excessive inflammation known also as systemic inflammatory response syndrome (SIRS) activates pathological signaling cascades that cause multiple organ failure (MOF). Interestingly, the organs of animals or patients, even those who died of MOF, often appear normal (34) with neither major necrotic areas nor a relevantly increased number of apoptotic cells (except for lymphocytes) (14, 42). This suggests that mechanisms other than cell death cause organ failure induced by SIRS. Mitochondrial dysfunction as a reason for MOF was suggested as an alternative to the cell death mechanism (33). However, results have been contradictory. While some investigators have reported impaired mitochondrial function (4, 10, 15, 21), others have failed to find any abnormality (9, 22, 39) or described even an improvement of mitochondrial respiration (20, 38, 40) in response to infection or bacterial endotoxin. Similar to mitochondrial function, reports on the ultrastructural examination of mitochondria appear contradictory. While some investigators have shown pathological changes in mitochondrial ultrastructure (3), others reported the absence of any remarkable changes in mitochondria both in animals and in patients who died from sepsis (42). Also, in our recent study, we have shown unchanged mitochondrial morphology in rat livers subjected to endotoxin challenge.

More recent observations suggest that endoplasmic reticulum (ER) stress can initiate inflammation (48). The ER is the major site in the cell for protein folding and trafficking and is central to many cellular functions. Failure of the ER's adaptive capacity results in activation of the unfolded protein response (UPR), which intersects with many different inflammatory and stress signaling pathways (13). It has been reported that the activity of cytochrome P-450 isoenzymes that reside in the ER is decreased upon endotoxic shock and that significant changes occurred in the proteome of the ER (18). Apart from mitochondria and ER, the cytoplasmic compartment may also contribute to inflammation-induced cellular dysfunction via upregulation of inducible nitric oxide synthase (iNOS) and excessive production of nitric oxide and/or its redox products (12). The aim of this study was to understand the mechanisms underlying liver failure induced by inflammation by analyzing ultrastructural and metabolic changes in diverse intracellular compartments.

Glossary

MATERIALS AND METHODS

Animal model.

All experiments were carried out in accordance with the Guide for the Care and Use of Laboratory Animals and were approved by the “Animal Protocol Review Board” of the municipal government of Vienna, Austria. A severe endotoxic shock, accompanied by signs of liver damage (high levels of alanine aminotransferase = high ALT), was induced by 8 mg/kg lipopolysaccharide (LPS) intravenously in adult male Sprague-Dawley rats. Similar and even slightly higher concentrations of LPS were previously used by many other research groups (41, 44, 45) and resulted in similar mortality. Originally 52 animals were used in this experiment. Control animals (n = 10) received saline intravenously, and 42 animals received 8 mg/kg LPS intravenously and were killed 2 (n = 7), 4 (n = 8), 8 (n = 8), and 16 (n = 9) h after LPS injection; several animals (n = 10) died between 8 and 16 h. The tissues of at least 3 of 9 animals that survived 16 h and 3 out of 10 control animals were randomly selected for further detailed examinations, including electron microscopy, proteome analyses in subcellular fractionation, and mRNA expression analysis. Mild endotoxic shock was induced by 8 mg/kg LPS intraperitoneally, with animals showing no signs of liver damage (normal ALT levels). Rats were killed by decapitation 16 h after injection. Medial, right lateral, and caudal liver lobes were used for preparation of subcellular fractions and proteome analysis. The left lateral lobe was used for all other examinations [RT-PCR, electron microscopy, histological examination, Western blot.

Blood analysis.

As a measure of liver damage, the ALT levels were determined as enzymatic activity (8).

Histological examination.

Liver tissues were fixed in 4% formaldehyde solution buffered with phosphate for 48 h and embedded in paraffin. Tissue sections were stained with hematoxylin and eosin and examined by light microscopy.

Electron microscopy.

Fresh liver tissue was immersed in 2.5% glutaraldehyde in 0.1 M sodium cacodylate and incubated at 4°C overnight. Afterward, the samples were washed three times with 0.1 M sodium cacodylate buffer with the same osmolarity as the primary fixative (adjusted with NaCl). Postfixation was carried out in 0.5% osmium tetroxide in 1% potassium ferrocyanide for 2 h. After the samples were rinsed, they were dehydrated in graded ethanol series and infiltrated in low-viscosity resin (Agar Scientific, Stansted, UK) with acetonitrile (Sigma-Aldrich, Munich, Germany) as an intermedium and polymerized at 60°C. Ultrathin sections were stained with uranyl acetate and lead citrate. The sections were examined on a Zeiss EM 902 electron microscope (Oberkochen, Germany).

Quantitative real-time PCR.

Total RNA from snap-frozen rat liver tissue (50–100 mg) was isolated using TriReagent (Sigma, St. Louis, MO). Reverse transcription was carried out with 1–2 μg total RNA by oligo(dT) priming (Invitrogen, Carlsbad, CA) using 100 units of SuperScript II (Invitrogen). Amplification of the cDNA targets was achieved by quantitative RT-PCR. Specific primers and hydrolysis probes (Applied Biosystems, Brooklyn, NY) were used for the amplification in an Applied Biosystems 7500 Fast real-time PCR machine (Applied Biosystems), as described elsewhere (30). SYBR GREEN I (0.5x; Sigma) was used with specific primer sets for the amplification in an iCycler (Bio-Rad, Hercules, CA), as previously described (7). All samples were run in duplicates, including RT-minus and NTCs as negative control on each plate. Amplification values were determined using the system built-in detection software, and the relative expression data were calculated by using the comparative ΔΔC_T method. The final data were elaborated by calculating a normalization factor using three reference genes (cyclophilin A, GAPDH, and HPRT).

Preparation of subcellular liver fractions.

Immediately after decapitation, liver was extracted and placed in ice-cold sucrose buffer (0.25 M sucrose, 10 mM Tris·HCl, 1 mM EDTA, and 0.1% ethanol, pH 7.4), diced, and rinsed with the same buffer to remove remaining blood. After being blotted dry with paper, the liver pieces were weighed, and the same buffer was added in a ratio of 1:6 liver-buffer (wt/vol). Liver was homogenized using a Potter-Elvehjem homogenizer. Subcellular fractions were prepared as previously described (17), with the exception that light and heavy mitochondrial fractions were not separated.

Two-dimensional difference gel electrophoresis.

The protein content in the three intracellular fractions was determined by the Coomassie G-binding assay (1). Samples were preelectrophoretically labeled with fluorescent CyDyes (GE Healthcare Life Sciences, Munich, Germany). Two-dimensional difference gel electrophoresis (2D-DIGE) was performed in accordance with existing protocols, using nonlinear pH 4–10 immobilized gradient strips of 10 cm length and 140 × 140 × 1.5 mm SDS-PAGE gels (T = 10–15% linear gradient, C = 2.7%) (23). After electrophoretic separation of the proteins, gels were scanned on a Typhoon 9400 Imager and evaluated with DeCyder Software (GE Healthcare Life Sciences). The ratios between volumes of single spots in the samples and the corresponding spots in the internal standard were calculated. Protein spots differentially expressed between groups were extracted, using volume ratios (treated/untreated group) and Student's t-test as selection criteria.

Protein identification by nano HPLC-MS/MS protein sequencing.

2D-DIGE gels were stained by a silver staining procedure compatible with mass spectrometry (MS) (23). Differentially regulated spots were excised and subjected to in-gel protein reduction, alkylation, and trypsinization (47). Nano HPLC separations were performed on an UltiMate system from Dionex online connected with an Ion Trap Mass Spectrometer (LCQ Deca XPplus; Thermo Finnigan) as described previously (47). Analysis of MS/MS spectra with respect to peptide identity was routinely performed by applying the MASCOT (28) (Matrix Science) search engine. A peptide was reliably identified only if the individual peptide score was ≥40. A protein was regarded as identified if at least two unique peptides were identified. All peptides were blasted against the UniProt KB.

Western blot.

Liver tissue was homogenized in 0.1 M Tris·HCl, pH 8.5, containing protease inhibitors (Complete Mini; Roche, Basel, Switzerland) and sonicated on ice. Next, 25 μg protein denatured in Laemmli's sample buffer (62.5 mM Tris·HCl, pH 6.8, 2% SDS, 5% β-mercaptoethanol, 4% glycerol, 1 mM phenylmethylsulfonyl fluoride, and 0.05% bromphenol blue) was separated by 12% SDS-PAGE and transferred to nitrocellulose. Membranes were blocked for 1 h in PBS containing 0.05% Tween 20 and 5% (wt/vol) nonfat dried milk. Membranes were then incubated with primary antibodies as follows: C/EBP-homologous protein (CHOP) (1:1,000), 78-kDa glucose-regulated protein (GRP78) (1:1,000), Bcl-2-interacting mediator of cell death (Bim) (1:1,000), homocysteine-induced ER protein (HERP) (1 in 4,000), iNOS (1:2,000; BD Biosciences Transduction Laboratories, Lexington, KY), or actin (1:1,000) overnight at 4°C. This was followed by incubation with appropriate horseradish peroxidase-conjugated goat secondary antibody. Protein bands were visualized using the Supersignal West Pico chemiluminescent detection kit (Pierce, Appleton, WI) and detected on an X-ray film (Agfa-Gevaert, Mortsel, Belgium). HERP antibody was a gift from Osamu Hori, Department of Neuroanatomy, Kanazawa University Graduate School of Medical Science, Japan.

Statistics.

The data are presented as means ± SE. The statistical significance was estimated by two-tailed unpaired t-test if two groups of variables were compared and by one-way ANOVA test followed by post hoc test for least-significant difference for multiple comparisons. Significance was set to the value of P < 0.05.

RESULTS

Time course of necrosis markers.

ALT was significantly increased in the plasma of animals receiving LPS intravenously 4 h postchallenge and continued to increase until 16 h. Lactate dehydrogenase (LDH) and creatinine levels were significantly increased by 2 h after LPS administration. Creatinine levels reached a steady state at 4 h, and LDH levels continued to increase until 16 h (Fig. 1, left). Ten out of 19 animals died between 8 and 16 h.

Time course of gene expression.

The spliced form of the X-box-binding protein 1 (XBP1) mRNA (Fig. 1, right, inset), a marker of ER stress and early upstream marker of UPR, was significantly increased at 4 h and remained increased until 16 h (Fig. 1D), suggesting ongoing ER stress. GRP78, a downstream marker of UPR that represents the prosurvival branch of UPR toward ER recovery, as well as CHOP, a marker of prodeath aspect of UPR, were significantly upregulated at the mRNA level only at 8 h (Fig. 1, E and F, respectively).

Three randomly selected animals from both control and LPS groups (16 h) were selected for further examinations (see materials and methods). For comparison, three animals were additionally challenged with the same dose of LPS intraperitoneally to obtain a mild form of shock. The ALT values in control, mild shock, and severe shock groups were 47 ± 10 (n = 3), 61 ± 3 (n = 3), and 2,500 ± 416 (n = 3) IU/l, respectively.

Histological examination.

Histological examination of the liver of animals treated with LPS intravenously revealed moderate focal necrosis and increased leukocyte infiltration. At the organ level, necrosis appeared preferentially in the periphery of the liver (Fig. 2, A and B), but more than 90% of the liver tissue appeared completely normal.

Ultrastructural evaluations.

Ultrastructural evaluations of the liver tissue revealed obvious alterations in the ER morphology of the hepatocytes. In contrast to the parallel membranes and a granular content that is observed in the liver from control animals (Fig. 3A), in the liver from LPS-challenged animals the ER cisterns were locally dilated and electron lucent. Both the smooth and the rough ER were affected. In the latter, the dilated regions were deprived of ribosomes. The degree of dilation varied within the liver and was highest in the surroundings of necrotic areas. In these regions, almost the entire ER was affected, the dilations were massive, and the ER sometimes even appeared vesiculated/fragmented (Fig. 3, B–D).

The fact that dilated ER appeared around necrotic areas (Fig. 3E) may suggest that both morphological alterations are associated phenomena. To address the question of whether the ER dilations occur in association with necrosis or not, we examined the liver ultrastructure in mild endotoxic shock, which was neither accompanied by increased ALT levels (ALT level between 42 and 85) nor by focal necrosis. Ultrastructural examination of these liver samples also revealed dilated ER, but to a much lesser degree compared with severe shock (Fig. 3, I–K). Again, rough as well as smooth ERs were affected. Importantly, in all cases, the dilations of the ER typically appeared adjacent to mitochondria or peroxisomes.

The other organelles, including mitochondria, peroxisomes, and Golgi apparatus, appeared intact. Necrotic areas were massively infiltrated by leucocytes, mainly granulocytes (Fig. 3, C and E). The majority of the granulocytes were neutrophils with large amounts of glycogen aggregates in their cytoplasm. Specific changes also appeared in the blood vessels. In the sinusoids, the endothelium was sometimes missing, and a fibrin-like material was observed in these areas (Fig. 3F).

In one animal, many apoptotic cells were detected in the liver. Apoptotic cells and apoptotic bodies were found in the sinusoids between intact erythrocytes and granulocytes (Fig. 3, G and H). Some of the apoptotic cells could be identified as neutrophils. Hepatocytes did not show signs of apoptosis, but they contained apoptotic bodies in their endosomes and large heterophagosomes in various stages of enzymatic degradation (Fig. 3, G and H). Heterophagosomes could either be formed from apoptotic bodies or from cell debris originating from neighboring necrotic cells or cell debris floating in the sinusoids. Heterophagic activity and material digestion were further found in granulocytes and Kupffer cells (data not shown).

Proteome analysis.

To better understand the LPS-mediated changes of the ER, we analyzed the protein composition of subcellular fractions, i.e., mitochondrial, ER, and cytosolic fractions, by 2D-DIGE. From the over 1,000 spots/fraction, the most abundant and the most differentially regulated spots (Fig. 4) were selected and subjected to mass spectrometric analysis. Differentially regulated proteins were found in all subcellular fractions. Most affected proteins of the mitochondrial fraction showed upregulated spot volumes and confirmed our previous findings (24). Similarly, we also found mainly upregulated protein spots in the cytosolic fractions. In addition, also cytosolic levels of iNOS were increased in LPS-treated animals (detected in Western blots, Fig. 5B, bottom). Conversely, in the ER fraction, we found that the most affected protein spots were downregulated (Fig. 4 and Table 1). The most dramatic changes were determined for proteins that are essential for proper ER function, such as assistance in proper protein folding (GPR78), formation of disulfide bonds (protein disulfide isomerase family A, member 3), and transport across ER membranes (transitional ER ATPase). These findings suggest that ER function may be critically affected and that the ultrastructural changes of the ER occurred as part of the UPR response to ER stress. This indicates that systemic inflammation affects primarily the function of the ER, a condition that we refer to as ER dysfunction or failure. Key proteomic findings were verified by Western blots.

Expression of genes relevant for cell death.

We also profiled the mRNA expression of the members of the Bcl-2 family, a protein family regulating apoptosis (Bim, Bcl-XL, Bak, Bmf, Hrk, Spike, Puma, BNIP3, Bax, Bcl-2; see definitions in glossary). BIM, Bcl-XL, and BAK were significantly upregulated, whereas Bmf was significantly downregulated (Fig. 5 top). Hrk, Spike, Puma, BNIP3, Bax, and Bcl-2 remained statistically unchanged. Bcl-2, however, had a strong trend to increase (P = 0.051). Thus, no clear proapoptotic phenotype was apparent in the livers of LPS-treated rats, confirming the findings of the ultrastructural analysis. Determination of selected proteins (GRP78, Herp, Bim, CHOP) by Western blot in nonfractionated liver samples showed no difference in protein levels except a slight trend to increase in the levels of GRP78 in LPS-treated animals (Fig. 5A, bottom).

DISCUSSION

The present study identifies the ER in the liver as the key target of inflammatory mediators induced by LPS in rats with endotoxic shock. Because the major liver functions such as protein synthesis and detoxification are centered in the ER, our findings can satisfactorily explain liver dysfunction that does not involve cell death, a phenomenon observed in different experimental models (34) and in patients with sepsis (42). There is accumulating evidence implicating prolonged ER stress in the development and progression of a number of diseases, including neurodegeneration, atherosclerosis, type 2 diabetes, liver disease, and cancer (27).

Irreversible ER stress, resulting in sustained or chronic ER dysfunction, activates the prodeath signaling pathway, which triggers apoptotic cell death (31). Interestingly, the apoptotic cell death could be prevented at multiple levels by interrupting the prodeath signaling pathway of UPR, e.g., CHOP deficiency (37). However, in this case, the initial cause for ER stress will persist, and the ER remains dysfunctional. Several models show that persistent ER dysfunction will lead to critical changes in the liver metabolism, such as steatosis or fibrosis, leading to liver cell damage and finally cell death (5, 29, 49). All this shows that ER stress results in a disturbed protein and lipid metabolism, thus representing cellular dysfunction, but does not induce cell death by itself. The cell death is triggered by the UPR. Thus, persisting ER stress will ultimately result in liver dysfunction, either because of an increased cell death or by the unresolved cellular dysfunction. We hypothesize that the pathological situation evoked by SIRS interferes with the UPR in a way that neither recovery nor apoptosis occurs sufficiently in response to ER stress. Consequently, ER dysfunction is remaining and liver failure will ultimately result. However, further studies are required to prove this assumption.

ER damage.

The major ultrastructural change in livers of rats subjected to LPS was dilation of the ER. In general, dilation of the ER may either result from a failure of synthetic function or from liquid accumulation as a result of the loss of osmotic equilibrium (25). The ER structural alterations in the liver samples of the LPS-treated rats were most visible in areas surrounding the necrotic zone, suggesting that ER dilations may be associated with necrotic cell death. However, the focal necrosis areas were too small to cause liver failure. ER dilations were not only the most dramatic but also relatively early alterations in the livers of LPS-treated rats. Dilated ER, although to a lesser extent, was also found in the livers of rats with mild endotoxic shock, which did not demonstrate any features of necrosis. Thus, ER dilation appears to be an outcome of ER dysfunction rather than from hepatocyte damage. We did not observe changes in the ultrastructure of mitochondria and in cytosol. Therefore, the ultrastructural findings identify the ER as the most susceptible compartment to LPS-induced damage. The proteome analysis of ER (Table 1) showed that the expression levels of most protein spots decreased in response to LPS. An important finding was that two ER luminal chaperone proteins [the chaperone GPR78 and two main spots of the protein disulfide isomerase A-3 (PDI A-3), a protein involved in the formation of disulfide bonds], both responsible for proper protein folding, were significantly downregulated, suggesting the impairment or at least considerable alteration of protein turnover in the ER. Specifically, the level of intact GPR78 was decreased as detected by 2-DE. This was concomitant with a marked appearance of GRP78 breakdown products, as shown by 2-DE Western blots. For PDI A-3, the two acidic spots of the 2-DE pattern were noticeably downregulated in LPS treatment samples, whereas the pattern of the more alkaline spots remained unchanged with the treatment. A marked downregulation of transitional ER ATPase, a protein facilitating the transport of proteins from ER to Golgi and subsequent secretion, was also noticed. These data suggest that not only synthesis and folding but also the transport of proteins to Golgi and overall secretory pathway may be impaired in LPS-treated animals.

ER stress, UPR.

Because ultrastructural and proteome analysis of the ER suggested impaired function, we questioned whether the ER homeostasis may be disturbed by sustained ER stress. To this aim, we determined the progression of known markers of ER stress at the mRNA and protein level in whole liver samples. ER stress is a condition in which the protein-folding capacity of the ER is exceeded, and which then leads to the accumulation of misfolded proteins (32, 36). Cells respond to ER stress by inducing UPR, autophagy, or cell death (11). ER stress is sensed and the responses are mediated by three ER transmembrane proteins (30, 36). One of these proximal sensors of ER stress is IRE1, which once activated splices the mRNA of XBP1 to generate a highly active transcription factor. In our model, we found increased splicing of XBP1 mRNA by 4 h, which remained at elevated levels until 16 h, a clear indication for sustained ER stress. Activation of IRE1α/XBP1 signaling results in increased expression of genes encoding chaperones needed to deal with unfolded proteins in the ER, especially GRP78 mRNA, which were found moderately but significantly upregulated at 8 h. GRP78 is an ER chaperone that not only assists proper folding but also keeps ER stress sentinels in an inactive form (36). Although we have found an upregulation of GRP78 at the mRNA level, this was not fully translated into increased levels of GRP78 protein in complete liver homogenates and resulted in even slightly decreased levels in ER preparations. Moreover, when analyzing this subcellular fraction by 2-DE, we found spots representing significantly downregulated intact GRP78, in addition to breakdown products. This observation may suggest that the ER contained less intact or less stable GRP78. Thus, despite activation of UPR, proper function was apparently not restored. A similar picture was obtained for another mediator of UPR, CHOP. The levels of CHOP mRNA were increased, implying activation of PKR-like ER kinase (PERK), another ER stress sentinel (36). Activation of PERK by autophosphorylation results in phosphorylation of eukaryotic initiation factor 2α and sequential activation of the transcription factors ATF4, ATF3, and CHOP. The PERK-ATF4-CHOP pathway is thought to mediate the prodeath response during ER stress (26, 50). Despite increased mRNA levels of CHOP, its protein levels were only moderately changed. Thus, the activation of apoptosis was also not achieved by UPR in this model.

The role of mitochondria.

Mitochondria appeared relatively intact as shown by electron microscopy. Some alterations in protein expression (upregulation of mitochondrial superoxide dismutase and several ATP synthase spots) appear to have a beneficial effect and are in line with our previously published report (24). We have observed that dilated ER regions are always located in close vicinity to mitochondria or peroxisomes. A common feature of these two subcellular compartments is that both generate reactive oxygen species (ROS). Increased generation of ROS may damage ER via local alteration of the ER Ca²⁺ flux (46). Modulations of Ca²⁺ flux are achieved by modifications of the Ca²⁺ transporters, i.e., ryanodine receptor, involving the close association with mitochondria. ROS were shown to modulate the activity of the ryanodine receptor (6), suggesting that mitochondria may affect ER function directly via increased generation of ROS. Bak and Bcl-XL were reported to cause ER dilations and formation of vesicles by modulating Ca²⁺ trafficking and Ca²⁺ flux across the ER membrane when the ryanodine receptor was blocked, or under conditions of ER stress (16, 25).

The role of cytosol.

Electron microscopy did not reveal significant changes in cytosolic fractions of hepatocytes. Proteome analysis (Table 1) revealed changes in the protein patterns of peroxiredoxin-1 (PRDX-1), both in cytosol and ER. PRDX-1 is known as a protein that exists as several spots in two-dimensional electrophoresis; the pattern (appearance and intensity of different spots) reflects the redox states. More acidic spots are usually interpreted as a sign of oxidation (2). We observed upregulation of acidic spots (as spot 13) in cytosol and two distinctly regulated PRDX-1 spots (spots 21a and b) in ER. Although both of the latter were downregulated, the more acidic spot 21a was decreased to a lower extent. Both observations are possibly a sign for increased oxidation, suggesting the involvement of oxidative stress. This supports the assumption that ROS may contribute to pathological signaling occurring in this model. In fact, increased levels of mitochondrial ROS have been shown in previous studies in similar models (19).

Protein expression changes.

Proteomic investigations by 2-DE and protein analysis by Western blotting cannot give hints about misfolding of proteins, since they reveal proteins in a denatured/reduced or a reduced state. Thus, they cannot replace functional or histological experiments but are good tools to complement them. Especially, 2-DE allows further insights into the substructure of proteins, displaying protein isoforms (e.g., due to posttranslational modifications that cause a change in charge on protein) or fragments, which allow more detailed information about ongoing (patho)physiological processes in the body as a reaction to LPS challenge. Thus, in the present case, spot shifts could be seen for PRDX-1, confirming/suggesting oxidative processes, and a higher degree of fragmentation in CPS-1, GRP78, and catalase suggest a higher susceptibility of those proteins in LPS-treated animals. Not all of these changes can be seen with other tests, e.g., SDS-PAGE, which is most likely related to different sample pretreatment or with analysis of the particular gene level, possibly because of delayed or inhibited protein expression. In our study, some of the investigated proteins were found in more than one cell organelle but usually were prominent in only one, for example, CPS-1 in rat liver mitochondria, GRP78 in ER. The separation of necrotic, apoptotic, and normal cells was not possible for this analysis; whole tissue was homogenized. However, taking into account that dead cells were <10%, we can conclude that >90% of the changes come from normally appearing cells.

Some proteins were found in more than one compartment, but, as visible in Fig. 4, with highly different intensity. For example, CPS-1 is the main protein in mitochondria and is only found in small amounts in cytoplasm. Similarly, GRP78 is mainly found in ER and only in traces in mitochondria. The localization of newly synthesized GRP78 in mitochondria has already been reported (35).

Necrosis and apoptosis.

Our findings show that neither necrosis nor apoptosis appears to be a sufficient explanation for liver dysfunction. However, apoptosis has been considered to be an important reason for organ dysfunction, in particular in sepsis models (43). To carefully consider this aspect, we additionally analyzed several other markers involved in apoptotic signaling processes. Interestingly, we found several such markers upregulated at the mRNA level, but not at the protein level. This suggests that apoptotic pathways, including mainly members of the Bcl-2 family, were activated but not yet executed. Alternatively, these proteins may be involved in osmotic changes of the ER, since recently Bcl-2 proteins were reported to cause ER dilation (25). Collectively, these data confirm our findings at the proteome and ultrastructural levels that show that apoptosis was clearly detected in leucocytes in the sinusoids but not in hepatocytes; this is in line with a recent paper by Watanabe (42).

Thus, these data indicate that not excessive cell death but possibly the loss of ER function is an explanation for liver failure (see Fig. 6). This interpretation is supported by the proteomic analysis of the three subcellular fractions, mitochondria, ER, and cytoplasm, showing that the most dramatic changes are found in the proteins of the ER fraction. Our data suggest that inflammatory mediators induce unresolved ER stress in liver cells with clear biochemical and morphological manifestations, creating a large pool of living but functionally impaired cells. Necrosis or apoptosis are of marginal significance for liver dysfunction.

GRANTS

The study was financed by the Ludwig Boltzmann Society of Austria and Allgemeine Unfallversicherungsanstalt research centre.

DISCLOSURES

The authors declare that they have no competing interests.

AUTHOR CONTRIBUTIONS

Author contributions: S.N., I.M., E.T.K., R.T.H., O.H., and B.G. performed experiments; S.N., I.M., J.C.D., E.T.K., S.G., O.H., B.G., A.S., A.K., and A.V.K. analyzed data; S.N., J.C.D., S.G., O.H., A.S., H.R., and A.V.K. interpreted results of experiments; S.N., I.M., J.C.D., E.T.K., S.G., R.T.H., A.K., and A.V.K. prepared figures; S.N., J.C.D., H.R., and A.V.K. drafted manuscript; S.N., I.M., J.C.D., E.T.K., S.G., R.T.H., O.H., B.G., A.S., A.K., H.R., and A.V.K. approved final version of manuscript; J.C.D., H.R., and A.V.K. conception and design of research.