The influence of oxygen tension on the structure and function of isolated liver sinusoidal endothelial cells

Background Liver sinusoidal endothelial cells (LSECs) are specialized scavenger cells, with crucial roles in maintaining hepatic and systemic homeostasis. Under normal physiological conditions, the oxygen tension encountered in the hepatic sinusoids is in general considerably lower than the oxygen tension in the air; therefore, cultivation of freshly isolated LSECs under more physiologic conditions with regard to oxygen would expect to improve cell survival, structure and function. In this study LSECs were isolated from rats and cultured under either 5% (normoxic) or 20% (hyperoxic) oxygen tensions, and several morpho-functional features were compared. Results Cultivation of LSECs under normoxia, as opposed to hyperoxia improved the survival of LSECs and scavenger receptor-mediated endocytic activity, reduced the production of the pro-inflammatory mediator, interleukin-6 and increased the production of the anti-inflammatory cytokine, interleukin-10. On the other hand, fenestration, a characteristic feature of LSECs disappeared gradually at the same rate regardless of the oxygen tension. Expression of the cell-adhesion molecule, ICAM-1 at the cell surface was slightly more elevated in cells maintained at hyperoxia. Under normoxia, endogenous generation of hydrogen peroxide was drastically reduced whereas the production of nitric oxide was unaltered. Culture decline in high oxygen-treated cultures was abrogated by administration of catalase, indicating that the toxic effects observed in high oxygen environments is largely caused by endogenous production of hydrogen peroxide. Conclusion Viability, structure and many of the essential functional characteristics of isolated LSECs are clearly better preserved when the cultures are maintained under more physiologic oxygen levels. Endogenous production of hydrogen peroxide is to a large extent responsible for the toxic effects observed in high oxygen environments.


Background
The liver sinusoids are lined by endothelial cells that have a unique structure and function essential for hepatic and systemic homeostasis. Much of our understanding of the biology of the LSECs has been generated in experiments curried out on cultured LSECs, mostly derived from rodents. However, in vitro preservation of functionally intact LSECs during isolation and culture has been a challenge because isolated LSECs have poor viability and rapidly loose many of their functional and morphological characteristics [1,2]. Some improvements have been achieved with autologous serum [3], hepatocyte-conditioned medium [2], VEGF or sophisticated synthetic serum-free medium [4].
Traditionally, most of the cell cultivation of today is performed in static culture systems maintained under atmospheric or hyperoxic oxygen levels (20%). As yet, the effects of different oxygen tensions on isolated LSECs have not been investigated. Oxygen is an important modulator of cellular function in both normal and disease states. Thus, hypoxic conditions (5-15 mmHg O 2 ) are characterized by a shift to more anaerobic metabolic processes in the cells, or to the expression of signalling molecules that promote oxygen delivery, such as pro-angiogenic switches [5]. In contrast, hyperoxic conditions (≥ 160 mmHg O 2 ) often results in the formation of reactive oxygen species that are directly implicated in the induction of cell injury via lipid peroxidation and expression of pro-inflammatory cytokines [6]. In the liver, baseline metabolism and functions occur typically in normoxic environments ranging from 30-90 mmHg O 2 . Thus although oxygen gradients occurs between the periportal and perivenous parts of the liver lobule, average oxygen tension is always significantly lower than atmospheric oxygen tension (160 mmHg O 2 ). Variation in oxygen levels could represent a critical element in LSEC viability because it drastically interferes with cellular energy metabolism and the generation of oxidative stress. LSECs are particularly sensitive to hyperoxia and oxidative stress induced either by hydrogen peroxide or tert-butylhydroperoxide [7,8]. Accordingly strategies to reduce oxidative stress such as lowering the oxygen tension might be useful in preserving funtional LSECs.
In this study we compared essential morphological and functional features of LSECs during in vitro culture using either atmospheric oxygen tension or more reduced oxygen conditions. The results indicate that most LSEC functions are better preserved when the cells are incubated under low oxygen tension.

In vivo and in vitro oxygen tension
Baseline oxygen levels were measured in blood samples from the portal vein, hepatic artery or the hepatic vein of anesthetized animals kept mechanically ventilated to stabilize body constants. Similarly, baseline measurements in culture supernatants were obtained after 24 h in CO 2 incubators adjusted to either 20% O 2 or 5% O 2 . Results in Table 1 show absolute values of oxygen measurements given in kilo Pascals (kPa). Of note, oxygen levels encountered in cultures maintained at 5% O 2 are slightly higher than the values found in venous blood entering and leaving the liver.

Cell viability assays and morphological analysis
All in vitro experiments in this study were carried out with an especially tailored serum-free medium which has been shown to preserve LSECs morphology and viability better than regular RPMI, DMEM or their serum-containing variants. Quickly after isolation the cells were placed in atmospheric or low oxygen environments and the morphologic development of the culture was monitored over time by conventional light microscopy. LSEC proliferation analysed by BrDU incorporation was undetectable at any culture condition (data not shown). No significant differences in the morphology were observed during the first 48 h of culture ( . 1a, 1d). However, the number of viable cells per well, as measured by the MTT assay significantly decreased at hyperoxic conditions already at 24 h (Fig. 2). From the 3 rd day in culture, LSECs maintained at atmospheric oxygen tension started to collapse gradually, as observed by the formation of small areas with rounded dying cells and detached cells scattered all over the cultures (Fig. 1b, 1f). The areas with dead cells and cell-remnants were more prominent at the 5 th day of cultures kept at high oxygen levels, representing about 85% of the total seeded area, Blood samples were collected from the indicated vascular beds and from 24 h-conditioned culture supernatants. Glass capillary tubes were used to avoid equilibration of the samples with atmospheric oxygen. The total oxygen content in the samples is given in kilo Pascals (kPa). The results are representative data obtained from three independent measurements.
whereas the LSEC cultures maintained at 5% oxygen preserved intact morphology at the end of the experiment (Fig. 1c, 1g). The MTT measurements also confirmed the faster decay of LSEC cultures incubated at high oxygen levels (Fig. 2c). Necrotic and late apoptotic cells detected by incorporation of propidium iodide were more abundant in cultures maintained at high oxygen tension three days after isolation (Fig. 2a, 2b).

Scanning electron microscopy
Fenestrations represent a specific morphological feature of LSECs. During the initial hours of culture, LSECs demonstrated a well differentiated fenestration pattern with large numbers of fenestrations clustered into liver sieve plates ( Fig. 3a-c). However fenestration was drastically reduced after day 1 and practically disappeared after day 2, regardless of the oxygen tension ( Fig. 3d-f). Average porosity of cells calculated by direct counting showed that fenestration is rapidly lost in plated LSECs independently of the oxygen levels ( Fig. 4). Interestingly, fenestrations appeared to be better preserved in LSECs seeded on collagen-coated dishes than on fibronectin-coated dishes (data not shown).

Scavenger receptor-mediated endocytosis
Under hyperoxia the endocytic capacity was reduced by approximately 50% within 24 h compared with freshly isolated cultures, and had decreased by about 75% by day 2 and 90% by day 3 (Fig. 5). Under normoxia, the loss of endocytic activity was attenuated, being reduced by 32% at day 1, 65% at day 2 and 75% by day 3 (Fig. 5). Of note, the degradation capacity measured in the cultures in terms of acid soluble radioactivity was largely lost within the first 24 h at either oxygen tensions (Fig. 5).

Expression of ICAM-1
Surface expression of ICAM-1 measured by flow cytometry on LSECs maintained for 24 h at diferent oxygen levels showed slightly higher scores in LSECs incubated at hyperoxia, compared with normoxia. Relative mean fluorescence values were 966 for 20% oxygen treatments and 819 for 5% oxygen treatments (Fig. 7) Lactate production and glucose consumption Regardless whether LSECs were cultivated under normoxic or hyperoxic conditions, LSECs consumed insignificant amounts of glucose (Table 2). In contrast, LSECs secreted large amounts of lactate to the supernatant. This lactate production was 2.5 times higher at normoxic conditions compared with hyperoxic oxygen levels.

Production of inflammatory cytokines
Immunoabsorbent assays performed with cell supernatants revealed that IL-1β is minimally expressed by LSECs and remains the same in both tested oxygen conditions (Fig. 6, upper pannel). IL-1β production by LSEC was induced in control experiments challenged to 10 µg/mL LPS (data not shown). In contrast, endogenous IL-6 is Morphological examination of LSEC cultures over time by light microscopy Figure 1 Morphological examination of LSEC cultures over time by light microscopy. Freshly isolated LSECs cultures were established on 24 well plates and incubated either at hyperoxia (a-c) or at normoxia (d-f). The general morphology of the cultures was monitored by light microscopy at day 1 (a, d), day 3 (b, d) and day 5 (c, f) after isolation. Decline of LSECs cultures may be observed in dishes maintained at atmospheric oxygen levels (a-c) after several days of culture.

Production of reactive oxygen and nitrogen species
There was little expression of NO during the first 48 h of culture at both oxygen tensions (Fig. 8a). Elevation of NO levels was observed in LPS-treated cultures that were used as positive controls (data not shown). LSECs cultured at hyperoxia generated nearly three -fold larger amounts of H 2 O 2 when compared to LSECs maintained at normoxia (Fig. 8b). The levels of H 2 O 2 approached the assay detection limit at 48 h of culture in LSECs kept at lower oxygen. Exogenous administration of 1000 U/mL of catalase from the beginning of cell culture, efficiently blocked the production of endogenous H 2 O 2 during the first 48 hours of culture ( Fig. 9a) and was able to revert the cell survival rates observed in cultures kept at hyperoxic conditions ( Fig. 9b).

Discussion
In this study, we demonstrate that atmospheric oxygen levels represent a deleterious environment for LSECs, and that different oxygen environments induce significant functional and structural changes in these cells. LSECs are known to be particularly vulnerable to variations of oxygen levels, as demonstrated by ischemia-reperfusion challenges of livers, or in anoxia-reoxygenation experiments in vitro [9]. During the re-oxygenation periods LSECs, KCs and hepatocytes produce large amounts of oxygen radicals, and LSECs gradually go into apoptosis induced by oxidative stress [8]. These findings suggest that LSECs are poorly equipped to adapt to hyperoxic conditions. It is therefore a curious fact that all published studies using cultured LSECs have been conducted with cells maintained under atmospheric or hyperoxic oxygen tension. We here demonstrate that LSECs cultured under moderately low oxygen tension exhibit improved survival and maintain their in vivo characteristics better as compared to LSECs cultured under atmospheric oxygen tension.
At physiologic conditions the ratio of portal vs. arterial blood flow entering the liver is around 4:1. In the rat system, most arterial blood reaches the sinusoids indirectly, via initial anastomosis between the terminal hepatic arteriole and the portal venule [10]. In average, the oxygen content of hepatic blood is rather low (~55 mmHg). However, oxygen gradients normally exist between the periportal and the perivenous areas of the liver lobule, ranging from 60- Several reports have shown that a major biological function of LSECs is to rid the blood of an array of naturally occurring soluble macromolecular and colloidal waste substances, via clathrin-mediated endocytosis (for review see [17]). Based on the knowledge that receptor-mediated endocytosis represents a characteristic function of LSECs, we compared the ability of the cells to internalize and degrade formaldehyde treated serum albumin (FSA), that is specifically taken up by the scavenger receptor of LSEC, in high and low oxygen environments. Although the endocytic capacity decreased gradually over time in both conditions, the total uptake measured at 24, 48 and 72 h was significantly higher under physiological oxygen conditions than under hyperoxic conditions. Yet, culturing of LSECs under low oxygen levels per se was not enough to maintain the endocytic capacity at the same level as measured in freshly prepared cultures. The reason for this is at least three-fold, based on the fact that LSEC monocultures lack: i) essential factors produced locally or brought to the cells via the portal circulation, ii) interaction with other liver cells, and/or iii) interaction with native extracellular matrix. Focusing in the present work on the impact of oxygen tension, we show here that a low oxygen level in vitro, approaching that of the local sinusoidal environment in vivo, significantly prolong the naturally high scavenger activity of LSECs when compare to traditional cultures established at atmospheric oxygen levels.
Another important physiological function carried out by LSECs in vivo is the filtration of numerous plasma components towards the liver parenchyma through a well organized net of transcytoplasmic holes called fenestrae. This fenestration represents a hallmark of intact mammalian LSEC. It is noteworthy that several reports conclude that the cells undergo a rapid defenestration after isolation and culture [18]. Assessing fenestration using scanning electron microscopy, we found that this feature is gradually lost over time independent of the culture conditions used. Of note, LSECs lost fenestration more rapidly when seeded on fibronectin-coated dishes than on collagencoated ones. This shows that the key factors which enable the maintenance of the LSEC fenestrae in the intact liver are lost upon cultivation, regardless the oxygen levels.
Adaptation to hypoxia is known to induce changes in the energy metabolic routes of cells, most commonly shifting from the oxidative phosphorylation pathways to the glycolytic routes. Morphometric studies of LSECs in the intact liver have shown that the cells contain unusually few mitochondria [19,20]. This observation, along with the fact that rat LSECs produce large amounts of lactate and acetate, even when cultured at high oxygen levels, strongly suggest that these cells are geared to a largely anaerobic type of metabolism. Conceivingly, LSECs perform less oxidative phosphorylation compared to most other cells types, and it has been suggested that in LSECs, glutamine and fatty acid oxidation are the main sources of energy [21]. An alternative path is the anaerobic conversion of pyruvate, originating from the catabolism of glucogenic aminoacids from the growth medium, into lactate [22]. Examining glucose consumption and lactate production under different oxygen tensions to explore possible variations in the energy sources of the cells, we found that glucose was not consumed by LSECs under hyperoxic or normoxic conditions. This suggests that the energy metabolic routes are independent of the oxygen levels. In contrast, the amount of lactate generated by LSECs under normoxic conditions was enhanced almost three times, suggesting that metabolic energy reactions are driven more efficiently under low oxygen.

Production of inflammatory cytokines by LSECs
As a rule, procedures used to isolate and cultivate cells induce cell activation to some extent. Desirable in vitro models should be based on non-activated or low-activated cells. In the present study we measured the production of inflammatory cytokines, reactive oxygen species and the expression of adhesion molecules after cell cultivation as indicators of cell activation. Our results confirm that LSECs produce high levels of IL-6 when cultured under "standard" high (20%) oxygen pressure. Notably, the expression of this cytokine was reduced by 50% when the cells were maintained at low oxygen levels. In contrast, the production of IL-10, an anti-inflammatory mediator, was enhanced when the cells were incubated at 5% oxy-  gen. Flow cytometric analysis of ICAM-1 expression at the cell surface show strong signal on LSECs cultured for 24 h at 20% oxygen. These values are however slightly reduced upon incubation of cells at low oxygen tensions. The overall results indicate that LSECs have a less activated phenotype when they are incubated at low oxygen levels.

ICAM expression on cultured LSECs
The transfer of LSECs from an in vivo low oxygen tension to an in vitro high oxygen tension may exert effects on the cells similar to those observed in LSECs during hypoxiareoxygenation of the intact liver. Indeed, we observed large production of hydrogen peroxide by LSECs when the cells were kept at atmospheric oxygen conditions. Of note, this production was much lower at low oxygen tension. Hydrogen peroxide induces toxic effects on LSECs, mostly because these cells are not well equipped to metabolize this reactive substance [8,9]

Conclusion
In this study we report that atmospheric oxygen tension has harmful effects on isolated rat LSECs during long term cultivation. These effects may be largely abrogated by incubation of the cells at physiological O 2 conditions. Our findings are compatible with a better preservation of essential morphologic and functional LSEC features under more physiological oxygen tensions. Based on these data, we recommend low oxygen environments for cultivation of LSECs, especially when long-term cultures are used.

Endocytosis measurements
Cultures of LSEC (0.5 × 10 6 ) were established in 2 cm 2 wells and maintained in serum-free medium. After experimental treatments, cells were cultured for 90 min at 37°C in 250 µl of RPMI, containing 1% Human Serum Albumin (HAS) and trace amounts (40.000 cpm, 50 ng/ml) of radioiodinated formaldehyde-treated bovine serum albumin ( 125 I-FSA). Endocytosis experiments were terminated by transferring incubation medium and two washing volumes to tubes containing 800 µl of 20% TCA, thereby inducing precipitation of non-degraded proteins. After centrifugation, precipitated and soluble radioactivity was measured using a Geiger counter. Acid soluble radioactivity was considered to indicate degraded FSA, and results were evaluated from total radioactivity added to cultures. Cell associated radioactivity was calculated by solubilisation of cell monolayers with 1% SDS, and measured with a γ-counter (Cobra II, Packard). Values were normalized for the total number of cells counted per well.

Scanning electron microscopy
Scanning EM was performed as previously described [24]. LSECs that had been cultured in hypoxic and normoxic conditions for 6, 24 and 48 h were fixed for 1 hour with 2.5% glutaraldehyde in 0.1 mol/l sodium cacodylate buffer (1% sucrose). Coverslips were treated with tannic acid (1% in 0.15 mol/l cac. buffer), osmicated (1% OsO4/ 0.1 mol/l cacodylate buffer), dehydrated in a series of ethanol gradients and finally incubated in hexamethyldisilazane for 2 min. Gold coated coverslips were viewed using a Jeol scanning microscope. Five representative cells from each time point were photographed (magnification 3,000-5,000 ×) and fenestral diameter and porosity (percentage of surface area occupied by fenestrations) analysed using ImageJ software. The results are expressed as mean ± S.D.

ELISA measurements of cytokine production
The production of the inflammatory mediators IL-1β, IL-6 and IL-10 in LSEC culture supernatants was determined with specific rat IL-1β, IL-6 and IL-10 ELISA kit (R&D Systems, Minneapolis, USA) according to manufacturers' instructions. Briefly, after the experimental incubations of cultures, the supernatants were collected and underwent high speed centrifugation, then kept at -20°C. 96 wellplates were coated with "capture" antibodies and 100 µl of 1:2 diluted supernatants were added to each well and incubated for 2 h at room temperature. The "detection" antibody was then applied for 2 h and the wells were ultimately subjected to peroxidase reaction. Absorbance was measured at 450 nm and the values were converted into µg/ml according to the standard curve. Values were normalized after the total number of cells counted per well.

ICAM expression by flow cytometry
For measurements of ICAM-1 expression at the cell surface, LSEC were incubated at different oxygen tension during 24 h, detached from wells by 20 min incubation in EDTA buffer, and immediately fixed in 4% paraformaldehyde. Fixative was removed by cell sedimentation and the cells were then resuspended in 500 µl of PBS containing 1% BSA. Specific monoclonal antibody against rat ICAM (Biodesign International, Saco, ME, USA) was added to tubes containing fixed LSEC and incubated during 30 min at room temperature. Unlabeled antibody was eliminated by a series of cell washings, followed by incubation with a secondary antibody against mouse IgG FITC-congugated (Dako Denmark A/S). A group of cells incubated only with secondary antibodies was used as negative controls. Fluorescent cells were then analyzed on a BD FACScan flow cytometer