Spatial hepatocyte plasticity of gluconeogenesis during the metabolic transitions between fed, fasted and starvation states

Abstract

Hepatocytes are organized along a spatial axis between the portal triad and the central vein to form functionally repetitive units known as lobules. The hepatocytes perform distinct metabolic functions depending on their location within the lobule. Single-cell analysis of hepatocytes across the liver lobule demonstrates that gluconeogenic gene expression is relatively low in the fed state and gradually increases in the periportal hepatocytes during the initial fasting period. As fasting progresses, pericentral hepatocyte gluconeogenic gene expression and gluconeogenic activity also increase and, following entry into a starvation state, the pericentral hepatocytes show similar gluconeogenic gene expression and activity to the periportal hepatocytes. In parallel, starvation suppresses canonical β-catenin signalling and modulates the expression of pericentral and periportal glutamine synthetase and glutaminase, respectively, resulting in enhanced incorporation of glutamine into glucose. Thus, hepatocyte gluconeogenic gene expression and glucose production are spatially and temporally plastic across the liver lobule, underscoring the complexity of defining hepatic insulin resistance and glucose production on a whole-organ level, as well as for a particular fasted or fed condition.

Extended Data Fig. 1 |. Day-time and night-time feeding does not significantly affect Pck1/Fasn gene expression or metabolic factors in either male or female mice.

(a, b) Relative mRNA levels of Pck1 (A) and Fasn (B) from total hepatocytes in the feeding time course. N = 3 mice per time point/condition. Ordinary one-way ANOVA followed by Dunnett’s multiple comparisons test was performed to compare −4h with different time points. p values of *p < 0.05, **p < 0.01, and ***p < 0.001 was considered statistically significant when comparing two groups. (N.S.=not statistically significant). (c, d) Relative mRNA levels of lipogenic (Fasn) (C) and gluconeogenic (Pck1) (D) genes from total hepatocytes in the fed (0 h), fasted (8 h) and starvation (24 h) state. N = 3 mice per time point/condition. Ordinary one-way ANOVA followed by Dunnett’s multiple comparisons test was performed to compare 0 h with different time points. p values of *p < 0.05, **p < 0.01, and ***p < 0.001 was considered statistically significant when comparing two groups. (N.S.=not statistically significant). (e-m) Glucagon/insulin ratio, Relative mRNA levels, body weight, blood glucose, and glycogen measurements from male/female mice (day-time fed vs. night-time fed trained) in the fasting time course; fed4h (0 h), fast8h (8 h), fast16h (16 h), fast20h (20 h), and fast24h (24 h). N = 3–4 mice per time point/condition. (E) Glucagon/insulin ratio in plasma from day-time fed vs. night-time fed trained male mice. (f, g) Relative mRNA levels of lipogenic (Pck1) (F) and gluconeogenic (Fasn) (G) genes from day-time fed vs. night-time fed trained male mice. (h) Body weight (g) from day-time fed vs. night-time fed trained male mice. (i) Blood glucose (mg/dl) from day-time fed vs. night-time fed trained male mice. (j, k) Relative mRNA levels of lipogenic (Pck1) (J) and gluconeogenic (Fasn) (K) genes from day-time fed vs. night-time fed trained female mice. (l) Body weight (g) from day-time fed vs. night-time fed trained female mice. (m) Blood glucose (mg/dl) from day-time fed vs. night-time fed trained female mice. (n-q) Glycogen content (mg/g) from total hepatocytes in the fasting time course; day-time fed vs. night-time fed trained male mice (N), day-time fed vs. night-time fed trained female mice (O), day-time fed vs. night-time fed untrained male mice (P), and day-time fed vs. night-time fed untrained female mice (Q). N = 3–4 mice per time point/condition. (r-u) Body weight and blood glucose measurements from male/female mice (day-time fed vs. night-time fed untrained) in the fasting time course; fed4h (0 h), fast8h (8 h), fast16h (16 h), fast20h (20 h), and fast24h (24 h). N = 3–4 mice per time point/condition. (R) Body weight (g) from day-time fed vs. night-time fed untrained male mice. (S) Blood glucose (mg/dl) from day-time fed vs. night-time fed untrained male mice. (T) Body weight (g) from day-time fed vs. night-time fed untrained female mice. (U) Blood glucose (mg/dl) from day-time fed vs. night-time fed untrained female mice.

Extended Data Fig. 2 |. Food entrainment effects on the liver local clock and plasma circadian hormone levels.

(A-H) Relative mRNA levels of Bmal1 and Clock from male/female mice in the fasting time course; fed4h (0 h), fast8h (8 h), fast16h (16 h), fast20h (20 h), and fast24h (24 h). N = 3–4 mice per time point/condition. Results aligned to zeitgeber time (ZT); ZT0–12 had the light on, whereas ZT12–0 had the light off. (a, b) Relative mRNA levels of Bmal1 from day-time fed vs. night-time fed trained male (A) and female (B) mice. (c, d) Relative mRNA levels of Clock from day-time fed vs. night-time fed trained male (C) and female (D) mice. (e, f) Relative mRNA levels of Bmal1 from day-time fed vs. night-time fed untrained male (E) and female (F) mice. (g, h) Relative mRNA levels of Clock from day-time fed vs. night-time fed untrained male (G) and female (H) mice. (i-l) Corticosterone and Growth Hormone levels in plasma from male mice in the fasting time course; fed4h (0 h), fast8h (8 h), fast16h (16 h), fast20h (20 h), and fast24h (24 h). N = 3–4 mice per time point/condition. Results aligned to zeitgeber time (ZT); ZT0–12 had the light on, whereas ZT12–0 had the light off. (I, J) Corticosterone and growth hormone levels in plasma from day-time fed vs. night-time fed trained (I) and untrained (J) male mice. (K, L) Growth hormone levels in plasma from day-time fed vs. night-time fed trained (K) and untrained (L) male mice.

Extended Data Fig. 3 |. Compass analyses of several gene pathways that are not altered in the fed, fasted or starvation state.

(a, b) Compass-score differential activity analysis of the Vitamin A pathway between 0 h and 8 h pericentral/periportal hepatocytes (A) and between 0 h and 24 h pericentral/periportal hepatocytes (B) based on Cohen’s D analysis. BCDO (BCMO1) is highlighted. (c, d) Compass-score differential activity analysis of the Vitamin B6 pathway between 0 h and 8 h pericentral/periportal hepatocytes (C) and between 0 h and 24 h pericentral/periportal hepatocytes (D) based on Cohen’s D analysis. PNPO is highlighted. (e, f) Compass-score differential activity analysis of the Starch and sucrose (glycogen) pathway between 0 h and 8 h pericentral/periportal hepatocytes (E) and between 0 h and 24 h pericentral/periportal hepatocytes (F) based on Cohen’s D analysis. GLPASE (PYGL) is highlighted. (g) Representative immunoblotting of BCMO1, PNPO, PYGL, CYP2E1, E-Cadherin, and Vinculin from pericentral and periportal isolated hepatocytes.

Extended Data Fig. 4 |. Quantitated target scRNA-seq analyses for zonated gene expression.

(a) Sequencing saturation plot of each library. X-axis represents mean reads per cell; Y-axis represents percentage of transcripts sequenced; dotted line is the approximate saturation point. (b) Violin plots of Log2 expression Pck1 and Fasn comparing standard and targeted scRNA-seq in 0 h and 16 h. (c) Ridge plot analysis of gluconeogenic (Pck1, G6pc) and lipogenic (Acly, Fasn) genes from targeted scRNA-seq. X-axis represents log2 expression; Y-axis represents each time point/condition, red dotted line shows the threshold for cells positive of each gene. (d) MFuzz analysis of the targeted scRNA-seq to cluster similar expression profiles. Once percentage of positive cells were defined based on ridge plot analysis (Fig. S4C), each gene expressed in pericentral and periportal hepatocytes were assigned to 6 main clusters; early fasting induced, late fasting induced, cycling, early fasting suppressed, late fasting suppressed, constitutively expressed. X-axis represents time points; Y-axis represents percentage of positive cells.

Extended Data Fig. 5 |. Quantification of Pck1 and Fasn transcription sites by spatial smFISH analyses.

(a, b) Representative smFISH images of pericentral (PC) and periportal (PP) areas selected for quantification of TS (Pck1 (A), and Fasn (B)) from 0 h, 4 h, and 24 h in male mice. TS are highlighted with arrows. (c, d) Histogram based on the intensity of each TS (Pck1 (C), and Fasn (D)). X-axis represents the intensity of TS (A.U.); Y-axis represents the number of TS within each intensity. Dotted line shows the mean value of the intensity of TS with SD.

Extended Data Fig. 6 |. smFISH spatial analyses of Pck1 and Fasn expression in female mice livers.

(a, b) Representative smFISH images of pericentral (PC) and periportal (PP) areas from 0 h, 4 h, and 24 h in female mice. TS are highlighted with arrows. (c-h) Relative mRNA levels of zonation markers (Cyp2e1 pericentral (C), Cyp2f2 periportal (D)), gluconeogenic genes (Pck1 (E), G6pc (F)), and lipogenic genes (Fasn (G), Acly (H)) from pericentral and periportal isolated hepatocytes in female mice. N = 3 mice for 8hPC/PP and 24hPP, N = 4 mice for 0hPC/PP and 24hPC. Data are presented as mean values +/− SD. Student’s t-test with unpaired two-tailed p values of *p < 0.05, **p < 0.01, and ***p < 0.001 was considered statistically significant when comparing two groups. (N.S.=not statistically significant).

Extended Data Fig. 7 |. Identification of nutrition-independent zonation mRNA markers that are unchanged in the fed, fasted and starvation states.

(a) Principal Component Analysis (PCA) of RNA-seq from pericentral and periportal isolated hepatocytes. Each dot represents each time point/condition based on coloring and shape of dot (0 h=red, 8 h=yellow, 24 h=blue; circle=pericentral, triangle=periportal). Green arrow shows how fasting shift the pericentral hepatocytes. (b) The number of differentially expressed genes in pericentral and periportal hepatocytes throughout the fed (0 h), fasted (8 h) and starvation (24 h) state in the RNA-seq from pericentral and periportal isolated hepatocytes. Differentially expressed genes were defined based on log2Fold Change <−1 for pericentral, >1 for periportal, both with a cutoff of p-adj<0.05. (c, e) Venn diagram of differentially expressed genes in pericentral (C) and periportal (E) hepatocytes throughout the fed (0 h), fasted (8 h) and starvation (24 h) state in the RNA-seq from pericentral and periportal isolated hepatocytes. (d, f) Metascape analysis on nutrition-independent upregulated pericentral genes (D) and periportal genes (F). Top 5 pathways in each zone shown (−log10(P)).

Extended Data Fig. 8 |. The entire mass isotopologue distribution (MID) of glucose from [U-¹³C] pyruvate or glutamine.

(a) Steady state gluconeogenic time course experiment from [U-¹³C]-pyruvate using total hepatocytes isolated from 24 h fasted mice. N = 3 mice. Glucose isotopologues with one ¹³C to six ¹³C carbons (M1 thru M6 isotopologues) are shown. Data are presented as mean values + SD. (b) Fed, fasting and starvation response of glucose isotopologues with one ¹³C to six ¹³C carbons (M1 thru M6 isotopologues) derived from [U-¹³C]-pyruvate using pericentral and periportal isolated hepatocytes. N = 4 for all conditions except 0hPC, 8hPP, 16hPP, and 24hPC (N = 3). Data are presented as mean values + SD. (c) Fasted 24 hr response of glucose isotopologues with one to three ¹³C carbons (% enrichment) derived from [U-¹³C]-glutamine comparing pericentral vs. periportal isolated hepatocytes using Control vs GLS2KO mice 24 h mice. N = 3 mice per condition. Data are presented as mean values + SD.

Extended Data Fig. 9 |. Canonical WNT target gene expression decreases in pericentral hepatocytes in the starvations state.

(a, b) RNA-seq data (counts) of pericentral zonation markers (Gulo and Cyp2e1) and periportal zonation markers (Hsd17b13 and Cyp2f2) (A), gluconeogenic genes (Pck1, G6pc) and lipogenic genes (Acly, Fasn) (B) from pericentral and periportal isolated hepatocytes. N = 3 mice per time point/condition. Data are presented as mean values +/− SD. Student’s t-test with unpaired two-tailed p values of *p < 0.05, **p < 0.01, and ***p < 0.001 was considered statistically significant when comparing two groups. (N.S.=not statistically significant). (c) Heatmap of WNT target genes in RNA-seq from pericentral and periportal isolated hepatocytes. Average expression of N = 3 mice from each time point/condition is shown. Coloring based on z-score. 4 main clusters show PC upregulated/fast reduced, fast induced, PP upregulated fast reduced, fast reduced (D) RNA-seq data (counts) of WNT signaling related genes from pericentral and periportal isolated hepatocytes. N = 3 mice per time point/condition. Data are presented as mean values +/− SD. Student’s t-test with unpaired two-tailed p values of *p < 0.05, **p < 0.01, and ***p < 0.001 was considered statistically significant when comparing two groups. (N.S.=not statistically significant).

Extended Data Fig. 10 |. HIPPO/YAP and sonic hedgehog zonation gene pathway changes in the fed, fasted and starvation states.

(a) Heatmap of Yap target genes in RNA-seq from pericentral and periportal isolated hepatocytes. Average expression of N = 3 mice from each time point/condition is shown. Coloring based on z-score. (b) RNA-seq data (counts) of Yap related genes from pericentral and periportal isolated hepatocytes. N = 3 mice per time point/condition. Data are presented as mean values +/− SD. Student’s t-test with unpaired two-tailed p values of *p < 0.05, **p < 0.01, and ***p < 0.001 was considered statistically significant when comparing two groups. (N.S.=not statistically significant). (c) Heatmap of Shh target genes in RNA-seq from pericentral and periportal isolated hepatocytes. Average expression of N = 3 mice from each time point/condition is shown. Coloring based on z-score. (d) RNA-seq data (counts) of Shh related genes from pericentral and periportal isolated hepatocytes. N = 3 mice per time point/condition. Data are presented as mean values +/− SD. Student’s t-test with unpaired two-tailed p values of *p < 0.05, **p < 0.01, and ***p < 0.001 was considered statistically significant when comparing two groups. (N.S.=not statistically significant).

Fig. 1 |. Time course of scRNA-seq of hepatocytes in the fed, fasted and starvation states.

a, Scheme of the feeding and fasting protocol. b, UMAP visualizations of the standard scRNA-seq based on feeding conditions: fed 4 h (0 h), fast 4 h (4 h), fast 8 h (8 h), fast 16 h (16 h) and fast 24 h (24 h) (n = 1 mouse per time point, n = 50,000 hepatocytes total; each dot represents one cell). c, Pericentral and periportal hepatocyte identification based on the expression levels of pericentral zonation markers (Cyp2e1 and Gulo) and periportal zonation markers (Cyp2f2 and Hsd17b13). d, Heatmap of metabolism-related KEGG pathways. Glycolysis/gluconeogenesis genes are in bold, fatty acid biosynthesis genes are in italics and fatty acid elongation genes are underlined. Pericentral (PC) and periportal (PP) hepatocytes are clustered based on panel c. Genes with low average counts are not included. e,f, UMAP plots of gluconeogenic genes (Pck1, G6pc) and lipogenic genes (Fasn, Acly) in the fed (0 h), fasted (8 h) and starvation (24 h) states.

Fig. 2 |. Compass analyses of the temporal changes of hepatocyte scRNA-seq in the fed, fasted and starvation states.

a–d, Compass analysis comparing pericentral and periportal fed (0 h, red), fasted (8 h, yellow) and starvation (24 h, blue) states on standard scRNA-seq. Each reaction (dot) is coloured and plotted based on Cohen’s d statistic and by Recon2 pathways. The top ten activated pathways in each condition are shown (rankings based on Cohen’s d). Comparison of differential activity of metabolic reactions between 0 h and 8 h periportal hepatocytes (a), 0 h and 8 h pericentral hepatocytes (b), 0 h and 24 h periportal hepatocytes (c) and 0 h and 24 h pericentral hepatocytes (d). e,f, Compass-score differential activity analysis of all pathways between 0 h and 8 h pericentral and periportal hepatocytes (e) and between 0 h and 24 h pericentral and periportal hepatocytes (f) based on Cohen’s d analysis. g,h, Compass-score differential activity analysis of the glycolysis/gluconeogenesis pathway between 0 h and 8 h pericentral and periportal hepatocytes (g) and between 0 h and 24 h pericentral and periportal hepatocytes (h) based on Cohen’s d analysis. PEPCK (Pck1) is highlighted.

Fig. 3 |. Quantitative targeted scRNA-seq analyses of hepatocytes in the fed, fasted and starvation states.

a, UMAP visualizations of the targeted scRNA-seq based on feeding conditions: fed 4 h (0 h), fast 8 h (8 h), fast 16 h (16 h), fast 24 h (24 h) and fast 30 h (30 h) (n = 1 mouse per time point, n = 47,473 hepatocytes total; each dot represents one cell). b,c, Pericentral and periportal hepatocyte identification based on the expression levels of pericentral zonation markers (Cyp2e1 and Gulo) and periportal zonation markers (Cyp2f2 and Hsd17b13). d,e, Quantitative analysis based on the percentage of gluconeogenic (Pck1, G6pc) (d) and lipogenic (Fasn, Acly) (e) gene-positive cells in the pericentral and periportal hepatocytes throughout the feeding and fasting time course. Data are presented as mean ± 99% confidence intervals, estimated by bootstrapping. Asterisks indicate significance when comparing two groups (***P < 0.001); N.S., not significant.

Fig. 4 |. Spatial imaging by smFISH of Pck1 and Fasn mRNA in the fed, fasted and starvation state.

a,d, smFISH image of Pck1 (a) and Fasn (d) in the fed (0 h), fasted (4 h) and starvation (24 h) states (representative image shown for each time point; n = 3 mice per time point). Lobules always show the central vein (CV) on the left and the portal vein (PV) on the right. Liver lobule images were taken by obtaining sequential images and stitching them together using FIJI. Magnified images show nuclei that have TSs. b,c, Quantitative analysis of TSs by FISH-quant (minimum of 800 nuclei in six veins were analysed per time point and zone) based on the percentage of Pck1 TS (b) and Fasn TS (c) positive nuclei in the pericentral and periportal hepatocytes throughout the feeding and fasting time course. The y axis represents the percentage of TS positive nuclei (black-outlined dots show average of n = 3; other dots represent each mouse). Statistics based on ordinary two-way ANOVA followed by Sidak’s multiple comparisons test; asterisks indicate significance (**P < 0.01 and ***P < 0.001) when comparing each time point; data are presented as mean values; error bars, s.d.

Fig. 5 |. Bulk RNA-seq analyses of isolated pericentral and periportal hepatocytes in the fed, fasted and starvation states.

a–f, Relative mRNA levels of zonation markers (Cyp2e1 pericentral (a), Cyp2f2 periportal (b)), gluconeogenic genes (Pck1 (c), G6pc (d)) and lipogenic genes (Fasn (e), Acly (f)) from pericentral and periportal isolated hepatocytes; n = 3–6 mice per time point and condition (n = 3 for 8 h, n = 4 for 0 h and 24 h in Cyp2e1, Cyp2f2, Fasn and Acly; n = 6 for 0 h and 24 h in Pck1 and G6pc). Data are presented as mean values; error bars, s.d. Comparison between groups was made with unpaired, two-tailed Student’s t-tests; asterisks indicate significance (*P < 0.05, **P < 0.01, ***P < 0.001). g–i, Heatmap of metabolic-related KEGG pathways of glycolysis/gluconeogenesis genes (g), fatty acid biosynthesis genes (h), fatty acid elongation genes (i) in RNA-seq from pericentral and periportal isolated hepatocytes. Average expression of n = 3 from each time point or condition is shown. Colouring based on z-score. Genes with low average counts are not included. j, Representative immunoblotting of PEPCK, CYP2E1, E-cadherin and vinculin from pericentral and periportal isolated hepatocytes. k, Quantification of PEPCK by vinculin (n = 3 mice per time point or condition). Data are presented as mean values; error bars, s.d. Comparison between groups was made with unpaired, two-tailed Student’s t-test (*P < 0.05). Gels and blots were processed in parallel to quantify all samples.

Fig. 6 |. Incorporation of labelled pyruvate carbons into glucose increases in both pericentral and periportal hepatocytes when the liver enters the starvation state.

a, Structure of glucose after acetic anhydride derivatization (for GC–MS). b, Structures of citrate and pyruvate after methoximation and MTBSTFA derivatization. c, Flux diagram of [U-¹³C]-pyruvate flowing into either PC or PDH on its way to citrate and clockwise transport in the TCA cycle, and via PEPCK to glucose (gluconeogenesis). d, $\mathrm{\Sigma }{\mathrm{M}}_n$ (average ¹³C carbons per molecule; see Methods) of glucose during the gluconeogenic time course (to steady state) experiment from [U-¹³C]-pyruvate to glucose in total hepatocytes isolated from 24 h-fasted mice, for n = 3 mice. Data are presented as mean values; error bars, s.d. Ordinary one-way ANOVA followed by Dunnett’s multiple comparisons test was performed to compare 30 min with different time points. e, $\mathrm{\Sigma }{\mathrm{M}}_n$ (average ¹³C carbons per molecule) of glucose derived from gluconeogenesis from [U-¹³C]-pyruvate during the transition between the fed, fasting and starvation states using pericentral and periportal isolated hepatocytes (n = 4 for all conditions except 0 h PC, 8 h PP, 16 h PP and 24 h PC (n = 3)). Data are presented as mean values; error bars, s.d. Comparison between groups was made with unpaired, two-tailed Student’s t-test (*P < 0.05, **P < 0.01). f, Pyruvate isotopologues with one to three ¹³C carbons (% enrichment) from [U-¹³C]-pyruvate during the fed, fasting and starvation time course in pericentral and periportal isolated hepatocytes at the end of the experiment (n = 4 for all conditions except 0 h PC, 8 h PP, 16 h PP and 24 h PC (n = 3)). Data are presented as mean values; error bars, s.d. g, Percentage (%) of ¹³C enrichment in citrate isotopologues derived from [U-¹³C]-pyruvate during gluconeogenic experiments using pericentral and periportal isolated hepatocytes in the fed (0 h) and starvation (20 h) state (m2 and m3 enrichment shown; n = 4 for all conditions except 0 h PC (n = 3)). Data are presented as mean values; error bars, s.d. Comparison between groups was made with unpaired, two-tailed Student’s t-test (*P < 0.05, **P < 0.01, ***P < 0.001). h, PC/PDH flux ratio comparing the responses from pericentral and periportal isolated hepatocytes in the fed (0 h) and starvation (20 h) states. The estimation of the PC/PDH flux ratio was based on the intensity ratios of m + 3 (PC-derived) and m + 2 (PDH-derived) citrate isotopologues (n = 4 for all conditions except 0 h PC (n = 3)). Data are presented as mean values; error bars, s.d. Ordinary one-way ANOVA followed by Tukey’s multiple comparisons test was performed.

Fig. 7 |. Fasting and starvation redistributes expression of glutamine synthetase and glutaminase.

a, Upstream regulator analysis with standard scRNA-seq on 0 h, 8 h and 24 h PC clusters. Signalling pathways ranked based on z-score. b–d, RNA-seq data (counts) of Gls2 (b), Glul (c) and Glud1 (d) from pericentral and periportal isolated hepatocytes; n = 3 mice per time point or condition. Data are presented as mean values; error bars, s.d. Comparison between groups was made with unpaired, two-tailed Student’s t-test (*P < 0.05, ***P < 0.001). e, Representative immunoblotting of GLS2 and vinculin from pericentral and periportal isolated hepatocytes. f, Representative immunofluorescence (IF) images of glutamine synthetase (GS) from fed (0 h) and fast 24 h (24 h). g, Quantification of GS immunofluorescence (y axis represents mean pixel intensity; dots represent n = 3 mice per time point). Data are presented as mean values; error bars, s.d. Comparison between groups was made with unpaired, two-tailed Student’s t-test (*P < 0.05). h,i, Proteomics (normalized intensity) of GS (h) and GLS2 (i) from total hepatocytes (n = 1 per time point, dots represent technical replicates). j–n, Relative mRNA levels of Gls2 (j), Glul (k), Glud1 (l) and zonation markers (Cyp2e1 pericentral (m), Cyp2f2 periportal (n)) from pericentral and periportal isolated hepatocytes in AAV-Tbg.Cre-NT-pgRNA (control) or AAV-Tbg.Cre-Gls2-pgRNA (Gls2-KO) injected Cas9-KI mice. Mice were fasted for 24 h; n = 3 mice per condition. Data are presented as mean values; error bars, s.d. Comparison between groups was made with unpaired, two-tailed Student’s t-test (*P < 0.05, **P < 0.01, ***P < 0.001). o, Representative immunoblotting of GLS2, CYP2E1, E-cadherin and vinculin from pericentral and periportal isolated hepatocytes in control or Gls2-KO mice fasted for 24 h. p, $\mathrm{\Sigma }{\mathrm{M}}_n$ (average ¹³C carbons per molecule) of glucose from [U-¹³C]-glutamine comparing pericentral versus periportal isolated hepatocytes from control or Gls2-KO mice after a 24 h fasting (dots represent three mice per condition). Data are presented as mean values; error bars, s.d. Comparison between groups was made with unpaired, two-tailed Student’s t-test (*P < 0.05).