Potential conflict of interest: Nothing to report.
Alpha-1-antitrypsin (a1AT) deficiency is caused by homozygosity for the a1AT mutant Z gene and occurs in 1 in 2000 births. The Z mutation confers an abnormal conformation on the protein, resulting in an accumulation within the endoplasmic reticulum of hepatocytes rather than appropriate secretion. The accumulation of the mutant protein is strikingly heterogeneous within the liver. Homozygous ZZ children and adults have an increased risk of chronic liver disease, which is thought to result from this variable intracellular accumulation of the a1AT mutant Z protein. Previous reports have suggested that autophagy, mitochondrial injury, apoptosis, and other pathways may be involved in the mechanism of hepatocyte injury, although the interplay of these mechanisms in vivo is unclear. In this study, we examine a well-characterized in vivo model of a1AT mutant Z liver injury, the PiZ mouse, to better understand the pathways involved in this disease. The results show an increase in the stimulation of the apoptotic cascade in hepatocytes, the magnitude of which strongly correlates to the absolute amount of the a1AT mutant Z protein accumulated within the individual cell. Increases in apoptotic regulatory proteins are also detected. Conclusion: These data, combined with previous work, permit for the first time the construction of a hypothetical hepatocellular injury cascade for this disease involving mitochondrial injury, caspase activation, and apoptosis, which takes into account the heterogeneous nature of the mutant Z protein accumulation within the liver. Further development of this hypothetical cascade will focus future research on this and other metabolic liver diseases. (HEPATOLOGY 2007.)
The genetic disease alpha-1-antitrypsin (a1AT) deficiency is caused by homozygosity for the a1AT mutant Z gene and occurs in 1 in 2000 births.1 The Z mutation confers an abnormal conformation on the nascent polypeptide, resulting in an accumulation of the mutant protein within the endoplasmic reticulum (ER) of hepatocytes rather than the appropriate, highly efficient secretion of the wild-type (WT) protein. Homozygous, ZZ individuals have an increased risk of chronic liver disease and hepatocellular carcinoma resulting from this intracellular accumulation of the a1AT mutant Z protein.
Studies of the a1AT mutant Z protein molecule have shown that the nascent polypeptide has a tendency to form unique protein homopolymers.1, 2 Although this loop-sheet insertion mechanism of a1AT mutant Z protein polymerization, in which the reactive site loop of one molecule inserts into a surface groove in a neighboring molecule, does not involve the formation of covalent bonds, physical-chemical studies of these polymers suggest that this conformation is highly favored. It is proposed and supported by some published data that the liver injury in humans with a1AT deficiency is directly related to the hepatic accumulation of the polymerized a1AT mutant Z protein.3–5
Our laboratory and others have reported a series of studies that have begun to examine the mechanism of liver cell injury in a1AT deficiency.3–9 The data suggest that an accumulation of the a1AT mutant Z protein in the ER of hepatocytes activates autophagy, causes mitochondrial injury and mitochondrial autophagy, and is associated with caspase activation and apoptosis. We have further shown that experimentally increasing the quantity of the polymerized a1AT mutant Z protein in the liver is directly related to increased injury.4 However, an analysis of both human liver and model systems shows that the accumulation of the mutant protein in the liver and the resulting injury are very heterogeneous. The basis for the heterogeneity within the organ in vivo is unclear but unusual, in that all of the hepatocytes carry the same homozygous mutant Z genetic complement. It has also been difficult to merge the in vitro data on the intracellular consequences of a1AT mutant Z intracellular protein retention with the clinical observation that this and many other metabolic liver diseases are slowly progressive, with a low grade of chronic injury present over many years in a given patient.
In this study, we test the hypothesis that heterogeneity in the accumulation of a1AT mutant Z protein among hepatocytes is directly related to variable hepatocellular injury within the liver in vivo. Furthermore, we investigate how important injury pathways, such as apoptosis, play a role in the progression of liver disease in an in vivo model of this metabolic liver disease.
PiZ mice were maintained on a C57Bl/6J background, and C57Bl/6J mice were used as controls as described.3, 5, 6, 10 Unless otherwise stated, male mice 2-5 months old were used. All experiments were approved by the animal studies committees of Washington University and St. Louis University and were conducted in accordance with the criteria outlined in the “Guide for Care and Use of Laboratory Animals.” Hepatocyte isolation and density gradient centrifugation were carried out as previously described.4 To minimize bacterial contamination, broad-spectrum antibiotics (GIBCO 15750-060 gentamicin, 1:1000, and GIBCO 15140-148 penicillin-streptomycin, 1:1000) were added to all solutions prior to the start of each experiment. Invitrogen liver perfusion medium and liver digestion medium solutions were used as described.
Cell suspensions were analyzed for scattering and other characteristics with a Beckman flow cytometer. For cell sorting experiments, approximately 1 × 106 cells were suspended in 2-3 mL of phosphate-buffered saline (PBS). Cells were sorted into 4 side-scatter populations, as described in the Results section, and the populations were collected in 2 mL of 1× PBS in 5-mL tubes prelubricated with bovine serum albumin. Cell fractions were spun at 200g for 5 minutes and washed, and the pellet was resuspended in a monomer-polymer buffer for further analysis. For flow cytometry sorting experiments, 5 mL of a cell suspension was added to 30 mL of a 1.096 g/mL Percoll solution (containing 140 mM NaCl and 10 mM D-glucose in 1× PBS) and centrifuged for 15 minutes at 1,100 g rpm in a Beckman Accuspin FR rotor. This produced a single cellular layer, which was collected with a pipette and washed. Hepatocytes were resuspended in 1× PBS for flow cytometry. A monomer-polymer assay was performed as previously described.11 Histologic and immunohistochemistry analysis was performed with standard techniques, as described.3–5, 11 Immunoblots were performed as previously described in triplicate with antibodies previously noted to be useful in these systems.4, 5, 8 For the Jo2 antibody treatment, mice were given a single intraperitoneal injection of 6 μm of anti-Jo2 antibody (DB Bioscience). Indomethacin was also used, as noted previously.4
In Vivo Hepatocellular Apoptosis Correlates with Increased a1AT Mutant Z Intracellular Accumulation Within Individual Hepatocytes Segregated by Flow Cytometry.
Several previous studies have shown that the intracellular accumulation of the a1AT mutant Z protein is associated with caspase activation and other markers of apoptosis in in vitro systems.4, 5, 8, 9 Furthermore, when in vivo model systems are examined, an increased rate of hepatocellular proliferation is present, suggesting a compensatory response to increased hepatocellular death.3 However, when either a homozygous ZZ human liver or a model transgenic mouse liver has been examined, widespread evidence of hepatocellular death has not been readily apparent. An example is shown in Fig. 1A, in which a liver from a PiZ mouse, a well-characterized C57Bl/6 model transgenic for the human a1AT mutant Z gene that recapitulates many features of the human liver injury, is examined by terminal deoxynucleotidyl transferase-mediated dUTP nick-end labeling (TUNEL) staining for apoptotic hepatocytes.3, 5, 6, 10 No significant increases in TUNEL-positive bodies are evident in comparison with WT C57Bl/6 mouse livers (3 PiZ and 3 WT 3-month-old livers were examined by TUNEL; P > 0.2 by a t test). A positive control is shown of a liver from a PiZ mouse treated with an intraperitoneal injection of Jo2 antibody to induce hepatocellular apoptosis to demonstrate the functionality of the TUNEL assay.
We proposed that one explanation for this conflicting result could be that the rate of apoptosis may be so low as to be difficult to detect by this method, especially if there were also a rapid rate of clearance of the TUNEL-positive bodies. Therefore, we examined a larger number of hepatocytes from the model liver by a more sensitive method. We isolated fresh, living hepatocytes from PiZ mice and from WT mice using standard collagenase digestion techniques, as previously described, and examined more than 1 million cells using flow cytometry, plotting the cell number on the vertical axis against the light side scatter on the horizontal axis.4 We proposed that, because the intracellular inclusions of the polymerized a1AT mutant Z protein, known as globules, are birefringent when examined optically, this method would effectively segregate cells with large amounts of mutant Z protein intracellular retention from those with small amounts. A photomicrograph of a PiZ mouse liver is shown in Fig. 1B as an example of the appearance of the globules, which is very similar to the appearance of a human ZZ liver. The globules are variable in size, and not all hepatocytes contain globules. We predicted that the larger the amount was of the intracellular mutant Z protein present (that is, the larger the globules), the greater the injury would be to the individual cell and the more likely it was that the cell would undergo apoptosis. The result in Fig. 1C shows that hepatocytes from the PiZ mouse (black curve) have increased side scatter in comparison with WT hepatocytes (gray curve). These are representative results of runs from a total of 4 PiZ mice and 3 WT mice. The cells from the PiZ mice were then segregated into 4 populations based on quartiles of the median side scatter and made into lysates. The first side-scatter quartile contained 30% ± 11% of the cells, the second contained 55% ± 3%, the third contained 11% ± 10%, and the fourth contained 4.2% ± 2%. With previously published biochemical techniques, the intracellular content of the a1AT mutant Z intracellular monomer and polymer protein was determined for the cells in each side-scatter quartile.11 The result is shown in Fig. 1D as a quantitative immunoblot of the a1AT mutant Z protein. Both the a1AT mutant Z monomer and the polymers are denatured in this technique to run at the monomeric molecular weight for comparison. This representative blot shows a progressive increase in the proportion of the intracellular a1AT mutant Z protein present as the protein polymer as side scatter increases [P < 0.04 by an analysis of variance (ANOVA) of densitometric scanning of the band density for the 4 quadruplicate livers for significantly increased polymer in Q4 versus Q1, and P < 0.01 for significantly decreased monomer in Q4 versus Q1]. The experiment was then repeated, and the proportion of apoptotic cells in the PiZ mouse livers (4 individual PiZ mouse livers were examined) was then determined with fluorescence-activated cell sorting through the counting of the cells on the basis of the detection of phosphotidyl serine changes in the outer membrane by fluorescein isothiocyanate–conjugated annexin-V antibody binding, a known marker for cells committed to apoptosis.9, 12 The result in Fig. 1E shows an increase in the proportion of apoptotic cells in the PiZ mouse liver directly correlated to the increased intracellular accumulation of a1AT mutant Z polymerized protein. The proportion of apoptotic cells detected by this method in WT mouse hepatocytes is much less then 1% in any part of the curve examined (3 individual WT mice were examined). These data suggest that a small population of hepatocytes with the largest intracellular accumulation of the globular a1AT mutant Z polymerized protein are the cells most frequently undergoing apoptosis.
Evidence of Activation of the Apoptotic Cascade Correlates with the a1AT Mutant Z Polymerized Protein Content of Hepatocytes Isolated on the Basis of Density Centrifugation.
Previous studies have shown that hepatocytes with large amounts of accumulated a1AT mutant Z protein can be isolated with Percoll gradient centrifugation.4 We proposed to develop this technique to provide a second, independent method to confirm the aforementioned results of the flow cytometry and to further evaluate the role of variable intracellular protein accumulation in apoptotic death. Hepatocytes were again isolated from WT mice and from PiZ mice by collagenase digestion and subjected to Percoll gradient centrifugation, as previously described, with density control beads run in parallel. WT mouse hepatocytes were almost exclusively isolated at a density of 1.020-1.040 g/mL. However, the PiZ mouse hepatocytes were widely distributed in layers: layer 1 (L1), 1.019-1.035 g/mL; layer 2 (L2), 1.035–1.045 g/mL; layer 3 (L3), 1.045–1.065 g/mL; and layer 4 (L4), >1.070 g/mL. A total of 11 PiZ livers and 5 WT livers were subjected to this analysis. Representative photomicrographs of cells from each layer are shown in Fig. 2A, and they can be compared to the appearance of the globules in the intact liver in Fig. 1B, illustrating that more abundant and larger globules are present in cells isolated from denser layers. L4 was generally not intact cells but was almost entirely naked, no longer membrane-bound globules of the polymerized a1AT mutant Z protein. The monomer and polymer a1AT mutant Z protein contents of the intact cells in L1, L2, and L3 were then analyzed by immunoblotting (Fig. 2B). The result is consistent with the microscopic examination showing a significant increase in the polymerized mutant Z protein in L3 (P < 0.02 by ANOVA for mean a1AT polymer band densitometry analyzed in 4 PiZ livers in L3 versus L1 and L2). Next, we compared these same lysates from L1, L2, and L3 for caspase cleavage by immunoblotting to determine if there was a correlation with the a1AT mutant Z protein polymer content (Fig. 2C). The results show a progressive cleavage of caspase proteins in L3 in comparison with the other layers. Caspase 12, thought to be related to ER stress–induced apoptosis, is cleaved, as are caspases 9, 3, and 7. These species delineate activation of the intrinsic apoptotic pathway in the cells in L3, but not in PiZ hepatocytes in the other layers with less mutant Z polymerized protein accumulation or in the negative control lysate of a whole WT liver. Positive controls to demonstrate the electrophoretic mobility of the cleaved caspase bands were generated with whole liver lysates made by the pretreatment of additional PiZ mice with indomethacin or Jo2. Indomethacin has been previously shown to increase the synthesis and intracellular accumulation of a1AT mutant Z protein polymers, leading to increased caspase cleavage, and Jo2 has also been previously shown to generate these cleavage products in murine hepatocytes.4 We also found increased cleavage of caspase 8 in L3 in comparison with the other layers (Fig. 2D). Caspase 8 is thought to be involved in the extrinsic pathway of apoptotic activation. To determine if this indicated an increased sensitivity to extrinsic pathway stimulation in vivo, we treated 5 PiZ mice with sublethal intraperitoneal injections of Jo2 antibody, which is known to stimulate extrinsic pathway apoptosis in hepatocytes in vivo. An analysis of the livers with TUNEL staining revealed increased apoptotic bodies in the Jo2-treated mice in comparison with the sham-treated controls, with more than 90% ± 9% (9% is the standard deviation of the mean of 4 PiZ livers counted) of the apoptotic hepatocytes containing globules (a mean in these livers of only 42% of the hepatocytes contained globules; Fig. 2E). This result indicates increased susceptibility in vivo of hepatocytes with large amounts of polymerized a1AT mutant Z protein accumulation to extrinsic apoptotic stimuli. Next, we repeated the flow cytometry on the cells recovered in L1, L2, and L3 for evidence of apoptotic death by an annexin-V signal, as described previously. The results showed only scant apoptotic hepatocytes in L1 and L2 but a significant increase in the occurrence of apoptotic hepatocytes in L3 (this was repeated on 2 PiZ livers from the samples subjected to centrifugation). Taken together, these data suggest that the few hepatocytes with the largest a1AT mutant Z protein polymer accumulation (the largest globules) are primed for apoptotic death and that they undergo apoptotic death in vivo more frequently.
Increased Expression of Inhibitors of Apoptosis Is Associated with Increased a1AT Mutant Z Intracellular Accumulation.
Although from the aforementioned data it is clear that a large globular accumulation of the mutant Z protein puts individual hepatocytes at risk for apoptosis, it is also clear from the data in Fig. 1 that globule-containing cells are abundant and that the majority are alive. Therefore, we proposed that there must be antiapoptotic factors active in the globule-containing cells preventing widespread death. Therefore, we again compared the cells in layers L1, L2, and L3, using the same lysates generated for the aforementioned assays, for evidence of up-regulation of antiapoptotic regulatory proteins (Fig. 3). The result shows an increase in the phosphorylation of Bad (p-Bad) in L3. Bad is a proapoptotic protein whose activity is inhibited when it is phosphorylated.13, 14 The results also show an increase in the antiapoptotic regulatory protein, cFLIP, in L3 but no change in the antiapoptotic species Bax inhibitor-I (BI-I; a densitometric analysis of replicates from 4 livers by ANOVA revealed P < 0.04 for cFLIP in L3 versus L2 and L1, P < 0.01 for pBAD in L3 versus L2 and L1, and P > 0.3 for Bax in L3 versus L2 or L1).15–18 These data indicate that an increasing hepatocellular globular accumulation of the a1AT mutant Z polymerized protein is associated with not only increased caspase activation but also an increase in antiapoptotic proteins. The presence of such a compensatory, antiapoptotic affect might be important to prevent widespread hepatocellular death and damage to the host organism under the chronic stimulation of a1AT mutant Z protein accumulation.
Only a little more than a decade ago, it was still unclear whether it was the accumulation of the a1AT mutant Z protein within hepatocytes or the loss of circulating antiprotease activity that was the basic lesion responsible for liver injury in a1AT deficiency. However, over this time, a large number of studies from our laboratory and from others have begun to elucidate the mechanisms of cell injury in this disease. First, it was shown that susceptibility to liver injury in PIZZ humans was associated with reduced efficiency of the intracellular proteolytic apparatus responsible for ER-associated protein degradation.19, 20 This presumably leads to an increased steady-state burden of accumulated protein and an increased risk of liver disease. Other detailed studies then described the unique intracellular polymerization of a1AT mutant Z molecules and began to postulate a critical role for this specialized protein conformation in the pathophysiology of this disease, especially considering the theoretical difficulty that the cell might have degrading insoluble, aggregated proteins. Subsequent investigations showed that the accumulation of the a1AT mutant Z protein polymers triggers a variety of cellular responses, including caspase activation, autophagy, and a low-grade regenerative response within the a1AT mutant Z liver.3–5, 8, 9 This led to the discovery that mitochondrial injury and autophagy are likely critical steps in the injury cascade and that they are linked to caspase activation and to the apoptotic death of hepatocytes accumulating the a1AT mutant Z protein intracellularly. However, understanding how these injury pathways culminate in a low-grade, chronic injury liver disease was until now unclear because of scant data that could relate these findings on the cellular level to the damage to the liver in vivo.
Our new data presented in this article indicate that the small population of in vivo hepatocytes with the largest burden of the a1AT mutant Z polymerized protein present as globules disproportionately demonstrates increased caspase activation, increased sensitivity to extrinsic apoptotic stimulation, up-regulated caspase inhibitors, and increased apoptotic hepatocellular death. When these new data are combined with the previous studies, it becomes possible, for the first time, to construct a hypothetical cascade for hepatocellular injury and death in a1AT deficiency. Figure 4 shows this proposed in vivo injury cascade, in the which a1AT mutant Z protein is synthesized and processed appropriately until the nascent polypeptide chain enters the ER lumen. At this point, the molecule folds inefficiently into its final conformation, is retained in the ER lumen rather then passed along the secretory pathway, and begins to form insoluble protein polymers. ER-associated degradation processes dispose of most of the protein retained in the ER. However, for reasons that are still unclear, some of the polymerized protein escapes degradation and large globules of polymerized protein begin to form within dilated areas of ER membranes. Autophagy is up-regulated, and mitochondrial injury and mitochondrial autophagy occur. It is still unclear whether up-regulated autophagy nonspecifically damages mitochondria and/or if there is a direct toxic effect on mitochondria of the ER accumulation of the a1AT mutant Z protein. In either or both cases, the result is that apoptotic pathways are triggered in the cells, with the largest burden of the mutant Z protein associated with cytochrome c release from mitochondria and with caspase activation. Apoptosis of the globule-containing hepatocytes could also be triggered by an unrelated stimulus to the extrinsic apoptotic pathway (that is, a second hit). Inhibitors of apoptosis (Bad and others, as shown in the results) are up-regulated to prevent widespread cellular death and loss of the overall organ, but the inhibition is insufficient in some cells, leading to a low but important and constant rate of hepatocellular death. An alternative pathway can be proposed that cannot be refuted or supported by existing data: in some cells, the mitochondrial injury might be so great as to cause cell death by a loss of energy supplies. The result of either scenario is a compensatory regenerative response within the liver.
Although this hypothetical cascade is consistent with the existing data at each step, it is still possible that new findings will significantly alter this scheme in the future. Proposing a cascade, however, that can now be tested at each step in a controlled and systematic way will likely speed the full understanding of liver injury in this and other metabolic liver diseases.