Hepatic lipid overload potentiates biliary epithelial cell activation via E2Fs

Reviewer #1 (Public Review):

This manuscript explores how biliary epithelial cells respond to excess dietary lipids, an important area of research given the increasing prevalence of NAFLD. The authors utilize in vivo models complemented with cultured organoid systems. Interesting, E2F transcription factors appear important for BEC glycolytic activation and proliferation.

Much of the work utilizes the BEC-organoid model, which is complicated by the fact that liver cell organoid models often fail to maintain exclusive cell identity in culture. The method used by the authors (Broutier et al., 2016) can lead to organoids with a mixture of ductal and hepatocyte markers. It would be helpful for the authors to further demonstrate the cholangiocyte identity of the organoid cells.

The authors suggest that BECs form lipid droplets in vivo by detecting BODIPY immunofluorescence of liver cryosections. While confocal microscopy would ensure that the BODIPY fluorescence signal is within the same plane as the cell of interest, the authors use a virtual slide microscope that cannot exclude fluorescence from a different focal plane. The conclusion that BECs accumulate lipids does not seem to be fully supported by this analysis.

Several mouse experiments rely heavily on rare BEC proliferation events with the median proliferation event per bile duct being 0-1 cell. While the proliferative effect appears consistent across experiments, a more quantitative approach, such as performing Epcam+ BEC FACS and flow cytometry-based cell cycle analyses, would be helpful.

Finally, it is not yet clear how relevant the findings in this study are to ductular reaction, which is a non-specific histopathologic indicator of liver injury in the context of severe liver disease. In NAFLD, the ductular reaction is uncommon in benign steatosis, and if seen at all, occurs in the setting of substantial liver inflammation and fibrosis (Gadd et al., Hepatology 2014). The authors use a dietary model containing 60 kcal% fat, which causes adipose lipid accumulation as well as subsequent liver lipid accumulation. This diet does not cause overt inflammation or fibrosis that would represent experimental NASH, which typically requires the addition of cholesterol in dietary lipid NASH models (Farrell et al., Hepatology, 2019). While the E2F-driven proliferation may be important for physiologic bile duct function in the setting of obesity, the claim that E2Fs mediate DR initiation would require an additional pathophysiologic model or human data to demonstrate relevance. The authors could clarify this point in their discussion.

Reviewer #2 (Public Review):

The manuscript by Yildiz et al investigates the early response of BECs to high fatty acid treatment. To achieve this, they employ organoids derived from primary isolated BECs and treat them with a FA mix followed by viability studies and analysis of selected lipid metabolism genes, which are upregulated indicating an adjustment to lipid overload. Both organoids with lipid overload and BECs in mice exposed to a HFD show increased BEC proliferation, indicating BEC activation as seen in DR. Applying bulk RNA-sequencing analysis to sorted BECs from HFD mice identified four E2F transcription factors and target genes as upregulated. Functional analysis of knock-out mice showed a clear requirement for E2F1 in mediating HFD induced BEC proliferation. Given the known function of E2Fs the authors performed cell respiration and transcriptome analysis of organoids challenged with FA treatment and found a shift of BECs towards a glycolytic metabolism.

The study is overall well-constructed, including appropriate analysis. Likewise, the manuscript is written clearly and supported by high-quality figures. My major point is the lack of classification of the progression of DR, since the authors investigate the early stages of DR associated with lipid overload reminiscent of stages preceding late NAFLD fibrosis. How are early stages distinguished from later stages in this study? Molecularly and/or morphologically? While the presented data are very suggestive, a more substantial description would support the findings and resulting claims.

Author Response

Reviewer #1 (Public Review):

This manuscript explores how biliary epithelial cells respond to excess dietary lipids, an important area of research given the increasing prevalence of NAFLD. The authors utilize in vivo models complemented with cultured organoid systems. Interesting, E2F transcription factors appear important for BEC glycolytic activation and proliferation.

We thank this reviewer for his/her comments and for finding the E2F-mediated mechanism of interest.

Much of the work utilizes the BEC-organoid model, which is complicated by the fact that liver cell organoid models often fail to maintain exclusive cell identity in culture. The method used by the authors (Broutier et al., 2016) can lead to organoids with a mixture of ductal and hepatocyte markers. It would be helpful for the authors to further demonstrate the cholangiocyte identity of the organoid cells.

We understand the concern of this reviewer. Indeed, this method can give rise to biliary cells or more hepatocyte-like cells. However, this choice depends on the culture media used. Our experiments used BEC-organoids in an undifferentiated state with a biliary expression profile. Please see point 1 above for a detailed answer.

The authors suggest that BECs form lipid droplets in vivo by detecting BODIPY immunofluorescence of liver cryosections. While confocal microscopy would ensure that the BODIPY fluorescence signal is within the same plane as the cell of interest, the authors use a virtual slide microscope that cannot exclude fluorescence from a different focal plane. The conclusion that BECs accumulate lipids does not seem to be fully supported by this analysis.

We fully agree with this criticism. To address this concern, we decided to use FACS analysis, a quantitative and independent method, to further confirm our initial findings. To this end, we stained sorted EPCAM+ BECs isolated from livers of CD- or HFD-fed mice with BODIPY, quantified the number of BODIPY+/EPCAM+ BECs in each experimental condition, and confirmed that these cells accumulate more lipids after HFD feeding (New Figure 1I, page 5, lines 112-115, and see also reply rebuttal to point 4).

Several mouse experiments rely heavily on rare BEC proliferation events with the median proliferation event per bile duct being 0-1 cell. While the proliferative effect appears consistent across experiments, a more quantitative approach, such as performing Epcam+ BEC FACS and flow cytometry-based cell cycle analyses, would be helpful.

Following this suggestion, we quantified proliferative EdU+ BEC cells by FACS in a new cohort of C57BL/6J mice fed CD or HFD. These data, now included in the revised manuscript (New Figure 2G, page 7, lines 143-147), strongly confirm that immunofluorescence quantification mirrors the FACS quantification and reinforce the initial finding that EPCAM+ BECs proliferate more in the livers of HFD-fed mice. Please see point 6 above for a detailed answer.

Finally, it is not yet clear how relevant the findings in this study are to ductular reaction, which is a non-specific histopathologic indicator of liver injury in the context of severe liver disease. In NAFLD, the ductular reaction is uncommon in benign steatosis, and if seen at all, occurs in the setting of substantial liver inflammation and fibrosis (Gadd et al., Hepatology 2014). The authors use a dietary model containing 60 kcal% fat, which causes adipose lipid accumulation as well as subsequent liver lipid accumulation. This diet does not cause overt inflammation or fibrosis that would represent experimental NASH, which typically requires the addition of cholesterol in dietary lipid NASH models (Farrell et al., Hepatology, 2019). While the E2F-driven proliferation may be important for physiologic bile duct function in the setting of obesity, the claim that E2Fs mediate DR initiation would require an additional pathophysiologic model or human data to demonstrate relevance. The authors could clarify this point in their discussion.

We agree with this reviewer that 15 weeks of HFD on C57BL/6J feeding are insufficient to trigger a ductular reaction. For this purpose, we used the term “BEC activation” in our manuscript, which refers to the first mandatory step for the ductular reaction to initiate. We apologize if our initial manuscript did not sufficiently emphasize this point. However, as suggested by the reviewer we investigated the ductular reaction in our model. In order to further characterize the livers after 15 weeks of CD or HFD feeding, we stained the bile ducts for pancytokeratin (PANCK) and osteopontin (OPN) and asked a pathologist (Dr. Christine Gopfert at EPFL) to evaluate these sections with a particular focus on the bile ducts. She concluded that the livers of HFD-fed mice showed steatosis and inflammation but no apparent fibrosis (New Figure 1 – figure supplement 1E). The shape of bile ducts was similar in the livers of CD- and HFD-fed mice (New Figure 1 – figure supplement 1I), concomitant with the absence of portal fibrosis and inflammation. In addition, we checked the expression levels of several established markers of ductular reaction in our RNA sequencing data and observed that, of all these genes, only Ncam1 was significantly upregulated with HFD feeding in EPCAM+-BEC cells (New Figure 2 – figure supplements 1D and 1E, Page 6, lines 127-131). Overall, these data support our conclusion that HFD triggers BEC activation without signs of an established ductular reaction and might suggest Ncam1 as a marker for this initial BEC activation process. Please see point 3 above for a detailed answer.

Reviewer #2 (Public Review):

The manuscript by Yildiz et al investigates the early response of BECs to high fatty acid treatment. To achieve this, they employ organoids derived from primary isolated BECs and treat them with a FA mix followed by viability studies and analysis of selected lipid metabolism genes, which are upregulated indicating an adjustment to lipid overload. Both organoids with lipid overload and BECs in mice exposed to a HFD show increased BEC proliferation, indicating BEC activation as seen in DR. Applying bulk RNA-sequencing analysis to sorted BECs from HFD mice identified four E2F transcription factors and target genes as upregulated. Functional analysis of knock-out mice showed a clear requirement for E2F1 in mediating HFD induced BEC proliferation. Given the known function of E2Fs the authors performed cell respiration and transcriptome analysis of organoids challenged with FA treatment and found a shift of BECs towards a glycolytic metabolism. The study is overall well-constructed, including appropriate analysis. Likewise, the manuscript is written clearly and supported by high-quality figures.

We appreciate that this reviewer finds our study well-constructed, clear, and with high-quality figures.

My major point is the lack of classification of the progression of DR, since the authors investigate the early stages of DR associated with lipid overload reminiscent of stages preceding late NAFLD fibrosis. How are early stages distinguished from later stages in this study? Molecularly and/or morphologically? While the presented data are very suggestive, a more substantial description would support the findings and resulting claims.

We thank the reviewer for the suggestion. We would like to emphasize that instead of ductular reaction, we used the term “BEC activation” in our revised manuscript, referring to the first mandatory step for initiating the ductular reaction. Both reviewers criticized the poor characterization of the ductular reaction process in the first version of our study; we put substantial effort into further clarifying this point. Our response to this point can be read in our reply to the last comment of reviewer 1 and point 3 of the rebuttal.