SEARCH

SEARCH BY CITATION

Abstract

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. REFERENCES

Fatty livers are particularly susceptible to mitochondrial alterations after cold preservation. We thus aimed to improve graft integrity by brief hypothermic oxygenation prior to warm reperfusion. Macrovesicular steatosis was induced in rat livers by fasting and subsequent feeding of a fat-free diet enriched with carbohydrates. Fatty livers were retrieved and stored ischemically at 4°C for 20 hours in histidine-tryptophan-ketoglutarate solution. Hypothermic reconditioning (HR) was performed in some livers by insufflation of gaseous oxygen via the caval vein during the last 90 minutes of preservation. Viability was assessed upon isolated reperfusion. HR resulted in a significant (approximately 5-fold) reduction of parenchymal (alanine aminotransferase and lactate dehydrogenase) and mitochondrial (glutamate dehydrogenase) enzyme release. Functional recovery (bile production, oxygen consumption, and tissue levels of adenosine triphosphate) was significantly improved by HR. In untreated grafts, cellular autophagy (cleavage of LC3B and protein expression of beclin-1) was significantly impaired (<50% of baseline) after preservation/reperfusion but was restored to normal values by HR. HR also increased cleavage of caspase 9 (P < 0.5) and caspase 3 enzyme activity (by a factor of 1.5). In contrast, histological signs of tissue necrosis were abundant after reperfusion in untreated livers and largely abrogated in reconditioned livers. In conclusion, HR limits mitochondrial defects and restores basal rates of cellular autophagy. This may represent a rescue mechanism for maintaining cellular homeostasis and tissue survival. Liver Transpl 15:798–805, 2009. © 2009 AASLD.

At this time, hepatic steatosis is a common clinical problem in industrialized countries, affecting almost 30% of donors for liver transplantation.1 Although functionally unapparent in the donor, steatotic livers, in contrast to normal livers, are often particularly susceptible to ischemia/reperfusion injury, leading to poor outcome following liver surgery2 or transplantation.3–5

The mechanistic basis for the reduced ischemic tolerance of steatotic livers is not yet fully understood. Different studies in experimental models of cold ischemia as well as human transplant biopsies have indicated severe deterioration of mitochondrial functions and energy metabolism in fatty livers during cold preservation and reperfusion.6–8

The observed impairment of postischemic oxidative phosphorylation has recently been linked to an increased loss of the functional integrity of the electron respiratory chain in fatty livers.8

Increased oxidative stress9 has also been reported to promote higher postischemic dysfunction rates in steatotic livers, and mitochondria appear to be a prominent cellular source of reactive oxygen species.10 Actually, fatty livers show a greater mitochondrial content of oxidized lipids and proteins than normal livers.11

The increased oxidative stress can alter mitochondrial functions, including key processes for adenosine triphosphate (ATP) synthesis,12 and if large amounts of radicals interfere with mitochondrial function, the cell dies because of a lack of energy.13

Depending on the amount of affected mitochondria and the energetic status of the cell, this results in autophagy, apoptosis, or necrosis.14

Autophagy occurs at low rates in cells to perform homeostatic functions (eg, lysosomal degradation of defective proteins, injured organelles, or depolarized mitochondria) and appears to be depressed after warm anoxia in isolated hepatocytes.15

In earlier studies, it was shown that gaseous oxygenation of the liver during ischemic storage maintains hepatic energy metabolism at physiological levels and reduces graft injury after transplantation.16, 17 The same technique was later proven to largely prevent mitochondrial dysfunction and ultrastructural alterations upon reperfusion of steatotic livers.18

In light of the observation that irreversible ultrastructural disintegration of mitochondria in fatty livers mostly occurs at the time of warm reperfusion but not during ischemic storage,18 we aimed in this study to evaluate the potential of using only a terminal aerobic reconditioning protocol for cold-stored liver grafts to counteract deleterious priming of the livers to succumb to major alterations upon warm reperfusion.

We show in this article that hypothermic reconditioning (HR) of cold-stored fatty livers can effectively limit mitochondrial dysfunction upon reperfusion and restore basal rates of cellular autophagy, which is likely to represent a rescue mechanism for maintaining cellular homeostasis and tissue survival.

MATERIALS AND METHODS

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. REFERENCES

All experiments were performed in accordance with the federal law regarding the protection of animals. The principles of laboratory animal care (NIH Publication 85–23, revised 1985) were followed.

Experimental Induction of Hepatic Steatosis

Male Wistar rats (250–300 g) were fed standard pellet food during an acclimatization period of at least 5 days under specific pathogen-free conditions according to Federation of European Laboratory Animal Science Associations regulations.

The rats were fasted for 2 days while having free access to tap water. Then, they received a special fat-free, carbohydrate-rich diet (C1000, Altromin, Lage, Germany) for another 2 days and were introduced to the experiment on the third day. The composition of the diet was as follows: saccharose (38%), starch (38%), casein (16%), and a mineral and vitamin mix (8%). This protocol induced mild-to-moderate, predominantly macrovesicular steatosis resulting in hepatocellular enlargement and subsequent disturbance of the trabecular architecture of hepatic cords, as documented in detail elsewhere.18, 19

Experimental Protocol

Rats were anaesthetized by an intramuscular injection of ketamine hydrochloride (90 mg/kg) and xylazine (10 mg/kg). The abdomen was opened by midline incision, and the liver was skeletonized and freed from all ligamentous attachments.

The portal vein was cannulated, and the livers were excised, rinsed via the portal vein with 20 mL of histidine-tryptophan-ketoglutarate solution, and then subjected to static cold storage at 4°C. The grafts were randomly assigned to one of the following groups (n = 6):

  • Group 1. Livers were cold-stored for 20 hours to simulate prolonged extension of the preservation time prior to reperfusion (cold storage [CS]).

  • Group 2. Livers were cold-stored for 20 hours as in group 1 but then were additionally subjected to endischemic HR by venous systemic oxygen persufflation, that is, insufflation of gaseous oxygen via the venous vascular system of the cold-stored organ as detailed previously.17 This persufflation period lasted 90 minutes in order to recondition the graft and improve liver integrity prior to reperfusion.

The viability of all livers was evaluated after the respective preservation protocol upon reperfusion in vitro. A total of 200 mL of Krebs-Henseleit buffer, including 3 g/100 mL bovine serum albumin (Fraction V, Sigma Chemical), was recirculated through thin-wall silicone tubing curled in a box, which was ventilated with a 95% O2/5% CO2 gas mixture, at a constant flow (3 mL/g min) and subsequently was passed through a heat exchanger (37°C) and a bubble trap before entering the liver.

The liver was placed swimming in Krebs-Henseleit buffer in a water-jacketed bath, the temperature of which was kept at 37°C with an external heating circuit. In this way, care was taken to prevent uneven liver perfusion of the organ ex vivo. This setup has been shown to allow the adequate approximation of tissue integrity and the detection of structural changes in rat livers after hypothermic preservation.20

To simulate the period of slow rewarming of the organ during surgical implantation in vivo, all livers were exposed to room temperature on a Petri dish for 20 minutes prior to reperfusion.

The portal venous pressure was measured during isolated perfusion by means of a water column connected to the portal inflow line and calibrated to the respective calculated flow with polyethylene catheters of a length and size identical to those of the catheter used for the perfusion of the livers.

Perfusate enzyme activities of alanine aminotransferase (ALT), lactate dehydrogenase, and glutamate dehydrogenase (GLDH) were assessed photometrically with standard commercial kits (Roche, Mannheim, FRG).

In order to assess the impact of reactive oxygen species on the tissue integrity, we measured the concentrations of lipid peroxides in the rinsing effluent and in the perfusate by fluorometric detection, as described previously.21

Metabolic activity of the livers was approximated by the calculation of hepatic oxygen utilization. Perfusate samples were taken at the portal inflow and from the venous effluent, and the respective contents of O2 were measured immediately with a pH-blood gas analyzer (ABL 500 acid-base laboratory, Radiometer, Copenhagen, Denmark). O2 uptake of the livers was calculated from the differences between the portal and venous sites and expressed as micromoles per gram of liver per minute according to the transhepatic flow and liver mass.

Hepatic clearance of ammonia was used as an additional functional parameter of liver viability after preservation. To this end, NH4Cl was added to the Krebs-Henseleit perfusion buffer at a concentration of 1 mmol/L. Concentrations of urea in the effluent perfusate were measured enzymatically, and urea production was calculated as micromoles per gram per hour.

Tissue Extraction and Assay of High-Energy Phosphates

Tissue specimens for the assessment of the energetic status after reperfusion were taken with precooled steel tongs, immersed in liquid nitrogen, and stored at −80°C until later analysis.

High-energy phosphates were determined enzymatically in the neutralized supernatant after protein extraction with perchloric acid of freeze-dried tissue samples, as described elsewhere,22 with minor modifications made to adapt the assay to a microtiter scale.

In brief, 225 μL of the reaction medium was prepared from fresh standard solutions [resulting in final concentrations of triethanolamine buffer (38 mM), ethylene diamine tetraacetic acid (2 mM), nicotinamide adenine dinucleotide phosphate (0.2 mM), MgCl2 (2.6 mM), glucose (1.3 mM), and As2O3 (12 μmol/mL)], and it was mixed and incubated with 20 μL of the sample.

Supplementation with glucose-6-phosphate dehydrogenase (final activity 1.8 U/mL) was followed by the continuous observation of absorbance at 340 nm and the addition of hexokinase (final activity, 7.5 U/mL).

The conversion rate of nicotinamide adenine dinucleotide phosphate to reduced-form nicotinamide adenine dinucleotide phosphate was determined after the addition of hexokinase following the increase in absorbance at 340 nm with a microplate reader (Tecan Safire2, Tecan Austria GmbH, Grödig, Austria) and was used to quantify tissue levels of ATP.

Readouts from freshly prepared external standards of ATP, run simultaneously with the samples, were used for the calculation of a standard curve, which served to quantify the respective sample concentrations. The accuracy of the measurements was confirmed by parallel analyses of standard aliquots in a 1-mL single-cuvette assay at 365 nm,23 the results of which were calculated with a molar extinction coefficient for reduced-form nicotinamide adenine dinucleotide phosphate of 3.30. Final results were corrected for the respective dry weight to wet weight ratio of the tissue samples and were expressed as micromoles per gram of dry weight.

The linearity of the assay was confirmed over the entire range under investigation (that is, 5–300 nmol/mL, which corresponds to 0.2–15 μmol/g of tissue; r2 = 0.99).

Adenosine diphosphate and adenosine monophosphate were quantified as detailed previously23 and transferred to a microtiter scale as already outlined.

In order to approximate the degree of phosphorylation of the adenine nucleotide system, the energy charge was calculated according to Atkinson24 as follows:

  • equation image

where ATP, ADP, and AMP represent the molar concentrations of adenosine triphosphate, adenosine diphosphate, and adenosine monophosphate, respectively, and EC is the energy charge.

Western Blot Analysis

Whole tissue lysates were prepared from frozen tissue obtained after reperfusion, separated by gel electrophoresis, and blotted onto nitrocellulose membrane as detailed previously.21

Expression of specific proteins was analyzed by incubation overnight at 4°C with the respective primary antibody and visualization on X-ray film via chemiluminescence (Phototope New England Biolabs, Inc., Schwalbach/Taunus, Germany). Subsequently, blots were stripped and probed with monoclonal anti-actin antibody (AB-1, Calbiochem, Darmstadt, Germany) to confirm equal amounts of protein loading.

Quantification of protein content was performed on the basis of the ratios of individual signals and actin, which were determined densitometrically with UN-SCAN-IT gel version 6.1 (Silk Scientific Corp., Orem, UT).

The antibodies used were anti-mtHSP70/GRP75 from Abcam (Cambridge, United Kingdom) and anti–cleaved caspase 9 (p35) from Cell Signaling/New England Biolabs (Frankfurt, Germany).

Autophagic activity was evaluated following the cleavage of LC3B by detection of the 14-kDa fragment of LC3B and protein expression of beclin-1 with polyclonal antibodies from Cell Signaling/New England Biolabs.

Apoptosis Assay

The functional activity of caspase 3 was analyzed from homogenized tissue lysates in 96-well plates with a fluorometric assay kit (Calbiochem) on the basis of the detection of the cleavage product 7-amino-4-trifluoromethyl coumarin, which emits yellow-green fluorescence (maximum at 505 nm) upon excitation at 400 nm.

Measurements were taken on a fluorescence microplate reader, and enzymatic activities of caspase 3 in the experimental groups are presented as percentage increases with respect to the baseline values obtained from nonischemic control livers.

Histology

Liver tissue was collected at the conclusion of the experiments, cut into small blocks (3 mm thick), and fixed by immersion in 4% buffered formalin. The blocks were embedded in paraffin and cut into 2-μm sections with a microtome. Hematoxylin and eosin staining was used to judge the morphological integrity of the parenchyma. Sections were examined at 200× magnification, and the extent of necrosis was graded semiquantitatively from 0 (no necrosis) to 3 (severe necrosis with disintegration of hepatic cods), as described elsewhere,25 in a blinded fashion by 2 independent investigators.

Statistics

All values were expressed as means ± the standard error of the mean. After the assumption of normality and equal variance across groups was proven, differences among groups were tested by analysis of variance followed by multiple comparisons of the means with the Student-Newman-Keuls test unless otherwise indicated. Statistical significance was set at P < 0.05.

RESULTS

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. REFERENCES

Functional Recovery During Liver Reperfusion

During 2 hours of isolated reperfusion, the total vascular resistance of cold-stored fatty livers slightly rose from 2221 ± 216 to 2842 ± 371 Pa s mL−1 but was found to be not affected by HR, with a maximal value of 2489 ± 158 Pa s mL−1.

Resumption of parenchymal metabolic activity after preservation was positively affected by HR, which entailed a notable improvement in secretory and detoxifying liver performance upon reperfusion as bile production (0.39 ± 0.20 versus 1.43 ± 0.01 μL/g h) and clearance of ammonia (174 ± 30 versus 231 ± 16 μmol urea/g h) were significantly elevated by HR (P < 0.05). However, there were no significant differences in the postischemic oxygen consumption of livers that were only cold-stored and those that were subjected to HR (1.40 ± 0.05 versus 1.67 ± 0.06 for untreated livers versus HR).

Energetic Recovery

The tissue concentrations of ATP and the energy charge potential were used to assess the energetic recovery of the livers upon warm reperfusion (Fig. 1A).

thumbnail image

Figure 1. (A) Tissue concentrations of ATP and EC in fatty livers at the end of 120 minutes of warm reperfusion after 20 hours of cold preservation without (untreated) or with an additional 90 minutes of HR. Baseline values were obtained from nonischemic control preparations perfused immediately after harvesting. Values are given as the mean ± standard error (n = 6). *P < 0.05 versus untreated livers and P < 0.05 versus baseline (analysis of variance and Student-Newman-Keuls test). (B) LPOs in the rinsing effluent after ischemic preservation (left) and in the circulation perfusate after 120 minutes of warm reperfusion as an index for the impact of reactive oxygen species on liver tissue. Values are given as the mean ± standard error (n = 6). *P < 0.05 versus untreated livers (Student t test). Abbreviations: ATP, adenosine triphosphate; EC, energy charge; HR, hypothermic reconditioning; LPO, lipid peroxide.

Download figure to PowerPoint

After 2 hours of oxygenated reperfusion, untreated livers still exhibited very low ATP levels, and their energy charge potential was significantly reduced to less than half of the baseline values. HR prior to reperfusion resulted in a net and significant amelioration of hepatic ATP levels and was followed by a significant rise of the energy charge potential to nearly normal values.

Reactive Oxygen Species Formation

The impact of oxygen free radicals on tissue integrity was approximated by fluorometric detection of lipid peroxidation in the rinse solution upon washout of the preservation fluid after cold storage or HR and in the circulating perfusate at the end of reperfusion.

Interestingly, 90 minutes of endischemic oxygen persufflation in hypothermia did not result in any increase of lipid peroxidation, and there were no differences in the washout solution between cold-stored and reconditioned livers. In contrast, HR significantly reduced lipid peroxidation during subsequent reperfusion (Fig. 1B).

HR Reduces Postischemic Tissue Damage/Necrosis

The release of cytosolic enzymes was investigated as a general readout of liver cell damage.

Perfusate concentrations of ALT showed a steep rise along with progression of reperfusion, but they were significantly reduced by HR to less than one-quarter of the values observed after cold storage alone (Fig. 2A).

thumbnail image

Figure 2. Release of (A) cytosolic ALT, (B) LDH, and (C) mitochondrial GLDH during normothermic reperfusion after 20 hours of ischemic preservation of fatty livers without (untreated) or with an additional 90 minutes of HR. Values are given as the mean ± standard error. *P < 0.05 versus untreated. Abbreviations: ALT, alanine aminotransferase; GLDH, glutamate dehydrogenase; HR, hypothermic reconditioning; LDH, lactate dehydrogenase.

Download figure to PowerPoint

A comparable pattern was disclosed concerning the leakage of lactate dehydrogenase, the activities of which in the perfusate were likewise reduced by HR (Fig. 2B).

Mitochondrial integrity was investigated on the basis of the leakage of the intramitochondrial enzyme GLDH into the circulation perfusate (Fig. 2C). Although untreated livers exhibited an ongoing loss of GLDH, the activity of which steeply rose in the perfusate during the whole observation period, a marked and significant attenuation of GLDH leakage was obtained with HR.

Histology

In line with the data on hepatocellular enzyme loss, hematoxylin and eosin histopathology after reperfusion of untreated grafts revealed major liver injury (Fig. 3), which was characterized by disruption of the general architecture, vacuolization, and disrupted cell and organelle membranes. HR resulted in less severe alterations and exhibited better preserved hepatic morphology. Score values for necrotic tissue injury were thus reduced by HR to approximately 50% in comparison with untreated livers (Fig. 4).

thumbnail image

Figure 3. Liver histology by light microscopy following 120 minutes of normothermic reperfusion: (A) control livers without ischemic injury, (B) after 20 hours of cold storage, and (C) after 20 hours of cold storage and 90 minutes of hypothermic reconditioning by gaseous oxygen persufflation. Note the markedly better preserved liver morphology in the hypothermic reconditioning group. (The arrows indicate necrotic areas, and the arrowheads mark nuclear pyknosis; the scale bar indicates 50 μm.)

Download figure to PowerPoint

thumbnail image

Figure 4. Histological injury score. Values are given as the mean ± standard error. *P < 0.05 versus untreated livers (Student t test). Abbreviation: HR, hypothermic reconditioning.

Download figure to PowerPoint

HR Restores Cellular Autophagy

The postischemic activation level of the autophagosomal pathway was investigated with the conversion of LC3 to LC3-II15 as well as the tissue content of the autophagy-related protein beclin-1 (Fig. 5).

thumbnail image

Figure 5. Cellular autophagy as approximated by the protein expression of beclin-1 and cleavage of LC3-II in fatty livers at the end of 120 minutes of warm reperfusion after 20 hours of cold preservation without (untreated) or with an additional 90 minutes of HR. Baseline values were obtained from nonischemic control preparations perfused immediately after harvesting. Values are given as the mean ± standard error (n = 6). *P < 0.05 versus untreated livers and ±P < 0.05 versus baseline (analysis of variance and Student-Newman-Keuls test). Abbreviation: HR, hypothermic reconditioning.

Download figure to PowerPoint

During ischemia/reperfusion of untreated livers, a loss of the autophagy protein beclin-1 occurred, resulting in a significant decrease to about 69% of pre-ischemic baseline levels.

Even more pronounced was the decline in the production rate of LC3-II, which was diminished to 40% of the baseline values.

HR, however, was effective at significantly maintaining cellular autophagy at normal rates, as evidenced by the occurrence of LC3-II and protein expression of beclin-1.

HR Stimulates Cellular Apoptosis

Because apoptosis is another environmentally less injurious form of cell death but depends on the energetic status of the tissue, we analyzed the cellular apoptosis cascade, including the cleavage of caspase 9 and the enzymatic activity of executioner caspase 3 (Fig. 6).

thumbnail image

Figure 6. Quantification of mitochondrial induction of apoptosis by cleavage of caspase 9 and enzymatic activity of caspase 3 in fatty livers at the end of 120 minutes of warm reperfusion after 20 hours of cold preservation without (untreated) or with an additional 90 minutes of HR. Baseline values were obtained from nonischemic control preparations perfused immediately after harvesting. Values are given as the mean ± standard error (n = 6). *P < 0.05 versus untreated livers and P < 0.05 versus baseline (analysis of variance and Student-Newman-Keuls test). Abbreviation: HR, hypothermic reconditioning.

Download figure to PowerPoint

After reperfusion of untreated livers, we observed an approximately 2-fold increase in caspase 9 cleavage with respect to baseline values, whereas the enzyme activity of caspase 3 was found to be concordantly enhanced.

In comparison, HR promoted an additional and significant increase in the cleavage rate of caspase 9. The enhanced cleavage of caspase 9 after HR was paralleled by a limited increase in the activity of caspase 3 at the end of postischemic reperfusion, which reached significance in comparison with baseline values.

DISCUSSION

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. REFERENCES

Ischemia/reperfusion injury is a major cause of primary liver dysfunction after transplantation26 and apparently is of particular importance following the preservation of fatty liver grafts.5

Expedited recovery of oxidative phosphorylation after ischemic storage is of pivotal importance for the resumption of energetic homeostasis and a prerequisite for organ viability after liver transplantation.27 As a result, mitochondrial dysfunction and consecutive failure to adequately comply with energetic demands upon reperfusion have redundantly been incriminated as the Achilles heel of fatty liver transplantation.5, 7, 28

In this study, we could confirm a serious defect of cold-stored fatty livers in restoring the ATP stores during warm reperfusion in line with signs of major mitochondrial membrane disintegration, as evidenced by the leakage of the intramitochondrial marker enzyme GLDH, the perfusate concentrations of which were even higher than those of the cytosolic ALT.

With similar models, it has been shown that in normal livers, the overall leakage of GLDH is much lower, and the ratio of GLDH/ALT leakage is inversed.20

Selzner and coworkers28 showed that retarded resumption of the hepatic energy charge potential after ischemia in fatty livers leads to an inability to organize cell death via the apoptotic pathway and shifts the consequences of lethal parenchymal injury to the more deleterious necrotic form in comparison with normal livers.

In line with this, we could observe in our study only limited activation of the mitochondrial apoptosis pathway, comprising activation of caspase 9 and caspase 3, which was slightly enhanced by HR. However, the less than 30% increase of caspase 3 activity would not explain the impressive reduction of tissue damage and liver dysfunction observed upon the reperfusion of fatty livers after HR.

In isolated hepatocytes, another mechanism of mitochondrial dysfunction during and after hypoxia has recently been disclosed to be impaired autophagy.15

Autophagy occurs at low basal levels in virtually all cells to perform homeostatic functions, including the removal of damaged organelles such as mitochondria, and is involved in the recycling of denaturated proteins or metabolic catabolites.29

Defective autophagy culminates in the onset of the membrane permeability transition and hepatocyte death after warm reperfusion.15

Autophagy, which can be detected by the assessment of the generation of LC3II,30 has been found to be severely impaired after cold storage of untreated fatty livers, and this is in line with the degradation of one of the key proteins of autophagy, Beclin-1. From this, the impairment of cellular quality control systems and mitochondrial dysfunction, including the membrane permeability transition, can be expected.15 In the absence of an adequate energy charge to enable apoptotic transformation, the membrane permeability transition will result in necrotic cell decay.

A potential basis for the beneficial effects of HR observed in this study is hence seen in preventing the breakdown of cellular autophagy. Through restoration of mitochondrial homeostasis and consecutively improved mitochondrial energy replenishment upon warm reperfusion, cellular autophagy is conjectured to counteract tissue necrosis in fatty livers. However, experimental evidence for the causal link between impaired autophagy and consecutive organ damage cannot yet be derived from the present data.

It is of special interest that continuous aeration of fatty livers during the whole period of ischemic storage18 is not required, but only the re-establishment of energetic homeostasis just prior to warm reperfusion is needed. HR by gaseous oxygen persufflation seems to be useful for regenerating cellular energy–dependent pathways under conditions of relatively low metabolic work load, representing an adequate and sufficient tool to prevent major cellular dysfunction upon warm reperfusion.

REFERENCES

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. REFERENCES