CORRESPONDENCE

mRNA was very low compared with the NEFA release from adipose tissue, and that by Tan et al.,rosiglitazone treatment did not level of FABP4 mRNA. this might account for the insulin-sensitiz- lower plasma levels of NEFAs. This differs We also did in vitro experiments on ing action of TZDs1.However, the contribu- from nearly all rodent models, as well as sev- human adipocytes. Levels of mRNA were tion of visceral adipose tissue to circulating eral controlled human trials reporting signif- measured using real-time PCR and the NEFA concentrations is only ∼7% (refs. icant lowering of NEFA levels by human glycerol kinase primers described by 9,10), with the overwhelming contribution rosiglitazone and other TZDs5,12,13.Human Guan et al.1 Expression of mRNA was nor- (74–78%) to circulating NEFAs being from responses are likely to be more heteroge- malized for 18S RNA. Glycerol kinase mRNA the upper body subcutaneous adipose neous than those of inbred rodent models. If, expression in isolated human adipocytes was depot9,10, the tissue studied here. as we have hypothesized, adipose glycerol very low compared with the levels found in Because we studied patients after 12 kinase induction is one factor contributing to human kidney and mouse brown adipose weeks of treatment, it could be that we have NEFA lowering by TZDs, then lack of glyc- tissue, two sites with high glycerol kinase missed early changes in glycerol kinase erol kinase induction may not be surprising activity (Fig. 1e). To assess the effect of expression and that we are studying com- in patient populations in which NEFA levels rosiglitazone on gene expression, differenti- pensatory changes to the insulin sensitiza- do not respond to TZDs. Larger studies will ated human adipocytes from primary cul- tion. It seems difficult to argue, however, be needed to determine the extent to which tures were treated for 48 h with 1 µmol/l that any process that is fundamental to TZD induction of adipose glycerol kinase is rosiglitazone. In contrast to the results by insulin sensitization could be switched off at variable in humans, and whether this corre- Guan et al.1,we found no increase in glycerol a time when insulin sensitization is so lates with reduction in NEFA levels upon kinase mRNA expression (P = 0.79; Fig. 1f). clearly observed. TZD treatment. As observed in vivo, FABP4 mRNA was In summary, a ‘futile metabolic cycle’ increased by rosiglitazone (P = 0.02). involving glycerol is not induced in adipose Hong-Ping Guan1 & Mitchell A Lazar2,3,4 If glycerol kinase were induced, glycerol tissue in humans treated with rosiglitazone, Departments of 1Medicine, 2Pharmacology and would be recycled in adipose tissue, resulting and thus does not contribute to the meta- 3Genetics, and 4The Penn Diabetes Center, http://www.nature.com/naturemedicine in a decreased glycerol output from adipose bolic actions of TZDs in humans. University of Pennsylvania School of Medicine, tissue. Rather than a decrease, we observed a Philadelphia, Pennsylvania 19104, USA. nonsignificant increase in adipose tissue ACKNOWLEDGMENTS e-mail: [email protected] This study was approved by the Oxfordshire Clinical glycerol output in humans in vivo,suggest- 1. Guan, H.P. et al. A futile metabolic cycle activated in Research Ethics Committee. GlaxoSmithKline adipocytes by antidiabetic agents. Nat. Med. 8, ing that glycerol is not reused in adipose tis- contributed to the funding of this study. We thank L. 1122–1128 (2002). sue after rosiglitazone treatment in humans. Dennis, J. Currie, V.Ilic and M. Clark for assistance 2. Frayn, K.N. & Coppack, S.W. Assessment of white The presence of a low glycerol kinase with the in vivo studies. adipose tissue metabolism by measurement of arteri- ovenous differences. Meth. Mol. Biol. 155, 269–279 activity in adipose tissue has been well doc- COMPETING INTERESTS STATEMENT (2001). 3 umented .We observed a low level of glyc- The authors declare competing financial interests (see 3. Chakrabarty, K., Tauber, J.W., Sigel, B., Bombeck, erol kinase mRNA expression in both the Nature Medicine website for details). C.T. & Jeffay, H. Glycerokinase activity in human adi- pose tissue as related to obesity. Int. J. Obes. 8, human adipose tissue and human 609–622 (1984). 1 2 3 adipocytes. Guan et al. presented data as rel- Garry D Tan ,Cyrille Debard , Claire Tiraby , 4. Maggs, D.G. et al. Metabolic effects of troglitazone Sandy M Humphreys1,Keith N Frayn1, monotherapy in type 2 diabetes mellitus. A random- ative inductions of glycerol kinase activity1. Dominique Langin3,Hubert Vidal2 & ized, double-blind, placebo-controlled trial. Ann. © Group 2003 Nature Publishing The threefold increase they described may Fredrik Karpe1 Intern. Med. 128, 176–185 (1998). 5. Miyazaki, Y. et al. Effect of rosiglitazone on glucose still be negligible in absolute terms. Our 1 Oxford Centre for Diabetes, Endocrinology and and non-esterified fatty acid metabolism in Type II findings of a lack of induction of glycerol Metabolism, University of Oxford, Churchill diabetic patients. Diabetologia 44, 2210–2219 kinase mRNA expression in response to Hospital, Oxford OX3 7LJ, UK. 2INSERM U449, (2001). rosiglitazone in human adipose tissue and R. Laennec Faculty of Medicine, Claude Bernard 6. Racette, S.B., Davis, A.O., McGill, J.B. & Klein, S. Thiazolidinediones enhance insulin-mediated sup- cultured human adipocytes show that more University of Lyon, 69372 Lyon Cedex 08, France. pression of fatty acid flux in type 2 diabetes mellitus. work needs to be done in this area before 3Unité de Recherches sur les Obésités INSERM Metabolism 51, 169–174 (2002). accepting any relationship of glycerol kinase U586, Institut Louis Bugnard, Centre Hospitalier 7. Fujiwara, T., Yoshioka, S., Yoshioka, T., Ushiyama, I. Universitaire de Toulouse, Université Paul & Horikoshi, H. Characterization of new oral antidia- to TZD-induced insulin sensitization. betic agent CS-045. Studies in KK and ob/ob mice Sabatier, 31403 Toulouse Cedex 4, France. Human studies using TZDs show small and Zucker fatty rats. Diabetes 37, 1549–1558 e-mail: [email protected] (1988). falls in NEFA concentrations in the range of 8. Finegood, D.T. et al. Beta-cell mass dynamics in 8–30% (refs. 4–6), not always statistically Zucker diabetic fatty rats. Rosiglitazone prevents the significant, as was true in the present study. Guan et al. reply: rise in net cell death. Diabetes 50, 1021–1029 (2001). In contrast, most rodent studies show a We are pleased that our report of TZD induc- 9. Basu, A. et al. Systemic and regional free fatty acid decrease in NEFA concentration of 50–90% tion of adipose glycerol kinase1 has metabolism in type 2 diabetes. Am. J. Physiol. 7,8 Endocrinol. Metab. 280, E1000–E1006 (2001). in response to TZD treatment . prompted further investigation by other 10. Jensen, M.D. Adipose tissue and fatty acid metabo- Accordingly, the conflicting data on glycerol groups. Tordjman et al. recently confirmed lism in humans. J. R. Soc. Med. 95 (suppl. 42), 3–7 kinase induction by rosiglitazone between our observation that rosiglitazone markedly (2002). 11. Tordjman, J. et al. Thiazolidinediones block fatty acid rodents and humans is consistent with the induces glycerol kinase in mouse release by inducing glyceroneogenesis in fat cells. J. species-specific changes in NEFA concen- adipocytes11.In their correspondence, Tan et Biol. Chem. 278, 18785–18790 (2003). trations by TZDs. al. did not find increased glycerol kinase 12. Kumar, S. et al. Troglitazone, an insulin action enhancer, improves metabolic control in NIDDM We are aware that depot-specific changes expression in rosiglitazone-treated human patients. Troglitazone Study Group. Diabetologia 39, in adipose tissue in response to TZD treat- adipocytes or in adipose samples from 701–709 (1996). 13. Chaiken, R.L. et al. Metabolic effects of darglitazone, ment have been described. Guan et al.pro- rosiglitazone-treated type 2 diabetics. It is an insulin sensitizer, in NIDDM subjects. posed that glycerol kinase induction lowers important to note that in the patients studied Diabetologia 38, 1307–1312 (1995).

812 VOLUME 9 | NUMBER 7 | JULY 2003 NATURE MEDICINE