Kim Cheng, Kenneth Ho, Rebecca Stokes, Christopher Scott, Sue Mei Lau, Wayne J. Hawthorne, Philip J. O’Connell, Thomas Loudovaris, Thomas W. Kay, Rohit N. Kulkarni, Terumasa Okada, Xiaohui L. Wang, Sun Hee Yim, Yatrik Shah, Shane T. Grey, Andrew V. Biankin, James G. Kench, D. Ross Laybutt, Frank J. Gonzalez, C. Ronald Kahn, Jenny E. Gunton
J Clin Invest.
ARNT protein levels decrease in islets from beta-Hif1a-null mice
Université catholique de Louvain, Institut de Recherches Clinique et Expérimentale, Pôle d'Endocrinologie, Diabète et Nutrition
Published June 28, 2010
We have read with great interest the recent paper published by Cheng et al in the Journal, about the effect of HIF1alpha KO in beta cells on their functional properties and in vivo glucose tolerance (Cheng K et al, J. Clin. Invest. 2010, 120(6):2171-83).
In figure 6, the authors provide a series of evidence suggesting that HIF1alpha is required for adequate expression of its dimerisation partner, ARNT, in beta cells. However, despite the fact that RT-PCR data show a 50% reduction of Arnt mRNA levels in beta-HIF1a-null islets and that Chip analysis indicates that HIF1a binds to the Arnt promoter, we believe that the immunofluorescence data shown in figure 6C do not really support the conclusion that ARNT protein levels are decreased in islets from beta-HIF1a-null vs. WT mice (It may well be correct but, in our opinion, the data do not support it).
Indeed, although the figure shows a difference in red fluorescent staining in beta-HIF1a-null vs. WT islets, the red signal in WT islets does not seem to occur in nuclei nor in the cytosol of insulin positive cells : by carefully comparing the single fluorescence images of the islets at high magnification, we find neither co-staining for the "DAPI blue signal" and the "ARNT red signal" nor co-staining for the "Insulin-green signal" and the "ARNT red signal" (We do not see the expected red and blue costained nuclei ; if the red staining were mostly in the cytoplasm of beta cells, which is admittedly possible for ARNT, there should be red and green costained areas which we do not see either).
In contrast, in an earlier paper on ARNT reduction in the beta cells from type 2 diabetic patients (Gunton et al. Cell 2005, vol 122, pp337-349), islet ARNT staining was mostly nuclear and was present in more than 50% of the beta cell nuclei (Using the same antibody, BD Biosciences mouse anti-HIF1beta/ARNT1, we observe a similar pattern of cytosolic and nuclear ARNT staining in islet cells that is different from that shown in Cheng et al.).
Actually, the red fluorescent staining in WT islets of figure 6c rather seems to correspond to either autofluorescent red blood cells or, although unlikely, to the endothelial cells within islet capillaries. Red blood cells are known to autofluoresce in the red due to their high content in hemoglobin/porphyrin derivatives - cfr Tristao et al, J Fluoresc. 2010, vol 20, pp665-9 and Nagabau et al, Free Radic Biol Med. 2000, vol 29, pp659-63). As to the reason why there would be less red blood cells or endothelial cells in beta-HIF1a-null vs. WT islets, we can only speculate that the phenotype of these islets could resemble that of islets from mice with VEGF-A KO in the pancreas : no change in total beta cell area, smaller islets on average, reduced intra-islet vascularization (Lammert et al, Curr Biol 2003, vol 13, pp1070-1074).
Mohammed Bensellam, Galena Rybachuk and Jean-Christophe Jonas