We read with interest the recent articles by Hattori et al. (1,2), which report activation of nuclear factor-κB (NF-κB) transcription by the recombinant globular domain of adiponectin. However, these findings contrast directly with a substantial body of published and ongoing work. As initially reported by Ouchi et al. and confirmed by other groups (3,4), a major effect of adiponectin is the suppression of proinflammatory endothelial responses, which include reversing the activation of NF-κB, reducing adhesion molecule expression, and enhancing nitric oxide bioavailability. We have shown that the globular domain of adiponectin also suppresses endothelial reactive oxygen species generation in response to various agonists, including oxidized LDL and high glucose (5,6). At the American Diabetes Association 66th Scientific Sessions, we also reported our confirmatory studies in vivo that globular adiponectin suppresses vascular inflammation, observed as a reduction in leukocyte-endothelial interactions in the mesenteric microvasculature (7). Consistent with these observations, we also recently reported (8) globular adiponectin–induced reduction in oxidative-nitrosative stress in postmyocardial ischemia-reperfusion injury.
The initial report from Hattori et al. (2), showing increased inflammatory gene expression in cultured endothelial cells treated with commercial globular adiponectin, was surprising to us because we have used similar materials in our own studies. Caution needs to be exerted when interpreting these studies in isolation because responses in endothelial cells isolated from different vascular beds should be considered. Furthermore, even within the same primary cell type, heterogeneity in proinflammatory gene expression has been observed (9,10). Given the inflammatory responses that were observed in their studies, even small amounts of lipopolysaccharide contamination remain a concern with bacterially expressed proteins and cannot be excluded as a confounding variable. Interestingly, they showed that globular adiponectin alone had a relatively small effect on the activation of NF-κB in a transfected SV40-transformed vascular endothelial cell system. However, a prominent dose-dependent suppression by globular adiponectin of tumor necrosis factor-α–stimulated NF-κB activation was also observed, representing an anti-inflammatory effect of globular adiponectin that is fully consistent with previous studies in this area. However, this was not highlighted in their article. As controls, it would also have been valuable to determine whether globular adiponectin from different sources affected more than one endogenous inflammatory signal, also using untransfected cells as well as cells transfected with control vectors.
The more recent article from Hattori et al. (1) uses primary cultured cardiac fibroblasts and shows relatively modest effects of globular adiponectin alone on transfected NF-κB and activating protein-1 promoter activity that is enhanced by costimulation with tumor necrosis factor-α. A small observed increase in extracellular signal–related kinase activity by globular adiponectin alone was also enhanced by costimulation with angiotensin II. However, using cardiomyocytes, Walsh et al. (11) reported that among a variety of pathways ameliorated by adiponectin in the postischemic heart, trimeric full-length adiponectin actually suppressed α-adrenoreceptor agonist stimulated in extracellular signal–related kinase activity activation, which is similar to our own finding in aortic endothelial cells that globular adiponectin suppressed p42/p44 mitogen-activated protein kinase activated by oxidized LDL (5).
Finally, in a C2C12 skeletal myoblast system, Tsao et al. (12,13) showed that only high–molecular weight oligomers of adiponectin (but not globular adiponectin) activated an NF-κB signaling cascade, in contrast to trimeric adiponectin that activated only AMP kinase. Although also performed in a different cell type, these findings also diverge from the results reported by Hattori et al.
In their recent papers, Hattori et al. have not offered any potential explanations that might help reconcile their findings with these multiple lines of research in the current literature. We hope that this letter might give them a chance to provide additional perspective on these issues.