Spatial mapping of hepatic ER and mitochondria architecture reveals zonated remodeling in fasting and obesity

Abstract

The hepatocytes within the liver present an immense capacity to adapt to changes in nutrient availability. Here, by using high resolution volume electron microscopy, we map how hepatic subcellular spatial organization is regulated during nutritional fluctuations and as a function of liver zonation. We identify that fasting leads to remodeling of endoplasmic reticulum (ER) architecture in hepatocytes, characterized by the induction of single rough ER sheet around the mitochondria, which becomes larger and flatter. These alterations are enriched in periportal and mid-lobular hepatocytes but not in pericentral hepatocytes. Gain- and loss-of-function in vivo models demonstrate that the Ribosome receptor binding protein1 (RRBP1) is required to enable fasting-induced ER sheet-mitochondria interactions and to regulate hepatic fatty acid oxidation. Endogenous RRBP1 is enriched around periportal and mid-lobular regions of the liver. In obesity, ER-mitochondria interactions are distinct and fasting fails to induce rough ER sheet-mitochondrion interactions. These findings illustrate the importance of a regulated molecular architecture for hepatocyte metabolic flexibility.

Fig. 1. Impact of feeding and fasting on the mitochondria 3D architecture in liver tissue.

A SEM images, 3D segmentation and reconstruction of individual mitochondria in mid-lobular hepatocytes in fed state. Scale bars from left to right: 500 nm, 5 µm, 1 µm, 500 nm, 500 nm. B SEM images, 3D segmentation and reconstruction of individual mitochondria in mid-lobular hepatocytes in fasted state. Scale bars from left to right: 500 nm, 5 µm, 1 µm, 500 nm, 500 nm. C Number of mitochondria from individual hepatocyte volumes in fed (n = 5 cells) and fasted (n = 6 cells) state. D Left: Mitochondria volume; Right: 95% of confidence interval for data in D. E Mitochondria volume frequency distribution. F Left: Mitochondria sphericity; Right: 95% of confidence interval for data in F. G Mitochondria sphericity frequency distribution. H Left: Mitochondria complexity index (MCI²) calculated based on formula described and in methods section; Right: 95% of confidence interval for data in H. I Bivariate plot of volume and MCI² for lean fed (gray) and lean fasted (pink). Each point represents a single mitochondrion. J Left: Quantification of % of mitochondria displaying interaction with lipid droplet (LD); Left: example of LD-mitochondria interaction in fasted state (left). Fed, n = 5 cells and fasted, n = 6 cells. Scale bars: 500 nm. For D–I quantifications were performed in 5 cells from fed data set and 6 cells for fasted conditions. The total mitochondria quantified were n = 14,855 in fed state and n = 6689 in fasted state. Fasted state corresponds to 20 h (overnight) food withdrawal. For the bar graphs, data are shown as mean ± s.e.m.; For C and J, two-tailed unpaired t-test ****p < 0.0001. For D, F, and H, two-tailed unpaired t-test ****p < 0.0001 and permutation test ****p < 0.0001.

Fig. 2. Impact of feeding and fasting on the endoplasmic reticulum (ER) 3D architecture and ER-mitochondria interactions in liver tissue.

A, B Reconstruction of ER network in fed (A) and fasted (B) state (1000 × 1000 × 400 voxels, 8 × 8 × 3.2 µm³). Scale bars: 500 nm. Images show SEM (left) and 3D rendering (right) of FIB-SEM data, rendered in Houdini software. In B, lower panel shows 3D rendering of ER and lipid droplet from FIB-SEM data using Houdini software. C Quantification of length of parallel organized ER stacks per total ER from TEM (n = 5 images per condition). D Ratio of ER tubules per total ER volume quantified from FIB-SEM data (n = 5 cells volumes for fed and n = 6 cell volumes for fasted). E Right: TEM image of mitochondria surrounded by rough ER. Scale bar: 100 nm. Left: quantification of distance between rough ER sheets-mitochondria membranes in TEM images. Number of mitochondria quantified n = 248 fed; n = 300 fasted. F Right: TEM image of mitochondria surrounded by smooth ER. Scale bar: 100 nm. Left: Quantification of distance between smooth ER-mitochondria membranes in TEM images. Number of mitochondria quantified n = 246 fed, n = 183 fasted. G 3D reconstruction of mitochondria volumes marked by areas of interactions with the ER at 0–24 nm distance (red) and 25–56 nm distance (purple) in fed (left, n = 14,855 mitochondria) and fasted (right, n = 6678 mitochondria) conditions. Scale bars: 1 µm. H, I Quantification of the mitochondria surface covered by ER at 24–56 nm distance in fed (n = 14,855 mitochondria) and fasted (n = 6678 mitochondria) conditions. Inset: 95% of confidence interval for data in H. J, K Quantification of the mitochondria surface covered by ER at 0–24 nm distance in fed (n = 14,855 mitochondria) and fasted (n = 6678 mitochondria) conditions. Inset: 95% of confidence interval for data in K. L Quantification of % of mitochondria surface covered by ER sheets at 0-56 nm in a whole hepatocyte volume (n = 2786 mitochondria for fed and n = 1275 mitochondria for fasted). ER sheets were segmented using the previously described algorithm. M 3D rendering of a mitochondrion from fasted cell covered by a single ER sheet (purple). Scale bar: 500 nm. For box-and-whisker plots in C and D, the line inside the box shows the median value. The bounds of the box represent the 25th–75th percentiles, with whiskers at minimum and maximum values, two-tailed unpaired t-test, ***p = 0.0002. For H and J, two-tailed unpaired t-test ****p < 0.0001 and permutation test ****p < 0.0001.

Fig. 3. Hepatic ER and mitochondria architectural organization in fed and fasted state across the liver lobule.

A Illustration of the different hepatic zones highlighting hepatocytes in peri-central region (PC), peri-portal region (PP) and intermediary (IT) region of the liver lobule in between PC and PP. B Representative TEM images from peri-central, intermediary, and peri-portal regions in fed state. Scale bars: 500 nm. C Quantification of mitochondria area from indicated zones in fed (n = 550 for PC, n = 530 for IT, n = 350 for PP) and fasted (n = 419 for PC, n = 231 for IT, n = 524 for PP) conditions. Two-way ANOVA, Tukey’s multiple comparisons test ****p < 0.0001. D Representative TEM images from peri-central, intermediary, and peri-portal regions in fasted state. Scale bars: 500 nm. E, F FIB-SEM images of mitochondria in hepatocytes located in peri-central (E) and peri-portal (F) regions of the liver from mice in fed state. Scale bars: 2 µm. G Quantification of length of parallel organized ER stacks per total ER from TEM of mice in fed (left) and fasted state (right) (n = 5 images per condition). For box-and-whisker plots, the line inside the box shows the median value. The bounds of the box represent the 25th–75th percentiles, with whiskers at minimum and maximum values, Two-tailed unpaired t-test, *p < 0.05, ***p < 0.0003. H, I FIB-SEM images of ER in hepatocytes located in peri-central (H) and peri-portal (I) regions of the liver from mice in fed state. Scale bars: 2 µm. J Quantification of the mitochondria surface covered by ER at 25-80 nm distance in fed and fasted conditions. Fed (n = 549 for PC, n = 529 for IT, n = 322 for PP) and fasted (n = 419 for PC, n = 227 for IT, n = 513 for PP) conditions. Two-way ANOVA, Tukey’s multiple comparisons test ***p < 0.0003 ****p < 0.0001. K Relative frequency distribution of % of mitochondria covered by ER at 25-80 nm in PC (left), IT (middle), PP (right).

Fig. 4. Impact of obesity on fasting-feeding dynamics of ER, mitochondria, and their interactions in liver tissue.

A Number of mitochondria in individual hepatocytes from mice in fed and fasted state (n = 5 cells in each condition). B Left: SEM image (scale bar: 500 nm); Right: 3D segmentation and reconstruction of individual mitochondria in fed state (scale bar: 1 µm). C Left: SEM image (scale bar: 500 nm); Right: 3D segmentation and reconstruction of individual mitochondria in fasted state (scale bar: 1 µm). D Left: Mitochondria volume; Right: 95% of confidence interval for data in D. E Mitochondria volume frequency distribution. F Left: Mitochondria sphericity; Right: 95% of confidence interval for data in F. G Mitochondria sphericity frequency distribution. H 3D Reconstruction of mitochondria volumes marked by areas of interactions with the ER at 0–24 nm distance (red) lean fed, lean fasted, obese fed and obese fasted state conditions. Scale bars: 1 µm. I Quantification of the % of mitochondria surface covered by ER at 0–24 nm distance in the same conditions described in H. J Frequency distribution of % of mitochondria covered by ER at distances indicated in the graphs. K 3D Reconstruction of mitochondria volumes marked by areas of interactions with the ER at 24-56 nm distance (purple) lean fed, lean fasted, obese fed and obese fasted state conditions. Scale bars: 1 µm. L Quantification of the % of mitochondria surface covered by ER at 24-56 nm distance in the same conditions described in K. M Frequency distribution of % of mitochondria covered by ER at distances indicated in the graphs. For (D–G) number of mitochondria are fed, n = 17,721 mitochondria and fasted, n = 12,712 mitochondria. For (I, L), lean fed (n = 3173, 3000, 3109, 2787, 2787), lean fasted (n = 894, 1183, 1275, 1246, 757, 1323), obese fed (n = 2084, 5054, 3854, 3521, 3201) and obese fasted (n = 3250, 2781, 2607, 2065, 1999) states. For (J, M), lean fed (n = 2766), lean fasted (n = 1275), obese fed (n = 3851), obese fasted (n = 2586) states. Obese: leptin-deficient mice, ob/ob. Fasted state corresponds to 20 h (overnight) food withdrawal. For the bar graph in A, data is shown as mean ± s.e.m; For D and F, two-tailed, unpaired t-test ****p < 0.0001 and permutation test ****p < 0.0001.

Fig. 5. RRBP1 expression in liver and primary hepatocytes in fed and fasting conditions.

A Immunoblotting and quantification analysis of RRBP1 protein in total liver lysates of indicated conditions. n = 3 in each group. B Immunostaining of RRBB1 and Glutamine synthetase (GS) in liver sections from lean mice in fed and fasted and from RRBP1 deficient mice. Left panel shows images pseudo colored by ‘Fire’ LUT. Scale bars: 100 µm. C Quantification of the immuno-stained tissue sections shown in B. For the quantification, a line was drawn from the edge of a central vein to the portal vein. The fluorescence intensity of RRBP1 and GS signals across each line was measured by plot profile and the distances were transformed into percentages. n = 9 fed, n = 7 fasted. Representative of 4 independent experiments. For C, two-way ANOVA, Sídák’s multiple comparisons test ****p < 0.0001. D Immunostaining of RRBP1 (green) and mitochondrial OXPHOS proteins (Oxphos cocktail antibody, red) in primary hepatocytes isolated from mice in fed and fasted state. Images were acquired with Lattice-SIM microscopy. Scale bars: 5 µm. E Quantification of RRBP1 fluorescence intensity in proximity with mitochondria. n = 5 cells fed and n = 7 cells fasted. Two-tailed, unpaired t test, p < 0.05. F Immunostaining of RRBP1 (green) and mitochondrial OXPHOS proteins (red) in primary hepatocytes isolated from mice in fed and fasted state. Images were acquired with STED microscopy. Scale bars: 2 µm. Representative image from 7 cells per group. G Lattice-SIM fluorescence images of primary hepatocytes isolated from lean mice in fed state exogenously expressing Adenoviurs-GFP-Sec61β (upper panel) and adenovirus-GFP-RRBP1 (lower panel), representative of three independent experiments. Scale bars: 10 µm. Mitochondria were stained with MitoTracker. For the line graph in C and bar graph in E, data are shown as mean ± s.e.m.

Fig. 6. Impact of hepatic RRBP1 deletion on ER-mitochondria interactions and mitochondria structure and function.

A, B Representative TEM images of livers from wild-type (WT) and RRBP1 deficient (RRBP1-KO) mice in fasted state. Scale bars: 500 nm. C Quantification of mitochondria surface covered by ER between 25-80 nm distance. n = 208 for Wt and n = 311 RRBP1-KO. D–F Quantification of mitochondria area (D), roundness (E) and aspect ratio (AR) (F), n = 208 for Wt and n = 308 RRBP1-KO. G ¹⁴C-palmitic acid-driven fatty acid oxidation in hepatocytes from wild-type control (n = 8) and RRBP1 KO (n = 8). H, I Oleate and palmitate driven lipid loading assay and quantification analysis in primary hepatocytes from wild type (Wt) and RRBP1 deficient cells (n = 107 cells for Wt, n = 133 cells for RRBP1 KO). Lipid droplets were stained with BODIPY (green) and RRBP1 was stained in red. Scale bars: 10 µm. J, K Oleate driven lipid loading assay in primary hepatocytes infected with adenovirus expressing GFP (ad-GFP) and adenovirus expressing RRBP1-GFP (ad-GFP-RRBP1), (GFP n = 6; RRBP1 NT n = 5, GFP Oleate n = 6, RRBP1 Oleate, n = 7). Scale bars: 5 µm. For the bar graphs data are shown as mean ± s.e.m.; For C, E, F, G, I, K, two-tailed unpaired t-test ****p < 0.0001, **p < 0.003.

Fig. 7. Exogenous expression of RRBP1 rescues rough ER—mitochondria interactions in obesity.

A Immunoblotting and quantification analysis of RRBP1 protein in total liver lysates of indicated conditions. n = 3 in each group. B Immunoblot analysis of RRBP1 in total liver lysates from mice exogenously expressing LacZ (n = 2) or GFP-RRBP1 (n = 2) through Adenovirus gene delivery. C Representative TEM images from obese mice exogenously expressing LacZ or RRBP1 though adenovirus (Ad). Scale bars: 500 nm. D FIB-SEM imaging and 3D segmentation and rendering of liver volume from obese mice expressing RRBP1 depicting ER morphology. Scale bar: 500 nm. E FIB-SEM imaging and 3D segmentation and rendering of liver volume from obese mice expressing either LacZ (left) and RRBP1 (right), depicting mitochondria surface covered by ER at 25-56 nm distance. Scale bars: 1 µm. F Quantification of mitochondria surface covered by ER at 25-80 nm distance. G Mitochondria surface covered by ER at 25-80 nm distance frequency distribution. H, I Mitochondria sphericity and frequency distribution in liver tissue from LacZ and RRBP1 overexpression. J, K Mitochondria volume and frequency distribution in liver tissue from LacZ and RRBP1 overexpression. For F–K, n = 23550 mitochondria for LacZ and n = 29235 for RRBP1. Mitochondria were quantified in the entire data set imaged and not segregated by cell. For the bar graphs data are shown as mean ± s.e.m; For F and H, two-tailed unpaired t-test. ****p < 0.0001.