MitoTracker Red for Isolation of Zone-Specific Hepatocytes and Characterization of Hepatic Sublobular Metabolism

Abstract

Background:

The liver lobule is divided into three zones or regions: periportal (PP or Zone 1) that is highly oxidative and active in ureagenesis, pericentral (PC or Zone 3) that is more glycolytic, and midzonal (MZ or Zone 2) with intermediate characteristics. Here, our Aim was to isolate and metabolically characterize hepatocytes from specific sublobular zones.

Methods:

Mice were administered rhodamine123 (Rh123) or MitoTracker Red (MTR) prior to intravital imaging, liver fixation, or hepatocyte isolation. After in vivo MTR, hepatocytes were isolated and sorted based on MTR fluorescence intensity, or E-cadherin (Ecad) and cytochrome P450 2E1 (CYP2E1) immunolabeling was performed in fixed liver slices. Ecad and CYP2E1 gene expression in sorted hepatocytes was assessed by qPCR. Oxygen consumption rates (OCR) of sorted hepatocytes were also assessed.

Results:

Multiphoton microscopy showed Rh123 and MTR fluorescence distributed zonally, decreasing from PP to PC in a flow-dependent fashion. In liver cross-sections, Ecad was expressed periportally and CYP2E1 pericentrally in association with high and low MTR labeling, respectively. Based on MTR fluorescence, hepatocytes were sorted into PP, MZ, and PC populations with PP and PC hepatocytes enriched in Ecad and CYP2E1, respectively. OCR of PP hepatocytes was ~4 times that of PC hepatocytes.

Conclusions:

MTR treatment in vivo delineates sublobular hepatic zones and can be used to sort hepatocytes zonally. PP hepatocytes have substantially greater OCR compared to PC and MZ. The results also indicate a sharp midzonal demarcation between hepatocytes with PP characteristics (Ecad) and those with PC features (CYP2E1). This new method to sort hepatocytes in a zone-specific fashion holds the potential to shed light on sublobular hepatocyte metabolism and regulatory pathways in health and disease.

Graphical Abstract

1. Introduction

The liver manages crucial functions in the body, including glucose metabolism, xenobiotic metabolism and detoxification, synthesis of plasma proteins, and biliary excretion, among others. Mammalian liver is organized into roughly hexagonal lobules that have intralobular heterogeneity in gene expression and function[2]. Heterogeneity arises in part from an intralobular oxygen gradient as oxygen is extracted during the transit of blood through the lobule leading to lower oxygen in pericentral (PC) regions than periportal (PP) regions [3,4]. Zonation is an important aspect of liver physiology and significantly impacts metabolism [5–11].

Based on the original proposal by Rappaport, three equally sized lobular zones or regions are generally recognized [12–14]. Zone 1 surrounds portal tracts and is highly oxidative and active in ureagenesis, whereas Zone 3 around central veins is more glycolytic. Zone 2 is between Zones 1 and 3 and is considered to have intermediate metabolism. Midzonal (MZ) hepatocytes of Zone 2 possess the highest proliferative activity and have recently been identified as having a primary role in liver regrowth [15]. PP hepatocytes situated near portal triads also primarily engage in nutrient absorption, gluconeogenesis, and ureagenesis. In contrast, PC hepatocytes closer to central veins specialize in glycolysis, lipogenesis, bile acid synthesis, and glutamine synthesis [5]. Overall, this spatial heterogeneity results in metabolic gradients within the hepatic lobule that influence the regulation of numerous hepatic functions.

Considerable research has contributed to our understanding of liver lobular heterogeneity in gene expression and metabolism, but further inquiry would benefit by use of live hepatocytes harvested from specific hepatic zones. Here, we describe a rapid practical method for sorting isolated hepatocytes into distinct sublobular populations by flow cytometry based on in vivo zone-dependent labeling with MitoTracker Red (MTR). Using such sorted hepatocytes, we characterize differences in oxygen metabolism between the different zones.

2. Materials & Methods

2.1. Animals

Male C57BL/6 mice (9–12 wk old) were purchased from Jackson Laboratories (Bar Harbor, ME, USA). All animals were given humane care using protocols approved by the Animal Care and Use Committee of the Medical University of South Carolina.

2.2. Loading of mitochondrial membrane potential-indicating fluorophores

At 1 h prior to liver fixation, intravital multiphoton microscopy, or hepatocyte isolation, mice were injected by tail vein with the covalent mitochondrial membrane potential (ΔΨ) indicator, MitoTracker Red CMXRos (MTR) (M7512, Thermo Fisher Scientific, Waltham, MA, USA) at either 0.05 or 0.5 μmol/mouse in 0.1 ml of saline. In some experiments, 0.01 μmol/liver of MTR in 10 mL phosphate buffered saline (PBS), pH 7.4 was infused retrogradely via the infrahepatic vena cava with clamping of suprahepatic vena cava at a rate of 2 mL/min followed by fixation with 4% paraformaldehyde (PFA) in PBS. For multiphoton microscopy, a second mitochondrial ΔΨ indicator, green-fluorescing rhodamine 123 (Rh123) (R302,Thermo Fisher Scientific), was administered via tail vein at 0.05 μmol/mouse 20 min prior to imaging [16].

2.3. Intravital multiphoton microscopy

After injection of fluorescent probes, anesthetized mice were laparotomized and placed in a prone position, as described [17–19]. An endotracheal intubation was performed using a 20-gauge catheter and connected to a respirator to control respiration during imaging. The liver was gently withdrawn from the abdominal cavity and placed over a No. 1.5 glass coverslip mounted on the stage of an inverted Olympus Fluoview1200 MPE multiphoton microscope (Olympus, Center Valley, PA, USA) equipped with a 30X 1.05 N.A. silicone oil immersion objective lens and a Spectra Physics Mai Tai Deep Sea tunable multiphoton laser (Newport, Irvine, CA). Rh123 and MTR fluorescence was imaged simultaneously using 840-nm multiphoton excitation. During image acquisition, the respirator was turned off 5–10 sec to eliminate breathing movement artifacts. Emission light was collected from planes of 10, 50, and 100 μm deep into the tissue and filtered through 495–540-nm (Rh123) and 575–630-nm (MTR) barrier filters. Pericentral and periportal areas were identified by Rh123 and MTR fluorescent intensity as well as by vasculature at 50 and 100-μm depths.

2.4. Hepatocyte Isolation

Primary hepatocytes were isolated from C57BL/6 mice by retrograde collagenase (C5138, Sigma-Aldrich, St. Louis, MO, USA) perfusion through the inferior vena cava, as described previously [20], and resuspended in modified Krebs-Ringer-Hepes buffer (KRH) containing: 115 mM NaCl, 5 mM KCl, 1 mM KH₂PO₄, 1.2 mM MgSO₄, 2 mM CaCl₂, 25 mM HEPES, and 2% bovine serum albumin (BSA,126575, Sigma Aldrich), pH 7.4. Cell viability was greater than 85% by trypan blue exclusion. The cell suspension was diluted to 8×10⁶ cells/mL for cell sorting or 1×10⁶ cells/mL for subsequent plating and imaging. For cell sorting, isolated hepatocytes were incubated 30 min in Fixable Violet LIVE/DEAD (L34963, Thermo Fisher Scientific) as a cell viability indicator during flow cytometry.

In some experiments, MTR was administered in vivo at 0.05 μmol/mouse prior to isolation of hepatocytes. For imaging, 10⁶ hepatocytes were plated on collagen-coated (C7661, Sigma-Aldrich) No. 1.5 coverslips in 35-mm Petri dishes (P35G-1.5–10-C, MatTek Life Sciences, Ashland, MA, USA) and centrifuged for 1 min at 65 x g, 4°C to adhere cells to the plate surface. In other experiments, hepatocytes were isolated and loaded in vitro with 200 nM MTR at 37°C in KRH for 30 min. Hepatocytes loaded in vivo or in vitro with MTR were subsequently loaded 30 min with 200 nM Rh123. Afterwards, a maintenance concentration of 50 nM Rh123 was used.

2.5. Immunohistochemistry

In anesthetized mice after clearing blood by perfusion with PBS at 37°C, livers were fixed by perfusion with 4% PFA in PBS for 5 min and placed in fixative at 4° C. The following day, liver tissues were mounted in 5% agarose (1613102, Bio-Rad, Hercules, CA, USA), and 100-μm sections were prepared with a VT1000P vibratome (Leica Biosystems, Nussloch, Germany). Subsequent steps at room temperature were separated by 3 times rinsing in PBS for 5 min. Blocking of liver sections was performed with 4% BSA in PBS for 1 h at room temperature. Sections were then incubated with anti-cytochrome P450 2E1 (CYP2E1) antibody (ab28146, Abcam, Waltham, MA, USA) for 72 h and/or anti-E cadherin (Ecad) antibody (ab11512, Abcam) to mark PP and PC regions, respectively. Slices were then incubated with Alexa Fluor^™ 488 goat anti-rat IgG (H+L) (A-11006, Thermo Fisher Scientific) and Alexa Fluor^™ 647 donkey anti-rabbit IgG (H+L) (A-31573, Thermo Fisher Scientific) at a dilution of 1:1,000 overnight at 4°C. For nuclear labeling, samples were incubated with a 1:2,000 dilution of 10 mg/mL Hoechst 33342 (62249, Thermo Fisher Scientific) for 20 min before coverslipping on slides with Prolong Diamond Antifade Mountant (P36961, Thermo Fisher Scientific) overnight to harden. Immuno-stained sections were imaged with a LSM 880 confocal microscope (Zeiss, Oberkochen, Germany) in super-resolution Airyscan mode. Tiled images were created from 64 images (2048×2048) of adjacent fields and further deconvolved with Zen software (black edition) v10.0.0.910 (Zeiss). Images of kidney, ileum and heart were deconvolved with Huygens Professional v23.10.0 (Scientific Volume Imaging, Hilversum, North Holland, Netherlands) software.

In other experiments, hepatocytes plated as described above were fixed with 4% PFA in PBS for 5 min. With PBS rinsing between steps, cells were permeabilized with 0.1% Triton-X-100 (HFH10, Thermo Fisher Scientific) in PBS for 10 min at 37°C followed by blocking with 4% BSA in PBS for 1 h at room temperature. Cells were then incubated with anti-CYP2E1 antibody at a dilution of 1:1,000 overnight at 4°C followed by Alexa Fluor^™ 488 goat anti-rabbit IgG (H+L) (A-11006, Thermo Fisher Scientific) at a dilution of 1:1,000 for 1 h in room temperature. Excess antibodies were washed out and replaced with PBS.

2.6. Flow cytometry

Hepatocytes isolated 1 h after tail vein injection of MTR (0.05 μmol/mouse) were sorted into three populations at 4°C based on fluorescence excited with a 561-nm laser using a BD FACS Aria III Flow Cytometer (BD Biosciences, Franklin Lakes, NJ, USA), 130-μm nozzle, and 1.5 neutral density filter with laser compensation manually corrected. To collect three populations, cells were sorted through four nested gates (see Section 3.5 and Fig. 6).

Fig. 6. Sorting of periportal, midzonal and pericentral hepatocytes based on MitoTracker labeling.

A mouse was injected by tail vein with MTR (0.05 μmol). After 1h, hepatocytes were isolated and subjected to cell sorting. (A) FSC-A and SSC-A were used for size selection of hepatocytes. (B) FSC-A and FSC-H were used to select hepatocyte singlets. (C) Fixable Violet LIVE/DEAD fluorescence was used to exclude non-viable cells. (D) The PP (gray), MZ (blue) and PC (red) populations were sorted corresponding to the highest, median, and lowest fluorescence intensity quintiles, respectively. (E) Distribution of the sorted cells by MTR intensity. (F) Normalized gene expression of Cyp2e1 and Cdh1 in PP, MZ, and PC cells. *, p< 0.05 vs PP, †, p < 0.05 MZ). Sort distributions of all mice were similar (n = 5).

2.7. Quantitative PCR Analysis

Total RNA was extracted from sorted hepatocytes using a Monarch^® Tola RNA Miniprep Kit (T2010S, New England Biolabs, Ipswich, MA, USA), reverse-transcribed using a High-Capacity cDNA Reverse Transcription Kit (4368814, Applied Biosystems, Waltham, MA, USA), and amplified with a C1000 Touch Thermal Cycler (Bio-Rad). qPCR plates were prepared using PowerUp^™ SYBR^™ Green Master Mix for qPCR (Applied Biosystems), and qPCR was performed using a CFX96 Touch Real-Time PCR Detection System (Bio-Rad) and normalized to glyceraldehyde 3-phosphate dehydrogenase (Gapdh). Custom primers for genes encoding Ecad (Cdh1) and CYP2E1 (Cyp2e1) are listed below (Integrated DNA Technologies, Coralville, IA, USA). All kits were used according to the manufacturer’s instructions. Primer sequences are listed in Table 1.

Table 1.

Primer sequences for qPCR analysis.

Name	Sequences (5' to 3')
Cyp2e1 - F	GGGGACATTCCTGTGTTCCA
Cyp2e1 - R	TTAGGGAAAACCTCCGCACG
Cdh1 - F	GAGACCAGTTTCCTCGTCCG
Cdh1 - R	CCATGGACTTCAGCGTCACT
Gapdh - F	GAAGGTCGGTGTGAACGGAT
Gapdh - R	AATCTCCACTTTGCCACTGC

2.8. Oxygen Consumption Rate Analysis

Sorted hepatocytes were plated on collagen-coated, 96-well Seahorse analyzer plates (103794–100, Agilent, Santa Clara, CA, USA) at a density of 5,000 cells per well in Beeson’s reduced serum (RS) buffer containing: 130 mM NaCl, 5.3 mM KCl, 0.5 mM Na₂HPO₄, 0.5 mM KH₂HPO₄, 1.8 mM CaCl₂, 0.6 mM MgCl₂, 1 mM sodium pyruvate, 10 mM glucose, 2 mM glutamine, 0.1% BSA, 1X MEM non-essential amino acids (11140076, Thermo Fisher Scientific), 1X MEM amino acids (11130051, Thermo Fisher Scientific) and 1X MEM vitamins (11120052, Thermo Fisher Scientific), pH 7.4 [21,22]. Seahorse plates were centrifuged for 1 min at 65 x g, 4°C to adhere cells. Oxygen consumption rates (OCR) of purified populations of hepatocytes were measured using a Seahorse XFe96 Analyzer (Agilent) at 37°C. Mitochondrial stress tests were performed from OCR measurements before and after sequential addition of 1 μM oligomycin (11342, Cayman Chemical, Ann Arbor, MI, USA), 1 μM carbonyl cyanide 4-(trifluoromethoxy) phenylhydrazone (FCCP, ab120081, Abcam), and 1 μM rotenone (557368, Sigma Aldrich)/1 μM antimycin a (A8674, Sigma Aldrich). Individual wells were imaged with TSView v6.1.3.2 software (Informer Technologies, Inc., Los Angeles, CA, USA) using a Nikon TMS-F inverted microscope (Nikon Corporation, Minato City, Tokyo, Japan) with a 4X objective lens. Cells situated within the three nodes located at the bottom of the Seahorse wells were quantified using ImageJ ’Analyze Particle’ function [23]. OCR values were corrected based on cell count in units of pmol/min/10³ cells.

2.9. Statistical analysis

All statistical analysis was calculated using GraphPad Prism v10.1.0 for Windows (GraphPad Software, Boston, MA, USA). Data are presented as means ± S.E. Differences in fluorescence were analyzed by the student’s t-test. Differences in gene expression and OCR between population subsets were calculated by one-way ANOVA. Linear regression analysis was performed to assess correlation of markers in primary cell culture. P < 0.05 was the criterion of significance. Images shown are representative of 3 or more mice.

3. Results

3.1. Intravital multiphoton microscopy shows lobular gradients of rhodamine123 and MitoTracker Red fluorescence

To assess hepatocyte zonation in vivo, tail vein injections of Rh123 (0.05 μmol/mouse) and MTR (0.05 μmol) were performed followed by intravital multiphoton microscopy. For analysis of fluorescence intensities, hepatic lobules were divided into portal and central halves, as identified by portal vein (PV) and central vein (CV) vasculature and demarcated by dashed lines in Fig. 1A (rightmost panels). Fluorescence quantification was performed by averaging mean intensities of multiple portal and central halves at depths of 10, 50 and 100 μm from the liver surface. Fluorescence intensity of Rh123, a membrane-permeant cationic ΔΨ indicator, was higher in the portal half of the liver lobule than the central half by a factor of 2.3 to 3.8 at all depths (Fig. 1A, left panels, and B). Notably, detection of Rh123 fluorescence by multiphoton microscopy decreased with increasing depth into the tissue. This is an artifact due to light scattering and re-absorbance of fluorescence, which increases with depth into the tissue and does not represent decreases of tissue Rh123 concentration. Accordingly, multiphoton images shown in Fig. 1A have been rescaled separately at each depth to better visualize intralobular gradients. By contrast, Fig. 1B plots mean absolute fluorescence in arbitrary units and shows similar drops of intensity with increasing depth in both lobular regions.

Fig. 1. Sublobular distribution of rhodamine123 and MitoTracker Red by intravital multiphoton microscopy.

(A) Representative intravital multiphoton images from a mouse liver 1 h after tail vein injection of Rh123 (0.05 μmol) and MTR (0.05 μmol). Images were collected 10, 50, and 100 μM from the liver surface. Dashed lines demarcate the boundary between the portal half (PH) and central half (CH) of liver lobules. Note bright fluorescence surrounding the portal vein (PV) and much weaker fluorescence around central veins (CV). Scale bar is 100 μm. (B) Quantification of Rh123 fluorescence in PH and CH regions at 10, 50, and 100-μm from the liver surface. Fluorescence is in arbitrary units (AU). (C) Quantification of MTR fluorescence in PH and CH regions at 10, 50, and 100 μm from the liver surface. Other details are as (B). (D) Overlay high power intravital multiphoton image of the mouse liver in (A) at a depth of 10 μm from the surface. Note colocalization of Rh123 and MTR in the overlay. Scale bar is 10 μm. *p < 0.05 versus PP.

MTR is another cationic membrane-permeant ΔΨ-indicating fluorophore. However, unlike Rh123, MTR binds covalently to mitochondrial proteins after uptake and is retained if mitochondria subsequently depolarize, as occurs, for example, after fixation. Nonetheless, MTR labeling visualized by intravital multiphoton microscopy was virtually identical to Rh123 and showed comparable zonation. Notably, fluorescence of MTR in the portal halves of liver lobules was 2.6 to 3.8 times that of central halves at 10 to 100-μm tissue depths (Fig. 1A, C). Importantly, higher power imaging showed nearly complete colocalization of the two fluorophores within mitochondria, confirming their expected localization (Fig. 1D).

3.2. Sublobular zonation of MTR is preserved after fixation throughout the liver and corresponds to zonation of E-cadherin and cytochrome P450 2E1

Since MTR fluorescence is retained after fixation, mice were injected with MTR (0.05 μmol/mouse), and their livers were fixed with PFA after 1 h to determine whether a similar lobular pattern of MTR fluorescence was present throughout the entirety of the liver. Vibratome sections were prepared and immunostained for Ecad and CYP2E1. Ecad is expressed on the plasma membrane of PP hepatocytes, whereas CYP2E1 is expressed in the endoplasmic reticulum and mitochondria of PC hepatocytes [24,25]. After fixation, MTR fluorescence (red) continued to show sublobular zonation throughout the interior of the liver (Fig. 2). Greater MTR fluorescence co-localized with Ecad immunostaining (green), whereas lesser MTR fluorescence co-localized with CYP2E1 (blue). Notably, hepatocytes expressed either Ecad or CYP2E1 but not both, and the boundary between CYP2E1- expressing central half regions and Ecad-expression portal half regions was sharp (Fig. 2, dashed lines in overlay images). Higher powered images highlighted the sharp boundary between the portal and central halves (Fig. 2B).

Fig. 2. Sublobular distribution of MitoTracker Red fluorescence in relation to E-cadherin and cytochrome P4502E1 immunostaining.

A mouse was injected with MTR, as described in Fig. 1. After 1 h, the liver was fixed. Tissue sections (100-μm) were prepared and stained for Ecad and CYP2E1 immunofluorescence before imaging by confocal microscopy, as described in Materials and Methods. (A) Low power of Ecad (green), CYP2E1 (blue), MTR (red), and overlay images. Note that MTR fluorescence decreases going from Ecad expressing PH to CYP2E1-expressing CH regions. Also note the sharp boundary between PH and CH, as shown by the dashed lines. Scale bar is 100 μm. (B) High power image. Scale bar is 50 μm.

We also assessed labeling of other tissues after MTR injection in vivo. Like the liver, MTR robustly labeled the kidney, ileum, and heart (Fig. 3) with higher power images confirming mitochondrial MTR localization (Fig. 3, right). However, unlike the liver, uptake of MTR in lower power images was relatively homogeneous for cells across these organs (Fig. 3, left).

Fig. 3. In vivo MitoTracker Red labeling of kidney, ileum and heart.

Mice were injected with MTR, as described in Fig. 1. After 1 h, kidney, ileum and heart were removed and fixed in PFA, and 100-μm vibrotome sections were prepared and counterstained with Hoechst 33342. (Left) Lower power images of MTR (red) and Hoechst 33342 (blue) fluorescence. Scale bar is 100 μm. (Right) High power images. Note mitochondrial pattern of MTR labeling. Scale bar is 10 μm.

3.3. Zonal patterning by MitoTracker Red is independent of amount administered

To determine the dose dependency of sublobular labeling with MTR, experiments were repeated by increasing MTR dosage from 0.05 μmol to 0.5 μmol per mouse. After 0.5 μmol, MTR fluorescence distribution retained the same zonal pattern (Fig. 4A–C). Importantly, increased dosage did not attenuate portal to central fluorescence gradients. Indeed, portal half to central half fluorescence ratios instead increased from 1.6 after 0.05 μmol MTR to 1.9 after 0.5 μmol, a small but statistically significant change (Fig. 4A, C). Somewhat surprisingly, hepatic MTR labeling did not scale with the dose of MTR administered. In portal halves, the intensity of MTR fluorescence labeling was 1.6-fold higher after 0.5 μmol MTR than after 0.05 μmol. In central halves, the increase after 0.5 μmol was smaller and not statistically significant (Fig. 4B).

Fig. 4. Zonal patterning after low versus high dose MitoTracker Red.

Mice were injected with either a low or a high dose of MTR (0.05 or 0.5 μmol, respectively, and 100-μm sections prepared and immunostained for CYP2E1, as described in Fig. 2. (A) Images of MTR (red), CYP2E1 (blue), and overlay fluorescence after low and high MTR. Note that both MTR dosages generated similar zonal patterning. Scale bar is 100 μm. (B) MTR fluorescence in PH compared to CH after low and high doses. *, p <0.05 vs PC; †, p < 0.05 vs PC; §, p < 0.05 vs 0.05 μmol MTR. (C) PP to PC ratios of MTR fluorescence. p < 0.05 vs 0.05 μmol MTR.

3.4. Intralobular gradients are flow-dependent

Differences of labeling by MTR and Rh123 may signify greater mitochondrial ΔΨ in hepatocytes in portal halves of the lobule. Alternatively, decreased labeling in central halves may be due to decreased blood concentrations of MTR and Rh123 as hepatocytes take up MTR and Rh123 during blood flow through the liver lobule. To assess the flow-dependence of intralobular labeling with MTR, MTR (0.01 μmol/liver) was infused retrogradely into mouse livers via the inferior vena cava. After retrograde perfusion, MTR labeling was greatest in central halves and much less in portal halves, a sublobular distribution opposite that observed after anterograde MTR perfusion after tail vein injection (Fig. 5A).

Fig. 5. Flow dependency of sublobular MitoTracker Red labeling.

(A) A mouse was retrogradely perfused with MTR at rate of 2 mL/min, and 100-μm vibratome sections were prepared after fixation, as described in Fig. 2. Note bright MTR fluorescence around the central vein (CV) and much less MTR fluorescence around the portal vein (PV). Scale bar is 100 μm. (B) MTR (0.05 μmol) was injected in vivo followed by isolation of hepatocytes. After plating, the freshly isolated hepatocytes were loaded in vitro with 200 nM Rh123. In the lower left and upper panels, an image of red, green and overlay fluorescence is shown. In the lower right panel, total fluorescence for MTR and Rh123 was quantified on a cell-by-cell basis and plotted. Note the absence of a correlation between the intensities of red MTR and green Rh123 labeling. Scale bar is 50 μm. (C) Hepatocytes from untreated mice were isolated, plated, and labeled with 200 nM MTR in vitro followed by CYP2E1 immunostaining. In the lower left and upper panels, an image of red, green and overlap fluorescence is shown. In the lower right panel, total fluorescence for MTR and CYP2E1 was quantified on a cell-by-cell basis and plotted. Note the lack of correlation between MTR and CYP2E1 labeling. Scale bar is 20 μm. (D) Hepatocytes were treated as in (C) except that MTR (0.05 μmol) was given intravenously 1 h before hepatocytes isolation. Note negative correlation between MTR labeling and CYP2E1 expression. Scale bar is 20 μm.

To confirm further that intralobular gradients of MTR labeling do not represent differences of mitochondrial ΔΨ, mice were injected with MTR in vivo. After 1 h, hepatocytes were isolated, plated, and labeled with Rh123 in vitro. No correlation was observed between in vivo labeling with red-fluorescing MTR and in vitro uptake of ΔΨ-indicating green Rh123 (r = 0.0959; p = 0.1781) (Fig. 5B).

To confirm again the flow gradient dynamic of MTR hepatic loading, immunohistochemistry for CYP2E1 was performed on hepatocytes labeled with MTR in vivo before isolation in comparison with hepatocytes loaded with MTR in vitro after isolation. In both conditions, CYP2E1 immunolabeling was at or near the surface of hepatocytes, as previously reported [26–28]. In hepatocytes labeled in vivo with MTR, CYP2E1 immunolabeling correlated negatively with MTR fluorescence intensity (r = 0.7221 in a log-log plot, p < 0.05) (Fig. 5C). After in vitro loading of MTR, CYP2E1 no longer correlated with MTR fluorescence intensity (r = 0.0787, p = 0.4625) (Fig. 5D). Overall, these findings indicate that lobular gradients of MTR fluorescence after in vivo administration are due to a flow-dependent blood concentration gradient of MTR decreasing from portal to central regions rather than to a difference in ΔΨ across the liver lobule (Fig. 5C, D).

3.5. Flow cytometric cell sorting based on in vivo MitoTracker Red labeling generates periportal, midzonal and pericentral populations of hepatocytes

A four-gate flow cytometry strategy was used to sort hepatocytes isolated from mice injected in vivo with MTR (0.05 μmol/mouse) into PP, MZ, and PC populations based on MTR fluorescence intensity. In the first gate, forward side scatter area (FSC-A) was plotted against side scatter area (SSC-A) to exclude large clusters and small debris (events outside the solid line in Fig. 6A). In the second gate, FSC-A was plotted versus forward side scatter height (FSC-H) to include only hepatocyte singlets inside the main cluster of events (Fig. 6B). The third gate excluded non-viable hepatocytes with increased Fixable Violet LIVE/DEAD fluorescence excited with a 375-nm laser (Fig. 6C). The final gate sorted hepatocytes into populations with the highest 20%, median 20%, and lowest 20% of MTR fluorescence as PP (gray), MZ (blue) and PC (red) hepatocytes, respectively (Fig. 6D). This procedure required about 1.5 h to perform and recovered between 40–50% of all sorting events into distinct equally sized hepatocyte populations (>300K cells) based on zone of origin (Fig. 6E).

qPCR analysis of the 3 populations revealed that PP hepatocytes exhibited 10.6 times the Cdh1 (Ecad) of PC hepatocytes, whereas PC hepatocytes exhibited 7.9 times the Cyp2e1 expression of PP hepatocytes (Fig. 6F, p < 0.05). MZ hepatocytes were intermediate in Cdh1 and Cyp2e1 expression. Overall, the qPCR results confirm the validity of using in vivo MTR labeling to sort hepatocytes into sublobular populations. Exploiting the MTR flow gradient within the liver, hepatocytes could be reliably and reproducibly sorted according to their zone of origin.

3.6. Periportal hepatocytes exhibit higher overall oxygen consumption rates compared with pericentral hepatocytes

We further explored differences in the bioenergetic profile of sorted PP, MZ and PC hepatocytes by Seahorse extracellular flux analysis. (Fig. 7A). Basal OCR and OCRs after sequential addition of oligomycin (ATP synthase inhibitor), FCCP (protonophoric uncoupler), and antimycin/rotenone (respiratory inhibitors) in PP hepatocytes were 3–4 times that of PC hepatocytes. MZ hepatocytes were in between. OCR normalized to basal OCR showed equivalent responses to oligomycin, FCCP, and rotenone/antimycin A for all three hepatocyte populations, signifying equivalent coupling in PP, MZ, and PC hepatocytes (Fig. 7B). Basal (Fig. 7C), non-mitochondrial (rotenone/antimycin-insensitive, Fig. 7D), maximal (Fig. 7E), proton leak-linked (Fig. 7F), and ATP-linked (Fig. 7G) OCR, as well as spare respiratory capacity (Fig. 7H), decreased stepwise from PP to MZ to PC hepatocytes. Overall, these findings showed a decrease in mitochondrial bioenergetic capacity but no change in coupling from PP to PC regions.

Fig. 7. Oxygen consumption rates of sorted hepatocytes: periportal > midzonal > pericentral.

Hepatocytes were isolated and sorted as described in Fig. 6. (A) Mitochondrial stress test. OCR of PP, MZ, and PC hepatocytes was measured before and after sequential additions of 1 μM oligomycin, 1 μM FCCP, and 1 μM Rot/ 1 μM AA. Values were corrected based on cell count with units of pmol/min/10³ cells. (B) Normalized OCR. The measurements shown of (A) were normalized to the third basal measurement. (C) Basal OCR. (D) Non-mitochondrial respiration. (E) Maximal respiratory capacity. (F) OCR associated with proton leak. (G) OCR associated with ATP production. (H) Spare respiratory capacity. Normalized OCR in PP, MZ, and PC cells. *, p< 0.05 vs PP, †, p < 0.05 MZ).

4. Discussion

4.1. Intralobular gradients of mitochondrial membrane potential-indicating fluorophores after tail vein injection

Here, we characterized the in vivo hepatic sublobular distribution of the mitochondrial ΔΨ-indicating fluorophores MTR and Rh123. After tail vein injection and visualization by multiphoton imaging near the liver surface, both fluorophores distributed in a similar fashion with mitochondrial uptake of both dyes decreasing progressively moving from portal tracts to central veins across the liver lobule (Fig. 1). MTR and Rh123 are membrane-permeant monovalent cations that enter mitochondria electrophoretically driven by mitochondrial ΔΨ in a fashion comparable to TMRM and tetramethylrhodamine ethylester (TMRE) [29–32]. Indeed, intravital multiphoton microscopy showed that the distribution of TMRM at depths of 10, 50, and 100 μm has a robust zonal patterning virtually identical to that observed with Rh123 and MTR (Supplemental Fig. 1).

Although high concentrations of Rh123 can inhibit OXPHOS [30], the Rh123 concentrations used in the present study (0.05 μmol/mouse) were below this inhibition level and the same or lower than used in previous in vivo mouse studies in which no apparent toxicity was observed [33]. Because red-fluorescing TMRM and the related fluorophore TMRE have similar excitation and emission spectra to MTR, TMRM and TMRE cannot be used together with MTR. Accordingly, green-fluorescing Rh123 was used instead, as chosen for its hepatic mitochondrial uptake identical to TMRM and for its compatibility in MTR fluorescence imaging. In contrast to Rh123 and TMRM, MTR has a reactive chloromethyl group that binds covalently to mitochondrial protein thiols to be retained inside mitochondria even after subsequent mitochondrial depolarization. Consequently, MTR is a ‘fixable’ dye whose mitochondrial fluorescence persists even after formaldehyde fixation. Notably, after in vivo treatment with MTR by tail vein injection followed by fixation, the gradient of MTR fluorescence across the liver lobule from portal to central veins persisted deep into the interior of the liver. Immunostaining for Ecad and CYP2E1, respective PP and PC markers, confirmed the decline of MTR labeling going from PP to PC regions (Fig. 2).

The concentrations (0.05 μmol/mouse) of MTR and Rh123 used here were 5–100 times less than used previously for intravital imaging to avoid possible toxicities [17–19,30,34]. Blood serum analysis from mice treated with either 0.05 μmol MTR or vehicle by tail vein injection showed that values were within normal ranges for C57BL/6J mice of comparable age according to standards provided by the Mouse Phenome Database at Jackson Laboratory (Supplemental Table) [35]. Specifically, liver enzymes and bilirubin were not increased consistent with an absence of hepatotoxicity. Moreover, multiphoton microscopy showed well polarized mitochondria (Fig. 1D), and mitochondrial stress tests in hepatocytes isolated from MTR-treated mice showed excellent coupling of respiration equivalent to that of hepatocytes isolated from mice without MTR treatment (Fig. 7) [36]. These findings indicate no or very minimal toxicity by MTR treatment, which is consistent with our previous studies using higher MTR dosing in vivo [34]. Even administered in vivo at 10-fold higher concentration (0.5 μmol/mouse), MTR created a sublobular gradient of fluorescence very similar to the lower dose (Fig. 4). By contrast, MTR labeling in other tissues, including heart, kidney, and intestine, was much more uniform (Fig. 3).

4.2. Blood flow dependency of MitoTracker Red intralobular gradients

Since PP regions of liver lobules are well known to have greater oxidative metabolism than PC regions, our findings initially suggested a gradient of mitochondrial ΔΨ across the liver lobule decreasing from portal to central veins. However, an alternate hypothesis is that blood MTR concentrations decrease progressively as blood flows through sinusoids from the portal areas to the central veins due to uptake of MTR into mitochondria. Since mitochondrial MTR uptake is proportional to extramitochondrial concentration, this flow-dependent effect might also account for the MTR gradient across the lobule.

Several findings lead to the conclusion that the latter hypothesis is correct, namely that intralobular gradients of MTR fluorescence are flow-dependent. First, when MTR was infused into the liver in a retrograde fashion via the inferior vena cava, the gradient of MTR reversed and became greatest around central veins and least around portal tracts (Fig. 5A). Second, when hepatocytes were labeled in vivo with MTR, isolated, plated, and then labeled with Rh123, the intensity of Rh123 fluorescence did not correlate with that of MTR (Fig. 5B). Rh123 fluorescence intensity under these conditions is a relative measure of mitochondrial ΔΨ. Thus after in vivo loading, hepatocytes with high MTR fluorescence had similar ΔΨ to those with low MTR fluorescence, at least after isolation.

Lastly, immunolabeling with the PC marker CYP2E1of hepatocytes isolated from MTR-treated mice revealed an inverse relation between CYP2E1 content and MTR staining, confirming that high MTR hepatocytes were PP, whereas low MTR cells were PC (Fig. 5C). By contrast, if hepatocytes were loaded with MTR after isolation, there was no correlation of CYP2E1 with MTR (Fig. 5D). Since after in vitro loading, MTR is a ΔΨ indicator, the result also indicates that PP and PC hepatocytes have similar ΔΨ. Although tail vein administration of MTR produced a marked fluorescence gradient between PP and PC regions, preliminary experiments showed that intraperitoneal (IP) injection of MTR (0.05 μmol/mouse) produced a much less pronounced MTR fluorescence gradient between PP and PC regions than tail vein injection (Supplemental Fig. 2). Notably, IP injections strongly labeled the peritoneal surface of the liver, indicating direct labeling that was independent of blood flow. Such uneven distribution would confound subsequent FACS sorting by falsely identifying surface-residing hepatocytes as PP. Accordingly, IP labeling was not pursued.

4.3. Cell sorting of subloblular populations of hepatocytes based on in vivo MitoTracker Red labeling

Based on the intensity of MTR fluorescence after tail vein injection, sorting hepatocytes became possible by flow cytometry into PP, MZ and PC populations after excluding cell clusters, debris, and non-viable cells not surviving the isolation procedure (Fig. 6). Sorting was robust and repeated several times. Cdh1 and Cyp2e1 gene expression by qPCR was greatest in the PP and PC populations, respectively, and intermediate in the MZ cell fraction, which confirmed the zonal specificity of the cell sorting. Although tail vein administration of MTR produced a marked fluorescence gradient between PP and PC regions, preliminary experiments showed that intraperitoneal (IP) injection of MTR (0.05 μmol/mouse) produced a much less pronounced MTR fluorescence gradient between PP and PC regions than tail vein injection (Supplemental Fig. 2). Notably, IP injections strongly labeled the peritoneal surface of the liver, indicating direct labeling that was independent of blood flow. Such uneven distribution would confound subsequent FACS sorting by falsely identifying surface-residing hepatocytes as PP. Accordingly, IP labeling was not pursued.

4.4. Greater mitochondrial metabolism in periportal hepatocytes compared to pericentral hepatocytes

Seahorse respirometry revealed that all populations of hepatocytes – PP, MZ and PC – were equivalently well coupled (Fig. 7B). However, the magnitude of both basal and maximal respiration (OCR) was 3–4 times greater in PP hepatocytes compared to PC with MZ in between (Fig. 7C), similar to recent findings in hepatocytes isolated by anterograde collagenase perfusion and sorted based on Ecad/CD37 [37]. This large difference is somewhat surprising, since mitochondrial volume percent and number of mitochondrial cristae are only modestly less by 25–35% in PC hepatocytes compared to PP hepatocytes [38]. Moreover, in vitro MTR labeling showed that ΔΨ in PP vs PC hepatocytes was also similar (Fig. 5D). Thus, other regulatory factors must be activating oxidative phosphorylation in PP compared to PC hepatocytes. Nonetheless, these results are in line with past investigations measuring O₂ uptake from PP and PC regions in perfused rat livers using mini-oxygen electrodes and miniature light guides [39].

4.5. Compartmentation of hepatic metabolism into portal and central halves of liver lobules

Immunolabeling with Ecad and CYP2E1 showed a relatively sharp demarcation between the portal and central halves of liver lobules. Data by others shows a similar sharp demarcation at midlobule for a variety of sublobular markers, including albumin, cytochrome P450 2f2, hydroxysteroid 17-beta dehydrogenase 13, and serine dehydratase for the portal half and aldehyde dehydrogenase 1a1 (Aldh1a1), aldehyde dehydrogenase 2 (Aldh2), and cytochrome P450 1a2 (Cyp1a2) for the central half [11,40–46]. Such observations belie the general assumption that metabolism gradually changes moving across the liver lobule. Rather, at least for some metabolic pathways, hepatic metabolism appears to be compartmentalized into two sublobular zones of roughly equal size – a portal half and a central half with a relatively sharp transition at mid-lobule. A subcompartment of the central half of liver lobule consists of hepatocytes lining central veins (not to be confused with the larger Zone 3 or PP region), which uniquely expresses glutamine synthetase. Future studies will be needed to develop antibody-free methods to selectively sort these central venular lining hepatocytes, perhaps by using retrograde MTR infusion where central venular lining hepatocytes are labeled most brightly (see Fig. 5A).

4.6. Methods for sorting hepatocytes into distinct sublobular populations

An early method for isolating PP and PC hepatocytes is the digitonin-collagenase perfusion technique, a method still in use [47,48]. In this approach, digitonin is perfused halfway into the liver lobule in an anterograde or retrograde fashion to permeabilize and make non-viable PP and PC hepatocytes, respectively. The digitonin is then washed out by perfusion in the opposite direction, and the remaining hepatocytes are harvested by collagenase digestion. Advantages are that digitonin-collagenase perfusion is a physical technique not reliant on antibody marking or fluorescent labels. Disadvantages are that PP and PC hepatocytes cannot be harvested simultaneously. Moreover, digitonin entry into lobules across the entire liver may not be uniform, leading to mixing of PP and PC hepatocytes in the final preparation.

More recently, hepatocyte sorting by flow cytometry has emerged, and our sorting based on MTR fluorescence represents an extension of these techniques [49,50]. Previous sorts were based on either in vivo acridine orange (AO) labeling or selective tagging of hepatocytes with zone-specific antibodies. In vivo AO labeling, like MTR labeling, is most likely flow-dependent resulting in brighter PP than PC fluorescence. For subsequent fluorescence imaging, AO has the drawbacks of fluorescing both red (acidic compartments) and green (nucleic acids) and of not being fixable. AO is also a phototoxic compound that poorly tolerates light exposure as during confocal imaging [51]. Zone-specific antibodies allow use of all colors of fluorophores during and after sorting, but antibody binding and internalization may induce phenotypic changes in target cells that can confound metabolic and functional studies of the sorted hepatocytes. Additionally, in disease models, antibody epitopes may change within the liver lobule. For example, disruptions in Wnt/β-catenin and hedgehog/Hippo signaling pathways alter Ecad/CD73 lobular zonation and thus their efficacy for surface marker sorting [52]. The additional step of labeling hepatocytes with antibodies also lengthens the time required to complete the sorting procedure. In a hepatocyte-specific EGFP transgenic mouse strain, EGFP expression occurs principally in PC regions, which allows FACS of zone-specific (PP and PC) hepatocytes [53]. This method is advantageous due to its speed and accuracy, as no additional fluorophore treatment or immunolabeling is needed to generate Zone 1 and Zone 3 hepatocytes. However, hepatocytes can only be sorted into PP and PC populations, since bright EGFP fluorescence is either present or absent in the liver lobule without an intermediate zone. The basis for selective PC expression of EGFP is unknown, and genetic or other change might alter the zonal distribution of EGFP. By contrast, the distribution of MTR fluorescence is flow-dependent, which allows sorting for MZ hepatocytes. Sublobular distribution of MTR fluorescence should also be independent of genetic and other manipulations unless sinusoidal blood flow is disrupted. Moreover, MTR allows sorting of hepatocytes from any mouse strain rather than a specific reporter strain.

Sorting based on MTR fluorescence avoids many of these shortcomings. MTR fluoresces only red such that blue, green, and far red-fluorescing probes can be used for post sorting experiments. MTR is also stable to confocal illumination. At the loading concentration used, MTR does not alter cellular and mitochondrial metabolism or function [29,34,54]. MTR sorting is also more rapid than antibody-based sorts. We typically complete our sorts within 1.5 h of harvesting livers.

5. Conclusions

In vivo treatment with MTR provides a rapid and effective method to display liver zonation. Cell sorting based on MTR fluorescence yields populations of several hundred thousand hepatocytes specific to PP, MZ and PC sublobular regions. Since MTR labeling is flow-dependent, sorting of hepatocytes will be independent of changes in expression of surface markers that might occur during development and disease. Thus, MTR-based sorting holds potential to elucidate sublobular differences in hepatocyte metabolism, regulatory pathways, and injury in many models of liver disease with implications for therapeutic intervention.

Supplementary Material

1

Supplemental Fig. 1. Sublobular distribution of tetramethylrhodamine methylester by intravital multiphoton microscopy. (A) Representative intravital multiphoton images from a mouse liver 1 h after tail vein injection of TMRM (0.05 μmol). Images were collected 10, 50, and 100 μM from the liver surface. Note gradient of red TMRM fluorescence decreasing from portal vein (PV) to central vein. Images as various depths are rescaled as described for Fig. 1. Scale bar is 100 μm.

2

Supplemental Fig. 2. Comparison of tail vein versus intraperitoneal injection on the hepatic distribution of MitoTracker Red fluorescence. Mice were injected with 0.05 μmol MTR by tail vein (upper panels) or intraperitoneally (lower panels). After 1 h, livers were fixed as described in Fig. 2, and confocal images of MTR fluorescence were collected. Note the absence of an intralobular gradient of fluorescence and the strong labeling of surface hepatocytes after IP injection. Scale bar is 100 μm. *, liver surface.

3

Supplemental Table. Blood serum analysis of mice treated with MitoTracker Red. C57BL/6J mice aged 9 wk were injected with 0.05 μmol MTR (n = 3) or vehicle (n = 2), and blood was collected after 1 h for serum analysis. Normal ranges are from the Mouse Phenome Database at Jackson Laboratory [35]. Shown are means ± SE.

Highlights.

• In vivo MitoTracker Red labeling generates a sublobular fluorescent gradient between periportal and pericentral zones.

• Fluorescence-activated cell sorting based on MitoTracker Red labeling provides a fast and reliable method to separate living hepatocytes based on sublobular zone.

• Sorted periportal hepatocytes exhibit substantially higher oxygen consumption rates compared to midzonal and pericentral hepatocytes.

Abbreviations used:

ΔΨ

membrane potential

Aldh1a1

aldehyde dehydrogenase 1a1

ALDH2

aldehyde dehydrogenase 2

AA

antimycin A

AO

acridine orange

CV

central vein

BSA

bovine serum albumin

Cyp1a2

cytochrome P450 1a2

CYP2E1

cytochrome P450 2E1

Ecad

e-cadherin

FACS

fluorescence-activated cell sorting

FCCP

carbonyl cyanide 4-(trifluoromethoxy)phenylhydrazone

FSC-A

forward scatter area

FSC-H

forward scatter height

Gapdh

glyceraldehyde 3-phosphate dehydrogenase

MASLD

metabolic dysfunction associated steatotic liver disease

MTR

MitoTracker Red

MZ

midzonal

n.s.

not significant

OCR

oxygen consumption rate

PBS

phosphate buffered saline

PC

pericentral

PFA

paraformaldehyde

PP

periportal

PV

portal vein

Rh123

rhodamine 123

Rot

rotenone

SIAM

swift increase in alcohol metabolism

SSC-A

side scattered area

TMRE

tetramethylrhodamine ethylester

TMRM

tetramethylrhodamine methylester