|Home | About | Journals | Submit | Contact Us | Français|
The inositol glycans (IGs) are glycolipid-derived carbohydrates produced by insulin-sensitive cells in response to insulin treatment. IGs exhibit an array of insulin-like activities including stimulation of lipogenesis, glucose transport and glycogen synthesis, suggesting that they may be involved in insulin signal transduction. However, because the natural IGs are structurally heterogeneous and difficult to purify to homogeneity, an understanding of the relationship between structure and biological activity has relied principally on synthetic IGs of defined structure.
This article briefly describes what is known about the role of IGs in signal transduction and reviews the specific biological activities of the structurally defined IGs synthesized and tested to date.
A pharmacophore for IG activity begins to emerge from the reviewed data and the structural elements necessary for activity are summarized.
The inositol glycans (IGs) are a class of naturally occurring, phosphorylated, inositol-containing pseudosaccharides first identified in 1986 from bovine liver treated with insulin . The finding that isolated natural IGs, released in response to insulin, are competent in stimulating insulin-sensitive cells in the absence of insulin led to the hypothesis that IGs are the long-elusive second messengers for insulin action, as implied by earlier experiments . However, it was quickly realized that the IG preparations obtained from natural sources were mixtures of closely related carbohydrates, and this microheterogeneity severely complicated the elucidation of the role of IGs in insulin signal transduction.
Using a combination of metabolic labeling and degradation studies, the basic structural features of the natural IGs were identified. All of the IGs contain a terminal inositol, but both chiro  and myo-inositol  isomers were found. Similarly, the second sugar is invariably an unacylated aminosugar, but both galactosamine  and glucosamine  were identified. The remaining structure of the glycan was less well established, but the presence of a number of mannose residues and one or more phosphate residues was reported. Given this information and the approximate molecular weight of 1400 Da , it became clear that the IGs are very closely related to the glycosylphosphatidylinositol (GPI) membrane anchors that many cells use to tether cell surface proteins to the plasma membrane and some of whose structures have been unequivocally identified. The close structural relationship between IGs and GPIs was confirmed by the ability of phosphatidylinositol-specific phospholipase C (PI-PLC) to release IGs from the cell surface  as well as by the observation that the trypanosome-derived variant surface glycoprotein GPI anchor, cleaved from the lipid with PI-PLC and from the protein with pronase, has significant insulin-like activity in rat hepatocytes and adipocytes [4,5].
The similarity between the partially known structures of the IGs and the known structures of the GPIs suggested a way of unraveling the relationship between IG structure and biological activity. The established GPI structures could be used as a template for the design and chemical synthesis of structurally defined IG analogues that could then be subjected to biological evaluation. The chemical synthesis of a bioactive IG analogue was first accomplished in 1992, when a simple pseudodisaccharide – comprising the glucosamine and myo-inositol terminus of the GPI anchor structure, but terminated by a cyclic phosphate as would be produced from PI-PLC hydrolysis of a glycolipid – was prepared and found to be able to mimic one of the actions of insulin: stimulation of lipogenesis in rat adipocytes . This led to a burst of synthetic activity and over the next 16 years, various research groups synthesized a large number of IG analogues and evaluated them for biological activity. This has provided an extensive collection of data from which a pharmacophore for insulin-like activity is emerging.
This article reviews the published information on the relationship between IG structure and biological activity. The review covers only the compounds for which a fully defined structure and biological activity have been reported. Therefore, neither synthetic IGs that have not been tested in any bioassays nor natural IGs whose structures are not completely elucidated are included here. Earlier publications have reviewed the role of the natural IGs in signal transduction [7-13] and the chemical synthesis of the IGs , but this is the first comprehensive survey of the biological activity of structurally defined IGs.
Type 2 diabetes mellitus is characterized by insulin resistance in the target tissues, effectively a failure in insulin signal transduction. Since the IGs are able to stimulate many insulin-like effects, but do not act via the insulin receptor (IR) they (or appropriate analogues) may be useful for the treatment of Type 2 diabetes. Any attempt to understand the role of IGs in producing insulin-like effects first requires a basic understanding of the molecular machinery of insulin signaling. A lot of information on the subject of insulin signal transduction has been published and several aspects have been reviewed elsewhere [15-18]. Insulin has two basic effects, mitogenesis and stimulation of anabolic metabolism. In this section we provide a brief and somewhat simplified overview of what is currently known about the stimulation of anabolism, including glucose transport, glycogen synthesis and lipogenesis. Each step in the process, described below, corresponds to a number in Figure 1:
(1) Insulin (Ins) delivers its signal by binding to the IR, a cell-surface receptor tyrosine kinase. A complex consisting of the IR and insulin is formed, leading to conformational changes of the IR. (2) The two intracellular domains of the IR, possessing kinase activity, phosphory-late each other. This leads to increased kinase activity as well as affinity toward other IR substrates. (3) Adaptor proteins (IRS-1, IRS-2) bind to the phosphorylated receptor and then (4) their tyrosine residues are phosphorylated by the IR kinase activity. IRS proteins are localized to the plasma membrane during this event, perhaps through the interaction with phosphatidylinositol-4,5-diphosphate . (5) The p85 regulatory domain of phosphatidylinositol 3 kinase (PI3K) docks with the phosphorylated sites of IRS1/2. (6) This leads to phosphorylation of the lipid phosphatidylinositol-4,5-diphosphate and generation of phosphatidylinositol-3,4,5-triphosphate (PIP3). An elevated concentration of PIP3 leads to a cascade of protein phosphorylations. (7) First, binding of phosphoinositide-dependent kinase 1 (PDK1) to PIP3 occurs. (8) Then PDK1 phosphorylates and activates protein kinase B (PKB), also referred to as Akt. In fact, two kinases are required for Akt activation and the nature of the second one was unknown for a long time; however, the putative kinase was designated PDK2. Recently, the mTOR (mammalian target of rapamycin)–RICTOR (rapamycin-insensitive companion of mTOR) protein complex has been suggested as PDK2 . Akt is believed to be responsible for the regulation of various metabolic pathways via phosphorylations. For example, (9) it catalyzes phosphorylation of glycogen synthase kinase (GSK)-3 that (10) leads to deactivation of GSK-3 and as a result (11) prevents the inactivation of glycogen synthase (GS) by phosphorylation, and hence increases glycogen synthesis . Moreover, PIP3 activates protein phosphatase 1 (PP1) localized on glycogen particles by an unknown mechanism , and (13) PP1 dephosphorylates inactive GS leading to (12) active GS and, therefore, to a further increase in glycogen synthesis.
Likewise, (14) deactivation of GSK-3 leads to a decreased level of phosphorylation of ATP citrate lyase (ACL) at one site. (15) Phosphorylation at a different site on ACL by PKB (16) increases the activity of that enzyme . ACL catalyzes conversion of citrate to acetyl-CoA in cytosol where it can be utilized in lipid biosynthesis. Citrate is actively transported from the mitochondria, where it is prepared from pyruvate (in two steps). Pyruvate in turn is a product of glycolysis. This path, therefore represents lipid biosynthesis from glucose.
In addition to activation of glucose and lipid metabolism, insulin greatly stimulates glucose uptake: the GLUT-4 glucose transporter is translocated from intracellular vesicles to the cell membrane. There are at least two signaling pathways that produce this result, but neither has been fully elucidated. In the first, PKB stimulates GLUT-4 exocytosis either (17) directly or (18) via activation of atypical protein kinases C: PKC-ζ and -γ (designated as aPKC in Figure 1). Several possible molecular regulators downstream of PKB and aPKC are under consideration . The second pathway of GLUT-4 translocation starts right downstream of the IR . Another IR substrate referred to as Cbl (existing in complex with Cbl-associated protein [CAP]) is (21) phosphorylated by IR kinase, however (19) binding to IR and (20) phosphorylation of adaptor protein APS is required before Cbl-CAP association with IR and APS and subsequent phosphorylation. Phosphorylated Cbl requires (22) adaptor protein CrkII together with guanine nucleotide exchange factor C3G. C3G (23) activates the TC10 protein, which belongs to the Rho family of proteins. TC10 (24) activates GLUT-4 vesicle translocation via an unknown mechanism, however it is hypothesized that this happens via actin cytoskeleton regulation – a function attributed to many of the Rho proteins. It is also believed that the two GLUT-4 translocation mechanisms described regulate different states of GLUT-4 transport and are not independent from each other.
Insulin mediates two more metabolic events through the inhibition of protein kinase A (PKA). The processes begin when (26) phosphodiesterase 3B (PDE3B) is activated at endoplasmic reticulum/Golgi . The exact mechanism of the activation is not known, however (25) PI3K might be indirectly involved (perhaps through PKB) [26,27]. Activated PDE3B (27) hydrolyzes cAMP. Low cAMP level (28) reduces the activity of PKA since cAMP activates PKA by binding to its regulatory domain. As a result (29) PKA does not phosphorylate hormone sensitive lipase (HSL) and HSL is not converted into active form. Lipolysis is, therefore, decreased.
Similarly, (30) inactive PKA does not phosphorylate fructose 2,6-bisphosphatase (Fru-2,6P2ase), thus failing to convert it into active form. Inactive Fru-2,6P2ase (31) cannot catalyze conversion of fructose-2,6-biphosphate (Fru-2,6P2) into fructose-6-phosphate (Fru-6P). Hence, Fru-2,6P2 accumulates and (32) inhibits fructose 1,6-bisphosphatase (Fru-1,6P2ase) – an enzyme that is involved in gluconeogenesis .
Another observation should be noted. At least some of the IRs are found in caveolae – small cave-like structures on the cell membrane. They represent a class of so-called lipid rafts or detergent/carbonate-insoluble glycolipid-enriched (DIG) microdomains. Like lipid rafts, the caveolae are rich in glycosphingolipids and cholesterol. Some proteins may contribute to DIG compositions; for example the IR is found to be associated with the protein caveolin (not shown in Figure 1), which is abundant in caveolae. If the caveolin-1 gene is disrupted, cells become insulin resistant.
Other molecular bases of insulin resistance have been found. It was shown that mutation of several proteins involved in insulin signal transduction, particularly in IRS-2 and the β isoform of PKB, cause the onset of diabetes or impair glucose metabolism and (or) insulin sensitivity . An excess of adipose tissue may also cause diabetes and there is a correlation between Type 2 diabetes and obesity. So-called visceral adipocytes that are found in these patients are relatively resistant to antilipolytic insulin signaling and as a result they secrete a lot of nonesterified fatty acids (NEFA) into the bloodstream. NEFAs can block insulin signaling in other cells via several possible mechanisms or they can impair glucose metabolism in liver .
How do IGs stimulate insulin-like effects in cells? Do IGs crosstalk with the insulin-signaling pathway? Are extracellularly-produced IGs transported into cells? The answer to this last question appears to be yes, at least in hepatocytes. The existence of an IG transporter is implied by the observation that a metabolically radiolabeled IG is transported into rat hepatocytes through an energy-requiring process , but whether this is due to a specific IG transporter or the normal detoxifying function of hepatocytes is unknown. However, both natural and synthetic IGs are able to modulate purified pyruvate dehydrogenase phosphatase [31,32] and pyruvate dehydrogenase kinase (PDK)  activities in cell free assays, suggesting not only an intracellular role for IGs, but also an intramitochondrial role.
Interestingly, IGs also appear to be able to stimulate cells by interacting with the exterior surface of the membranes. For example, a fluorescently-labeled synthetic IG analogue was found to stimulate lipogenesis in rat adipocytes despite the fact that it was not able to enter the cell . Müller has suggested that the dynamics of plasma membrane microdomains may be a part of the insulin-signaling pathway and has collected considerable evidence suggesting that IGs act via this mechanism [34,35]. Müller’s theory is summarized in Figure 2 and the numbered description below.
There is evidence that DIG microdomains are heterogeneous: there are cholesterol-rich (hcDIGs) and cholesterol-depleted micro-domains (lcDIGs). Major fractions of certain GPI anchored proteins exist at hcDIG. Other lipid-modified signaling proteins (fatty acid-anchored or prenylated) may exist preferentially in different DIG areas depending on the structure of their lipid anchor. Another targeting signal for proteins that defines their location at plasma membrane microdomains is association with caveolin or other hcDIG proteins. Moreover, association of caveolin with a signaling protein (e.g., the kinase pp59Lyn) seems to be a way of regulating the latter’s activity. It was found that pp59Lyn in its inactive state resides in hcDIG bound to the caveolin scaffolding domain (CSD) of caveolin through caveolin binding domains (CBDs). One explanation for the loss of activity due to caveolin binding is that the interaction with caveolin may cause some conformation change in pp59Lyn, leading to its inactivation. On the other hand, this binding may keep the kinase away from its substrates. However, it was shown that CSDP directly interacts with CBD, blocking pp59Lyn activity, and therefore supporting the first hypothesis.
Receptors for the GPI epitope of natural GPI-anchored proteins (as well as probably for free GPIs) exist in hcDIG microdomains. It was shown that GPI-anchored proteins are freely distributed with some preference to lcDIG in membrane models. Accordingly, it is the binding to the receptor that favors accumulation at hcDIG. The nature of the receptor is not well established other than that it is a protein with a molecular weight of 115 kDa.
Müller’s hypothesis, illustrated in Figure 2, is that (1) insulin stimulates a glycosylphosphatidylinositol-specific phospholipase C (GPI-PLC) that (2) lipolytically cleaves GPI-anchored proteins (and, perhaps free GPIs) on the outer leaflet of the cell membrane. (3) The lipolytically cleaved protein–IG conjugate (and IGs) promotes GPI-anchored protein(s) to (4) dissociate from the receptor; GPI-anchored proteins are found to translocate to lcDIG microdomains after insulin treatment. (5) Redistribution of acylated non-receptor tyrosine kinases (NRTKs) such as pp59Lyn on the inner leaflet of the plasma membrane between hcDIGs and lcDIGs is also observed. (6) Kinase pp59Lyn activates another kinase pp125Fak that is able to (7) recruit IRS proteins for (8) phosphorylation by pp59Lyn at the sites that are recognized by PI(3)K. This event, which was discussed above, turns on the phosphorylation cascade and is a point of convergence between the IG mechanism of action and the insulin-signaling pathway described above.
There are several explanations for the fact that GPI-anchored proteins on the exterior leaflet and acylated proteins on the cytoplasmic side move at the same time, namely because there is communication between the leaflets within the microdomain . The preference of GPI-anchored proteins for lcDIG could be considered a driving force. The lipid part of a migrating GPI may bring some other components of hcDIG (perhaps, cholesterol) with it, thus changing hcDIG composition on the cytoplasmic side. This in turn may somehow cause a protein–protein interaction-mediated signal to caveolin leading to its conformational change and to a decreased affinity toward pp59Lyn. Another possibility is that it is a change in the lipid composition of the inner leaflet that must occur after lipid rearrangement on the outer leaflet that drives pp59Lyn away.
Almost two dozen distinct assays or measurements have been used to probe how IG structure affects the molecule’s ability to interact with cells. To help the reader connect the mechanisms of action described above with Tables Tables11--77 correlating IG structure to specific activity in published assays, we have compiled a list of the assays with a brief description of each. The assays are listed in alphabetical order and, where appropriate, the step in Figure 1 or Figure 2 that is being probed is noted. Finally, the specific structurally defined IG(s) tested in the assay are listed.
The structures and the biological activities of all of the published IGs that have been tested to date are compiled in Figures Figures11 & 2 and Tables Tables11--7.7. They are organized by the number of sugar (or pseudosugar) units and span the range of monosaccharides to heptasaccharides. The tables list our compound number (designated with an IG-prefix) to allow easy correlation with the structures, the assay performed (see the preceding section for a brief description of each assay), the results of that assay, the compound number in the literature (for easy comparison when reading the primary source) and the reference.
The large collection of structures listed above (largely due to the prolific work of Dr Müller’s team) suggests a preliminary IG pharmacophore for insulin-like action. Structural elements that appear to most often contribute to high activity are a cyclic phosphate on the inositol, though the inositol isomer is less important, a free (unacylated) amino group on the second sugar, and a distal anionic group on one or more of the mannose residues (especially the fourth residue from the inositol terminus) though the nature of the ionic group (phosphonate, phosphate, sulphate) is less important. Curiously, the anomeric configuration between the aminosugar and the inositol is relatively unimportant. The observation that a hybrid compound, IG-5, bearing a distal anionic group pendant on a noncarbohydrate scaffold attached to a disaccharide is more active than the disaccharide, IG-2, from which it was derived, suggests the possibility of noncarbohydrate or partially carbohydrate-based structures with IG receptor agonist properties. This may lead to the development of compounds with useful pharmacological properties that are also considerably simpler to synthesize than the currently most active IGs such as IG-78 or IG-39.
The role of the inositol glycans in insulin signal transduction has been controversial since the IGs were first identified. While there is no doubt that IGs are produced in response to insulin (and indeed to other biological signals [61-70]) and that the IGs are able to mimic the metabolic effects of insulin on insulin-sensitive cells, questions remain regarding the extent to which IGs are necessary and/or sufficient for insulin’s metabolic action.
Regardless of their role in insulin signaling, the fact that the IGs elicit insulin-like effects downstream of the IR suggests their use in the treatment of diabetes and possibly other metabolic diseases. Several pharmaceutical companies with interest in developing therapies for Type 2 diabetes have dabbled in the IG area, but most have abandoned this avenue of research, presumably daunted by the lack of ’drug-like’ structures among the IGs and the extremely lengthy syntheses required to prepare the most active substances. The field is still in need of considerable basic research to elucidate better the nature of the IG receptor on the cell surface, develop more efficient syntheses, establish structure–activity relationships differentiating mitogenic and metabolic activity, as well as to identify simpler and synthetically more accessible IG receptor agonists that may have greater promise as drug candidates.
The basic chemistry and biology of the IGs remains a fascinating area for study. The ubiquitous presence of GPI-anchored proteins on eukaryotic cell surfaces, the highly conserved core structure of the GPI anchors, and the strong kinship in structure between the GPI protein anchors and the hormonally released IGs all suggest a significant role for this class of compounds in cell biology, certainly in hormone signal transduction, and possibly in autocrine and paracrine signaling. It is quite probable that in the next decade we will see an expanding appreciation of the role of IGs in controlling cellular function and that this will be accompanied by a renewal of interest in the use of IGs or their analogues in the treatment of human diseases.
Financial & competing interests disclosure The authors gratefully acknowledge the NIH (DK-44589 and GM-84819) for support of the work in Marc d’Alarcao’s laboratory cited in this review. The authors have no other relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript. This includes employment, consultancies, honoraria, stock ownership or options, expert testimony, grants or patents received or pending, or royalties.
No writing assistance was utilized in the production of this manuscript.
Papers of special note have been highlighted as:
■ of interest
■■ of considerable interest