Transcriptional Profiling of the Left and Median Liver Lobes of Male F344/N Rats Following Exposure to Acetaminophen

Abstract

The liver is a common organ for transcriptional profiling because of its role in xenobiotic metabolism and because hepatotoxicity is a common response to chemical exposure. To explore the impact that sampling different lobes may have on transcriptional profiling experiments we have examined and compared gene expression profiles of the left and median lobes of livers from male F344 rats exposed to toxic and nontoxic doses of acetaminophen. Transcript profiling using micorarrays revealed clear differences in the response of the left and median liver lobes of F344 rats to acetaminophen exposure both at low doses as well as doses that caused hepatotoxicity. Differences were found in the total number of differentially expressed genes in the left and median lobes, the number and identity of genes that were differentially expressed uniquely only in the left or median lobe, and in the patterns of gene expression. While it is not possible to generalize these results to compounds other than acetaminophen or other strains of rat, these results highlight the potential impact of sampling differences on the interpretation of gene expression profiles in the liver.

Introduction

Although the liver has a uniform gross appearance there is increasing evidence of functional heterogeneity among individual liver lobes, and within different regions within the same lobe. For example iron accumulates more in the left hepatic lobe in infants (Faa et al., 1994), a pattern also seen in iron storage disease in adults (Ambu et al., 1995). Copper tends to accumulate more in the left hepatic lobe in newborn human infants (Faa et al., 1987) but more in the right hepatic lobe in Wilson’s disease (Faa et al., 1995). Diethyl-nitrosamine administration causes a higher incidence of carcinoma in the left and median lobe as compared to the right in the rat (Richardson et al., 1986). DNA damage is greater in the left and median lobes while cell replication is higher in median and right anterior lobes (Richardson et al., 1986). The progression of cirrhosis also shows lobe specific differences progressing more rapidly in the right lobe compared to the left (Matsuzaki et al., 1997).

In the developing mouse all components to the left of the falciform ligament give rise to the left lobe while all components to the right give rise to the right, median and caudate lobes (Kaufman and Bard, 1999). There is also left-right asymmetry in portal blood flow during development (Germain et al., 1987) with the fetal left liver lobe receiving more nutrients than other regions of the liver (Zhang and Bryne, 2000). Moreover, adult offspring of dams (Wistar rats) fed a low protein diet during pregnancy exhibited reduced fibrinogen mRNA levels and reduced fibrinogen protein only in the left lobe of the liver (Zhang and Bryne, 2000). These results suggest that differences between liver lobes may be “hard wired” during development.

To explore the impact that sampling different lobes may have on transcriptional profiling experiments we have examined and compared gene expression profiles of the left and median lobes of livers from male F344 rats exposed to toxic and nontoxic doses of acetaminophen. Since the development of hepatocellular necrosis in response to acetaminophen exposure appears to be similar in the left and median lobe of F344 rats, differences in gene expression may reflect intrinsic differences in the molecular response of each lobe.

Materials and Methods

Animals

Male Fischer 344 rats approximately 36 +/− 3 days old were supplied by Taconic laboratory animals (Germantown, NY) and were approximately 90+/− 4 days old when placed on test. The studies were conducted at Battelle International, Inc., Columbus, Ohio and the protocol was approved by the Battelle IACUC and followed the standards outlined in the Guide for the Care and Use of Laboratory Animals (NRC, 1996). Rats were randomized to experiments by body weight partitioning using the PATH/TOX SYSTEM (Xybion Medical Systems Corp., Cedar Knolls, NJ) algorithm. The rats were housed 3 per cage in 22″ L × 12.5″ W × 8″ H polycarbonate cages (Lab Products, Inc., Seaford, DE) with polyester cage filters (Snow Filtration Co., Cincinnati, OH). Animal room temperature and humidity were continuously monitored and varied between 71° and 75°F and 36% to 48% relative humidity. Animal room lighting followed a 12-hour light period from 8 AM to 8 PM with a corresponding 12-hour dark (night) period from 8 PM to 8 AM. Acetaminophen was administered from a continually stirred 0.5% aqueous methylcellulose vehicle at 5 ml/kg body weight by gavage at noon with necropsies at 6 PM (6 hours), 6 AM (18 hours) and 24 and 48 hours (12 PM). The acetaminophen dose levels included both 50 and 150 mg/kg as nontoxic doses and 1500 mg/kg as moderately toxic and 2000 mg/kg severely toxic doses. The controls received methylcellulose alone.

The rats had ad libitum access to irradiated NTP-2000 wafer feed (Ziegler Brothers, Gardners, PA) during the 12-hour dark period with no food present in their cage during the 12-hour light period of the daily light cycle during the experimental period. The rats had ad libitum access to city water (Columbus, OH) at all times. The room light intensity during the 12-hour light period ranged from 38–40 foot-candles measured 5 feet from the floor. The study was conducted in replicate groups of 3 to assure prompt and timely handing of the tissues for RNA analysis. The necropsies took place within 1 hour of the scheduled period of time and each liver lobe was placed in RNA stabilizer within 4 minutes of sacrifice.

The rats were anesthetized with CO₂/O₂, blood samples collected for clinical chemistry by cardiac puncture, the abdominal cavity opened, and the portal vein severed before necropsy.

Radioactive Acetaminophen Studies

Studies using radiolabeled acetaminophen were done at NIEHS under a NIEHS-approved animal protocol. Acetaminophen was administered to groups of 5 rats either by gavage, intraperitoneal injection (ip), or intravenous injection (iv), and the animals sacrificed with CO₂ asphyxiation at the designated times after dosing. The oral dose was 2000 mg/kg ³H-acetaminophen administered as a suspension in 0.5% aqueous hydroxyethylcellulose at a dose rate of 10 ml/kg. The ip dose was 50 mg/kg in 0.5% aqueous hydroxyethylcellulose administered at 10 ml/kg, and the IV dose was 0.171 mg/kg in physiological saline. All rats received 48–55 μCi/kg ³H-acetaminophen. At 10 minutes (IV and IP study) or 10 and 30 minutes (oral study), the rats were euthanized with CO₂ and O₂ and the livers were promptly removed by lobe. The left and median lobes were divided into left and right halves, the right lobe separated in anterior and posterior sublobes and both caudate sublobes collected and promptly frozen. At the time of scintillation counting, the liver sections were removed, weighed, and triplicate 100 mg samples were oxidized in a Packard 307 Sample Oxidizer. Samples were counted in a Beckman LS 6500.

Histological Methods

Cross-sections of the left and median lobes of the liver were embedded, sectioned, stained with H&E and examined for histological changes. The diagnoses and severity grades were subjected to peer review as described previously (Boorman et al., 1986, 2002).

RNA Isolation

The portions of the left and median hepatic lobes for RNA isolation were chopped with sharp razors into 0.5 cm cubes or smaller in RNALater (Ambion Inc., Austin, TX) within 4 minutes of necropsy. The liver cubes were stored in RNALater overnight at 4+/− 3°C, then in RNALater at −20+/− 1°C until RNA isolation (within 60 days). For RNA isolation 130–150 mg of tissue from left or median lobe was weighed out, 1 ml of chilled lysis buffer was poured over the tissue which was then minced into 1 mm pieces. Then 7 ml of lysis buffer was then added and the tissue homogenized with a handheld Omni tissue homogenizer with a disposable plastic 7 mm diameter Omni generator probe (Omni # 34750, Omni International, Marietta, GA) at no more than 1/2 maximum speed for 45 seconds. The resulting tissue lysate was centrifuged for 5 minutes at 4000 × g, the supernatant divided into 2 tubes, 3.5 ml of 50% absolute ethanol added and the tubes vigorously shaken for 1 minute. RNeasy midi spin columns (Qiagen, Valencia, CA) were used for RNA isolation. The RNA was concentrated using Millipore Microcon centrifugal filter devices (Millipore Corporation, Billerica, MA), analyzed using a spectrophotometer (capable of reading absorbance at 260 nm and 280 nm with UV light), frozen at minutes −70°C and shipped to the NTP repository until transfer to Paradigm Genetics Inc. (RTP, NC) for hybridization.

Microarray Hybridizations

The study design included 4 dose groups of 6 rats sacrificed at 4 time points. Pathological diagnoses and clinical chemistry were available from the 6 rats but RNA analysis was conducted on both the median and left hepatic lobe from 4 rats selected at random from each group. The 4 individual rats were hybridized against a pool made of equal amounts of RNA from three time matched control rats. The hybridizations were lobe specific, with median lobes from treated rats hybridized against a pool made from median lobe controls, and left lobes from treated rats hybridized against left lobe controls. One μg of total RNA from either an individual rat or from a pooled sample was amplified and labeled with a fluorescent dye (either Cy3 or Cy5) using Agilent Low RNA Input Linear Amplification labeling kit (Agilent Technologies, Palo Alto, CA) following the manufacture’s protocol. The resulting fluorescently labeled cRNA was quantitated and qualitated using a Nanodrop ND-100 spectrophometer and an Agilent Bio-analyzer. The fluors were reversed for a second (dye reversal) hybridization. Four individual rats at 4 time points, with 4 doses, 2 hepatic lobes, and fluor reversals resulted in 256 hybrizations.

Data Analysis

Data from dye reversal hybridizations representing the same individuals were combined in the microarray analysis software package Rosetta Resolver version 4.0.0.1.1 (Rosetta Biosoftware, Seattle, WA) using a weighted average. From this combined data, up- or down-regulated genes that were significantly differentially expressed (Rosetta error model, p ≤ 0.005) for both the median and left hepatic lobe for each individual rat were identified and served as a basis for all additional analyses. Hierarchical clustering in which Pearson’s Correlation was used as the similarity metric, and Venn Diagrams used in the lobe specific gene expression, were produced in GeneSpring 6.2.2 (Silicon Genetics, Redwood City California). Pearson correlation coefficient was used to compare lobe histology results with serum ALT activity.

Results

Acute centrilobular necrosis and inflammation were observed only at the 1500 and 2000 mg/kg doses and are consistent with acetaminophen toxicity (Table 1). Each lobe was given a severity grade depending on the overall amount of necrosis within a section. As shown in Table 1, there were some differences in severity of necrosis between the median and left lobe, however no particular pattern was evident, and there was no indication of a preference for the left or median lobe. What is not represented in the table is the often patchy distribution of necrosis not involving the entire section. Patchy hepatocellular necrosis following a single dose of acetaminophen has been reported by others (McLean and Day, 1975)

Results of clinical chemistry evaluations were also consistent with acute hepatocellular necrosis at the 1500 and 2000 mg/kg doses (Table 2). There was considerable variability of response for individual rats: for instance rats dosed with 1500 mg/kg had ALT values ranging from 96 to 6756 IU/L at 18 hours. Similarly rats dosed with 2000 mg/kg had ALT activities varying from 244 to 15,180 IU/L. A similar degree of variability was also seen at 24 hours and 48 hours (Table 3). This type of variable response to acetaminophen exposure has also been by others with unfasted animals (McLean and Day, 1975).

In an effort to compare the response of the left and median lobes we examined the correlation between histological severity grade and ALT level for each lobe. As indicated in Figure 1, the response of the median lobe was less variable than that of the left, and the correlation between severity grade and ALT levels was substantially better for the median lobe than for the left lobe.

We also evaluated the distribution of radioactive acetaminophen to the various hepatic lobes following oral, intravenous, and intraperitoneal exposure. Although some significant differences were found, no pattern was evident either between different lobes or between the right and left parts of the same lobe (Table 3).

The results of transcriptional profiling of the left and median lobes differed with respect to the number of genes differentially expressed in each lobe, the identity of the genes, and the patterns of gene expression between the 2 lobes. The total number of differentially expressed genes (genes differentially expressed at the p < 0.005 level in at least one animal in the dose-time group) at each time and dose were substantially different for the left and median lobe. When the stringency was increased by selecting the set of genes differentially expressed by all animals in a group (genes differentially expressed at the p < 0.005 level in each animal in the dose-time group), the differences were still present (Table 4). Although no overall pattern is apparent, in general more genes are differentially expressed in the median lobe at each time and dose than in the left lobe.

The difference in response between lobes becomes even more apparent when comparing the number of genes differentially expressed in the left lobe only or the median lobe only to the number of genes expressed in common in both lobes (Table 5). In the 50 and 150 mg/kg dose groups the left lobe and median lobe express very few genes in common. At 1500 mg/kg a greater percentage of genes are expressed in common, however each lobe also expresses genes not expressed in the other lobe. At 2000 mg/kg all genes expressed in the left lobe are also expressed in the median lobe, however there are also genes expressed in the median lobe that are not expressed in the left lobe. Again the data also show that at nearly all doses and times, more genes are expressed uniquely in the median lobe than in the left lobe.

There were also differences in the patterns of gene expression between the left and median lobes in response to acetaminophen exposure. To examine patterns of expression we performed hierarchical clustering using Pearson’s Correlation Coefficient as a metric. We first performed unsupervised clustering with all animals from each dose group. This invariably resulted in the grouping together of all animals necropsied at the same time; thus all animals necropsied 6 hours after dosing clustered together, all animals necropsied 18 hours after dosing clustered together, etc. Therefore at this level the major factor determining membership in a cluster was the time after dosing when the animals were necropsied. However among animals necropsied at the same time, hierarchical clustering demonstrated clear differences in the patterns of gene expression between the left and median lobes. An example is shown in Figure 2 for rats necropsied 6 hours after administration of 150 mg/kg. This dendrogram shows that the left lobes from each animal form one cluster and the median lobes a second cluster. Thus, the gene expression patterns of the left or median lobes of different animals are more similar to one another than are the gene expression patterns of the left and median lobes from the same animal.

This distinction was most apparent for animals necropsied 6 hours after dosing where clustering was strictly by lobe. However even at later times, clustering by lobe predominated over clustering by animal for animals necropsied at the same time (Figure 3). In this figure the left and median lobe of animal 404 and 405 are clustered together, however for the other animals in this group, left lobes clustered with left lobes and median lobes with median lobes.

Discussion

Transcriptional profiling has revealed clear differences in the response of the left and median liver lobes of F344 rats to acetaminophen exposure with respect to the total number of genes differentially expressed in each lobe, the number of uniquely expressed genes in each lobe, and the pattern of expression. This was true at doses of 50 and 150 mg/kg that were not associated with any overt toxicity as well as doses of 1500 and 2000 mg/kg that caused obvious hepatotoxicity. Based on the number of genes identified as differentially expressed, the median lobe was more responsive to acetaminophen exposure than the left lobe.

Although there were differentially expressed genes in common in both lobes, there were also genes that were differentially expressed uniquely only in the left or median lobe. The most interesting difference occurred at the 2000 mg/kg dose where at each necropsy time, all genes differentially expressed in the left lobe were also differentially expressed in the median lobe, with an additional group of genes differentially expressed only in the median lobe. If only the left lobe had been sampled, this latter group of genes would not have been identified as differentially expressed.

Histopathologic analysis revealed the presence of hepatocellular necrosis in both the left and median lobes from animals exposed to 1500 or 2000 mg/kg, however the distribution of necrosis was patchy and not uniform across the section. There were some differences in lesion severity between the left and median lobes, but no obvious preference for either lobe was apparent. A similar response to acetaminophen exposure with focal areas of necrosis in one section of a lobe while other parts of the same lobe appear to be unaffected has previously been described by (McLean and Day, 1975). When the correlation between ALT levels and lobe specific severity grade was examined a significantly better correlation was found for the median lobe than for the left lobe. Thus, severity grades scored in the median lobe were more highly correlated to ALT levels than those of the left lobe.

There were also minor differences in the distribution of radioactivity among the four separate lobes, and between halves of the same lobe. However in general each of the lobes of the liver received comparable amounts of radioactive acetaminophen whether administered orally, intravenously, or intraperitoneally. There did not appear to be a differential lobe distribution based on route including oral gavage. Therefore it seems unlikely that unequal distribution of acetaminophen among the different lobes of the liver could account for all of the lobe specific differences in gene expression.

Total blood flow to each lobe is related to lobe size (Wheatley et al., 1993) however, each lobe is organized into 3-dimensional vascular units that exhibit biochemical gradients from the portal vessels entering the liver to the periphery (Teutsch, 1985; Teutsch et al., 1999). The left lobe has 1 primary portal branch but the median lobe appears to have 2 portal branches (Kogure et al., 1999) one of which it shares with the left lobe (Duchen, 1961). Therefore, nutrient levels, oxygen tension and gene expression may not be uniform across the hepatic lobe. Differences in blood flow to the various areas of the human liver (Sherriff et al., 1977) and within the left hepatic lobe in the rat (Andersson et al., 1987) suggest that not only should one address which lobe to sample for gene expression but perhaps specify whether the sample is from the hilar or peripheral part of the lobe.

Although it is not possible to generalize these results to compounds other than acetaminophen or other strains of rat, these results highlight the potential impact of sampling differences on the interpretation of gene expression profiles in the liver, or any organ for that matter, and underscore the importance of careful experimental design and complete description of exactly what tissue/organ was sampled in experiments involving gene expression profiling. Comparison of data sets from different experiments will be extremely difficult, and may be fruitless, without such information.