Glucose Homeostasis in the Teleost Fish Tilapia: Insights from Brockmann Body Xenotransplantation Studies1

Abstract

Certain teleost fish have large anatomically discrete islet organs called Brockmann bodies (BBs). When transplanted into streptozotocin diabetic athymic nude mice, tilapia BBs provide long-term normoglycemia. This has afforded us the opportunity to examine tilapia islet in vivo function in a non-piscine environment and compare this with in vivo function in the donor species. As expected, fasting and non-fasting glycemic levels in long-term murine recipients of tilapia BBs were analogous to corresponding values in donor tilapia, but, surprisingly, tilapia BB grafts provided mammalian-like glucose tolerance profiles. Teleost fish, in general, are severely glucose intolerant. When glucose tolerance tests were performed in tilapia, the mean glucose disappearance rates were very low; however, diabetic nude mice bearing long-term tilapia BB grafts were extremely glucose responsive. This suggested a severe or absolute peripheral resistance to the glucostatic effects of insulin. Using Western blotting with polyclonal antibodies and then confirmed by Northern analysis, tilapia peripheral tissues appear to be devoid of GLUT-4, the insulin-sensitive glucose transporter responsible for the hypoglycemic effect of insulin in mammals, but not GLUT-1, the insulin independent glucose transporter. This may explain why tilapia, and possibly other teleost fish, are severely glucose intolerant after pharmacologic glucose-loading. Because tilapia do not tend to consume large quantities of glucose in the wild, it is not surprising that they have evolved without a mechanism to move glucose rapidly from the bloodstream into muscle and fat. Nevertheless, insulin still appears to play an important role in maintaining normoglycemia in tilapia; however, this is mostly likely a result of its effect on glucose uptake in the liver. We also present comparative data on tilapia beta cell function, quantification of islet cell numbers and types, islet products, insulin gene structure and expression, and beta cell sensitivity to the diabetogenic drug streptozotocin.

One possible treatment for insulin dependent diabetes mellitus in humans would be to transplant pancreatic islets. However, the number of potential islet graft recipients vastly exceeds the number of human cadaveric pancreatic donors. Therefore, widespread implementation would require the development of clinical islet xenotransplantation (i.e., transplantation across species) programs (Lacy, 1995; Smith and Mandel, 1998). Seven years ago, our laboratory developed a novel islet xenotransplantation model utilizing Oreochromis niloticus (tilapia), a teleost fish, as islet donors (Wright, 1992; Wright and Yang, 1997). Many teleost fish, including tilapia, have large anatomically discrete pancreatic islets called Brockmann bodies (BBs) which, unlike mammalian islets, can be easily harvested without expensive “islet isolation” procedures (Wright, 1994; Wright and Schrezenmeir, 1995; Yang and Wright, 1995). We and others have demonstrated that tilapia islets transplanted under the renal capsules of streptozotocin (STZ)-diabetic athymic nude mice will induce long-term normoglycemia (Wright et al., 1992; Morsiani et al., 1995). Using euthymic rodents as recipients, we have used this model to examine the mechanisms of xenograft rejection (Wright et al., 1994b; Wright et al., 1997) as well as methods of preventing islet xenograft rejection between discordant species. We have explored, and continue to explore, various methods of immunosuppression (Wright et al., 1994a; Wright and Kearns, 1995b), graft immunomodulation (Wright and Kearns, 1995a; Yang et al., 1995; O'Hali et al., 1997), and transplantation into immunoprivileged sites (Coddington et al., 1997). Our most promising results, to date, have been achieved with islet encapsulation (i.e., enclosing the grafts within semi-permeable membranes with very small “pores” allowing diffusion of glucose, insulin, and oxygen, but preventing entry of larger molecules such as immunoglobulins and cells of the immune system) (Lanza and Chick, 1994; Lanza and Chick, 1997; Kuhtreiber et al., 1999). Using these modalities, we have now achieved uniform long-term function of encapsulated fish islet grafts in euthymic diabetic rodents (Yang et al., 1997b; Yang and Wright, 1999a; Yang and Wright, 1999b).

Surprisingly, tilapia islets transplanted into diabetic athymic nude mice not only provide long-term normoglycemia, but also mammalian-like glucose tolerance profiles (Wright et al., 1992; Yang et al., 1997a). Therefore, we are also exploring the possibility that fish islets might have a future in the treatment of clinical type I diabetes. Unfortunately, the structure of tilapia insulin differs significantly from that of human insulin and would be only partially active, and probably immunogenic, in man (Nguyen et al., 1995). As a result, we are currently attempting to produce transgenic tilapia with “humanized” tilapia insulin genes as potential donors for clinical islet xenotransplantion (Wright and Pohajdak, 1995; MacKenzie, 1996). We have cloned and sequenced the tilapia insulin gene (Mansour et al., 1998) and modified it by site-directed mutagenesis so that it codes for human insulin (Wright and Pohajdak, 1995). We are now attempting to reintroduce the hybrid gene through homologous recombination using two approaches, gene microinjection of fertilized tilapia eggs at the single cell stage and embryonic stem cell technology. If successful, offspring with appropriate gene incorporation will be selectively bred to produce homozygosity. Because of the very short generation times of tilapia (fertilized egg to sexual maturity in 5–6 months), it should be possible to produce homozygous strains within two years of achieving homologous recombination. We expect that islets harvested from these fish will physiologically secrete humanized insulin, but this will be tested both in vitro and in vivo.

Because we are attempting to produce and characterize humanized tilapia islets for clinical transplantation, we have now begun studying normal tilapia islet structure and function. In the past couple of years, we have made a number of original observations providing some important insights into piscine glucose homeostasis. The remainder of this review will summarize what we have learned about tilapia islet morphology, glucose homeostasis, islet peptide structures, insulin gene, and islet ontogeny.

TILAPIA ISLET STRUCTURE

Tilapia have roughly 10–15 BBs scattered throughout the adipose tissue surrounding the common bile duct; this lies within a roughly triangular region outlined superiorly by the stomach, anteriorly by the liver, and inferiorly by the spleen and gallbladder (Wright, 1994). Each BB is composed of relatively pure islet tissue surrounded by a very thin rim of exocrine pancreas. The remainder of the exocrine pancreatic tissue in tilapia is located within the liver as a hepatopancreas. BBs vary in size depending upon the size of the fish, but most measure between 0.5 and 5.0 mm in diameter. There is a direct, relatively linear relationship between donor body weight and total islet endocrine cell number; therefore, donor body weight can be used to measure islet graft volume (Dickson et al., 1998). Tilapia islet morphology has been extensively studied by us (Yang et al., 1999). Based on immunoperoxidase staining of dispersed islet cell preparations cytocentrifuged onto glass slides and then quantification by computerized image analysis, tilapia islets are composed of roughly 42.3% insulin positive (beta) cells, 11.5% glucagon positive (alpha) cells, 23.1% somatostatin-28 positive (delta) cells, 21.8% somatostatin-14 positive (delta) cells, and 1.3% peptide tyrosine-tyrosine (PYY) positive cells. Islet endocrine cell topography was evaluated by staining histologic sections of whole endocrine pancreas including large, medium, and small BBs; this is described in detail elsewhere (Yang et al., 1999). In general, alpha, beta, and delta cells are scattered, often in clusters, throughout the BBs regardless of size; PYY cells are peripherally located. Some ultrastructural features have been previously reported as well (Wright, 1994).

GLUCOSE HOMEOSTASIS IN TILAPIA

Studying in vivo islet physiology and glucose homeostasis in teleost fish is problematic because little is known about peripheral tissue responses to insulin. For instance, many decades of evidence have shown that teleost fish, in general, are severely glucose intolerant (Palmer and Ryman, 1972; Ince, 1983). Initially, it was thought that this was due to glucose being a poor insulin secretogogue and that insulin was produced primarily in response to amino acid secretogogues. This seemed very plausible as most of the studies examining glucose metabolism in fish were performed in carnivores. However, Plisetskaya was able to show, using species-specific radioimmunoassays, a brisk insulin response to glucose even in carnivorous teleosts (Plisetskaya et al., 1976; Hilton et al., 1987) suggesting that the state of glucose intolerance was due to a blunted peripheral response to the glucostatic effects of insulin such as seen in human type II diabetes mellitus rather than inadequate insulin secretion (Mommsen and Plisetskaya, 1991).

What is normoglycemia for tilapia?

In man, blood glucose levels are tightly controlled. There is considerable evidence that tight control is not the norm in fish (Mommsen and Plisetskaya, 1991; Wright, 1992). To address this issue in tilapia, we determined terminal non-fasting plasma glucose levels for all tilapia, male and female, killed for experimental studies in our laboratory over a one year long period. Mean non-fasting plasma glucose levels (±SEM) in tilapia, maintained in the constant internal environment of the Dalhousie University Aquatron aquatic animal facility, were 91.9 ± 3.3 mg/dl (n = 202), and there was no apparent seasonal variation (Yang et al., 1997a). Mean fasting (>48 hr) plasma glucose levels (±SEM) in tilapia were 75.4 ± 3.0 mg/dl (n = 140) (Wright, 1992). These values compare favorably with mean non-fasting and fasting plasma glucose values in man (i.e., 90 mg/dl and 63 mg/dl, respectively) (Aoki, 1985).

How is normoglycemia maintained in tilapia?

This is a very difficult question to address. Unlike mammalian systems where much is known about the mechanism of glucose homeostasis, to some extent studying the in vivo response to insulin in fish is analogous to describing the contents of a black box. In some of the in vivo studies described below, we have attempted to gain new insights into tilapia glucose homeostasis by harvesting their islets from the black box and transplanting them into a more transparent box. Because pancreatic islets have species-specific set points for normoglycemia, islet xenografts regulate glycemia in recipients according to the norms for the donor rather than the recipient species (Sullivan et al., 1987; Carroll et al., 1992; Yang et al., 1997a). Therefore, STZ-diabetic athymic nude mouse recipients bearing long-term tilapia islet grafts can be used to gain insights into the in vivo function of tilapia islets in a non-piscine environment. The significance of this is that peripheral tissue responses to the glucostatic effect of insulin are better understood in mice than in fish. In this series of studies, we have compared these results with the results of similar in vivo function studies in the intact islet donor species.

In a recent study, we compared the function of CD-1 mouse, Lewis rat, and tilapia islet grafts transplanted under the renal capsules of STZ-diabetic nude mice. Equal volumes of islet tissue were transplanted for each of the three species and non-fasting blood glucose levels were followed for 30 days. Over this period, mean non-fasting blood glucose levels (±SEM) in recipients of tilapia, rat, and mouse islets were 78.8 (±4.4), 77.0 (±1.3), and 115 (±4.5) mg/dl, respectively. After day 30, the mean fasting blood glucose levels in recipients of tilapia, rat, and mouse islets were 72.7, 89.8, and 113.3 mg/dl, respectively. Analysis of variance followed by paired comparisons demonstrated that, for both mean fasting and nonfasting blood glucose levels, recipients of murine islets were significantly higher than in recipients of either rat or piscine islets and that there was no significant difference between rat and fish islet recipients (Yang et al., 1997a). Interestingly, fasting and non-fasting glycemic levels in long-term murine recipients of tilapia BBs were roughly analogous to corresponding values in donor tilapia (see above); the same was also true of Lewis rat donors and recipients of Lewis rat islets as well as of CD-1 mouse donors and recipients of CD-1 mouse islets (Yang et al., 1997a).

We then performed intraperitoneal glucose tolerance tests on recipient mice. Glucose disappearance rates (K values) were calculated using the equation K = 70/t ½, where t½ is the number of minutes for the blood glucose level to drop to 50% of its level at 10 minutes (Marble and Ferguson, 1985). All three groups of recipient mice had similar glucose tolerance profiles with mean glucose disappearance rates (K values) of 4.3 (mouse islets), 5.0 (rat islets), and 5.7 (tilapia islets). It was very surprising that tilapia islets appear to be at least as glucose-responsive as either of the rodent islets and, on a cell for cell basis, appear to produce an insulin response that is in the same order of magnitude as rodent islets. Because it is well-known that rodent islets produce much more insulin per cell than human or porcine islets (Dionne, 1994), it seems likely that tilapia islet cells actually are more efficient insulin producers in response to glucose than human or porcine islets. This is consistent with the observation that it requires several fold more human or porcine islets than rodent or tilapia islets to achieve stable, long-term normoglycemia in STZ-diabetic nude mouse recipients (Yang et al., 1997a).

Are tilapia as glucose intolerant as other teleosts and, if so, how can that be?

It is well-documented that teleost fish, in general, are severely glucose intolerant; however, the vast majority of these studies were performed on carnivorous, cold-water species. We speculated that tilapia, because they are an omnivorous tropical species, might not be glucose intolerant. However, when glucose tolerance tests were performed in intact tilapia, the mean glucose disappearance rates (K-values) were roughly 0.05, while diabetic nude mice bearing long-term tilapia islet grafts were extremely glucose responsive with K-values of 6.0 (Wright et al., 1998a). As shown in Figure 1, tilapia mean glycemia levels plotted over time formed a relatively straight line with such a minimal slope that it took more than three days to return to baseline levels after glucose loading; in fact, the rate of glucose clearance from the blood was so slow that it could probably be accounted for entirely on the basis of either renal or hepatic clearance or possibly by passive diffusion into peripheral tissues. Clearly, our initial hypothesis was wrong and the data suggested to us a severe or absolute peripheral resistance to the glucostatic effects of insulin in tilapia. In mammals, glucose uptake in peripheral tissues such as skeletal muscle, heart and adipose tissue is facilitated by GLUT-1 and GLUT-4 glucose transporters which are co-expressed in these tissues. In skeletal muscle, GLUT-1 is insulin-independent and facilitates basal glucose transport while GLUT-4 is translocated to the cell surface by insulin and it is this insulin-sensitive transporter which is responsible for the hypoglycemic effect of insulin in mammals. Because nothing was known about the existence of glucose transporters in fish tissues, we examined a number of tilapia tissues for the presence or absence of GLUT-1- and GLUT-4-like transporters. Using Western blotting with polyclonal antibodies and then confirmed by Northern analysis, tilapia peripheral tissues appear to be devoid of GLUT-4, the insulin-sensitive glucose transporter, but not GLUT-1, the insulin independent glucose transporter (Wright et al., 1998a). Because glucose would be, at best, a minimal component of the natural diet of most teleost species, it would be logical that teleosts could have evolved without GLUT-4 transporters in their peripheral tissues. This could explain why tilapia, and possibly other teleost fish, are severely glucose intolerant after glucose loading yet their islets are glucose-sensitive, and, thus, explain the paradox that has plagued our fish islet transplantation research for years. Apparently, tilapia islets have evolved along mammalian lines to be glucose sensitive while tilapia peripheral tissues have diverged widely (Wright et al., 1998a).

Does glucose play a significant role in piscine metabolism?

Although we were able to demonstrate the presence of GLUT-1 transporters in tilapia tissues, these insulin-insensitive glucose transporters that are responsible for basal glucose uptake in most mammalian tissues were found to have a very limited tissue distribution in tilapia (Wright et al., 1998a). The very limited tissue distribution of GLUT-1, and the total absence of GLUT-4, suggests that the metabolic role of glucose in most tilapia tissues may be much less significant than in mammalian tissues. However, heart and brain appear to be exceptions. GLUT-1 levels in tilapia heart were about ten-fold higher than in rodent heart while glycogen levels were 17-fold greater; these findings suggest that the working tilapia heart may be very efficient at using glucose (Wright et al., 1998a). If our results in tilapia can be generalized to other piscine species, it could explain an observation by West et al. (1993) who reported that glucose utilization by rainbow trout cardiac muscle is about ten-fold greater than in skeletal muscle.

Why do fish tolerate hypoglycemia so much better than mammals?

The most important role of insulin in mammalians is glucose homeostasis which is required to maintain the functional integrity of the brain. As mentioned above, fish do not appear to require as tight control of blood glucose levels as mammalians require. It is well documented that teleost fish can tolerate, without any apparent neurological consequences, exceedingly low glucose levels that would quickly cause lethal hypoglycemic brain damage in mammalians (Wright, 1992). It has been suggested that high glycogen levels in teleost brain may be responsible for this ability (Mommsen and Plisetskaya, 1991). Interestingly, brain was the only other tilapia tissue we tested that had GLUT-1 and brain glycogen levels in tilapia were five-fold greater than in rodent brain (Wright et al., 1998a). These findings seem to support this theory and may also provide insights into a mechanism for maintaining these high glycogen levels.

What is insulin's role in maintaining normoglycemia in fish?

We have shown above that tilapia, like mammals, maintain blood glucose levels in a relatively narrow zone and that their islets are glucose responsive. We have also shown that tilapia peripheral tissues such as skeletal muscle and fat, the primary targets of insulin in mammals, do not appear to possess GLUT-4-like glucose transporters, which are responsible for rapid insulin-induced glucose uptake in mammalian peripheral tissues. The paradox is that tilapia islets produce insulin in a very glucose sensitive manner but simultaneously appear to be peripherally insensitive to insulin. The obvious question is what does the insulin do? As we do not have data to directly address this question, we can only speculate.

It is well documented that tilapia peripheral tissues (i.e., skeletal muscle) possess functional insulin receptors (Parrizas et al., 1994) and that a very important role for insulin in teleosts is promotion of somatic growth (Mommsen and Plisetskaya, 1991; King and Kahn, 1981). We speculate that this may be its major role in teleost peripheral tissues. If so, this creates another apparent paradox, as it is well-known that a chronic high glucose diet, which would be expected to elevate plasma insulin levels, does not promote growth in tilapia (Lin and Shiau, 1995; Shiau and Chuang, 1995). However, growth promotion does not occur because chronic hyperglycemia causes hepatomegaly and other health problems which would negate any growth promoting effects of hyperinsulinemia. In general, teleost diets are low in simple carbohydrates, and teleosts do not appear to use glucose efficiently as a general source of energy (Wilson, 1994). This contention is supported by the very limited tissue distribution of GLUT-1 glucose transporters, the absence of GLUT-4 glucose transporters, and the adverse health effects precipitated by high glucose diets. Furthermore, because fish do not tend to receive large oral glucose loads, it is not surprising that they have evolved without a mechanism to move glucose rapidly from the bloodstream into muscle and fat.

Nevertheless, it appears that insulin still plays an important role in glucose homeostasis in tilapia. This contention is strongly supported by our observation that tilapia islets, regardless of whether they are in intact tilapia or transplanted into diabetic nude mice, regulate blood sugar levels at a species-specific set point. Because of feedback loops, islet beta cells turn insulin secretion on and off in response to changes in blood glucose much like a thermostat regulates room temperature. Since, our data suggests that it is unlikely that insulin plays a large role in the maintenance of normoglycemia by its actions on peripheral tissues, it seems most likely that insulin acts at the liver, probably by stimulating glucose uptake.

OTHER ASPECTS OF TILAPIA ISLET METABOLISM

Are tilapia islets sensitive to STZ?

Transplantation of tilapia islets into STZ-diabetic nude mice has also provided us some new insights into the effects of STZ on tilapia islets, and, hence, glucose uptake into tilapia beta cells. STZ, a methylnitrosourea with a 2-substituted glucose, causes beta cell necrosis and insulin-dependent diabetes in a wide variety of vertebrate species. The mechanism of action of STZ in rodent beta cells is well-characterized. STZ, an alkylating agent, is taken up by beta cells via the GLUT-2 transporter (Schnedl et al., 1994) and causes DNA strand breaks which activate poly (ADP-ribose) synthase, an enzyme present in eukaryotic cells which polymerizes the ADP-ribose moiety of NAD to form poly (ADP-ribose). This reaction, which uses NAD as its substrate, causes lethal NAD depletion in beta cells. Beta cell necrosis and severe, permanent hyperglycemia ensue over the next few days (Okamoto, 1985).

Several prior studies using primarily carnivorous species suggested that teleosts are relatively resistant to STZ. Because we know that tilapia islets are highly glucose-responsive, we hypothesized that tilapia islets might be highly susceptible to the diabetogenic effects of STZ. Initially, we performed a dose-response study using intact tilapia. Fasted tilapia received 100, 150, 200, 250, 300, or 350 mg/kg STZ intravenously (i.v.). Plasma glucose levels were followed for 72 hr and then the fish were killed and necropsied. Severe diabetes was present in 20%, 80%, and 100% of fish receiving 250, 300, and 350 mg/kg, respectively; no diabetes was seen in tilapia receiving doses under 250 mg/kg. However, histologic sections of tilapia islets, stained with hematoxylin & eosin or by immunoperoxidase for tilapia insulin, showed only partial degranulation, but no evidence of beta cell necrosis, as is characteristic of STZ toxicity in mammalian species. Furthermore, there was histological evidence of hepatic and renal toxicity. Another group of fish was treated with 350 mg/kg i.v. and was followed longer to determine whether beta cell necrosis eventually ensued and whether diabetes was transient or permanent. All fish died or had to be killed within 9 days because of severe hepatic failure characterized by hepatic necrosis, jaundice, and ascites. Terminal blood glucose levels were within normal limits and islet histology was relatively normal (Wright et al., 1999). Simultaneously, we also performed a study in which nude mouse recipients of long-term tilapia islet grafts were treated with STZ at doses up to 400 mg/kg i.v.; not only did STZ not cause graft beta cell necrosis, diabetes, or death in recipient mice, it did not even have an adverse effect on glucose tolerance profiles of the mice when compared to the results of glucose tolerance tests performed on each tilapia islet recipient prior to STZ treatment (unpublished data, Yang, O'Hali, and Wright). The magnitude of STZ resistance of tilapia islets can be better appreciated if one considers the following: (1) that STZ causes global beta cell necrosis in the native pancreata of many mammalian species at doses well under 100 mg/kg i.v. (Wright and Lacy, 1988), (2) that STZ causes global beta cell necrosis in the native pancreata of mice at a doses of 150–225 gm/kg i.v. (Wright et al., 1988), (3) that doses >250 mg/kg i.v. are uniformly lethal in most strains of mice (including nude mice without tilapia islet grafts), and (4) that a dose of 350 mg/kg is uniformly lethal in intact tilapia. Clearly, mice with tilapia islets are more resistant to the diabetogenic and/or lethal effects of STZ than either the donor or the recipient species; we believe that the resistance of tilapia islets to the diabetogenic effects of STZ exceeds that of any species studied to date. Clearly, our initial hypothesis was incorrect, again.

The extreme degree of resistance of tilapia islets to the diabetogenic effects of STZ could be explained: (1) if tilapia islets were deficient in GLUT-2, (2) if DNA in tilapia beta cells was resistant to damage from alkylating agents, or (3) if STZ does not readily cause depletion of NAD in tilapia beta cells, possibly due to greater NAD stores than in rodent islets. Clearly, tilapia islets would be a very useful model to study the mechanisms of action of the beta cell toxin STZ. Alternatively, STZ can possibly be used to gain important insights into differences in glucose uptake and metabolism by mammalian and piscine beta cells (Wright et al., 1999). We are currently attempting to gain further insights into the mechanism of glucose uptake into tilapia beta cells.

Oxygen requirements of tilapia islets

One of the major problems with islet encapsulation devices used to prevent islet xenograft rejection is that the grafts are not directly vascularized and thus islet attrition occurs over time due to chronic hypoxia. We speculated that, because tilapia are adapted to live in hypoxic environments (i.e., hot, stagnant water), tilapia islets might tolerate much lower oxygen tensions than mammalian islets, thus making them more suitable for encapsulation. Therefore, we directly compared the viability of tilapia and mammalian islets by co-culturing them under severely hypoxic conditions (i.e.,similar to those encountered in the centers of encapsulated islets); tilapia islets survived (and functioned) many-fold longer than mammalian islets under these conditions (Wright et al., 1998b). To date, no one has directly measured oxygen consumption in tilapia islet cells, but the above study suggests that their oxygen requirements are low relative to those of mammalian islets.

TILAPIA INSULIN AND OTHER ISLET PEPTIDES

We have also gained considerable insight into the amino acid sequences of peptides produced by tilapia islets as well as the structure of the tilapia insulin gene. The former studies were performed because knowledge of the structure of the other hormones produced by tilapia BBs would provide important insights into their potential cross-reactivities with human peptides, and hence, their potential function in man after xenotransplantation of tilapia islets. This is crucial knowledge as islet physiology is complicated with multiple feedback loops between the various islet cell types. The latter studies characterizing the tilapia insulin gene provided the foundation which we used to commence production of transgenic tilapia bearing a “humanized” tilapia insulin gene (see above).

Tilapia islet peptides

All known insulin genes code for one continuous protein (B-C-A) that is later processed into three fragments; the mature insulin protein (B-A) is linked by disulfide bonds and the C-peptide is removed during processing. In comparison to the human insulin, the tilapia A chain has 13/21 identical amino acids (61.9%) and five of the substituted changes are conservative. The B chain has 22/29 identical amino acids (75.8%) and three of the changes are conservative (Nguyen et al., 1995). Although the C chain, which may be catabolized during processing, was not isolated, its structure, deduced from the cloned tilapia insulin gene, is markedly different from the human C chain but is more similar to the other fish C chains (Mansour et al., 1998). Interestingly, while the B chain length in fish and humans is identical (30 amino acids), the alignment shows that the tilapia B chain has an extra amino acid (valine) at its N-terminus while it lacks the C-terminal threonine that is found in human insulin. The tilapia insulin protein also shows a uniquely substituted fifth amino acid in its A chain. In this position, tilapia has glutamic acid while almost all other organisms have glutamine (Nguyen et al., 1995). A single base pair mutation (C to G transversion) could have accounted for this change during evolution. Until very recently, this amino acid position was thought to be critically involved in binding to the insulin receptor and, thus, substitution of a charged residue for an neutral one might have been expected to drastically alter the potency of tilapia insulin in mammalian systems. Other residues formerly considered to be important for insulin receptor binding, such as A1, A4, A19, A21, and B23–B27, as well as residues considered important for maintaining tertiary structure are conserved in tilapia (Nguyen et al., 1995). However, this traditional view was based primarily on x-ray crystallographic data, and it now is believed that crystalized insulin is an inactive conformation. Very recently, Kristensen et al. (1997) have suggested an alternative model based on using alanine scanning mutagenesis to identify specific side chains of insulin which strongly influence binding to the insulin receptor. These data suggest that the insulin receptor binding domain consists of five residues (IleA2, ValA3, TyrA19, GlyB23, and PheB24) and exists as a patch on the surface of the insulin molecule; interestingly, these five residues are invariant in all species yet studied (Conlon, 2000). Additional analysis of other insulin analog constructs suggests that the positions LeuB6, GlyB8, LeuB11, GluB13, and TyrB26, although not part of the binding epitope, are important in maintaining the receptor-binding conformation of the insulin (Kristensen et al., 1997); all of these residues, except for GluB13, are maintained in tilapia insulin. Interestingly, LeuB6 and LeuB11 are invariant in all species yet tested and only the conservative substitution of aspartic acid for glutamic acid, present in tilapia insulin, is seen at B13 (Conlon, 2000). In general, teleost insulins exhibit roughly 30–50% of the activity of human insulin in, for example, stimulating lipogenesis in dispersed rat adipocytes (Nguyen et al., 1995).

As is true of other teleosts, tilapia have two non-allelic glucagon genes, proglucagon I and proglucagon II. Alternative pathways of post-translational processing result in the synthesis of two glucagons, one with 29 and another with 36 amino acids, derived from proglucagon I and two glucagons, one with 29 and another with 32 amino acids, derived from proglucagon II. In both instances, the longer peptide was a C-terminally extended version of the shorter. Although C-terminally extended forms of glucagon have been previously isolated from islets of more primitive fish, this appears to be unique among teleost BBs. Two glucagon-like peptides (GLPs) with 30 and 34 amino acid residues were also isolated and are derived from proglucagon II (Nguyen et al., 1995).

Tilapia somatostatin-14 is identical to all known mammalian somatostatins and almost all known piscine somatostatin-14s; however, tilapia somatostatin-28, derived from the prosomatostatin II gene, does not correspond closely with the large somatostatins of other teleost species. The prosomatostatin II gene is either not present or not expressed in mammals and the potential biological activity of tilapia somatostatin-28 in mammalian systems is currently unknown (Nguyen et al., 1995).

As in all known fish islets, the tilapia islet cells analogous to the pancreatic polypeptide (PP) cells in mammalian islets (Yang et al., 1999) produce a peptide that resembles mammalian PYY and neuropeptide Y more closely than mammalian PP. The biological activity of tilapia PYY is currently unknown (Nguyen et al., 1995).

Tilapia insulin gene

To clone the tilapia insulin gene, we began by searching Genbank for all known insulin proteins isolated from various animals. A computer program (Clustal V) was then used to align these sequences. Based on this alignment we designed DNA primers that were degenerate (multiple codons) and used these in the polymerase chain reaction (PCR) technique to isolate the tilapia insulin gene, a method we have used to isolate other tilapia genes (Dixon et al., 1993). We completely sequenced the DNA and confirmed that it indeed coded for a protein similar to other fish insulins. Independently and roughly simultaneously, one of us (JMC) had extracted insulin from tilapia BBs and had sequenced the protein (Nguyen et al., 1995). Both protein and DNA sequencing gave the identical sequence indicating that we indeed had the tilapia insulin gene (Mansour et al., 1998). To obtain upstream and downstream tilapia insulin DNA (non-coding), we used another PCR based strategy developed commercially by TaKaRa. Using this method we isolated approximately 1.5 kilobases of DNA at both ends. The general structure of the tilapia insulin gene (three exons and two introns) is identical to all previously reported insulin genes. However, the first intron (located in the upstream non-coding region) is the smallest isolated (73 bp) to date. Interestingly, the upstream region also contains a microsatellite. These regions are composed of repetitive DNA (usually a dinucleotide) that often shows size polymorphisms (number of repeats). Our tilapia insulin gene's microsatellite has the dinucleotide CA repeated 17 times. How polymorphic this site is in other tilapia is presently unknown. Remarkably, humans also have a repetitive element, a microsatellite, in a nearly identical position upstream and the length of the repeat has been linked to the susceptability to insulin-dependent diabetes in man (Mansour et al., 1998).

Recently, we have examined the expression and development of insulin production in tilapia embryos. Interestingly, the insulin gene is expressed very early, during the period of embryonic segmentation, which is several days before hatching (Tait et al., in preparation). In fact, in a different study characterizing the embryological and larval development of tilapia by morphology (Morrison et al., 2000), an endocrine pancreatic islet with insulin immunopositive cells was identified histologically by 76 hours after fertilization (i.e., a full day before hatching).

In mammals, insulin production is affected by glucose levels and is regulated at the level of transcription. The turning on or off of transcription depends on multiple proteins (transcription factors) that are activated and bind to regions of the DNA, usually within 100 to 300 bp upstream from the coding region, which contain promoter/enhancer regions. Without these factors binding, transcription cannot start. To determine whether the tilapia insulin gene contains similar regions or elements suggesting a similar control mechanism as in mammals, we analysed this region and have detected several transcriptional DNA elements that are similar to those found in mammalian insulin genes. Most striking is the location of an insulin control element (ICE) in the tilapia promoter region at an identical position as that found in the rat insulin gene. All mammalian insulin genes contain this ICE element. To confirm that this region contains all the necessary regulatory elements we have fused this DNA to a reporter gene. We are presently transfecting this chimeric or hybrid gene into mammalian beta cells to determine if the tilapia insulin promoter shows mammalian glucose responsiveness. We will also transfect this gene into other non-islet cell lines to determine if this promoter can function in different tissues or is this tilapia insulin promoter beta cell specific?

CONCLUSIONS

In summary, we have accumulated considerable and diverse information on the structure and function of tilapia islets and their peptides, glucose homeostasis, and peripheral tissue responses to insulin in tilapia. All of these insights resulted, directly or indirectly, from our tilapia islet xenograft studies. As a result, tilapia are now one of the better characterized teleost species. The purpose of this review is to summarize this information in one location that is readily accessible to the experimental zoologist.

1

From the symposium A Tribute to Erika M. Plisetskaya: New Insights on the Function and Evolution of Gastroenteropancreatic Hormones presented at the Annual Meeting of the Society for Integrative and Comparative Biology, 6–10 January 1999, at Denver, Colorado.

Fig. 1. Glucose tolerance profiles of tilapia (n = 13), non-diabetic athymic nude mice (n = 10), and STZ-diabetic athymic nude mice bearing tilapia islet grafts (n = 3) after IP glucose loading. Mean plasma glucose disappearance rates (K values) were 0.05, 0.8, and 6.0, respectively. * Mean plasma glucose values (±SD) for tilapia at 24, 48, and 72h were 362 (±196), 233 (±168), and 107 (±33.1) mg/dl (data not shown in fig.). Reproduced with permission from Gen. Comp. Endocrinol. 111:20–27, 1998 (Academic Press).

The authors would like to acknowledge the financial support of the Canadian Diabetes Association (in granted in honour of Marion C. Dill and George Goodwin), the Juvenile Diabetes Foundation International, the Medical Research Council of Canada, the Natural Sciences and Engineering Council of Canada, Sandoz Canada, VivoRx Canada, and the IWK-Grace Health Centre Research Foundation. We would also like to acknowledge the contributions of the following students (Doug Coddington, Brendan Dickson, Allison Lobsinger, Marc Mansour, Heather McLean, T.M. Nguyen, Aaron Tait), post doctoral fellows (Eileen Donovan-Wright, X-X Han, Carol Morrison, Hua Yang), pathology or surgery residents (Cherrie Abraham, Wael O'Hali, Weiming Yu), and technicians (Gabrielle Girourd, Heather Kearns, Danielle Paquet, Colleen Pelly, Sherilyn Polvi) to various of these studies.

We would particularly like to thank a very special consultant, Dr. Erika Plisetskaya, for all of her help and advice throughout many of these studies.