We thank Dr. Takahashi for his interest (1,2). First, the main reason for using 100 nmol/l adrenomedullin was to compensate for the uncontrolled catabolism of exogenously added adrenomedullin. Several reports point out that human adipocytes express neprilysin, one of the main enzymes that degrade adrenomedullin (3,4). The metalloprotease is also expressed by 3T3-F442A preadipocytes during differentiation (R.H., F.S., unpublished data). Moreover, we observed in preliminary experiments that omapatrilat, a dual vasopeptidase inhibitor of neprilysin and ACE, increased adipogenesis in 3T3-F442A preadipocytes.
We also reported weak adrenomedullin secretion by adipose cells in a previous study (5), but one should keep in mind that traditional Sep-Pak C18 purification procedures before radioimmunoassay do not take into account the adrenomedullin fraction bound to complement factor H (6). This has been demonstrated as a particular bias in the estimation of circulating adrenomedullin levels. We also think that it may affect adrenomedullin assay of cell culture media containing serum (7).
Second, electroporation of differentiated adipocytes is difficult because they are prone to collapse, and other efficient transfection methods that utilize cationic lipid reagents are suspected of inhibiting insulin signaling (8). Isolated human adipocytes treated with insulin showed that adrenomedullin expression is downregulated in fully differentiated cells (see Fig. 5B in ref. 2).
Third, the discrepancy between murine adipose cell lines and human mesenchymal stem cells may be due to variations in the treatments used to induce adipogenesis; however, it suggests that adipocytes have mechanisms to both upregulate and downregulate adrenomedullin gene expression during differentiation. Recent advances in the field of adipogenesis point out the many facets of the effect of insulin. In addition to the recent discovery of cyclic AMP response element–binding protein activation (9), we think that adrenomedullin downregulation could contribute to the proadipogenic action of insulin, although this will effectively require further investigation.
Finally, we agree with Dr. Takahashi that the relation between plasma adrenomedullin and insulin levels might have been different with insulin-resistant obese subjects. We believe that the development of insulin resistance during obesity may lead to an effective increase in adrenomedullin plasma concentration because the peptide would no longer be negatively controlled by insulin. This hypothesis is supported by recent results, demonstrating a positive correlation of adrenomedullin levels in omental adipose tissue with BMI and homeostasis model assessment index in obese patients with no endocrine, cardiovascular, hepatic, renal, or inflammatory disease (10).
Data and results presented in the article are drawn from 10 patients in each group: 5 male and 5 female patients in the lean group and 4 male and 6 female patients in the obese group.
