is driven by islet-selective or islet preferred promoter/enhancer elements (e .g. , the insulin, GLUT-2 or glucokinase promoter elements) . It is contemplated that restoration of GLUT-2 and/or glucokinase in β cells of animals or humans with NIDDM-like syndromes may restore normal β cell function, thus correcting the diabetic state. Since the virus is replication incompetent, there will be no deleterious effect of the virus itself on subject health.

Prior to the present invention, the concept of gene therapy in β cells was generally complicated by the fact that surgical resection of islet tissue from prospective patients is impractical, if not impossible, and by the lack of a targeted gene delivery system analogous to the newly developed glycoprotein complex strategy that targets DNA to the asialoglycoprotein receptor of liver (Wu et al . , 1989). However, the recombinant adenovirus strategies envisioned by the inventors also address these points. Importantly, simple intravenous injection of adenovirus is sufficient to result in viral infection of tissues, at sites distant from the injection (Stratford- Perricaudet et al . , 1990; Herz and Gerard, 1993). (See also Figure 8) . Also, perhaps more direct physical targeting of the recombinant adenovirus could be employed if desired, in an analogous manner to the intratracheal administration of the cystic fibrosis transmembrane conductance regulator which resulted in expression of this gene in rat airway epithelial cells (Rosenfeld et al . , 1992).

It is contemplated that the adenovirus system will have other distinct advantages, for example, in its use with primary cells as the expression of foreign genes is not restricted to replicating cells, as large DNA inserts can be accommodated, and in that surprisingly persistent expression is maintained despite the fact that the viral DNA integrates often into chromosomal DNA with limited efficiency (Stratford-Perricaudet et al . , 1990; Van Doren and Gluzman, 1984) .

Progress has already been made in that the inventors have determined that recombinant adenovirus represents an efficient means of introducing genes into primary islets. In studies employing the marker gene β-galactosidase, it was found that 75-80% of islet cells infected with the recombinant virus expressed this gene and that the virus per se had no detrimental effect on islet cell function (Becker et al., 1993; Becker et al . , 1994b). Recombinant adenoviruses have been constructed which contain genes involved in glucose sensing, particularly GLUT-2 and glucokinase; but also mutant glucokinases associated with MODY, GLUT-1 and sense and anti-sense hexokinase. Viruses containing GLUT-l or hexokinase in sense orientation can serve as important controls and have also served to validate the concept that overexpression of one or both of these low Km glucose metabolizing proteins predisposes to hypersensitivity to glucose when introduced into normal islets. B: ADENOVIRUS-MEDIATED GENE THERAPY FOR IDDM

As discussed above, IDDM is caused by autoimmune destruction of the islet β cells, leading to complete insulin deficiency and uncontrolled glucagon secretion. As an alternative strategy to insulin injection, the transplantation of appropriately encapsulated insulin- secreting islet tissue into patients has been proposed. The major problem with this approach is that it is very difficult to obtain large numbers of islets, in part because isolated islets do not proliferate in vitro. The implantation of recombinant cells engineered to have regulated insulin secretion, such as those disclosed herein, is one particularly advantageous manner in which to replace islet β cells in IDDM patients. However, the present inventors contemplate other ways in which this may be achieved, for example, by utilizing adenovirus- linked protocols, such as those described below, to expand islet cells in culture after their isolation and to thus increase the number of cells available for transplantation.

This aspect of the present invention was developed from a consideration of work on neurotrophic factors and their receptors. A number of neurotrophic factor receptors, including molecules named gpl30, LIFRβ, CNTFR, TrkA, TrkB, and TrkC have been cloned. These receptors, and their corresponding peptide ligands such as brain- derived neurotrophic factor (BDNF) , nerve growth factor (NGF) , neurotrophin-3 (NT-3) , and neurotrophin-4 (NT-4) appear to be involved in survival, growth and differentiation of cells derived from the nervous system. Furthermore, the introduction of one of the receptor neurotrophin receptor molecules (TrkB) into NIH 3T3 fibroblasts, a cell type not normally expressing TrkB, has been shown to result in a proliferative response and enhanced survival exhibited by engineered fibroblasts upon administration of BDNF or NT-3 (Glass et al . , 1991) .

The essence of this novel idea is to isolate islets by standard techniques from either human or large animal sources, and infect these cells in vitro with a recombinant adenovirus containing the gene or cDNA for one of the neurotrophin receptors listed above. Genes or cDNAs encoding other growth factor or cytokine receptors or cloned islet cell-specific receptors could also be employed. This efficient mode of gene transfer would result in a large percentage of islet cells expressing the desired receptor molecule. It is proposed that administration of the corresponding ligand would then lead to a large expansion of the number of islet cells in the culture, thus increasing the number of cells available for transplantation and overcoming the obstacle imposed by repeated islet isolation procedures.

A further aspect of this new concept is that islet cells infected with one of these receptors will exhibit enhanced survival in the presence of the cognate peptide ligand. If so, a supply of the peptide may be maintained in close proximity to the transplanted cells expressing the receptor when the cells are transplanted into diabetic patients. It is proposed that this could be achieved by transfecting endocrine cells, such as cells derived from islet cells or anterior pituitary tissue with the cDNA encoding the appropriate peptide ligand. The recombinant secretory cells of the present invention are ideally suited for delivery of neurotrophic peptides. The cells engineered for expression of the peptide could be co-transplanted with primary islet cells expressing the neurotrophin receptor, thus maintaining a supply of the necessary ligand in close proximity to the target tissue. USE OF ENGINEERED CELLS FOR DIAGNOSIS OF IDDM PRIOR TO ONSET

As discussed above, antibodies against islet proteins have been identified in individuals with new- onset IDDM. The appearance of these antibodies likely precedes the period of islet β cell destruction and consequent loss of insulin production. In recent years, significant progress has been made in the identification of the specific proteins that are recognized by the immune system. Expression of one such potential antigen, the GLUT-2 islet β cell glucose transporter, in non-islet cell lines, as described herein now allows us to test the immune response of patient sera with a specific islet antigen. Other particular epitopes contemplated by the inventor as being preferred include epitopes of cytoplasmic and surface islet cell antigens (Lernmark, 1982), insulin (Srikanta et al . , 1986), proinsulin (Kuglin et al . , 1988), islet 64 Kd and 38 Kd protein (Baekkeskov et al . , 1982), immunoglobulins (DiMario et al . , 1988), mammalian 65 Kd heat shock protein (Elias et al . , 1991), and even insulin receptors (Ludwig et al . , 1987) .

The inventors propose that cells engineered for specific expression of one of the foregoing epitopes, or for any epitope that may subsequently be identified in autoimmune diabetes, may be employed in diagnostic tests for diabetes. The principle of such a test involves reaction of the antibodies in a patients' serum with cells expressing the antigen(s) of choice, or epitope(s) of such an antigen, and subsequent detection of the antigen/antibody complex by reaction with a second antibody that recognizes human immunoglobulins (antibodies) . A test would be scored as positive if the serum being tested reacts with the cells engineered for expression of the antigen of interest, but not with the parental (non-engineered) cell line. The reaction of the patient's serum with the expressed antigen is measured indirectly by virtue of the fact that the anti- immunoglobulin antibody used is "labeled" or "tagged" with a molecule that readily allows its detection by direct inspection or mechanical measurement. The most common "tags" that are linked to commercially available preparations of anti-human immunoglobulin are fluorescent molecules such as fluorescein or tetramethyl rhodamine or enzymes such as horseradish peroxidase or alkaline phosphatase.

As disclosed herein below, the use of engineered cells expressing the GLUT-2 antigen in diagnostic assays is greatly advantageous in that it allows rapid, efficient and reproducible analyses of patients' sera. Engineered GLUT-2-expressing cells, such as GLUT-2- expressing AtT20_ins cells may be used in diagnostic assays based either on immunocomplex formation, or on the inhibition of glucose uptake, for example, using 3-0- methyl-β-D-glucose. However, it will be understood that engineered cells expressing the GLUT-1 antigen will also have utility. In particular, they may be used as control' cells in diagnostic tests since no reaction of IDDM sera is detected with GLUT-1-expressing cells in these assays.

Regarding immunocomplex formation, two methodologies are available for measuring the fluorescent signal resulting from formation of an antigen-antibody-anti- antibody complex. The first is simple direct inspection of cells by fluorescence microscopy. In this procedure, cells are adhered to poly-L-lysine coated microscope slides or cover slips. The cells are then fixed lightly by treatment with 0.5% paraformaldehyde or left untreated. Treatment of the cells with paraformaldehyde will cause changes in membrane structure of cells, resulting in changes in the conformation of antigen molecules. For some, but not all antibodies, alteration of antigen conformation in this way will allow a tighter association of the antibody and antigen. Engineered and control cells are then exposed to either crude serum or purified immunoglobulins (IgGs) from patients to be tested for antibodies against the expressed antigen. After washing, the cells are exposed to an appropriately "tagged" or labeled antibody recognizing human IgGs and the antigen/antibody/anti-antibody complexes are visualized in a microscope by excitation of the fluorescent tag by exposure to light of an appropriate wavelength.

An alternative and more quantitative approach is to use a fluorescence activated cell sorter (FACS) to score immune complex formation. In this procedure, cells are treated with patient serum and labeled second antibody much as described for the microscope slide approach except that the incubations are done with the cells in suspension rather than attached to a slide. After treatment with the anti-human IgG antibody, cells are loaded into the FACS, which passes the cells one-by-one past a light source set at a wavelength that will excite the fluorescent marker of the second antibody. The cells then pass a detector which measures the fluorescence emission from the cells. Data are plotted as a histogram of fluorescence intensity. A positive antibody/antigen/anti-antibody reaction will result in an increase in fluorescence in most of the cells in a test. In contrast, exposure of cells to sera that lack antibodies against the specific antigen being presented will result in little fluorescence. The utility of the FACS is that it provides a display of the fluorescence intensity of all of the cells in a sample and plots the data as the distribution of fluorescence intensities. Thus a positive sample will have a peak in cell distribution at a position on the graph that is shifted to the right (corresponding to a greater fluorescence intensity) relatively to a sample that is not reactive.

To date the inventors have observed a noticeable increase in the fluorescent signal in GLUT-2 transfected AtT-20_ins cells treated with sera from patients with IDDM compared with normal sera with both the microscopic and FACS techniques. Importantly, an antibody raised against an exposed (extracellular) region of the GLUT-2 molecule has been found by the inventors to cause a shift (increase) in fluorescence that is similar to the shift caused by the diabetic sera. Thus, GLUT-2 appeared to be a particularly useful epitope for the identification of new-onset IDDM patients and even prediction of diabetes onset.

In copending application US Serial Number 07/483,224, filed February 20, 1990, it is demonstrated that the sera of IDDM patients includes autoantibodies that are capable of inhibiting the uptake of glucose by /3-cells. This observation led to the development of a bioassay for identifying individuals at risk for the development of IDDM. Unfortunately, this method is somewhat cumbersome. Accordingly various approaches were taken to simplify and improve this diagnostic assay, centering on the development of an immunological-based assay.

Among the approaches studied included ELISA- and Western blot-based assays, as opposed to measurement of glucose transport rates. Attempts at using these techniques were successful, but the problem at this level was the numbers of false positive normal individuals that were identified. Since there was a much better separation of the normal and diabetic populations observed using the glucose transport assay, it was hypothesized that the use of intact cellular protein in the transport assay, as opposed to the use of denatured protein in the Western blot and ELISA techniques, might account for the difference.

To test this hypothesis, artificial 3-cells of the present invention were tested in the glucose transport assay. In these studies, it was shown that IgG from IDDM patients effectively inhibited glucose transport in the artificial _/3-cells, while no effect was seen with IgGs from normal individuals. Moreover, no effect of IgGs from new-onset Type 1 individuals on glucose uptake was observed against cells that did not contain the GLUT-2 protein, and instead were engineered for overexpression of the Glut-1 protein.

These data led to the development of a flow cytometry-based immunofluorescence assay for antigen- antibody interaction between the patient's autoantibodies and the glucose transporter mechanism. Initial attempts to develop such a system met with variable success. It was suspected that this variability might be due to the day-to-day handling of samples. Accordingly, a protocol was developed to ensure uniform growth of the cells, harvesting of the cells and treatment of the cells under conditions as close as possible to the transport assay. These conditions were as follows:

1. AtT 20 GT6 cells were grown for 72 hours following a 1:10 split at confluence of the seed culture.

2. Cells were harvested from plates by scraping with a rubber policeman into Dulbecco's phosphate-buffered saline at pH 7.6. 3. The cells were resuspended to a density of approximately 10" cells/ml, washed by centrifugation at 500 xg in Dulbecco's phosphate-buffered saline, and incubated with shaking for 15 minutes at 37°C followed by 1 hour at 4°C in 150μl of patient serum.

4. Following two washes by centrifugation at 500 xg in Dulbecco's phosphate-buffered saline, the cells were resuspended in 200 μl of R- phycoerythrin-labeled goat antihuman IgG (heavy chain specific) , vortexed lightly, and incubated for 1 hour at 4°C on a dual action shaker.

5. Following two washes by centrifugation at 500 xg in Dulbecco's phosphate-buffered saline, the cells were resuspended in 500 μl of Dulbecco's phosphate-buffered saline and analyzed for antigen-antibody interaction using a flow cytometer.

As is discussed in greater detail in Example III, this flow cytometry-based immunofluorescence assay was found to be particularly useful in distinguishing the sera of patients with new-onset IDDM from non-diabetic subjects. It was found that 29 of 31 (94%) of the nondiabetic population were negative for IgG binding to GLUT-2 while 23 of 30 (77%) of sera from IDDM patients were positive. Thus, 81% of negative results were from nondiabetic patients and 92% of positive results were from patients with IDDM (Table 2) . USE OF ENGINEERED CELLS IN THE IDENTIFICATION OF SPECIFIC EPITOPES

The present inventors have recently discovered that GLUT-1 transfected AtT-20_ins cells do not discriminate diabetic from normal sera in FACS-based diagnostic tests, providing strong evidence that diabetic sera contain an antibody specific for the islet GLUT-2 glucose transporter (Inman et al . , 1993). It is therefore envisioned that the artificial _/3-cells of the present invention will be of use in the identification of the specific epitope or segment of protein within GLUT-2 that is responsible for interacting with the antibody. Comparison of the GLUT-1 and GLUT-2 sequences reveals that the 2 putative membrane spanning regions in the two molecules are highly hydrophobic and of very similar sequence. These hydrophobic segments are connected by "loops" of amino acids that have much less sequence conservation (Bell et al . , 1990) In particular, GLUT-2 contains a very large extracellular loop between membrane spanning regions 1 and 2, while GLUT-1 contains a much smaller loop with little sequence homology to the GLUT-2 loop.

The inventors propose that construction of chimeric GLUT molecules in which individual or multiple "loop" regions are substituted could lead to identification of the specific epitope of GLUT-2 that reacts with diabetic sera. Thus, for example, the DNA encoding the large extracellular loop of GLUT-2 can be inserted in place of the small extracellular loop of GLUT-1 in the GLUT-1 cDNA sequence, and this chimeric molecule expressed in AtT- 20^ cells. If the chimera reacts with diabetic serum (as the native GLUT-1 molecule does not) , the added GLUT- 2 extracellular loop would be the specific epitope. Once such an epitope is identified by the procedure outlined above, synthetic peptides corresponding to this region of the protein sequence can be produced and used to develop simpler diagnostic procedures. Examples would include a simple test in which the peptide epitope is reacted with test serum and the formation of an antibody/peptide complex is monitored by well established techniques such as ELISA or RIA.

An alternative means of identifying the reactive epitope of GLUT-2 is to synthesize the individual peptide sequences that constitute the entire GLUT-2 sequence as individual segments of 15-50 amino acids in length. This analysis would initially be focused on the regions that are most dissimilar in comparing GLUT-2 and GLUT-1, such as the extracellular "loop" region described above.

Once the GLUT-2 epitope responsible for reactivity with IDDM sera is identified, the peptide or peptides corresponding to this epitope can be synthesized and tested for utility in screening of diabetic (and pre- diabetic) sera. This can be approached by rapid solid- phase assays such as ELISA, in which the peptide is aliquoted into a multi-well plastic plate and tested for reactivity with diabetic and non-diabetic sera by using a "tagged" second antibody and well-established colorimetric procedures. The intensity of the colorimetric reaction (and thus, the reactivity of the serum) would then be evaluated with an automated plate scanner that reads the optical density of the colorimetric solution at desired wavelengths. This solid-phase assay would be expected to be less expensive and generally easier in practice than an assay involving fluorescence-activated cell sorting, although it should be emphasized that the current cellular technology may be crucial to the identification of the epitope. INSULIN PRODUCTION FROM HIGH INSULIN-CONTENT ENGINEERED CELLS

GLUT-2 transfection is herein shown to cause an increase in intracellular insulin of approximately 5-fold in the AtT-20_ins cell line, CGT-6. This finding demonstrates that batch extraction of insulin directly from these or related cells is an alternative strategy for isolation and purification of human insulin for use in IDDM therapy. CGT-6 cells contain approximately 1 mUnit/10° cells of human insulin when grown on gelatin beads in solution. The average IDDM patient requires approximately 30 Units of insulin per day for control of blood glucose levels. Cell densities of 5 x 10^ cells/liter cell culture media are readily achieved in the current liquid culture configuration, meaning that 5 Units of insulin/liter can be produced. Much higher densities can be achieved using currently commercially available bioreactor technology (e.g., that available from New Brunswick Scientific) . Such instruments are reported to support cell densities of 10^/ml for a variety of well known cell lines.

Furthermore, it is highly likely that the intracellular insulin content of the cells can be further increased by one of the following methods: 1) Retransfection of AtT-20 cells with the Rous sarcoma virus/human proinsulin cDNA plasmid that was originally used to generate the AtT-20_ins cell line. The level of expression of a transfected gene appears to be dependent on the site of insertion of the plasmid in the chromosome. Thus, it is highly likely that higher levels of insulin expression will be achieved by simply reintroducing the plasmid and isolating new clones, that are rendered resistant to neomycin by the neomycin resistance gene in the plasmid. 2) Construction of plasmids in which human proinsulin cDNA expression is directed by alternate promoters. Examples include the CMV promoter, which was used to achieve very high levels of expression of GLUT-2 in the creation of the CGT-6 cell line in the inventor's laboratory, or 3) Amplification of the viral promoter/human proinsulin cDNA (Sambrook et al . , 1989) by cloning next to a resistance gene such as dihydrofolate reductase (DHFR) , adenosine deaminase, or glutamine synthetase (Cockett et al . , 1990). Of these, DHFR is the most commonly used system, but is generally of limited usefulness in cell lines that have endogenous expression of DHFR (this is true of the AtT- 20_ins cell line) . The glutamine synthetase system allows amplification of the gene of interest even in the presence of endogenous expression of glutamine synthetase.

Cells are stably transfected with a plasmid containing the transcription unit (i.e., viral promoter fused to the human proinsulin gene) adjacent to the hamster glutamine synthetase coding sequences. Selection of clones and amplification of the integrated transcription unit/GS gene is then carried out by addition of methionine sulfoxide to the tissue culture media (Cockett et al . , 1990). Resulting clones contain greatly increased copy numbers of the transcription unit, by virtue of its association with the amplified glutamine synthetase gene. As a result, much greater quantities of insulin are produced by the recombinant cell, making it an even more viable source for human insulin production. 4) Use of a strong and ubiquitously active cellular promoter, such as that for elongation factor elα.

LIQUID CULTURE OF ENGINEERED CELLS FOR INSULIN PRODUCTION

In considering the drawbacks of the methods currently employed for insulin production, the invention contemplates that correctly-folded human insulin could be produced relatively simply and rapidly using clonal cells that secrete insulin in response to glucose.

The most appropriate method to accomplish this has been found by the inventors to be the perfusion of a column containing engineered AtT-20_ins cells adhered to gelatin beads. Passing a glucose-containing buffer, such as Hanks Balanced Salt solution, with 5mM glucose, pH 7.4, over such a column of insulin producing β cells has been found to stimulate the increased secretion of insulin into the surrounding media, which can then be collected and used as a starting material for the purification of recombinant insulin.

It is anticipated that purification of insulin from the perfusion media can be rapidly achieved by one or a combination of the following approaches: 1) Affinity chromatography, for example, passage of the insulin containing media over a column containing anti-insulin antibodies. After removal of non-insulin proteins and other impurities by washing of the column, insulin can be specifically eluted by using a buffer with an increased salt concentration or decreased pH. 2) Preparative high performance liquid chromatography. 3) Size selection by conventional size-exclusion column chromatography.

The following examples are included to demonstrate preferred embodiments of the invention. It should be appreciated by those of skill in the art that the techniques disclosed in the examples which follow represent laboratory techniques discovered by the inventor to function well in the practice of the invention, and thus can be considered to constitute preferred modes for its practice. However, those of skill in the art should, in light of the present disclosure, appreciate that many changes can be made in the specific embodiments which are disclosed and still obtain a like or similar result without departing from the spirit and scope of the invention.

EXAMPLE I ENGINEERING OF GLUCOSE-STIMULATED INSULIN

SECRETION IN NON-ISLET CELLS

A. Methods

1. AtT-20_jns cell culture and tissue isolation

The AtT-20_ins cells used were provided by Dr. Regis Kelly, University of California San Francisco, and were similar to the line that was originally described (Moore et al . , 1983) except that the Rous sarcoma virus long terminal repeat was substituted for the SV40 early gene promoter for directing insulin cDNA expression. The cells were grown in Dulbecco's modified Eagles' medium (DMEM) , supplemented with 10% fetal calf serum, 100 μg/ml streptomycin, and 250 μg/ml neomycin. Anterior pituitary and liver samples were excised from normal ad-lib fed Wistar rats, and islets were isolated from groups of 10- 20 animals as previously described (Johnson et al . , 1990a, 1990c) and pooled for RNA extraction or homogenization for glucose phosphorylation assays.

2. Stable transfection of Atτ-20_ins cells with

GLUT-2

The rat islet GLUT-2 cDNA (Johnson et al . , 1990a) was cloned into the vector pCB-7, a derivative of vector pCMV4 (Andersson et al . , 1989), immediately downstream of its cytomegalovirus (CMV) promoter. The cDNA was cleaved at its 3' end with Hind III, resulting in the removal of 635 base pairs of 3' untranslated region. AtT-20_ins cells were transfected with this construct using electroporation. Cells were harvested from pre-confluent plates by light trypsinization, washed twice in phosphate buffered saline, and resuspended at 3 x 10⁶ cells/ml in a solution containing 20 mM Hepes (pH 7.05), 137 mM NaCl, 5 mM KCl, 0.7 mM Na₂HP0₄, 6 mM glucose, and 0.5 mg/ml salmon testis DNA. After equilibration of the cells to room temperature in electroporation cuvettes (Bio-Rad Labs; electrode gap width 0.4 cm), a single pulse was delivered using a capacitance setting of 960 μF and voltage settings between 0.2 and 0.3 kV. The cells remained in the buffer for five minutes and were then plated onto tissue culture dishes. Stable transfectants were selected with hygromycin, since the plasmid also contains a resistance gene for this drug. Four colonies were obtained and passaged several times in the presence of hygromycin to obtain pure cultures.

3. RNA blot hybridization analysis

RNA was prepared by guanidinium isothiocyanate extraction, resolved on a formaldehyde/agarose gel and transferred to a nylon membrane (Micron Separations Inc.) as previously described (Newgard et al . , 1986). Blots were hybridized sequentially with *P labeled antisense GLUT-2 or 18S rRNA probes, prepared as described (Chen et al . , 1990), with stripping of the blot between hybridizations by boiling in 0.1% SDS for 30 minutes.

4. immunoblot analysis

Liver plasma membranes were prepared by the method of Axelrod and Pilch (1983) and only the light plasma membrane fraction was used. Islet and AtT-20_ins cell membranes were prepared as previously described (Johnson, 1990b) , except that the sucrose gradient was deleted and the homogenization buffer consisted of 50 mM Tris, pH

7.4, 5 mM EDTA, 0.1 mM phenylmethylsulfonyl fluoride, 10 mM benzamidine, and 1% Trasylol. The samples were transferred onto nitrocellulose and immunoblotted exactly as described (Hughes et al . , 1991; Quaade et al . , 1991), either with a 1:1000 dilution of the anti GLUT-2 polyclonal antiserum (Johnson et al . , 1990b), or with the diluted antiserum after 10 minutes of preincubation with an equal volume of a 1 mg/ml solution of the antigenic peptide dissolved in PBS. The second antibody was 125j_ labeled goat anti-rabbit anti-IgG, and the resultant immune complexes were visualized by autoradiography.

5. GLU -2 immunofluorescence in AtT-20_jns cells

Parental AtT-20_jns cells or transfected lines CGT-5 and CGT-6 were grown to a density of 5 x 10° cells per 100 mm dish and harvested by incubation at 37°C with a solution of 0.02% EDTA in PBS. After three washes in DMEM containing 20 mM HEPES, approximately 1.5 x 10⁵ cells were transferred onto 12 mm poly-L-lysine coated glass coverslips, to which they adhered during a 30 minute incubation at 37°C. The cells were then fixed with 3% paraformaldehyde in PBS for 30 minutes at room temperature, and incubated with 0.1 M NH₄C1 in phosphate buffered saline (PBS), pH 7.9 for 30 minutes. After 4 rinses with PBS, cells were permeabilized with 0.1% Triton X-100 for 5 minutes, then rinsed 3 times with PBS. After a pre-incubation with 2% BSA, GLUT-2 antiserum (1:2500) was applied in the presence or absence of an equal volume of the antigenic peptide (1 mg/ml) . Slides were incubated overnight, and excess antibody removed by washing 5 times with 0.1% BSA in 0.1 M phosphate buffer, pH 7.9. Cells were then incubated with FITC-conjugated goat anti-rabbit IgG for two hours at 37°C and washed sequentially with BSA/phosphate buffer and water. After application of coverslips, the slides were visualized by fluorescent light microscopy. 6. Glucose transport measurements

Cells were harvested by scraping with a rubber policeman, washed in Hanks balanced salt solution by centrifugation at 600 x g, and resuspended in Dulbecco's modified Eagles's media with 10% fetal calf serum and 5 mM glucose. Cells were incubated at 37°C for 30 minutes, washed, resuspended in phosphate buffered saline and assayed for 3-O-methyl glucose uptake as previously described (Johnson et al . , 1990c). Results were expressed as mmoles 3-O-methyl glucose uptake/min/liter cell space. Initial velocities of uptake were derived from duplicated measurements at 3, 6, and 15 seconds for each concentration of glucose with the transfected cell lines and 3, 15, and 30 seconds for the parental cell line (due to slower transport rate in these cells) .

7. Glucose phosphorylation assays

Glucose phosphorylation and glucokinase activities were measured by conversion of U-¹⁴C glucose to U-*⁴C glucose-6-phosphate, as previously described (Method "B" in Kuwajima et al . , 1986). Cultured cells or tissues were homogenized in 5 volumes of buffer containing 10 mM Tris, 1 mM EDTA, 1 mM MgCl₂, 150 mM KCl, and 1 mM DTT, pH 7.2. The homogenate was cleared by centrifugation at 12,000 x g and the supernatant used for assays of glucose phosphorylation. Reactions were carried out at 37°C in a total volume of 150 μl, and initiated by addition of 10- 30 μl of extract to a reaction mix containing 100 mM Tris, 5 mM ATP, 10 mM MgCl₂, 100 mM KCl, 1 mM DTT, pH 7.2, 15 or 50 mM glucose, and 6.2 μCi of U-¹⁴C glucose (300 mCi/mmol; New England Nuclear) . In order to discriminate glucokinase and hexokinase activities, assays were performed in the presence and absence of 10 mM glucose-6-phosphate, which potently inhibits hexokinase but not glucokinase activity. Reactions were carried out for 90 minutes and terminated by addition of 50 μl of reaction mix to 100 μl of 3% methanol in 95% ethanol. An aliquot of this mixture was transferred to nitrocellulose filter circles (Grade NA 45, Schleicher & Schuell) , which bind phosphosugars, and after air drying, washed extensively in water to remove labeled glucose. Radioactivity on the paper was then detected by liquid scintillation counting, and glucose phosphorylating activities are expressed in terms of the total protein content of the extracts.

8. Insulin secretion from Atτ-20_ins cells in response to secretagogues

Parental or GLUT-2 transfected lines CGT-5 and CGT-6 were removed from growth plates by light trypsinization and reseeded in 6 well dishes (Costar) at a density of 5 x 10⁵ cells per well. The cells were then grown for three days in culture media containing lmM glucose (see above) . On the third day, cells were washed twice for 10 minutes each in HEPES balanced salt solution containing 1% BSA (HBSS) , but lacking glucose. Secretion studies were initiated by addition of HBSS plus a range of glucose concentrations (0 - 2OmM) or in the presence of one of three non-glucose secretagogues, forskolin

(0.5μM), dibutyryl cAMP (5mM) , or isobutylmethylxanthine (IBMX, O.lmM), in the presence or absence of glucose. Cells were incubated with secretagogues for 3 hours, after which media was collected for insulin radioimmunoassay.

9. Assay of intracellular insulin

Cells were collected in 1ml of 5M acetic acid, lysed by three cycles of freeze-thawing, and lyophilized. The dried lysate was then reconstituted in 5ml of insulin assay buffer (50mM NaH₂P0₄, 0.1% BSA, 0.25% EDTA, 1% aprotinin, pH 7.1) and aliquots were assayed for insulin by radioimmunoassay.

B. Results

1. Expression of GLUT-2 mRNA in transfected Atτ-20_ins cells

Expression of GLUT-2 mRNA was evaluated by blot hybridization analysis of AtT-20_ins cells, either transfected or untransfected with a cytomegalovirus (CMV) promoter/GLUT-2 hybrid gene, and in extracts of rat liver, islets of Langerhans, and anterior pituitary tissues. The radiolabeled GLUT-2 antisense RNA probe (Johnson et al . , 1990a; Chen et al . , 1990) was hybridized to a blot containing equal amounts of RNA from four GLUT- 2 transfected AtT-20_ins cell lines (CGT-1, CGT-2, CGT-5, CGT-6) , untransfected AtT-20_ins cells, and the three primary tissues (Figure 1) . Steady state levels of GLUT- 2 mRNA were highest in CGT-5 and CGT-6; the former contained approximately half as much and the latter an equal amount of GLUT-2 mRNA as rat islets, and they contained 10 and 16 times as much, respectively, as rat liver, measured by densitometric scanning and normalization to the signal obtained with an 18S rRNA probe. The transfected lines contained a smaller GLUT-2 transcript than liver or islets (2.2 versus 2.8 kb) because 635 base pairs of the 3' untranslated region were removed in the course of cloning the GLUT-2 cDNA into the pCB-7 vector. Lines CGT-1 and CGT-2 exhibited less active expression of GLUT-2. Untransfected AtT-20j__c cells and primary anterior pituitary cells did not contain detectable amounts of GLUT-2 mRNA, consistent with previous studies (Hughes et al . , 1991). 2. Expression of GLUT-2 protein in tissues and cell lines

In order to evaluate the levels and molecular status of the expressed GLUT-2 protein in transfected AtT-20^ cells, crude membrane fractions were resolved by SDS/PAGE, transferred the proteins to nitrocellulose, and detected GLUT-2 protein with an antibody raised against its C-terminal hexadecapeptide sequence (Johnson et al . , 1990b) . The antibody recognized two distinct bands in liver and islets, with apparent molecular weights of 70 and 52 kd in liver and slightly different sizes of 72 and 56 kd in islets (Figure 2, left) . Consistent with the RNA blot hybridization data, untransfected AtT-20^ cells were found to lack GLUT-2 protein, while a single intense band of approximately 70 kd was observed in extracts from either of the transfected lines CGT-5 and CGT-6. The specificity of the antibody is demonstrated by the fact that all bands were blocked by preincubation of the antibody with the antigenic peptide (Figure 2, right). Thorens (Thorens et al . , 1988) have previously reported that a similar anti-peptide antibody recognizes GLUT-2 proteins of distinct molecular weights in liver (53 kd) and islets (55 kd) , despite the fact that the cDNA sequences for GLUT-2 are identical in liver and islets in both rat (Johnson et al . , 1990a) and man (Permutt et al . , 1989) . They did not report on the larger bands shown herein, possibly because of differences in the protocols used for membrane preparation.

3. Immunocytochemistry of GLUT-2 in transfected AtT-20_ins cells

Expression of GLUT-2 protein in transfected AtT-20_ins cells was studied by immunofluorescent staining techniques, using an antibody raised against the C- terminal hexadecapeptide of GLUT-2 (Johnson et al . , 1990b) . In the lines with highest GLUT-2 mRNA levels (CGT-5 and CGT-6) , abundant GLUT-2 immunofluorescence was detected at the cell membrane as well as some intracellular signal that was mostly localized to regions of cell-cell contact. The fluorescent signal was blocked by preincubation of the antibody with the antigenic peptide and was not seen in untransfected cells or in cells transfected with the vector lacking the GLUT-2 insert. Expression of GLUT-2 protein in transfected AtT- 20^ cells and its absence in the untransfected parental line was confirmed by immunoblot analysis. Thus, AtT- 20^ cells not only have the capacity to produce GLUT-2 mRNA and protein but also sort the protein to the cell membrane, as occurs in both islets and liver (Thorens et al . , 1988; Orci et al . , 1989; Tal et al . , 1990; Orci et al . , 1990). Preferential expression at regions of cell- cell contact is in keeping with a recent report (Orci et al . , 1989) showing that GLUT-2 expression in islet β cells is not homogenous and is most abundant in regions of membrane enriched in microvilli and facing adjacent endocrine cells, as opposed to regions facing capillaries or empty spaces between cells. The functional significance of this phenomenon is currently not understood.

4. Glucose transport measurements in parental and GLUT-2 expression AtT-20_ins cells

The GLUT-2 cDNA has been cloned from both liver (Thorens et al . , 1988) and islets (Permutt et al . , 1989; Johnson et al . , 1990a), two tissues with high Km glucose transport activity. Although the cDNA has been expressed in bacteria (Thorens et al . , 1988) and oocytes (Permutt et al . , 1989), these systems have not been used for kinetic studies. Thus, direct evidence that the GLUT-2 cDNA encodes a protein that confers the high Km glucose transport activity has not been presented to date. Dramatic differences in glucose transport kinetics were found between AtT-20_ins cells transfected with the pCB7/GLUT-2 expression vector and untransfected AtT-20_jns cells. Figure 3A shows a plot of the concentration dependence of glucose uptake in the AtT-20_ins cell lines, and demonstrates the dramatically increased rates of glucose transport in lines CGT-5 and CGT-6 relative to the untransfected (parental) AtT-20_ins cells. Lineweaver- Burke analysis of the data showed that the CGT-5 and CGT- 6 lines had apparent Kms for glucose of 16 and 17 mM and Vmax values of 25 and 17 mmoles/min/liter cell space, respectively (Figure 3B) . In contrast, the untransfected parental AtT-20^ line had an apparent Km for glucose of 2 mM and a Vmax of 0.5 mmoles/min/liter cell space (Figure 3C) , consistent with its expression of the GLUT-1 mRNA (Hughes et al . , 1991), which encodes the low Km glucose transporter found in most clonal cell lines (Flier et al . , 1987; Birnbaum et al . , 1987). The transfected AtT-20^ cells have glucose transport kinetics that are remarkably similar to isolated, dispersed islets of Langerhans, which have a Km of 18 mM for glucose and a Vmax of 24 mmoles/min/liter cell space (Johnson et al . , 1990a). Thus, the GLUT-2 cDNA clearly encodes the protein responsible for the high Km glucose transport activity in islets and liver, and is capable of transferring this activity into the AtT-20_ins cell line.

5. Glucose-stimulated insulin secretion from AtT-20_ins cells

Insulin secretion from GLUT-2 transfected and untransfected cells was measured over a range of glucose concentrations from 0-2OmM. Figure 4A compares glucose-stimulated insulin release from AtT-20ins cells and CGT-6 cells, expressed as mU insulin released/mg total cellular protein. Consistent with previous results (Hughes et al . , 1991), glucose had no significant effect on insulin release from parental AtT-20ins cells. AtT-20ins cells transfected with the pCB7 vector lacking a GLUT-2 insert were also found to be unresponsive to glucose. GLUT-2 transfected cells, in contrast, are clearly glucose responsive (data are shown for line CGT-6 only; results for line CGT-5 were qualitatively identical) . A submaximal but statistically significant (p = 0.002) increase in insulin release relative to insulin release at OmM glucose was observed at the lowest concentration of glucose studied (5μM) ; maximal stimulation of approximately 2.5-fold was observed at all higher concentrations over the range 10μM-20mM (p < 0.001). It is highly unlikely that these results can be attributed to clonal selection of glucose responsive subpopulations of the parental AtT-20ins cells, since cells transfected with vector lacking GLUT-2 failed to respond, while two independent GLUT-2 expressing lines (CGT-5 and CGT-6) gained glucose sensing.

In normal islets, glucose potentiates the insulin secretory response to various /3-cell secretagogues, including agents that increase intracellular cAMP levels (Ullrich and Wollheim, 1984; Malaisse et al . , 1984). The potentiating effect of glucose on insulin secretion in the presence of forskolin, dibutyryl cAMP, and IBMX was therefore studied. Glucose had a modest stimulatory effect on forskolin stimulated insulin release from parental AtT-20ins cells, expressing the data either as insulin release/mg cellular protein (Figure 4A) or as insulin release/mg cellular DNA (Figure 4B) . In contrast, glucose had a powerful potentiating effect on forskolin stimulated insulin release from transfected CGT-6 cells. The response was unchanged by glucose concentration over the range of l-5mM, and similar potentiating effects of glucose on dibutyl cAMP and IBMX induced secretion were also observed.

Insulin secretion studies involved static incubation of cells with the secretagogue for three hours, and thus provided little information about the dynamics of insulin release. The inventors succeeded in growing the parental and transfected AtT-20ins cell lines on gelatin beads in liquid culture, thus allowing their secretory properties to be studies by perfusion with glucose containing media. Cells grown in this configuration released insulin within minutes of glucose stimulation. Furthermore, the insulin secretory response exhibits a first intense and a second less intense but sustained phase, as is characteristic of normal β cells.

6. Insulin Content of Native and Engineered AtT-20ins cells

A remarkable finding of this study is that transfection of AtT-20ins cells with the GLUT-2 cDNA results in a substantial increase in intracellular insulin content, despite the fact that insulin gene expression is driven by the glucose insensitive Rous sarcoma virus long-terminal repeat enhancer/promoter in these cells. Native AtT-20ins cells and the GLUT-2 transfected CGT-6 cells were grown for 3 days in media supplemented with low (lmM) or high (25mM) glucose. The CGT-6 cells were found to contain 3.6-fold and 5.4-fold more insulin than the AtT-20ins cells when studied at low and high glucose, respectively (p < 0.001 for both comparisons) . Furthermore, insulin content was approximately double in the CGT-6 cells grown at high glucose compared with the same cells grown at low glucose (p < 0.001). In contrast, in the untransfected AtT-20ins cells, high glucose caused only a 20% increase in insulin content. 7. Glucose phosphorylation in AtT-20_ins cells

AtT-20_ins cells transfected with GLUT-2 secrete insulin at glucose concentrations that are substimulatory for islets. The enhanced sensitivity to glucose is not explained by the kinetics of glucose transport, since both the CGT-5 and CGT-6 lines transport glucose with a velocity and concentration dependence that is virtually identical to islets. Alternatively, stimulation of insulin secretion at low glucose concentrations might be explained by differential regulation of glucose phosphorylation in AtT-20_ins cells relative to β cells. The ratio of hexokinase:glucokinase activity in these cells was therefore compared with activities found in normal islets of Langerhans and liver. Studies from this and other laboratories (Iynedjian et al . , 1989; Magnuson and Shelton, 1989; Newgard et al . , 1990; Hughes et al . , 1991) have shown that the single glucokinase gene is alternatively regulated and processed in liver and islets, resulting in distinct transcripts that predict proteins with unique N-termini; the Km for glucose of both isoforms is in the range of 8-10 mM. AtT-20^_ns cells express the islet isoform of glucokinase (Hughes et al . , 1991) .

A radioisotopic glucose phosphorylation assay was performed (Method "B" in Kuwajima et al . , 1986) that allows discrimination of glucokinase and hexokinase activities when performed in the presence and absence of 10 mM glucose-6-phosphate, since this metabolite is a potent inhibitor of hexokinase, but not glucokinase (Wilson, 1984) . As shown in Table 1, total glucose phosphorylating capacity and glucokinase activity are not significantly different in transfected (line CGT-6) versus untransfected (parental) AtT-20_ins cells. Both lines have a total glucose phosphorylating capacity that is similar to that in liver and islets. However, glucokinase activity in AtT-20_ins cells is only 32% of the glucokinase activity in islets and 10% of that in liver. Moreover, glucokinase represents only 9% of the total glucose phosphorylating activity of AtT-20^_ns cells (the remaining 91% is presumably due to hexokinase activity) , as compared to 24% in normal islets and 86% in normal liver. The altered hexokinase:glucokinase ratio in AtT- 20^ cells may result in low Km glucose metabolism that accounts for the insulin secretory response at low glucose concentrations.

Table 1. Glucose Phosphorylating Activities in Tissues and Cell Lines.

Cell Type Total Glucose* Glucokinase# Glucokinase Phosphorylation (U/gram (% of (U/gram protein) protein) total)

AtT-20_ins 9.19 + 0.27 0.63 + 0.06^a 6.8% (parental) 0.43 + 0.08^b

AtT-20^ 8.09 + 0.20 0.86 + 0.18^a 10.6% (line CGT- 6)

Islet 9.61 ± 2.10 2.31 ± 0.35^a 24.0%

Liver 8.42 + 1.09 7.19 + 1.31* 85.4%

* Total glucose phosphorylation was measured in 14,000 x g supernatant of crude homogenates, at 50 mM glucose, using an assay that monitors 1¹4⁴,C glucose conversion to ¹⁴C glucose-6-phosphate ("Method B" in Kuwajima, et al . ,

1986) . # Glucokinase activity was determined with the same assay as used for total glucose phosphorylation at 50 (^a) or 15 (^b) mM glucose, except in the presence of 10 mM glucose-6-phosphate to inhibit hexokinase. Values represent the means + SEM for 3 independent determinations for liver and islets and 4 independent determinations for untransfected (parental) and GLUT-2 transfected (line CGT-6) AtT-20_ins cells.

EXAMPLE II DIAGNOSIS OF IDDM

A. Methods

1. Direct inspection of immunoreactive cells by fluorescence microscopy

Parental and engineered AtT-20_ins cells are grown to a density of 5 x 10" cells per 100 mm dish and harvested by incubation at 37°C with a solution of 0.02% EDTA in phosphate buffered saline (PBS) . After washing the cells in DMEM media containing 20 mM Hepes, approximately 1.5 x 10^ cells are transferred onto 12 mm poly-L-lysine coated glass coverslips, to which they adhere during a 30 minute incubation at 37°C. The cells are then fixed for 30 minutes with varying amounts (0.5-3.0%) of paraformaldehyde, depending on the extent of fixation that is desired. For studies with anti-GLUT-2 antibodies or serum, the inventors have found a light fixation (0.5% paraformaldehyde) to be most appropriate. After preincubation with 2% BSA, a serum sample (usually diluted 1:1 in BSA) is added to the sample in sufficient volume to cover the cells. As a positive control, an antibody (designated X617) raised against the unique extracellular loop peptide of the rat GLUT-2 transporter is used, diluted 1:100 in PBS (the antibody is raised against a peptide with sequence DAWEEETEGSAHIV (SEQ ID NO:l), as found at amino acids 64-77 of the rat GLUT-2 primary structure) .

Slides are incubated overnight with serum or antibody, and excess antibody is removed by washing with 0.1% BSA in 0.1M phosphate buffer, pH 7.9. Cells are then incubated with FITC-conjugated goat anti-human IgG (in the case of human serum samples) or FITC-conjugated goat anti-rabbit IgG (in the case of antibody X617, which was raised in rabbits) . After application of coverslips, the slides are visualized by fluorescent light microscopy. A test is scored as positive if for a particular serum sample, a clear fluorescent signal is seen at the membrane surface of GLUT-2 expressing AtT-

20_ins cells but not in parental AtT-20_ins cells . A positive response with antibody X617 further proves that the GLUT- 2 protein is expressed in proper orientation and that epitopes that are expected to reside at the cell surface are indeed recognizable.

2. Use of a fluorescence activated cell sorter (FACS) to score immune complex formation

Cells are prepared for FACS analysis essentially as described for the microscope slide approach except that incubations are done with cells in suspension rather than attached to microscope slides. Briefly, near-confluent tissue culture plates containing parental AtT-20_ins cells or GLUT-2 expressing CGT-6 cells are washed with PBS, and then exposed to 0.02% EDTA for 15 minutes at 37°C to dislodge cells from the plate. The dispersed cells are washed with culture media followed by PBS and used as intact, live cells or fixed gently in 0.5% paraformaldehyde/PBS for 15 minutes at room temperature. The live or fixed cells are then incubated in 100 μl of patient serum: PBS in a ratio of 1:1, with 0.002% EDTA added to keep the cells dispersed. After a one hour incubation at 4°C, the cells are washed 3 times with PBS and incubated with anti-human IgG or anti-human globulin fraction labeled with phycoerythrin for 1 hour at 4°C. Subsequently, the cells are washed with PBS and run through a flow cytometer in the red channel. Phycoerythrin is chosen as the fluorescent marker because it was found that the AtT-20_ins cells have a natural fluorescence in the green channel that is used for FITC- labeled antibodies. B. Results

1. Microscope slide technique

Use of the antibody raised against the external loop peptide of GLUT-2 in the inventor's laboratory (X617) results in a clear fluorescent staining at the surface of engineered AtT-20_ins cells that express GLUT-2, but gives no such signal in parental cells that have not been engineered for GLUT-2 expression. Furthermore, the signal in GLUT-2 transfected cells can be blocked by preincubation of antibody X617 with the peptide to which it was raised. These results indicate that formation of an immune complex with an external (extracellular) epitope of the GLUT-2 protein can occur and is readily detectable. In preliminary studies with sera isolated from new-onset Type I diabetic patients (ranging in age from 10-20 years old) , and age matched normal controls, the diabetic sera, but not the normal sera show a greater immunoreactivity against the GLUT-2 transfected cells relative to the untransfected controls.

2. FACS technique

The FACS method was found to be appropriate for detecting the presence of a specific immune complex. GLUT-2 expressing AtT-20_ins cells were treated with the anti-GLUT-2 antibody X617 and with anti-rabbit IgG second antibody labeled with phycoerythrin (set one) . Cells were also incubated with antibody X617 after it had been preincubated with GLUT-2 expressing AtT-20ins cells (set two) . The cells are loaded into the FACS, which passes the cells one-by-one past a light source set at a wavelength that will excite the fluorescent marker of the second antibody. The cells then pass a detector which measures the fluorescence emission from the cells. Data are plotted as a histogram of fluorescence intensity. Cells from set one gave curves which were shifted to the right relative to those of set two, indicating a greater fluorescence intensity in those cells. A similar study was performed with parental AtT-20_ins cells not expressing GLUT-2. In these cells, no difference is seen between the naked antibody and antibody preabsorbed with GLUT-2 expressing cells. Taken together, these data serve to validate the technique, in that a specific response can be measured to an antibody known to react with an extracellular domain of GLUT-2.

This method was used in the preliminary analysis of serum from a diabetic patient. A fluorescence spectrum of GLUT-2 transfected AtT-20_ins cells incubated with the second antibody (phycoerythrin labeled anti-human globulin) alone was generated (set one) . GLUT-2 transfected cells incubated with serum isolated from a normal patient, resulted in a shift in the fluorescence intensity relative to the set one cells. Importantly, cells incubated with serum from a patient with new-onset Type I diabetes (set three) had a much more pronounced rightward shift in fluorescence relative to the normal or nonserum controls.

EXAMPLE III

INTERACTIONS OF SERA FROM DIABETIC PATIENTS

WITH ISLET CELLS AND ENGINEERED AtT20_jns CELLS

The following example is directed to an analysis of serum samples from diabetic patients and non-diabetic subjects. In particular, the interactions of purified IgG samples with rat islet cells and engineered AtT20_ins cells was investigated using both binding assays and assays based on the inhibition of glucose uptake. The following results demonstrate the usefulness of such analyses in diagnostic and prognostic tests. A. Immunofluorescence/Flow Cvtometric Methods

AtT20_ins cells and GLUT-2-expressing AtT20_jns cells were harvested by removal of cells from plates with a rubber policeman in Dulbecco's phosphate-buffered saline, pH 7.6. Following two washes in Dulbecco's phosphate- buffered saline by sedimentation at 500 x g for 30 seconds at room temperature, the cells were divided into

1.5 ml microfuge tubes at a density of approximately 10⁵ cells per tube. Cells were incubated for 1 hour at 4°C in 150μl of patient sera with occasional agitation. The cells were then washed twice by centrifugation at 500 x g for 30 seconds in Dulbecco's phosphate-buffered saline pH

7.6 and resuspended in R-phycoerythrin-labeled goat antihuman, heavy chain-specific IgG (R-PEAb) (Fisher

Scientific) and incubated for 1 hour at 4°C with occasional shaking. Following an additional two washes by centrifugation at 500 Xg for 30 seconds in Dulbecco's phosphate-buffered saline, pH 7.6, the cells were resuspended in 500μl of Dulbecco's phosphate-buffered saline, pH 7.6, and analyzed for IgG binding using flow cytometry.

Flow cytometry was performed on a FACScan (Becton Dickinson) flow cytometer. Forward scatter threshold was set at 100 using the E-01 forward scatter detector. Linear amplifier gains were 6.18 for forward scatter and 1.22 for 90° angle light scatter with a photomultiplier setting of 274 volts. Forward and 90° angle light scatter were read on linear scale and fluorescence measurements were made on logarithmic scale. Setting adjustments were made by using a sample of unstained cells and increasing the photomultiplier voltage so that events were on-scale during observation of 530 ± 15 nm (FL1) histogram. A sample of cells stained only with R- phycoerythrin-labeled goat antihuman IgG (R-PEAb) was then used to adjust the photomultiplier voltage so that events were on-scale during measurement of a 575 ± 13 n (FL2) histogram. A control specimen was then used to adjust the FL2 photomultiplier tube voltage such that FL2 histogram events remain minimally on scale. The FL2-FL1 compensation was adjusted to minimize fluorescence overlap and for these cells a setting of 45.9% was used. Acquisition of 10⁴ events per specimen were required and data were stored on floppy discs for analysis.

B. Results

1. Effects of IgG from Diabetic Patients and

Nondiabetic Subjects on 3-0-Methyl-/3-D-Glucose Uptake by Islet Cells and GLUT-2-Expressing and GLUT-l-Expressing AtT20_ins Cells

The following assays were performed to investigate the effects of human IgG on glucose uptake by intact cells. The assays were performed as described by Johnson and Unger, PCT Patent Application WO 91/13361, incorporated herein by reference.

Examination of the effects of purified IgG from 7 patients with new-onset IDDM and 6 nondiabetic individuals revealed that 3-0-methyl-/3-D-glucose uptake by rat islet cells was significantly inhibited in the presence of IgG from patients with IDDM. Initial rates of uptake averaged 15 mmoles 3-0-methyl-glucose/min/litre islet cell space in the presence of IgG from nondiabetic patient sera versus 9 mmoles 3-0-methyl-glucose/min/litre islet cell space in the presence of IgG from sera of patients with IDDM (p < 0.05). These rates translate into a 40% inhibition of glucose transport in the presence of IgG from patients with IDDM.

If this inhibition is the result of an antibody effect on GLUT-2 activity, it should also be manifest on the GLUT-2-transfected AtT20_ins cell line but not on the GLUT-1-transfected cells. The expression of GLUT-2 in this cell line confers them with glucose transport characteristics remarkably similar to those found in islet cells. Purified IgG from the same patients with new-onset IDDM reduced the initial rate of glucose transport in GLUT-2-expressing AtT20_ins cells from 15.5 mmoles 3-0-methyl-glucose/min/litre cell space to 6.2 mmoles 3-0-methyl-glucose/min/litre cell space, p < 0.05. This represents a 60% reduction in glucose transport in the presence of IgG from patients with IDDM compared to uptake in the presence of IgG from nondiabetic subjects.

Rat islet cells exhibit two kineticall'y distinct facilitated diffusion glucose transporter functions, a high Kr_ø function ascribed to GLUT-2 and a low I^ transport function attributed to unidentified transporter. Results from a detailed kinetic analysis of the inhibition of glucose transport into islet cells induced by diabetic IgG indicated that the inhibition was directed against the high J^ or GLUT-2 mediated function. As a test of the specificity of inhibition of GLUT-2, additional measurements in GLUT-1 expressing AtT20_jns cells were made. Although nontransfected AtT20_ins cells express GLUT-1 constitutively, the GLUT-1-transfected cell line overexpresses this protein and exhibits a greater than 10-fold increase in the velocity of glucose uptake which increases the accuracy of the transport measurement. Glucose uptake in GLUT-1-transfected AtT20_ins cells treated with IgG from new-onset IDDM patients was indistinguishable from transport in the presence of IgG from nondiabetic individuals. These data indicate that IgG from new-onset IDDM patients does not inhibit glucose transport in AtT20_ins cells that express a facilitative glucose transporter other than GLUT-2. 2. Specificity of nteractions of Sera from

Patients with New-Onset IDDM and Nondiabetic Patients for GLUT-2-Expressing AtT20_|ns cells

It was important to establish the specificity of IgG binding to intact GLUT-2-expressing AtT20_ins cells by performing analyses of IgG binding to the parent AtT20_ins cells. Subtraction of the percentage of cells found in R₂ using the nontransfected AtT20_ins cell line from the percentage of cells found in R using the GLUT-2- expressing AtT20_jns cell line would be expected to reflect the specific binding of IgG to GLUT-2. Such analyses were performed for each individual serum and a positive interaction was defined as an increase in IgG binding greater than two standard deviations from the mean observed in the nondiabetic patient population. It was found that 29 of 31 (94%) of the nondiabetic population were negative for IgG binding to GLUT-2 while 23 of 30 (77%) of sera from IDDM patients were positive. Thus, 81% of negative results were from nondiabetic patients and 92% of positive results were from patients with IDDM (Table 2) . The Youden index of these results gave J = 0.73 (Table 2), and the level of significance of the separation between the two populations was p < 0.0001.

Table 2. Regional Analysis of IgG Binding from Sera of Nondiabetic Children and Patients with IDDM to GLUT-2- Expressing AtT20_jns Cells after Subtractions of IgG Binding to Nontransfected AtT20_jns Cells.

Individual with Peak Fluorescence in R₂

Patient Group >2 Standard <2 Standard

Deviation Shift Deviation Shift

IDDM 23/30 (77%)* 7/30 (23%)

Nondiabetic 2/31 (6%) 29/31 (94%) Sensitivity 23/30 (77%)

Specificity 29/31 (94%)

False 2/25 (8%)

Positive Rate

False 7/36 (19%) Negative Rate

Youden Index: J = l - (0.08 + 0.19) = 0.73

Defined as an increase in the peak number of fluorescent cells in R fluorescence of greater than two standard deviations from the mean of the number of fluorescent cells in R after treatment with sera from nondiabetic children.

^* p < 0.0001 compared to the nondiabetic population

EXAMPLE IV PERFUSION OF A COLUMN CONTAINING CGT-6 CELLS FOR INCREASED INSULIN PRODUCTION

A. Methods

Insulin secretion from CGT-6 (GLUT-2 expressing AtT-20_jns) cells was evaluated using a column perfusion technique (Knudsen et al . , 1983). Cells were grown in liquid culture on microcarrier beads (InvitroGen) .

Approximately 50 x 10⁶ cells were harvested by gentle centrifugation (500 rpm in a Sorvall RT6000B desk top centrifuge) , resuspended in 4ml Krebs-Ringer salt (KRS) solution, pH 7.4, and loaded onto a Pharmacia PlO/10 column. A cell count was obtained immediately before loading the column in the following manner. An aliquot of cells was taken, the beads digested with 1.2 U/ml Dispase (Boehringer Mannheim) , the cell clumps were dispersed by extrusion through a 25 gauge needle and the cells were counted directly.

After the beads settled in the column, the top plunger of the column was gently inserted and the whole apparatus was submerged in a 37°C water bath. The cells were then perfused as described below.

B. Results

In early secretion studies, a static incubation procedure was used in which cells were grown in tissue culture dishes and exposed to secretagogue-containing media over relatively long time periods (3 hours) . While this technique was found to be valuable for screening new cell lines, it provides no information concerning the dynamics of insulin release. Perfusion studies were therefore carried out to address this concern, and to evaluate whether glucose-stimulated insulin secretion from GLUT-2 expressing AtT-20_ins cells occurs in a similar time frame as the rapid islet β cell response.

Native AtT-20^ cells, as well as GLUT-2 and GLUT-1 transfected lines were grown in liquid culture on microcarrier beads (InvitroGen) , harvested into a Pharmacia PlO/10 column, and washed with HBSS lacking glucose for 15 minutes. The capacity of lines CGT-6 (GLUT-2 transfected) , CGT1-15 (GLUT-1 transfected) and the parental AtT-20_ins cells to secrete insulin in response to glucose was compared. Perfusion with HBSS lacking glucose was continued after the 15 minute wash-out for an additional 25 minutes. During this period, there was a gradual decline in insulin release from all three cell lines. Phase II was initiated by switching to HBSS buffer containing 5mM glucose. A 10-fold increase in insulin release from CGT- 6 cells was noted in the first sample, collected in the first 2.5 minutes after the switch to glucose-containing buffer. This increase was sustained in 2 samples (representing a total of 5 minutes) , after which insulin secretion declined to a second plateau that was 3-fold above the pre-glucose level. This biphasic pattern of insulin release is similar to that observed upon glucose stimulation of normal islets. Only small changes in insulin release were observed in phase II for either the parental AtT-20_ins cells or the GLUT-1 transfected CGT1-15 line.

After 25 minutes of perfusion with 5mM glucose, the cells were switched back to HBSS lacking glucose.

Insulin secretion from the CGT-6 cells persisted at the glucose-stimulated level for approximately 10 minutes after the switch to buffer lacking glucose, but then declined rapidly. The low level of insulin release from parental AtT-20^ cells and CGT1-16 cells was further reduced during perfusion with glucose free media. In phase IV, cells were switched back to buffer containing 5 mM glucose. The CGT-6 cells again showed a much stronger secretory response to glucose, but the response was less rapid (requiring 15 minutes to reach maximum) , and was without an obvious first phase and second phase.

Switching back to buffer lacking glucose in phase V again resulted in a dramatic albeit delayed reduction in insulin release from CGT-6 cells. In the last 25 minute phase, cells were perfused with HBSS containing the combination of 5 mM glucose and 0.5 μM forskolin. In _keeping with the results from earlier static incubation studies, GLUT-2 expressing CGT-6 cells exhibited a stronger insulin secretory response to glucose + forskolin than either the parental cells or the GLUT-1 transfected cells. The response of line CGT-6 to glucose + forskolin was sustained until the end of the study, suggesting that the cells were not depleted of insulin during the perfusion study. Consistent with this interpretation, no changes in insulin content were noticed in any of the cell lines isolated before and after perfusion studies.

EXAMPLE V ADENOVIRUS-MEDIATED GENE TRANSFER INTO ISLET CELLS

A. Methods

1. Preparation of Recombinant Adenovirus

Recombinant adenovirus (Gluzman et al . , 1982) containing distinct cDNAs (AdCMV-cDNA) were prepared according to the following method, which relates particularly to the preparation of recombinant adenovirus containing /3-galactosidase (AdCMV-βGAL) (Herz and Gerard, 1993) . A cDNA encoding the E. coli _/3-galactosidase carrying the SV40 T antigen nuclear targeting signal (Bonnerot et al . , 1987) was inserted into pACCMV to create a novel construct. The resulting expression cassette comprises the cytomegalovirus (CMV) promoter, the _/3-galactosidase cDNA and the mouse protamine polyadenylation signal, and is flanked by adenovirus type 5 sequences. In this construct, the El region of adenovirus is replaced by the foreign gene.

The resulting plasmid was cotransfected into 293 cells (Graham et al . , 1977) together with a plasmid carrying the complete adenovirus type 5 genome (pJM17) (McGrory et al . , 1988). Plasmid sequences conferring ampicillin and tetracycline resistance are inserted into the virus genome at map position 3.7. The molecular strategy employed to produce recombinant adenovirus is based upon the fact that, due to the packaging limit of adenovirus, pJM17 cannot efficiently form plaques on its own. Therefore, homologous recombination between the pAC-cDNA plasmid and pJM17 within a transfected cell results in a viable virus that can be packaged and form plaques only on 293 cells which express adenovirus ElA proteins.

Co-transfection was performed as follows: 293 cells were cultured in Dulbecco's modified Eagle's medium (DMEM, from GIBCO-BRL) containing 10% fetal bovine serum (FBS, from Hyclone) in a humidified 5% C0₂ atmosphere. Confluent 10 cm dishes were split to three 6 cm dishes the day before calcium phosphate cotransfection of an appropriate amount of DNA, such as 4 μg pJM17, 4 μg pACCMV-βGAL, and 12 μg HeLa DNA as carrier. Six hours after addition of the DNA to the cells, a 15% glycerol shock was used to boost transfection efficiency and the cells were overlaid with 0.65% Noble agar in DMEM containing 2% FBS, 50 μg/ml penicillin G, 10 μg/ml streptomycin sulfate, and 0.25 μg/ml fungizone (GIBCO) .

Monolayers were incubated for approximately 10 days until the appearance of viral plaques.

These plaques were picked, suspended in DMEM containing 2% FBS, and used to infect a new monolayer of 293 cells. When greater than 90% of the cells showed infection, viral lysates were subjected to a freeze/thaw cycle and were designated as primary stocks. Recombinant virus with the correct structure was verified by preparation of viral DNA from productively-infected 293 cells, restriction analysis, and Southern blotting. Secondary stocks were subsequently generated by infecting 293 cells with primary virus stock at a multiplicity of infection of 0.01 and incubation until lysis.

The large scale production of recombinant adenovirus can be performed in 293 cells grown either in 15 cm culture dishes or in suspension using Joklik's calcium-free MEM (GIBCO) supplemented with 10% FBS, as follows. Infected cells may be lysed 48 hours post¬ infection with Dulbecco's PBS (GIBCO) containing 1 mM MgCl₂ and 0.1% NP-40. Virus-containing extracts should then be centrifuged, such as at 12,OOOxg for 10 minutes, to remove debris before precipitation of the virus particles by addition of 0.5 vol 20% polyethylene glycol (PEG) 8000, 2.5 M NaCl and incubated on ice' for 1 hour. Virus can be collected by centrifugation at 12,OOOxg for 10 minutes, resuspended in isotonic saline (135 mM NaCl, 5 mM KCl, 1 mM MgCl₂, 10 mM Tris-HCl, pH 7.4), and dialyzed against the same buffer overnight before sterilization through a 0.22 μm filter.

Alternatively, PEG precipitated virus may be resuspended in 50 mM Tris-HCl pH 7.8 containing CsCl (d=1.10 g/ml) , layered over a step-gradient formed of 2 ml CsCl (d=1.40) and 3 ml of CsCl (d=1.30), and centrifuged 2 hours at 20,000 rpm at 10°C in a Sorvall

TH641 rotor. Virus can then be collected from the lower interface and dialyzed overnight at 4°C versus isotonic saline.

2. Gene Transfer into Islet Cells

To investigate the utility of recombinant adenoviruses in gene transfer into the islets of Langerhans, cultured rat islets were exposed to a replication-defective, infectious adenovirus containing β-galactosidase as a marker gene. Islets were incubated with the recombinant virus, termed AdCMV-βGAL, for 1 hour at a multiplicity of infection of approximately 10. A stock solution of the recombinant virus, termed AdCMV/3gal, with a titer of approximately 5 x 10⁷ pfu/ml was used and virus was applied to cells at a multiplicity of infection of approximately 10:1. β-galactosidase expression was scored following dispersal of the. islets and incubation with substrate and colorimetric reagents.

B. Results

1. Construction of Recombinant Adenoviruses

In addition to the β-galactosidase adenovirus, recombinant adenoviruses containing genes involved in glucose sensing, including GLUT-2, GLUT-1, glucokinase and mutant glucokinases associated with Maturity-Onset Diabetes of the Young (MODY) were prepared according to the methodology set forth above. The said genes are under the control of the CMV promoter and adenovirus containing the insulin promoter have also been prepared. The structures of the recombinant viruses generated were verified by restriction enzyme digestion and Southern blotting. Note that the inventors' recognized expertise in the construction of recombinant adenoviruses led to an invitation to write a chapter on the subject which is now in press (Becker et al . , 1994) and which provides further details pertaining to these procedures.

2. Marker Gene Transfer into islet Cells

Four to six days after exposure to the replication- defective adenovirus containing the marker gene β-galactosidase, 75-80% of the islet cells were found to express β-galactosidase as indicated by their blue color. At the 6 day time point, infected and uninfected islets had a similar glucose-stimulated insulin secretion response as assayed by perfusion, indicating that the recombinant virus has no detrimental effect on islet cell function. It can thus be concluded that recombinant adenovirus represents an efficient means of introducing genes into primary islet cells.

/3-galactosidase expression in isolated rat islets treated with AdCMV-/3GAL recombinant adenovirus is demonstrated in Figure 5. Light microscopic view of a representative intact islet four days after transduction with AdCMV-jSGAL and treatment of the islets with chromogenic substrate as described above, viewed at 20 x magnification, is shown in Figure 5A. /3-galactosidase expressing cells are indicted by their blue color. In Figure 5B, an islet 21 days after viral transductions treated similarly to that in Figure 5A, as viewed at 20 x magnification, is pictured. Figure 5C shows a multicell aggregate that was prepared by trypsin-mediated dispersal of intact islets prior to incubation with the chromogenic substrate to allow quantitation of the efficiency of gene transfer, viewed at 40 x magnification. In Figure 5D, a control islet treated with _/3-galactosidase substrate after 4 days in culture without exposure to AdCMV-/3GAL, is viewed at 20 x magnification.

Adenovirus-mediated delivery of /3-galactosidase to islets in vivo was also investigated. Recombinant adenovirus containing the /3-galactosidase gene directed by the CMV promoter was infused via the jugular vein into normal rats over 2 four hour periods at a concentration of approximately 1 x 10' pfu/ml. Pancreas tissue was removed, frozen and sliced into thin sections. Sections were fixed with 0.5% glutaraldehyde and then incubated with the /3-galactosidase chromogenic substrate X-gal. Animals were infused with the AdCMV-/3gal virus for five days. Blue cells are clearly located in discrete areas of the pancreatic sections which match the areas detected by the insulin antibody. Some quenching of the insulin immunofluorescence by the pre-existing blue color is evident. Five days after termination of the infusion, fresh-frozen 15μm pancreatic sections were prepared, incubated with the _/3-galactosidase substrate solution (Figure 8A and 8C) and then treated with an anti-insulin antibody (Figure 8B and 8D) . In Figure 8E, a representative islet isolated from a virally infused animal that was treated with j3-galactosidase substrate and then embedded and sectioned into 5 μm sections is displayed. Multiple blue nuclei are evident. In contrast to Figure 8E, islets isolated from a control animal are completely devoid of blue color (Figure 8F) .

3. GLUT-2 Gene Transfer into Diabetic Islet cells

Islets from Zucker diabetic fatty (ZDF) rats, which completely lack GLUT-2 expression and which fail to respond to levels of glucose that stimulate normal islets to secrete insulin, were isolated for use in gene transfer studies. Recombinant adenovirus containing the GLUT-2 cDNA was administered to cultured islets from such animals allowing the effects of this maneuver on the insulin secretion response to glucose to be assayed.

4. Adenovirus-mediated Expression of Variant

Kinases.

Adenovirus-mediated expression of variant glucokinases in CV-1 cells was studied (Figure 6) . CV-1 cells were infected with recombinant adenoviruses at a multiplicity of infection of 10. Glucokinase expression was assayed by western blot hybridization analysis using 100 μg islet protein and an antibody that detects rat islet glucokinase (Antibody U343, Quaade et al . , 1991). Lanes contain the following samples: Lane 1, Islet glucokinase expressed in bacteria (Quaade et al . , 1991); Lane 2, Uninfected islets; Lane 3, Islets infected with recombinant adenovirus containing the 3-galactosidase cDNA; Lane 4, Islets infected with recombinant adenovirus containing cDNA encoding a glucokinase variant containing an amber mutation (premature stop codon) in place of glutamic acid 279 that is not synthesized as a stable protein; Lane 5, Islets infected with recombinant adenovirus containing cDNA encoding a glucokinase variant with a point substitution that results in replacement of amino acid glycine 261 with arginine; Lane 6, Islets infected with recombinant adenovirus containing cDNA encoding a glucokinase variant with a point substitution that results in replacement of amino acid glutamic acid 279 with glutamine (Figure 6A) .

The glucokinase enzymatic activity measured in crude homogenates of the CV-1 cells is described in the western blot (Figure 6B) . Expression of the Gly261Arg and Glu270Gln variants increase glucokinase activity in CV-1 cells in accord with their reported enzymatic properties as assayed in bacteria (Gidh-Jain et al . , 1993). These data show that recombinant adenovirus allows highly efficient glucokinase gene transfer into cell lines such as CV-1.

Adenovirus-mediated expression of variant glucokinases in primary rat islets was also investigated (Figure 7). On the day of isolation, approximately 2,000 rat islets were mixed with recombinant adenoviruses at a multiplicity of infection of 10. Glucokinase expression was then assayed by blot hybridization analysis using 100 μg islet protein and an antibody that detects rat islet glucokinase (Antibody U343, Quaade et al . , 1991). Lanes contain the following samples: Lane 1, Uninfected islets; Lane 2, Islets infected with recombinant adenovirus containing the /3-galactosidase cDNA; Lane 3, Islets infected with recombinant adenovirus containing cDNA encoding a glucokinase variant containing an amber -Ill- mutation (premature stop codon) in place of glutamic acid 279 that is not synthesized as a stable protein; Lane 4, Islets infected with recombinant adenovirus containing cDNA encoding a glucokinase variant with a point substitution that results in replacement of amino acid glycine 261 with arginine; Lane 5, Islets infected with recombinant adenovirus containing cDNA encoding a glucokinase variant with a point substitution that results in replacement of amino acid glutamic acid 279 with glutamine; Lane 6, empty; Lane 7, Islet glucokinase expressed in bacteria (Quaade et al . , 1991). This figure shows that recombinant adenovirus allows highly efficient glucokinase gene transfer into freshly isolated islets of Langerhans.

EXAMPLE VI HUMAN GENE TRANSFER PROTOCOLS

This prophetic example describes some of the ways in which the present invention is envisioned to be of use in the treatment of diabetes via adenovirus-mediated gene therapy.

Diabetic patients for whom the medical indication for adenovirus-mediated gene transfer treatment has been established would be tested for the presence of antibodies directed against adenovirus. If antibodies are present and the patient has a history of allergy to either pharmacological or naturally occurring substances, application of a test dose of on the order of 10^ to 10^ recombinant adenovirus under close clinical observation would be indicated.

For the treatment of NIDDM, recombinant adenovirus expressing a suitable candidate gene or genes involved in glucose sensing, particularly either the GLUT-2 and/or glucokinase genes, under the control of β cell specific promoter/enhancer elements, such as the insulin promoter, would be prepared and purified according to a method that would be acceptable to the Food and Drug Administration (FDA) for administration to human subjects. Such methods include, but are not limited to, cesium chloride density gradient centrifugation, followed by testing for efficacy and purity.

Virus may be administered to patients by means of intravenous administration in any pharmacologically acceptable solution, either as a bolus or as an infusion over a period of time. Generally speaking, it is believed that the effective number of functional virus particles to be administered would range from 1 x 10¹⁰ to 5 x 10 1 . If warranted or desi.red, recombi.nant adenovi.rus could be delivered closer to the site of the target cells using a catheter.

Since the adenovirus employed will be replication incompetent, no deleterious effect of the virus itself on subject health is anticipated. However, patients would remain hospitalized during the treatment for at least 48 hours to monitor acute and delayed adverse reactions. Glucose-tolerance tests would be monitored twice daily to follow the efficacy of the gene transfer.

Further possible follow-up examinations include pancreatic biopsy, in which the pattern of expression of the transferred gene could be directly assessed. This would also supply information about the number of islet cells that have taken up the transferred gene and about the relative promoter strength in the human pancreas. Based on the data obtained adjustments to the treatment may be desirable. These adjustments might include adenovirus constructs that use different promoters or a change in the number of pfu injected to ensure a infection of more, or all, islet cells without unphysiological overexpression of the recombinant genes.

It is contemplated that the expression of exogenous genes transferred in vivo by adenovirus can persist for extended periods of time. Therapeutically effective long-term expression of virally transferred exogenous genes will have to be addressed on a case by case basis. Marker genes are limited in their usefulness to assess therapeutically relevant persistence of gene expression as the expression levels required for the amelioration of any given genetic disorder might differ considerably from the level required to completely cure another disease. For example, the relatively high expression necessary to treat α_j-antitrypsin deficiency, which results from consumption of molecule whilst executing its desired function, has not yet been obtained (Jaffe et al . , 1992). In contrast, correction of normal islet cell function in NIDDM patients will require considerably lower expression of the transferred gene(s), as such intracellular molecules fulfill their physiological roles without any consequent inactivation.

Safety-related concerns of the use of replication deficient adenovirus as a gene transfer vehicle in humans have been addressed in the past (Rosenfeld et al . , 1992; Jaffe et al . , 1992). In agreement with previous studies the present inventors have not been able to detect any noticeable histological changes in the liver morphology of mice injected with recombinant adenovirus at doses below about 3xlθ" pfu. However, at higher doses of approximately lxl0¹" pfu or greater, an extensive infiltration of monocytic immune cells into the livers of injected mice has been noted at four days after injection. This infiltration phenomenon has been previously observed in the mouse lung following intranasal inoculation of 10 pfu of adenovirus type 5 (Ginsberg et al . , 1991). Plasma levels of the liver marker enzymes alanine aminotransferase and aspartate aminotransferase show a 20 fold increase following administration of 2x10 pfu intravenously. Therefore, the dose of adenovirus to be administered must be appropriately controlled so as to minimize untoward side effects of the gene therapy regimen, and a more extensive careful evaluation will be necessary to ensure that adenovirus is safe for human in vivo gene therapy.

While the compositions and methods of this invention have been described in terms of preferred embodiments, it will be apparent to those of skill in the art that variations may be applied to the composition, methods and in the steps or in the sequence of steps of the method described herein without departing from the concept, spirit and scope of the invention. More specifically, it will be apparent that certain agents which are both chemically and physiologically related may be substituted for the agents described herein while the same or similar results would be achieved. All such similar substitutes and modifications apparent to those skilled in the art are deemed to be within the spirit, scope and concept of the invention as defined by the appended claims.

REFERENCES

The references listed below are incorporated herein by reference to the extent that they supplement, explain, provide a background for or teach methodology, techniques and/or compositions employed herein.

PCT Application WO 9110470.

PCT Application WO 9110425.

PCT Application WO 9015637.

PCT Application WO 9002580.

PCT Application WO 890167.

US Patent 5,011,472.

US Patent 4,892,538.

US Patent Application 07/823,747.

Altman et al. (1986) Diabetes, 35:625-633.

Andersson et al. (1989) J. Biol. Chem., 264:8222-8229.

Andreone et al. (1989) J. Biol. Chem., 264:363-369.

Arora et al. (1990) J. Biol. Chem., 265:6481-6488.

Ashcroft (1980) Diabetologia, 18:5-15.

Baekkeskov et al. (1982) Nature, 298:167-169.

Baekkeskov et al. (1990) Nature, 347:151-156. Barr and Leiden (1991) Science, 254:1507-1509.

Becker et al. (1993) Diabetes, in press.

Becker et al. (1994a) Cell Biology, 43:161-189.

Becker et al. (1994b) J. Biol. Chem., in press.

Becker et al. (1994c) J. Cell. Biochem. Suppl., 18A:133.

Becker et al. (1994d) Methods Cell Biology, 43:161-189.

Bell et al. (1990) Diabetes Care, 13:198-208.

Bell (1991) Diabetes, 40:413-422.

BeltrandelRio et al. (1994) Diabetes, in press.

Birnbaum et al. (1987) Science, 235:1495-1498.

Bonnerot et al. (1987) Proc. Natl. Acad. Sci. USA, 84:6795-6799.

Bright (1987) Diabetes, 36:1183-1186.

Canonico (1989) Endocrinology, 125:1180-1186.

Capecchi (1989) Trends in Genetics, 5:70-76.

Chen et al. (1990) Proc. Natl. Acad. Sci. US. A., 87:4088- 4092.

Clark et al. (1990) Endocrinology, 127:2779-2788.

Cockett et al. (1990) Bio /Technology, 8:662-667. Cone et al. (1984) Proc. Natl. Acad. Sci. U.S.A., 81:6349-6353.

Danos et al. (1988) Proc. Natl. Acad. Sci. U.S.A., 85:6460-6464.

Dhawan et al. (1991) Science, 254:1509-1512.

DiMario et al. (1988) Diabetes, 37:462-466.

Doberson et al. (1980) N. Engl. J. Med., 303:1493-1498.

Efrat et al. (1988) Proc. Natl. Acad. Sci. USA, 85:9037- 9041.

Efrat et al. (1994a) Proc. Natl. Acad. Sci. USA, 91:2051- 2055.

Efrat et al. (1994b) Transplantation Proceedings, 26:363- 365.

Elahi et al. (1982) N. Engl. J. Med., 306:1196-1202.

Elias et al. (1991) Proc. Natl. Acad. Sci., U.S.A., 88:3088-3091.

Ferber et al. (1993) Diabetes, 42 (Suppl.l) :11A.

Ferber et al. (1994) J. Biol. Chem., 269:11523-11529.

Flier et al. (1987) Science, 235:1492-1495.

Forster and Symons (1987) Cell, 49:211-220.

Fritschy et al. (1991) Diabetes, 40:37.

Frougel et al. (1992) Nature 356:162-164. Fukumoto et al . (1988) Proc. Natl. Acad Sci, U.S.A., 85:5434-5438.

Gazdar et al. (1980) Proc. Natl. Acad. Sci. U.S.A., 77:2519-2523.

Gerard and Meidell (1991) J. Cell Biochem. Suppl., 15C133.

German and Rutter (1991) Diabetes, 40:163A.

Ghosh-Choudhury and Graham (1982) Biochem. Biophys. Res. Comm., 147:964-973.

Gidh-Jain et al. (1993) Proc. Natl. Acad. Sci. USA, 90:1932-1936.

Ginsberg et al. (1991) Proc. Natl. Acad. Sci. USA, 88:1651-1655.

Giroix et al. (1984) Biochem. J., 223:447-453.

Giroix et al. (1985) Arch. Biochem. Biophys., 241:561- 570.

Glass et al. (1991) Cell, 66:405-413.

Gleichmann et al. (1987) Diabetes, 36:578-584.

Gluzman et al. (1982) in Eukaryotic Viral Vectors

Gluzman, Ed., pp. 187-192, Cold Spring Harbor Press:Cold Spring Harbor, New York.

Gomez-Foiz et al. (1992) J. Biol. Chem., 26:25129-25134.

Graham et al. (1977) J. Gen. Virol., 36:59-72. Grandison (1990) Endocrinology, 127:1786-1791.

Halban et al. (1983) Biochem. J., 212:439-443.

Hasellof and Gerlach (1988) Nature, 334:585-591.

Hedeskov (1980) Physiol. Rev., 60:442-509.

Herz and Gerard (1993) Proc. Natl. Acad. Sci. USA, .

Hughes et al. (1991) J. Biol. Chem., 266:4521-4530.

Hughes et al. (1992) Proc. Natl. Acad. Sci. USA, 89: 688- 692.

Inman L.R. et al. (1993) Proc. Natl. Acad. Sci. USA, 90:1281-1284.

Irvine et al. (1980) In: Immunology of Diabetes, Teviot Scientific Publications, pp. 117-154.

Iynedjian et al. (1989) Proc. Natl. Acad. Sci. U.S.A., 86:7838-7842.

Jaffe et al. (1992) Nature Genetics, 1:372-378

Johnson et al. (1990a) J. Biol. Chem., 265:6548-6551.

Johnson et al. (1990b) Science, 250:546-549

Johnson et al. (1990c) N. Engl. J. Med., 322:653-659.

Karlsson et al. (1986) EMBO J., 5:2377-2385.

Kuglin et al. (1988) Diabetes, 37:130-132.

Kuwajima et al. (1986) J. Biol. Chem., 261:8849-8853. Lacy et al . (1986) Ann. Rev. Med., 37:33-40.

Lacy et al. (1991) Science, 254:1782-1784.

Ledley (1990) J. Inher. Metab. Dis., 13:597-616.

Lenzen et al. (1987) Acta Endocrinologica, 115:514-520.

Lernmark et al. (1981) Diabetology, 21:431-35.

Lernmark et al. (1982) Diabetes Med., 4:285-292.

Ludwig et al. (1987) Diabetes, 36:420-425.

Lynch et al. (1991) J. Cell Biol., 112:385-395.

Maclaren et al. (1975) Lancet, 1:997-1000.

Madsen et al. (1988) Proc. Natl. Acad. Sci. U.S.A., 85:6652-6656.

Magnuson and Shelton (1989) J. Biol. Chem., 264:15936- 15942.

Malaisse et al. (1984) Endocrinology, 115:2015-2020.

Malaisse et al. (1990) Biochem. Soc. Trans., 18:107-108.

Mansour et al. (1988) Nature, 336:348-352.

McGrory et al. (1988) Virol., 163: 614-617.

Meglasson and Matchinsky (1986) Diabetes /Metabolism Rev., 2:163-214.

Meglasson et al. (1987) Diabetes, 36:477-484. Miller (1992) Nature, 357:455-460.

Moore et al. (1983) Cell, 35:531-538.

Moriarity (1978) Life Sci., 23:185-189.

Mueckler et al. (1985) Science 229: 941-945.

Newgard et al. (1986) Proc. Natl. Acad. Sci. U.S.A., 83:8132-8136.

Newgard et al. (1990) Biochem. Soc. Trans., 18:851-853.

Newgard et al. (1994a) Diabetes, 43:341-350:

Newgard et al. (1994b) J. Lab. Clin. Med., 122:356-363.

Newgard et al. (1994c) In Molecular Biology of Diabetes, Part I, Draznin and LeRoith, Eds., Humana Press Inc.:Totowa, NY, pp.1-33.

Nishi et al. (1988) Biochem. Biophys. Res. Comm., 157:937-943.

O'Shea and Sun (1986) Diabetes, 35:943-946.

Ogawa et al. (1990) J. Clin. Invest, 85:973-976.

Ora et al. (1976) Proc. Natl. Acad. Sci., USA, 73:1338- 42.

Orci et al. (1989) Science, 245:295-297.

Orci et al. (1990) Proc. Natl. Acad. Sci. USA, 87:9953- 9957. Permutt et al. (1989) Proc. Natl. Acad. Sci. USA, 86:8688-8692.

Praz et al. (1983) Biochem. J., 210:345-352.

Prentki et al. (1987) Physiol. Rev., 67:1185-1248.

Quaade et al. (1991) FEBS Lett., 280:47-52.

Rosenfeld et al. (1992) Cell, 68:143-155.

Rossini et al. (1985) Ann. Rev. Immunol., 3:289-320.

Rothman et al. (1992) J. Clin. Invest., 89:1069-1075.

Sambrook et al. (1989) Molecular Cloning: A Laboratory Manual, Cold Spring Harbor Press, pp. 16.30-16.55

Sarkar et al. (1988) Proc. Natl. Acad. Sci. USA, 85:5463- 5467.

Sato et al. (1962) Proc. Natl. Acad. Sci. USA, 48:1184- 1190.

Selden et al. (1987) The New England Journal of Medicine 317:1067-1076.

Schwab et al. (1989) Proc. Natl. Acad. Sci. USA, 86:2563- 2567.

Sen et al. (1988) J. Biol. Chem. 263:19132-19136.

Shelton et al. (1992) Mol. Cell. Biol., 12:4578-4589.

Shimizu et al. (1988a) Diabetes, 37:563-568.

Shimizu et al. (1988b) Diabetes, 37:1524-1530. Srikanta et al. (1983) N. Engl. Jrnl. Med., 308:322-25.

Srikanta et al. (1986) Diabetes, 25:139-142.

Stoller et al. (1989) J. Biol. Chem., 264:6922-6928.

Stossel T. (1987) in The Molecular Basis of Blood Diseases, Chapter 14, pp. 499-533, . B. Saunders Co.: Philadelphia, PA.

Stratford-Perricaudet et al. (1990) Hum. Gene. Ther., 1:241-256.

Stratford-Perricaudet et al. (1992) J. Cliή. Invest., 90:626-630.

Sullivan et al. (1991) Science, 252:718-721.

Tal et al. (1990) J. Clin. Invest., 86:986-992.

Takeda et al. (1993) Diabetes, 42:773-777.

Thomsen et al. (1984) Proc. Natl. Acad. Sci. USA, 81:659- 663.

Thorens et al. (1988) Cell, 55:281-290.

Thorens et al. (1990a) Diabetes Care, 13:209-218.

Thorens et al. (1990b) Proc. Natl. Acad. Sci. USA, 87:6492-6496.

Trus et al. (1981) Diabetes, 30:911-922.

Turk et al. (1987) Prog. Lipid Res., 26:125-181.

Unger (1991) Science, 251:1200-1205. Ullrich and Wollheim (1984) J. Biol. Chem., 259:4111- 4115.

Van Doren and Gluzman (1984) J. Virol., 50:606-614.

Vera et al. (1989) Mol. Cell Biol., 9:4287-4295.

Vionnet et al. (1992) Nature, 356:721-722.

Vischer et al. (1987) -Sioσhe-n. J., 241:249-255.

Walder (1988) Genes & Development, 2:502-504.

Weinhouse (1976) Curr. Top. Cell. Regul., 11:1-50.

Welsh et al. (1990) Proc. Natl. Acad. Sci. USA, 87:5807- 5811.

Wilson (1984) Regulation of Carbohydrate Metabolism, ed. , Beitner, R. (CRC, Boca Raton, FL) , pp. 45-85.

Wolff et al. (1990) Science, 247:1465-1468.

Wu et al. (1989) J. Biol. Chem., 264:16985-16987.

Zheng et al. (1990) Nature, 344:170-173.

SEQUENCE LISTING

(1) GENERAL INFORMATION: (i) APPLICANTS: NAME: BOARD OF REGENTS,

THE UNIVERSITY OF TEXAS SYSTEM STREET: 201 West 7th Street CITY: Austin STATE: Texas COUNTRY: United States of America

POSTAL CODE: 78701 TELEPHONE NO: (512)499-4462 TELEFAX: (512)499-4523

(ii) INVENTORS: NEWGARD, Christopher B.

GERARD, Robert D.

(iii) TITLE OF INVENTION: VECTORS FOR GENETICALLY

ENGINEERED CELLS THAT PRODUCE INSULIN IN

RESPONSE TO GLUCOSE

(iv) NUMBER OF SEQUENCES: 1

(V) CORRESPONDENCE ADDRESS:

(A) ADDRESSEE: Arnold, White & Durkee

(B) STREET: P.O. BOX 4433

(C) CITY: Houston

(D) STATE: Texas (E) COUNTRY: USA

(F) ZIP: 77210

(Vi) COMPUTER READABLE FORM:

(A) MEDIUM TYPE: Floppy disk (B) COMPUTER: IBM PC compatible

(C) OPERATING SYSTEM: PC-DOS/MS-DOS/ASCII (vii) CURRENT APPLICATION DATA:

(A) APPLICATION NUMBER: UNKNOWN

(B) FILING DATE: CONCURRENTLY HEREWITH

(C) CLASSIFICATION: UNKNOWN

(viii) PRIOR APPLICATIONS DATA:

(A) APPLICATION NUMBER: USSN 08/084,742

(B) FILING DATE: 28 JUNE 1993 (28.06.93)

(C) CLASSIFICATION: UNKNOWN

(ix) ATTORNEY/AGENT INFORMATION:

(A) NAME: PARKER, DAVID L.

(B) REGISTRATION NUMBER: 32,165

(C) REFERENCE/DOCKET NUMBER: UTFD416P—

(ix) TELECOMMUNICATION INFORMATION:

(A) TELEPHONE: 512/320-7200

(B) TELEFAX: 713/789-2679 (B) TELEX: 79-0924

(2) INFORMATION FOR SEQ ID NO:l:

(i) SEQUENCE CHARACTERISTICS:

(A) LENGTH: 14 amino acid residues (B) TYPE: amino acid

(C) STRANDEDNESS: single

(D) TOPOLOGY: linear

(ii) MOLECULE TYPE: peptide

(xi) SEQUENCE DESCRIPTION: SEQ ID NO:l:

sp Ala Trp Glu Glu Glu Thr Glu Gly Ser Ala His lie Val 1 5 10

Claims (29)

1. An adenovirus vector construct comprising a recombinant insert including an expression region, wherein the expression region is under the control of a promoter and includes a coding region that encodes at least one glucose transport protein, glucose phosphorylating protein or a fragment of a glucose transport protein or glucose phosphorylating protein.

2. The vector construct of claim 1, wherein the expression region of the recombinant insert is positioned under the control of a β cell-preferential promoter.

3. The vector construct of claim 2 , wherein the β cell- preferential promoter comprises an insulin, GLUT-2 or glucokinase promoter.

4. The vector construct of claim 1, wherein the coding region is in the antisense orientation relative to the promoter.

5. The vector construct of claim 1, wherein El region of the adenovirus vector has been removed, and the recombinant insert is introduced in its place.

6. The vector construct of claim 1, wherein the recombinant insert includes an SV40 or protamine gene polyadenylation signal.

7. The vector construct of claim 1, wherein the recombinant insert is flanked by adenovirus type 5 sequences.

8. The vector construct of claim 1, wherein the recombinant insert includes an expression region comprising at least one cDNA insert.

9. The vector construct of claim 1, wherein the recombinant insert comprises an expression region encoding a GLUT-2 glucose transporter or a glucokinase enzyme.

10. The vector construct of claim 9, wherein the expression region encodes a GLUT-2 glucose transporter.

11. The vector construct of claim 10, wherein the GLUT-2 glucose transporter is an islet cell GLUT-2 glucose transporter.

12. The vector construct of claim 9, wherein the expression region encodes a glucokinase enzyme.

13. The vector construct of claim 12, wherein the glucokinase enzyme is an islet cell glucokinase enzyme.

14. The vector construct of claim 12, wherein the glucokinase enzyme is a mutant glucokinase enzyme associated with Maturity-Onset Diabetes of the Young (MODY) .

15. The vector construct of claim 9, wherein the expression region encodes a GLUT-2 glucose transporter and a glucokinase enzyme.

16. The vector construct of claim 9 , wherein the expression region encodes a GLUT-2 glucose transporter protein and a glucokinase enzyme or portion thereof in the antisense orientation relative to the promoter.

17. A recombinant host cell incorporating a vector construct in accordance with claim 1.

18. The recombinant host cell of claim 17, further defined as a eukaryotic host cell.

19. The recombinant host cell of claim 18, further defined as a cell which is capable of secreting insulin in response to glucose.

20. The recombinant host cell of claim 19, further defined as a mammalian pancreatic islet β cell.

21. The recombinant host cell of claim 20, further defined as a pancreatic islet β cell derived from a diabetic animal or cell line.

22. An adenoviral virion containing a vector construct as in claim 1.

23. A composition comprising a vector construct in accordance with claim 1, dispersed in a pharmacologically acceptable buffer.

24. A method of providing glucose-responsive insulin secreting capability to a cell, the method comprising the steps of:

(a) obtaining an insulin producing cell; and

(b) expressing a GLUT-2 glucose transporter or a glucokinase enzyme or both in said cell.

25. The method of claim 24 further comprising inhibiting the hexokinase activity in said cell.

26. The method of claim 24, wherein said cell is an islet β cell.

27. The method of claim 26, wherein said islet β cell is located within a mammal with a non-insulin dependent diabetes-like (NIDDM-like) syndrome and the vector construct is administered to the mammal in a pharmacologically acceptable form.

28. The method of claim 24, wherein the GLUT-2 glucose transporter or the glucokinase enzyme or both are expressed from a vector in accordance with claim 1.

29. The method of claim 25, wherein the hexokinase activity is inhibited by the expression, in the cell, of a vector encoding a hexokinase or fragment thereof in an antisense orientation relative to the promoter.