Comment on: HMGB1-dependent and -independent autophagy

HMGB1 (high mobility group box 1), a ubiquitously expressed DNA-binding nucleoprotein, has not only been attributed with important functions in the regulation of gene expression but is thought to function as an important damage-associated molecular pattern in the extracellular space. Recently, conditional Hmgb1 deletion strategies have been employed to overcome the perinatal mortality of global Hmgb1 deletion and to understand HMGB1 functions under disease conditions. From these studies, it has become evident that HMGB1 is not required for normal organ function. However, the different conditional ablation strategies have yielded contradictory results in some disease models. With nearly complete recombination in all transgenic mouse models, the main reason for opposite results is likely to lie within different targeting strategies. In summary, different targeting strategies need to be taken into account when interpreting HMGB1 functions, and further efforts need to be undertaken to compare these models side by side.We appreciate the thoughtful analysis on HMGB1-dependent and -independent autophagy by Sun and Tang.¹ However, we disagree with several statements in this review. Sun and Tang write “Mice with hepatocyte-specific deletion of Hmgb1 from Robert Schwabe''s lab are not complete conditional knockout mice; the protein level of HMGB1 in the liver is decreased by about 70%,” as well as “a major difference between Robert Schwabe''s engineered HMGB1 mice and other groups is the tissue-level expression of HMGB1 after knockout.”¹We would like to point out that livers are not solely composed of hepatocytes and that albumin-Cre mediated deletion of target genes in the liver cannot result in complete loss of hepatic mRNA or protein of target genes due to the presence of unrecombined nonparenchymal cells, unless the target gene is exclusively expressed in hepatocytes and/or cholangiocytes. The reduction of hepatic HMGB1 in our studies—reaching 90% and 72% at the mRNA and protein level, respectively—is precisely at the expected level for this conditional strategy, and similar to other studies that employed albumin-Cre for hepatocyte-specific knockout of other target genes.^2-5 Hepatocytes account only for approximately 52% of cells in the liver, with other cell types including Kupffer cells (∼18% of liver cells), hepatic stellate cells (˜8% of liver cells), endothelial cells (∼22% cells of liver cells) and cholangiocytes (<1 % of liver cells) contributing to the remainder.⁶ Accordingly, albumin-Cre-mediated reduction of mRNA and protein levels of target genes (i.e., Hmgb1 and HMGB1 in our study) in the liver cannot exceed the amount of mRNA and protein expressed by hepatocytes and cholangiocytes (which is typically about 70–90%,^2-5 due to higher mRNA and protein levels in hepatocytes than in other hepatic cell types). The high efficacy of our conditional approach is best demonstrated by almost complete loss of HMGB1 expression in the hepatocellular compartment of albumin-Cre mice—as evidenced by loss of HMGB1 expression in all HNF4α-positive cells and in isolated primary hepatocytes—whereas HMGB1 expression is retained in nonparenchymal cells, as demonstrated by costaining for Kupffer cell marker F4/80, endothelial cell marker endomucin, and hepatic stellate cell marker desmin.^7,8 The nearly perfect recombination rate in our mice was further confirmed by experiments that employed Mx1Cre for Hmgb1 deletion, which resulted in almost complete loss of hepatic Hmgb1 mRNA and HMGB1 protein.^7,8 Moreover, our transgenic mice show early postnatal mortality when bred with a germline Cre deleter,⁷ thus reproducing the phenotype of the global HMGB1 knockout.⁹In summary, our transgenic mouse model results in nearly perfect recombination efficiency with virtually complete loss of Hmgb1 mRNA and HMGB1 protein in all targeted cell types, and constitutes a valid tool for the assessment of HMGB1 functions in vivo. Findings from this model need to be taken into account for proper interpretation of the role of HMGB1 in the normal and diseased liver, and cannot be interpreted as a result of incomplete deletion efficiency. Hence, differences in targeting strategies (exons 2–4 by our approach, exons 2–3 in mice from Tang and colleagues) are likely to explain opposite findings, e.g. improvement of ischemia-reperfusion injury in our hands, but aggravation of liver damage in the study by Huang et al.^8,10 Further analysis needs to be performed to determine whether ablation of exons 2–3 versus exons 2–4 leads to complete loss of HMGB1 function.