Porcine liver sinusoidal endothelial cells contribute significantly to intrahepatic ammonia metabolism


  • Potential conflicts of interest: none.


Ammonia metabolism in the liver has been largely credited to hepatocytes (HCs). We have shown that liver nonparenchymal cells that include liver sinusoidal endothelial cells (LSECs) produce ammonia. To address the limited knowledge regarding a role for LSECs in ammonia metabolism, we investigated the ammonia metabolism of isolated LSECs and HCs under three different conditions: (1) bioreactors containing LSECs (LSEC-bioreactors), (2) bioreactors containing HCs (HC-bioreactors), and (3) separate bioreactors containing LSECs and HCs connected in sequence (Seq-bioreactors). Our results showed that LSEC-bioreactors released six-fold more ammonia (22.2 nM/hour/106 cells) into the growth media than HC-bioreactors (3.3 nM/hour/106 cells) and Seq-bioreactors (3.8 nM/hour/106 cells). The glutamate released by LSEC-bioreactors (32.0 nM/hour/106 cells) was over four-fold larger than that released by HC-bioreactors and Seq-bioreactors (<7 nM/hour/106 cells). LSEC-bioreactors and HC-bioreactors consumed large amounts of glutamine (>25 nM/hour/106 cells). Glutaminase is known for catalyzing glutamine into glutamate and ammonia. To determine if this mechanism may be responsible for the large levels of glutamate and ammonia found in LSEC-bioreactors, immunolabeling of glutaminase and messenger RNA expression were tested. Our results demonstrated that glutaminase was present with colocalization of an LSEC-specific functional probe in lysosomes of LSECs. Furthermore, using a nucleotide sequence specific for kidney-type glutaminase, reverse-transcription polymerase chain reaction revealed that this isoform of glutaminase was expressed in porcine LSECs. Conclusion: LSECs released large amounts of ammonia, perhaps due to the presence of glutaminase in lysosomes. The ammonia and glutamate released by LSECs in Seq-bioreactors were used by hepatocytes, suggesting an intrahepatic collaboration between these two cell types. (HEPATOLOGY 2009.)

Since it was revealed that the liver plays a role in nitrogen metabolism (ureagenesis), ammonia detoxification, and amino acid metabolism,1 nitrogen metabolism has been largely credited to hepatocytes (HCs).2 Later electron microscopical studies in the early 1970s revealed the existence of liver sinusoidal endothelial cells (LSECs).3 Improved cultivating techniques and the advent of specific and functional probes for LSECs in recent decades have opened a new era for the study of the functions of LSECs.4, 5 In a previous study, we showed that cultured liver nonparenchymal cells (NPCs) that include LSECs produce significant amounts of ammonia (ammonium-ions), most likely from glutamine.6 On a single cell level, NPCs used four times more glutamine than HCs. A plausible explanation for this difference was reflected in a study that showed different expressions of glutamine transporter systems in LSECs and HCs.7 The same study revealed that LSECs expressed kidney-type glutaminase as opposed to liver-type glutaminase in HCs. The kidney-type glutaminase is probably also expressed in endothelial cells other than LSECs.8 The ammonia production by NPCs, may have implications for a better understanding of the pathophysiology of hyperammonemia in relation to liver disease.9

Under normal conditions, the intrahepatic turnover of glutamine is excessive, requiring both a rapid breakdown and a high capacity for resynthesis.10 This phenomenon provides the basis for efficient and highly adaptive regulation of body nitrogen balance.2, 9 In the periportal region, glutamine is hydrolyzed to ammonia and glutamate to fuel the urea synthesis. Accordingly, the concentration of ammonia declines during downstream passage through the sinusoids, and in the pericentral region a small population of the total HC population (5%–7%) scavenge residual ammonia to resynthesize glutamine from ammonia and glutamate.

Today, most in vitro studies on liver cells are performed in incubators with atmospheric oxygen tensions (atmosO2) that result in oxygen tensions around 139 mm Hg, far above the physiological oxygen tensions (physO2) being between 91 and 10 mm Hg (<10 mm Hg is hypoxia).11 The oxygen tensions within the liver lobules are believed to be between 70 mm Hg in the periportal region and decrease to 25 mm Hg in the pericentral region of the liver acini.12 Cell types such as embryonic stem cells have been shown to grow better in incubators at physO2 than at atmosO2.13 Primary rat LSECs cultured at physO2 sustained better survival and maintained their differentiated features better than LSECs cultured at atmosO2.14

In vivo, LSECs have a large capacity for clearing the blood of numerous physiological and foreign waste macromolecules by receptor-mediated endocytosis.4, 15 The final degradation products of the waste macromolecules (hyaluronan and serglycin) are lactate and acetate.16, 17 During anaerobic conditions, most mammalian cells release lactate, mainly via glycolysis. NPCs produced large amounts of lactate, indicating that they are highly metabolically active cells, even at atmosO2. NPCs favored an anaerobic oxidation of monosaccharides.6

In the present study, bioreactors containing porcine LSECs (LSEC-bioreactors), bioreactors containing HCs (HC-bioreactors), and separate bioreactors containing LSECs and HCs connected in sequence (Seq-bioreactors) were cultivated at physO2 (91 mm Hg) in simulated microgravity that has been shown to maintain metabolic functions of HCs.18 We found that the LSEC-bioreactors used approximately the same amount of glutamine as the HC-bioreactors, but the LSEC-bioreactors released six-fold and four-fold more ammonia and glutamate than the HC-bioreactors into the growth media, respectively. Immunolabeling and reverse-transcription polymerase chain reaction (RT-PCR) confirmed the presence of the enzyme glutaminase in the lysosomes of the LSECs. Furthermore, LSEC-bioreactors released five-fold more lactate into the growth media than HC-bioreactors. The ammonia, glutamate, and lactate released by LSECs were used by HCs in the Seq-bioreactors.


AST, aspartate aminotransferase; atmosO2, atmospheric oxygen tensions; FITC, fluorescein isothiocyanate; GAPDH, glyceraldehyde-3-phosphate dehydrogenase; HC, hepatocyte; HC-bioreactor, bioreactor containing hepatocytes; KC, Kupffer cell; LSEC, liver sinusoidal endothelial cell; LSEC-bioreactor, bioreactor containing LSECs; NPC, nonparenchymal cell; OUR, oxygen uptake rate; physO2, physiological oxygen tensions; RT-PCR, reverse-transcription polymerase chain reaction; SC, stellate cell; Seq-bioreactor, sequentially connected LSEC-bioreactor and HC-bioreactor.

Materials and Methods

Chemicals were obtained from Sigma-Aldrich (St. Louis, MO) unless stated otherwise. Collagenase P was obtained from Roche, Norway. One liter of growth medium, DM110/SS, was composed of 6.0 g DME powder (BioWhittaker, Walkersville, MD), 6.8 g MCDB 110-powder, 2.6 g NaHCO3 (Supporting Table 1), and antibiotics (penicillin, streptomycin, and gentamycin) supplemented with a prototype version of 5% tissue synthetic serum.19 Formaldehyde-treated serum albumin (0.8 mg/mL) was incubated with the fluorochrome fluorescein isothiocyanate (FITC) in sodium carbonate buffer (0.5 mol/L [pH 9.5]) in a protein/dye weight ratio of 1:1 at 4°C overnight. Unreacted fluorochrome was removed by way of gel filtration through a PD-10 column equilibrated and eluted with phosphate-buffered saline without Ca2+ or Mg2+. The final concentrations of the growth factors in the growth media were for fibroblast growth factor 20 ng/mL, insulin-like growth factor l 10 ng/mL (both from R&D Systems, Minneapolis, MN), and porcine endothelial growth factor 1 ng/mL (a gift from L. Erickson, Guelph, Canada). Collagen-coated dextran microcarriers were obtained from Pharmacia Biotech (Sweden). The microcarriers were inoculated at a concentration of 4.0 mg/mL. Two mouse polyclonal antibodies (clones 2G6 and 2B10) against porcine macrophages were kindly provided by Dr. Angela Berndt (Jena, Germany). Monoclonal mouse anti-human desmin (clone D33) was obtained from DAKO (Denmark). Polyclonal mouse anti-human glutaminase was obtained from Novus Biologicals (Littleton, CO). Monoclonal goat anti-mouse immunoglobulin G, TRITC-conjugate, was obtained from Zymed (San Francisco, CA).

Cell Isolation and Bioreactors.

All animals received care according to the criteria outlined in the Guide for the Care and Use of Laboratory Animals prepared by the National Academy of Sciences and published by the National Institutes of Health (publication 86-23, revised 1985). LSECs and HCs were isolated from male pigs (Sus scrofa domesticus) weighing 7–8 kg.20 Briefly, the livers were perfused with a HEPES buffer to wash out blood cells before perfusion with a HEPES buffer containing collagenase. To pellet the HCs, the resulting cell suspensions were subjected to velocity centrifugation (50g). The resulting supernatants were concentrated by centrifugation (850g), thereafter enriched in a 21% Optiprep (Nycomed, Oslo) density solution at 3,300g. The enriched NPCs were subjected to elutriation centrifugation to isolate LSECs. The viabilities of the cells were >95% as judged by a trypan blue exclusion test. No pyrogenes were detected on Limulus amebocyte lysate assay.

Perfusion bioreactors with volumes of 250 mL and 500 mL with oxygenators were obtained from Synthecon (Houston, TX) (Fig. 1A). The 250-mL bioreactors were inoculated with 6.9 × 108 ± 8.2 × 107 LSECs (LSEC-bioreactors [n = 7]), and the 500-mL bioreactors were inoculated with 2.0 × 109 ± 2.9 × 106 HCs (HC-bioreactors [n = 7]). Separate bioreactors containing 6.9 × 108 ± 8.2 × 107 LSECs and 2.0 × 109 ± 2.9 × 106 HCs were connected in sequence (Seq-bioreactors [n = 7]) (Fig. 1B). The inner cores of the bioreactors were covered with a nylon mesh with a diameter of 5 μm obtained from F. Burmeister (Oslo, Norway). The nylon mesh was sutured with polypropylene sutures (Ethicon, NJ). The flow rate was kept at 25 mL/minute with a peristaltic pump (Allmed, Sweden), and the rate was confirmed with an inline flow probe (Medi-Stim, Norway). Extra channels were drilled in the inner cores of the bioreactors to ensure more uniform oxygenation of the cells (dotted horizontal lines in the outer part of the inner core). Fresh DM110/SS was replenished between periods 1 and 2. The cell samples for DNA and protein measurements were sampled directly from the cell compartments of the bioreactors. The pH was between 7.3 and 7.4.

Figure 1.

Illustration of the perfusion bioreactor and outline of the experiments. (A) The perfusion bioreactor was rotated horizontally around its axis, thereby simulating microgravity. The cells were attached to microcarriers maintained in a suspension by balancing their sedimentation-induced gravity with centrifugation caused by the rotation of the cell compartment (30–40 rpm). The nylon mesh (with an average diameter of 5 μm) prevented the cells from escaping the cell compartment. (B) The experiments consisted of bioreactors allocated into three groups: (1) LSEC-bioreactors (n = 7), (2) HC-bioreactors (n = 7), and (3) Seq-bioreactors (n = 7).


Lactate and glucose concentrations were measured with a YSI-analyzer (Cobas, Switzerland). Albumin was measured with a pig albumin enzyme-linked immunosorbent assay kit obtained from Bethyl Laboratories (Montgomery, TX). Pyruvate was measured with a Pyruvate-726 kit (Sigma-Aldrich). Ammonia, amino acids, and urea were analyzed as described.21 Immediately after sampling, the samples were precipitated in 20% trichloroacetic acid and centrifuged, and we then performed affinity and ion exchange chromatography. The ammonia concentrations were confirmed with an automated enzymatic method (Cobas Fara II, Roche, Switzerland). Protein was assayed as described by Lowry et al.22 Lactate dehydrogenase, aspartate aminotransferase (AST), and alanine aminotransferase released into growth media were analyzed by standard automated techniques at the Department of Clinical Chemistry, University Hospital of Northern Norway. For the measurement of the intracellular contents of lactate dehydrogenase, AST, and alanine aminotransferase, the purified LSECs and HCs were thawed from −70°C to 22°C, centrifuged, washed in phosphate-buffered saline, and sonicated before analysis. DNA concentrations in the bioreactors were determined by the binding of mithramycin to DNA.23 All biochemical values were normalized to the cellular DNA content.

RNA Isolation and RT-PCR.

The total RNA was extracted from the different liver cell populations using TRIZOL reagent (Invitrogen, Carlsbad, CA) according to the manufacturer's protocol. Complementatry DNA was synthesized using 2.0 μg total RNA that was reverse-transcribed to a final volume of 50 μL using the SuperScript Preamplification kit (Invitrogen). PCR amplifications were performed in a PTC-200 Peltier Thermal Cycler (Bio-Rad, Hercules, CA). Primer pairs were: glutaminase, 5′-CTGGATGCTTCTGGTTTGGCCATT-3′ (sense) and 5′-ACCCTACTCCAACAGTAAGTGCGT-3′ (antisense) where a 334-bp fragment was expected; glyceraldehyde-3-phosphate dehydrogenase (GAPDH), 5′-GCAAATTCCACGGCACAGTCA-3′ (sense) and 5′-TCACGCCACAGTTTCCCAGAG-3′ (antisense) where a 433-bp fragment was expected. PCR products were analyzed by agarose gel (1.5%) electrophoresis and photographed under ultraviolet light.


To determine the purity of the cultures of LSECs, the LSEC fraction after the centrifugal elutriation was seeded on fibronectin-coated glass coverslips and labeled as described.20 The purity of the LSEC cultures varied between 80% and 95%. The other cell types in the LSEC cultures were stellate cells (SCs) (3%-11%) as judged by staining with anti-desmin antibody, Kupffer cells (KCs) (1%-7%) by staining with anti-porcine macrophage antibodies, and HCs (1%-2%) as judged by simple morphology.

To investigate the localization of glutaminase, we cultivated NPCs and HCs on fibronectin-coated glass coverslips. The cells were incubated with a specific and functional probe for LSECs, FITC-labeled formaldehyde-treated serum albumin (10 μg/mL), for 10 minutes. The cells were then fixed in cold methanol for 5 minutes. The fixed cells were incubated for 1 hour with polyclonal mouse anti-human glutaminase (1:100 dilution), and thereafter additional 1-hour incubation with TRITC-conjugated monoclonal goat anti-mouse immunoglobulin G (1:200 dilution). The examination of the specimens was performed with an Axiophot photomicroscope equipped with fluorescence optics (Zeiss Axiophot, Germany).

Oxygen Uptake Rate.

The experiments were performed at physO2 in a humidified CO2 incubator with supplemental O2 control (12% O2) (Galaxy R, RSBiotech, UK). The partial pressures of oxygen were measured by sampling growth media in capillary glass tubes taken simultaneously from the arterial and the venous ports of the bioreactors (Fig. 1A) and analyzed immediately in a blood-gas analyzer (Rapidlab 865, Chiron Diagnostics, UK).

Oxygen uptake rate (OUR) (nmol/minute) = Δ arterial-venous pO2 × flow × 1.18 × 10−3, 1.18 × 10−3 μmol/mL/mmHg was a constant derived from the Bunsen solubility coefficient for oxygen 37°C.24 The results were adjusted for the OUR measured in cell-free controls.


We used the SPSS 11.0 statistical package to establish significance between groups (SPSS, Chicago, IL). Two-way (time and group) analysis of variance for repeated measurements was applied to test whether any statistical significance existed between the three groups, followed by the LSD post hoc comparison test. An overall significance in analyses of variance for repeated measurements (F-test; P < 0.05) may be attributable to either the effect of time (PT) or the interaction of group and time (PGT). PGT < 0.05 denoted a significant difference between the groups dependent of time, whereas PG < 0.05 denoted a significant difference between the groups independent of time. If both PGT and PG were significant, then only PGT was denoted, because PGT is more robust than PG. PT < 0.05 denoted a significant change in time for the groups.


Ammonia Metabolism

Ammonia and Urea.

The ammonia released into the growth media by LSEC-bioreactors (22.2 ± 5.1 nM/hour/106 cells) was six-fold larger than that of HC-bioreactors (3.3 ± 1.8 nM/hour/106 cells) in period 1 (PGT < 0.05) (Fig. 2A). In period 2, the ammonia release by LSEC-bioreactors was four-fold larger (17.0 ± 2.8 nM/hour/106 cells) than that of LSEC-containing Seq-bioreactors (3.8 ± 1.8 nM/hour/106 cells) (PGT < 0.05). The urea production in the HC-bioreactors was 11.5 ± 2.0 nM/hour/106 cells, whereas there was no urea production as expected in the LSEC-bioreactors in period 1 (Fig. 2B). There was a significant urea production in both the HC-bioreactors and the Seq-bioreactors in period 2: 14.3 ± 2.5 nM/hour/106 cells and 14.5 ± 5.0 nM/hour/106 cells, respectively (PT < 0.05).

Figure 2.

Ammonia metabolism in the bioreactors. The net release of (A) ammonia and (B) urea, (C) the net removal of glutamine, and (D) the net release of glutamate were presented as differences between the last and the first measurement divided by the number of hours each period lasted per 106 cells (mean ± standard deviation). The positive values denoted net release into the growth media, and the negative values denoted net removal of glutamine from the growth media. PGT < 0.05 denoted a significant difference between the groups dependent of time, whereas PT < 0.05 denoted a significant change for the groups in time. The ammonia figures were corrected for ammonia generated in the cell-free controls. Urea was below detectable limits in LSEC-bioreactors. White bars represent LSEC-bioreactors; black bars represent HC-bioreactors; gray bars represent Seq-bioreactors.

Glutamine and Glutamate.

We attempted to identify the source of the ammonia release. There was a large removal of glutamine both in LSEC-bioreactors (27.7 ± 11.8 nM/hour/106 cells) and HC-bioreactors (28.8 ± 1.7 nM/hour/106 cells) (PT < 0.05) (Fig. 2C). The largest removal of glutamine in period 2 was by the LSEC-bioreactors (25.1 ± 2.3 nM/hour/106 cells) (PGT < 0.05), whereas the glutamine removal by HC-bioreactors and Seq-bioreactors was 8.8 ± 2.8 nM/hour/106 cells and 10.0 ± 2.0 nM/hour/106 cells, respectively. The corresponding glutamate release by the LSEC-bioreactors in period 1 was 32.0 ± 8.5 nM/hour/106 cells, which were over four-fold larger than the HC-bioreactors (6.7 ± 1.5 nM/hour/106 cells) (Fig. 2D) (PGT < 0.05). Moreover, the LSEC-bioreactors released 26.1 ± 1.1 nM/hour/106 cells glutamate in period 2 (PGT < 0.05), which was over nine-fold more than the HC-bioreactors (2.7 ± 1.1 nM/hour/106 cells) and Seq-bioreactors (1.3 ± 0.6 nM/hour/106 cells). For the net removal/release of amino acids, urea, and ammonia (Supporting Table 2). Cell-free control was also performed to determine the rate of autohydrolysis (Supporting Table 3).

Localization of Glutaminase in Isolated Liver Cells.

We then attempted to determine the mechanism of ammonia release by LSECs. Kidney-type glutaminase messenger RNA was present in all examined liver cells, including HCs, as revealed on RT-PCR analysis (Fig. 3). We next investigated the intracellular localization of glutaminase by immunolabeling-cultivated NPCs (LSECs, SCs, and KCs). The LSECs showed strong immunolabeling of glutaminase, and anti-glutaminase colocalized with the LSEC-specific functional probe, FITC-labeled formaldehyde-treated serum albumin, that accumulated in lysosomes of LSECs (Fig. 4). There was weaker labeling of SCs as compared with LSECs. The labeling of HCs revealed punctuated cytoplasmic structures.

Figure 3.

Expression analysis of kidney-type glutaminase and GAPDH in LSECs, SCs, KCs, and HCs. RT-PCR products detected by way of gel electrophoresis of expected size were obtained. N was the negative control sample, and GAPDH was used as a positive control. Note the weaker signal in the KC band due to the smaller amount of isolated KCs.

Figure 4.

Fluorescence micrographs of immunolabeled glutaminase in cultured NPCs and HCs. (A) Red fluorescence was from immunolabeled anti-glutaminase antibody (large arrows) residing in the lysosomes of LSECs. SCs revealed weak anti-glutaminase labeling (small arrows). (B) Green fluorescence was from endocytosed LSEC-specific probe (FITC-labeled modified albumin) (large arrows). Red and green fluorescence colocalized in the lysosomes of LSECs. (C) Anti-glutaminase labeling was evenly distributed in the cytosol of the HCs. (D) Negative control without anti-glutaminase antibody showed no labeling (scale bars, 20 μm).

Lactate Metabolism

At physO2, the release of lactate in LSEC-bioreactors (87.5 ± 8.9 nM/hour/106 cells) was over five-fold larger than for HC-bioreactors (15.9 ± 1.9 nM/hour/106 cells) in period 1 (PGT < 0.05) (Fig. 5). There was also a large release of lactate by LSEC-bioreactors in period 2 (42.2 ± 6.3 nM/hour/106 cells) (PGT < 0.05), whereas there was a rather small lactate release by LSEC-containing Seq-bioreactors (7.4 ± 1.7 nM/hour/106 cells). The large lactate release coincided with large glucose removal by LSEC-bioreactors (34.8 ± 9.3 nM/hour/106 cells) as compared with HC-bioreactors (3.9 ± 7.3 nM/hour/106 cells) in period 1 (PGT < 0.05) (Table 1). The glucose consumption was larger by the LSEC-bioreactors (28.3 ± 5.7 nM/hour/106 cells) as compared with both the HC-bioreactors and the Seq-bioreactors in period 2 (PGT < 0.05). In period 2, there was a three-fold larger pyruvate removal by LSEC-bioreactors (28.2 ± 10.0 nM/hour/106 cells) than for both the HC-bioreactors and the Seq-bioreactors (7.7 ± 1.7 nM/hour/106 cells and 5.2 ± 1.5 nM/hour/106 cells, respectively) (PGT < 0.05).

Figure 5.

Net release of lactate. The results are presented as differences between the last and the first measurement divided by the number of hours each period lasted per 106 cells (mean ± standard deviation). PGT < 0.05 denoted a significant difference between the groups dependent on time. White bars represent LSEC-bioreactors; black bars represent HC-bioreactors; gray bar represents Seq-bioreactors.

Table 1. Net Release and Removal of Glucose, Pyruvate, and Amino Acids in Bioreactors
 Period 1Period 2
  • Only the amino acids (out of the total 21 amino acids except for glutamine and glutamate) that showed a significant release or removal are presented. Results are presented as the differences between the last and the first measurement divided by the number of hours each period lasted per 106 cells (mean ± standard deviation). Positive values denoted net release into the growth media; negative values denoted net removal from the growth media.

  • Abbreviation: NA, not available.

  • *

    Significant difference between the groups dependent on time (PGT < 0.05).

  • Significant difference between the groups independent of time (PG < 0.05).

  • Significant change for the groups in time (PT < 0.05). The amino acid figures were corrected for autohydrolysis based on cell-free control. Unit: nM/hour/106 cells.

 LSEC-bioreactors−34.8 ± 9.3*−28.3 ± 5.7*
 HC-bioreactors−3.9 ± 7.3−4.9 ± 8.3
 Seq-bioreactorsNA−5.0 ± 8.9
 LSEC-bioreactors−36.2 ± 11.8−28.2 ± 10.0*
 HC-bioreactors−20.6 ± 1.1−7.7 ± 1.7
 Seq-bioreactorsNA−5.2 ± 1.5
 LSEC-bioreactors−20.4 ± 9.5*−12.2 ± 4.7*
 HC-bioreactors−10.8 ± 1.4−11.5 ± 1.2*
 Seq-bioreactorsNA−6.5 ± 0.9
 LSEC-bioreactors22.7 ± 5.85.6 ± 2.1
 HC-bioreactors10.9 ± 0.69.4 ± 1.1*
 Seq-bioreactorsNA5.5 ± 0.6

Functional Features of Bioreactors

The bioreactors maintained their cellular performance throughout the periods: the results are summarized in Table 2. The oxygen uptake rate by LSEC-bioreactors was 25-29 nmol/minute (≈2% of the HC-bioreactors and the Seq-bioreactors). The release of AST (4.3 ± 0.5 mU/hour/106 cells) into the growth media in the LSEC-bioreactors in period 1 was an unexpected finding. We therefore measured the intracellular content of AST by freezing and then thawing isolated LSECs and HCs; the amount of AST was 49.1 ± 13.1 mU/hour/106 cells and 529.4 ± 130.9 mU/hour/106 cells, respectively (data not shown).

Table 2. Functional Features of Bioreactors
 Period 1Period 2
  1. Results are presented as differences between the last and the first measurement in each period divided by the number of hours each period lasted per 106 cells (mean ± standard deviation)—except for the oxygen uptake rate, which is presented as nmol/minute (mean ± standard deviation). Positive values denoted net release to the growth medium. ALT, AST, and LDH units: mU/hour/106 cells. Albumin units: ng/hour/106 HCs.

  2. Abbreviations: ALT, alanine aminotransferase; AST, aspartate aminotransferase; BDL, below detectable limit; LDH, lactate dehydrogenase; NA, not available.

Oxygen uptake rate  
 LSEC-bioreactors29 ± 1725 ± 16
 HC-bioreactors1,246 ± 1031,227 ± 222
 Seq-bioreactorsNA1,337 ± 173
 HC-bioreactors1.7 ± 0.61.6 ± 0.5
 Seq-bioreactorsNA1.7 ± 0.5
 LSEC-bioreactors4.3 ± 0.51.8 ± 0.9
 HC-bioreactors4.9 ± 0.81.9 ± 0.5
 Seq-bioreactorsNA1.1 ± 0.4
 LSEC-bioreactors0.0 ± 0.00.0 ± 0.0
 HC-bioreactors0.1 ± 0.10.0 ± 0.0
 Seq-bioreactorsNA0.0 ± 0.0
 LSEC-bioreactors3.7 ± 1.41.5 ± 0.8
 HC-bioreactors4.6 ± 1.21.7 ± 0.6
 Seq-bioreactorsNA1.3 ± 0.7


Ammonia Metabolism.

Intrahepatic ammonia metabolism is largely credited to HCs. Since the discovery of LSECs in the 1970s,3 no research regarding the role of LSECs in ammonia metabolism has been conducted. Our results show that LSEC-bioreactors released six-fold and four-fold more ammonia and glutamate, respectively, than HC-bioreactors. The LSEC-bioreactors used approximately the same amount of glutamine as HC-bioreactors. In Seq-bioreactors, where LSEC-bioreactors and the HC-bioreactors are connected in sequence, the amounts of ammonia and lactate were as low as the levels observed in HC-bioreactors. RT-PCR analysis revealed the presence of kidney-type glutaminase in porcine LSECs. Immunolabeling revealed colocalization of glutaminase with the specific LSEC-probe (FITC-labeled modified albumin) in the lysosomes of LSECs; however, we do not know the catalytic activity of this enzyme in these acidic vesicles. A study on metabolic acidosis and kidney-type glutaminase suggested that this enzyme has a high activity at a low pH,25 indicating that its lysosomal localization may maintain its enzyme activity.

LSEC-bioreactors used approximately 29 nM glutamine/hour/106 cells, released 31 nM glutamate/hour/106 cells, and released 23 nM ammonia/hour/106 cells. This indicated that after hydrolysis of glutamine by LSEC-bioreactors, little or no further metabolism of ammonia and glutamate occurred by LSEC-bioreactors, as opposed to the HC-bioreactors and Seq-bioreactors. Only 3 nM ammonia/hour/106 cells were released by HC-bioreactors into the growth media together with 12 nM urea/hour/106 cells. Because two ammonia molecules are incorporated into one urea molecule, a total of 24 nM ammonia/hour/106 cells can be explained stoichiometrically. Likewise, glutamate was most likely further metabolized by the HCs, because only 8 nM glutamate/hour/106 cells was released by HC-bioreactors, indicating a true release of 24 nM glutamate/hour/106 cells. The low levels of glutamate suggested a further breakdown of this amino acid in HC-bioreactors.26 The removal of arginine by LSEC-bioreactors was 20.4 nM/hour/106 cells, whereas the release of ornithine into the growth media was 22.7 nM/hour/106 cells. Arginase hydrolyses arginine to urea and ornithine in the urea cycle. Because there was no urea production in the LSEC-bioreactors, the conversion of arginine to ornithine must occur by an alternative enzyme—namely, arginine:glycine aminotransferase.27 This enzymatic reaction uses glycine. The LSEC-bioreactors showed no net release (or production) of glycine. Glycine, however, is also a substrate for several other enzymes that may mask its release.

Glutaminase Activity in Endothelial Cells.

The only previous study of kidney-type glutaminase did not reveal the activity of this enzyme in LSECs.7 However, Wu et al.8 and Leighton et al.28 reported maximal glutaminase activities of 260 and 634 nmol/minute/mg protein lysates in venular and pulmonary endothelial cells, respectively. These studies demonstrated the activity of kidney-type glutaminase, assuming that the kidney-type variant existed in endothelial cells other than LSECs. Calculation of glutamine degradation in intact cells in period 1 showed a cellular glutaminase activity of 10.9 and 0.5 nmol/minute/mg in LSEC-bioreactors and HC-bioreactors, indicating that the activity was 21-fold larger for LSECs than HCs. This difference in glutaminase activity can be explained by different Km values: the Km for the kidney-type variant was 2-5 mM glutamine, and the Km for the liver-type variant was 8–40 mM,29 indicating that LSECs contribute more than HCs to the degradation of glutamine.

The NPCs of the liver occupy only 3% of the liver volume.30 In contrast, NPCs constitute approximately 33% of the total number of liver cells; LSECs constituting 21% of the number.31 The glutamine removal per cell was similar for LSECs and HCs in the Seq-bioreactors group, where the LSEC/HC number ratio was 1:3, resembling the in vivo ratio between these two cell types. Can this be taken to reflect the situation in the intact liver? Considering the cellular architecture of the intact sinusoid, LSECs provide a large and efficient interaction surface through which HCs stay in contact with the blood. This will to a large extent make up for the fact that the HCs constitute three times as many cells as LSECs in the sinusoid.

Lactate Metabolism.

The release of lactate by LSEC-bioreactors in the present study was substantial (87.5 nM lactate/hour/106 cells), and the removal of glucose was 34.8 nM/hour/106 cells. The products of glycolysis are two pyruvate per glucose. The stoichiometric conversion of glucose to lactate yielded 69.6 nM lactate; therefore, approximately 18 nM lactate must come from a different source. One probable source is pyruvate, which under anaerobic conditions is known to be reduced to lactate in vigorously contracting muscles or in erythrocytes. LSEC-bioreactors consumed 36 nM pyruvate/hour/106 cells at physO2; half of this pyruvate was perhaps completely oxidated in the tricarboxylic acid cycle, whereas the other half was reduced to lactate. Recently, we showed that the NPCs released large amounts of lactate with no glucose use at atmosO2.6 In the present study, we improved the culture conditions by adding the growth factors insulin-like growth factor 1, endothelial growth factor, and fibroblast growth factor in a serum-free synthetic growth medium.32 With this growth medium, the LSEC-bioreactors consumed 34.8 nM glucose/hour/106 cells at physO2. Rat LSECs cultivated both at physO2 and atmosO2, with the same growth medium but without the growth factors, released more than two-fold lactate at physO2.14 The large release of lactate at physO2 was not due to anaerobic oxidation of glucose, because the glucose in the growth media in this study was unaltered.

Functional Features of the Bioreactors.

The bioreactors maintained their cellular performance throughout the periods (the OUR and the synthesis of albumin). The OURs of the HC-bioreactors and Seq-bioreactors were nearly the same; the Seq-bioreactors had a slightly higher OUR, perhaps due to the presence of LSECs. There was a surprisingly large amount of AST in the medium of the LSEC-bioreactors, suggesting that this enzyme resides intracellularly in LSECs. This assumption was confirmed when the intracellular content of AST in LSECs was measured. The AST amounts in LSEC-bioreactors cannot be explained by HC contamination, because only 1%–2% of HCs were found in LSEC-bioreactors.

In conclusion, we demonstrate that LSECs release significant amounts of ammonia that may be due to the presence of kidney-type glutaminase in the lysosomes of LSECs.


We express our gratitude to Ellinor Hareide for analytical service, Hege Hagerup for technical assistance, and Tom Wilsgaard for statistical advice.