Division of Biological Sciences and Department of Nutrition, Harvard School Of Public Health, Boston, Massachusetts, USA
Address correspondence to: Gökhan S. Hotamisligil, Division of Biological Sciences and Department of Nutrition, Harvard School Of Public Health, 665 Huntington Avenue, Boston, Massachusetts 02115, USA. Phone: (617) 432-1950; Fax: (617) 432-1941; E-mail: firstname.lastname@example.org.
First published January 15, 2003 - More info
Fuel sources have historically triggered many battles throughout the globe and will likely continue to do so. The design of biological systems is no different. Adipose tissue holds the richest energy sources and exerts substantial influence on the organism. While this fundamental truth has eluded scientists for a long time, it is now clear that adipocytes have established effective communication networks and gained control over important trade routes to maintain their influential status. To this end, these innocent-looking cells appear to utilize two main tools: fatty acids and polypeptide hormones. It is likely that they have additional, yet unknown, signaling mechanisms.
It has been known for some time that obesity is associated with insulin resistance and diabetes. Yet how and why these problems are so closely linked at the molecular level has not been clear. As in most such cases, lack of knowledge led to many conspiracy theories — my favorite one being that agents produced by fat cells affect metabolic control at other sites. An example of such an agent is the inflammatory cytokine TNF-α, which is produced in excess in obesity and causally linked to insulin resistance (1). Many other such agents have now been added to this list, although unlike TNF-α, some do not mediate adverse effects on metabolism. In fact, some, such as leptin and adiponectin, work to improve the metabolic status of their fellow organs (2).
The most recent addition to this adverse family of adipocyte-derived polypeptides is resistin. During a search for targets of thiozolidinediones (TZDs), a class of insulin sensitizers, Steppan et al. discovered that expression of the gene encoding this secreted protein was suppressed by TZD treatment in adipocytes (3). Subsequently, additional members of this family, resistin-like molecule–α (RELMα) and RELMβ, and alternative biological activities were reported (4, 5). In their original report, Steppan et al. showed that levels of this protein in the blood of obese mice are increased and that TZDs acted to suppress this obesity-related increase in resistin expression (4). Moreover, in vivo administration of recombinant resistin to otherwise normal mice caused insulin resistance, whereas administration of an antiresistin antibody increased insulin sensitivity of obese and insulin-resistant animals. Finally, cultured adipocytes exposed to resistin showed a reduction of insulin-stimulated glucose uptake, whereas the antiresistin antibody produced the opposite effect (4). This discovery generated much interest and excitement, not only as a mechanism of action for this class of drugs but also as a potential general mechanism by which obesity is linked to insulin resistance and diabetes, not to mention the potential therapeutic applications of the discovery.
Yet it became rapidly clear that the story might be much more complicated than originally envisioned. First, expression of resistin mRNA and protein was found to be strongly suppressed in adipose tissues in many models of experimental obesity (6–8). This raised the question of why expression of a protein that mediates insulin resistance would be strongly suppressed in conditions of insulin resistance. Could protein levels in obesity still be elevated in the circulation due to increased adipose tissue mass? This issue was also recently addressed, and it appears that serum protein levels in experimental obesity, for example in db/db mice, are significantly reduced when compared to those in lean control mice, similarly to resistin mRNA and protein levels in adipose tissue (9). In addition, treatment of animals with insulin-sensitizing drugs such as metformin or several different categories of synthetic PPAR-γ ligands, including TZDs, resulted in a complex and inconsistent pattern of resistin expression (3, 6–8, 10). On the other hand, agents that cause insulin resistance, such as TNF-α, negatively regulate expression and secretion of resistin in cultured adipocytes (11). Finally, unlike in the mouse gene, the human homologue of resistin was not detected in human adipocytes in any appreciable amount, and the quantities of mRNA found in adipose tissue were not found to be related to obesity or diabetes (12). To complicate things further, one of the two homologues of resistin, RELMα, is absent from the human genome.
This picture is certainly complicated. In this issue of the JCI, Rajala et al. provide much-needed clarification of the biological functions of this family of proteins (13). In an elegant set of in vivo experiments, the authors demonstrated that administration of recombinant resistin and RELMβ to rats results in acutely impaired hepatic insulin sensitivity and glucose metabolism. The primary pathway underlying changes in hepatic glucose metabolism appears to be increased glucose production. Interestingly, no effect was observed on peripheral glucose disposal under the clamp conditions, effectively ruling out a role for resistin in this part of the insulin action, at least under the experimental paradigms used in the study. This is a straightforward examination of the acute effects of resistin and RELMβ on insulin action, and the results are quite clear. These results support the findings of Steppan et al., who observed impaired glucose metabolism upon administration of another version of recom-binant resistin in mice (3). Furthermore, the findings also suggest that resistin and the closely related RELMβ may act to establish links among adipose tissue, the intestine, and the liver. It now remains to be determined whether these pathways are operational under physiological or pathophysiological conditions.
Overall, it is clear that resistin and RELMβ have the capacity to impair insulin action in the liver (13). The question is whether this capacity is important in the insulin resistance of obesity or the insulin-sensitizing effects of synthetic PPAR-γ ligands. It is not easy to answer any of these questions. It is challenging to place resistin in a straightforward mechanistic model as a major mediator of the insulin-sensitizing effects of PPAR-γ agonists. While the experiments with PPAR-γ agonists in cell-based systems have been relatively consistent, the in vivo experiments examining their effects on resistin expression have generated complicated results. Existing data suggest that suppression of resistin might not be a prerequisite to observing the insulin-sensitizing effects of PPAR-γ agonists. However, the kinetics of regulation of resistin expression and protein synthesis and secretion have not yet been correlated with the onset of the insulin-sensitizing effects of PPAR-γ agonists in vivo. Also of interest is the potential divergence among the effects of agonists with different structure and potency. Definitive answers to these issues will require in vivo systems with and without endogenous resistin and/or the related molecules. Testing the ability of PPAR-γ agonists to regulate insulin sensitivity in these systems should provide useful insights. As for RELMβ, it is not clear whether it is secreted into the circulation to act on the liver under physiologically relevant conditions.
The findings thus far do not support the idea that elevated resistin levels play a critical role in the insulin resistance associated with experimental obesity. This is not to say that resistin has no role in insulin resistance. First, the correlation between circulating levels of a bioactive peptide and its effects is not always linear. The receptor systems for resistin, their signaling, and their sensitivity state are not known. Studies towards identification of these pathways will be of great interest. Second, insulin resistance is not necessarily associated with obesity. The impact of resistin and related proteins could be most important, or easier to test, in different biological settings, perhaps in the absence of severe obesity. Genetic loss-of-function models will be essential to addressing these questions.
What about the role of resistin in human biology? In my view, it is premature to comment on this point, since there is no information regarding the biology of resistin in human health and disease. There are only two isoforms in humans as opposed to three in rodents. There are also striking differences in the cellular source of resistin in humans and rodents, which might imply that the biology has evolved differently among species. If and how these differences relate to the biological functions of resistin in humans is simply not known at the moment.
A final relevant point is the potential role of resistin and related proteins in the inflammatory response. This family of proteins was first identified in experimental models of asthma and named FIZZ as a gene regulated during inflammation (found in inflammatory zone, ref. 14), thus suggesting a potential role in inflammation. Their structure also suggests that they might represent a new group of cytokines. Since inflammation has central mechanistic significance in obesity and insulin resistance, this might be of importance to the contribution of resistin and related molecules in these diseases (15, 16).
In closing, perhaps not every biological paradigm could be viewed from an intransigent and conventional perspective. Sometimes things just don’t fit. And, if they don’t, science must not quit.
See the related article beginning on page 225.
Conflict of interest: The author has declared that no conflict of interest exists.
Nonstandard abbreviations used: thiozolidinedione (TZD); resistin-like molecule (RELM).