Gene Expression Profiling of Acute Liver Stress During Living Donor Liver Transplantation

Authors


Abstract

During liver transplantation, the donor graft is subjected to a number of acute stresses whose molecular basis is not well-understood. The effects of surgical stress, preservation and reperfusion injury were studied in 24 consecutive living donor liver transplant (LDLT) operations. Liver biopsies were taken early in the donor operation (OPENING), after transection of the donor liver (PRECLAMP) and following implantation of the graft (post hepatic artery, [PHA]); these were evaluated for histology, tissue glutathione content and gene expression using a 19K-human cDNA microarray. LDLT was associated with an ischemia/reperfusion injury, with accumulation of small numbers of neutrophils and decreased glutathione in the PHA biopsies. Following reperfusion, the expression of 129 genes increased and 106 genes decreased when compared to OPENING levels (> or <2-fold, p < 0.01). By real-time PCR a subset of 25 genes was verified (15 increased, 10 decreased). These genes were similarly altered in another condition of acute liver stress (the response to braindeath), but not in three chronic liver disease states (HCV, HBV and PBC). This study has identified a set of genes whose expression is altered in acute, but not chronic, liver stress, likely to play a central role in the pathogenesis of acute liver injury of liver transplantation.

Introduction

During a liver transplant, the graft is subjected to a number of innate and antigen-independent stresses. These include the hepatic response to systemic cytokines (stress of surgery, braindeath), the hypothermia and preservation injury of cold ischemia and the hepatic ischemia/reperfusion injury (HIRI) that follows implantation of the graft. These stresses impact the graft and affect clinical function; in particular, HIRI increases early graft failure and both acute and chronic rejection (1,2). However, the molecular basis underlying the clinical response to acute liver stress remains to be defined.

Much of the current understanding of hepatic transplantation injury comes from small animal models. Cold ischemia results in a pro-apoptotic state and is associated with alterations in mitochodrial respiration, ATP depletion, cellular ion shifts and susceptibility to non-lysosomal proteases (3,4). Hepatocyte apoptosis may augment neutrophil transmigration and act as a pro-inflammatory event (5–7). Following liver reperfusion, an ordered sequence of inflammatory events is driven by oxidant stress generated in Kuppfer cells (KC) and hepatocytes (8–12). Local (reactive oxygen species [ROS], complement, TNFα, IL-1, IL-6, PAF) and distant (gut-derived endotoxin) factors lead to KC activation, resulting in increased ROS generation, inflammatory mediators and chemokines. Upregulation of sinuisoidal cell adhesion molecules (ICAM-1, E-selectin) promotes the recruitment of circulating inflammatory cells (1,2). Neutrophils accumulate within the sinusoids within the first few hours of reperfusion but probably do not become important to hepatic damage until at least 6 h post-reperfusion (13,14). Recent work has also defined a role for T cells in HIRI (well-reviewed in Ref. [1]).

There is relatively little direct human data on the mechanisms of HIRI-associated transplantation injury. Electron microscopic evidence supports the early activation of KC and SEC (15). Increased levels of the ICAM-1 adhesion molecule, IL-1 cytokine and endothelin-1 vasoconstrictor have been documented early post-reperfusion (16,17). The NO precursor l-arginine is depleted 30 min post-reperfusion, suggesting that low levels of NO persist post-reperfusion (18). Plasma levels of TNFα, IL-1 and IL-6 are increased after hepatic ischemia and reperfusion (19,20). Apoptosis has been documented in both SEC and hepatocytes (21). None of the human data to date examines the molecular basis for the effects observed.

The living donor liver transplant (LDLT) operation offers a unique opportunity to examine the effects of HIRI directly. Normal and reasonably unstressed liver can be represented by a liver biopsy taken at the start of the donor operation (the OPENING liver biopsy), and compared to biopsies taken after division of the liver (PRECLAMP liver biopsy) and to biopsies taken at the end of the implantation of the graft in the recipient (post hepatic artery, PHA, liver biopsy). Comparing the gene expression changes in these different biopsies sheds light on the stresses subjected by the right lobe graft during liver transplantation. In order to extract functional connections among these genes in an impartial manner, we developed a method similar to the EASE method (22). In this method, enrichment for functional categories is compared statistically, eliminating investigator bias.

Materials and Methods

Patients and biopsy specimens

Living donor patients:  All living donors are healthy individuals with no significant comorbid illness and no evidence of prior or current liver disease. Thorough serological testing is performed, and a donor is excluded if hepatitis B or hepatitis C serology is positive. Donors are also excluded if there is evidence of significant steatosis on biopsy, or if the body-mass index exceeds 32.

Living donor liver transplant operation:  At the University Health Network, Toronto, Canada, LDLT operations are performed with staggered donor and recipient operations. The donor right hepatectomy is performed without vascular inflow occlusion (23). In the recipient operation, the donor right hepatic vein (and possibly middle hepatic vein) is first anastomosed to the inferior vena cava, followed by anastomosis of the donor right portal vein to the recipient portal vein. The graft is reperfused via the portal vein, and the donor right hepatic artery is anastomosed to the recipient right hepatic artery using a microvascular technique.

Living donor biopsy schedule:  From June 2002 through November 2003, 24 consecutive living donor transplant patients gave informed consent for this study. Normal liver tissue was biopsied as the first step of the right hepatectomy operation performed on living transplant donors (OPENING). Whenever feasible, a second biopsy was taken immediately prior to placing the vascular clamps and flushing the right lobe graft (PRECLAMP), and a third biopsy as the last step of the recipient operation, following reperfusion (venous and arterial) of the right lobe graft (PHA). Portions of each biopsy were promptly immersed in RNAlater (Qiagen, Mississauga, Canada), left at –4°C for 12 h and then stored at −20°C pending RNA extraction (see below). Portions of the remaining biopsy sample were immersed in 10% formaldehyde for histological analysis or were snap frozen in liquid nitrogen for GSH determination (see below).

Deceased donor liver biopsies:  From June 2002 through September 2003, liver biopsies were taken immediately after the initial laparotomy incision during 10 deceased donor (DD) operations. Portions of each biopsy were promptly immersed in RNAlater (Qiagen), left at −4°C for 12 h and then stored at −20°C pending RNA extraction (see below).

Chronic liver disease liver biopsies:  Liver biopsies from 48 patients with chronic hepatitis C viral infection (HCV) were taken and processed for gene expression profiling as previously described (24). Similar biopsies were obtained from 18 patients with chronic hepatitis B viral infection (HBV) and 31 patients with primary biliary cirrhosis (PBC).

Ethical review

In all cases, consent was given for use of human tissue for research using protocols reviewed by the hospital research ethics board.

GSH assays

Quantification of hepatic non-protein sulfhydryls was assessed as previously described using a 5,5′ dithiobis-2-nitrobenzoic acid (DNTB)-based assay (25,26). In brief, roughly 20 mg of homogenized (5% SSA) liver tissue was sonicated for 30 s and centrifuged for 10 min at 10 000g. Non-protein sulfhydryls were assayed from the acid thiol extract by spectrophotometrically quantifying the reduction of DNTB to 5-thio-2-nitrobenzoic acid at 412 nm. Sample values were calculated from a GSH standard curve and expressed as GSH equivalents per gram of liver tissue.

RNA extraction and amplification

A portion of each liver biopsy was immersed in RNAlater (Qiagen). Total RNA was extracted, and 2 μg of total RNA from each biopsy or from Universal Human Reference RNA (Stratagene, La Jolla, CA) was amplified using the MessageAmp aRNA kit (Ambion, Austin, TX) (24,27). Gene expression profiles from amplified RNA were highly correlated to those developed from non-amplified RNA (correlation coefficient ≥0.85, data not shown).

cDNA microarrays

Human single spot microarrays comprising 19 000 human clones were used (UHN Microarray Center, http://www.microarrays.ca/support/glists.html). For each array, 5 μg of liver aRNA was compared to 5 μg of reference aRNA. After reverse transcription, liver cDNA was labeled with Cy5 and reference cDNA with Cy3 (27). Hybridization was performed overnight at 37°C (DIGEasy, Roche, Basel, Switzerland). Arrays were read with a GenePix 4000A laser scanner and quantified with GenePix Pro software (Axon Instruments, Chicago, IL). Microarray data were normalized using an R-based, intensity-dependent lowess scatter plot smoother (http://142.150.56.35/~LiverArrayProject/home.html) (28–30).

Real-Time PCR

Two-step real-time PCR was performed after reverse transcription of 5 μg of aRNA with 5 μg pd(N)6-Random Hexamer primer (Amersham, Piscataway, NJ). The resulting cDNA was used as a template for real-time PCR quantification with the QuantiTect SYBR PCR Kit (Qiagen), and real-time PCR (normalized to β-actin) was performed using the DNA Engine Opticon 2 cycler (MJ Research, Hercules, CA). A description of the primers used is found in Table 1. All real-time PCR primers were designed with Primer 3 open-source software (http://frodo.wi.mit.edu/cgi-bin/primer3/primer3_www.cgi). Although not specifically designed to span introns or cross intron/exon boundaries, DNAse I was incorporated into the process of RNA amplification (for amplified RNA) or used before reverse transcription (for unamplified RNA). Genomic DNA contamination was not noted in our real-time PCR assay as no signal was obtained from the parallel control samples without reverse transcriptase (−RT). A primer control (without cDNA template) was included for each tested primer pair in order to monitor for ‘primer-dimer’ formation.

Table 1. Real-time PCR primers
cloneIDGene Nameforward primerreverse primerproduct length (bp)
137360transmembrane 4 superfamily member 1 CATGAAAACTGTGGCAAACG GCCGAGGGAATCAAGACATA 138
488548pre-B-cell colony enhancing factor 1 GCCAGCAGGGAATTTTGTTA TGATGTGCTGCTTCCAGTTC 168
222041cyclin-dependent kinase inhibitor 1A (p21, Cip1) GGAAGACCATGTGGACCTGT GGCGTTTGGAGTGGTAGAAA 146
428077syndecan 4 (amphiglycan, ryudocan) TCGATCCGAGAGACTGAGGT CCAGATCTCCAGAGCCAGAC 142
140447DEAD (Asp-Glu-Ala-Asp) box polypeptide 21 GCAGCAGTTATTGGGGATGT GCTGAATTCTGGGACAGCTC 107
262808thrombospondin 1 AGATGGCCACCAGAACAATC GTCATCATCGTGGTCACAGG 106
146483transglutaminase 2 TCAACCCCAAGTTCCTGAAG TCATCGTTGCAGTTGACCAT 103
501851hypothetical protein FLJ10116 CTGAGGGACTGGCTGAGAAC CATCCCTCACCAGAGTTGCT 135
291191nicotinamide N-methyltransferase GCGAGATCACCTCAAACCAT AGCACACTGTGTCTGGATGC 235
52027transglutaminase 2 (C polypeptide) TCAACCCCAAGTTCCTGAAG TCATCGTTGCAGTTGACCAT 103
24838Exostoses (multiple) 1 – heparan sulfate elongation TGCTGGTATTCAAGGGGAAG ACAGCGAGAATCCTTGTGCT 149
143439microtubule-associated protein 1 light chain 3 beta CTGTTGGTGAACGGACACAG CTGGGAGGCATAGACCATGT 108
4552577dual specificity phosphatase 5 AGGTCCTGGTCCACTGTGAG GAGACCATGCTCCTCCTCTG 130
130269serine (or cysteine) proteinase inhibitor, clade E (nexin, plasminogen activator inhibitor type 1), member 1 GACATCCTGGAACTGCCCTA GGTCATGTTGCCTTTCCAGT 138
274305SC5DL TCGTGTCCCCCAAATCTTAC CCCCTCAAAGGATGAAGGAT 148
120559metallothionein 1F (functional) TCCTGCAAGTGCAAAGAGTG CACTTCTCTGACGCCCCTTT 119
502897syndecan 4 (amphiglycan, ryudocan) TCGATCCGAGAGACTGAGGT CCAGATCTCCAGAGCCAGAC 142
49029FBJ murine osteosarcoma viral oncogene homolog B (FOSB) AGGAAGAGGAGAAGCGAAGG CGACTCCAGCTCTGCTTTTT 140
416118cytochrome P450, family 4, subfamily A, polypeptide 11 TTCAGCACGTCTCCTTGATG ATGGCCTGGATGTAGGACTG 106
2882043histamine N-methyltransferase CCTCACTCAGATGCTGGACA TGCTCTGAGATCAGGTGGTG 172
230477protein kinase, AMP-activated, beta 2 non-catalytic subunit GCGTTTCGATCTGAGGAAAG GGAGTAAGGCTGGGTCACAA 109
23778polycystic kidney disease 2 (autosomal dominant) GTTCCACGAATACGGCAACT CCAAAGGGAGCCCTATCTTC 125
44477vascular cell adhesion molecule 1 AAGATGGTCGTGATCCTTGG GGTGCTGCAAGTCAATGAGA 138
123730src family associated phosphoprotein 1 GCTCCTGGAAGATGCTGAAG CATCAGAGCTGTCCTGTCCA 173
41563solute carrier family 6 (neurotransmitter transporter, GABA), member 1 CATGTCCTGTGTGGGCTATG CGCAAAGATGAGTGTCAGGA 118
124365glycine-N-acyltransferase GTGGACAAGTGGCCTGATTT CAGTTTTGGGGATCTTTGGA 119
2728822alanine-glyoxylate aminotransferase 2 TCTTCTTCCACCCTCCAATG ATTGGCTTCTGAGCCACTGT 104
240729similar to RIKEN cDNA 0610008P16 gene (MGC15937) TCGAGAGCACTCCTCTTGGT GACATCCAGCTGGGCATACT 147
241037bile acid Coenzyme A TCAGGCATCACTGGAAGATG TGGGGTGGACTCCCATATAA 131

Statistics and clustering analyses

Comparisons between continuous variables were performed using the two-sample Welch t-statistic with the multtest package, which includes an estimation of adjusted p-values by permutation (31). Unsupervised hierachicial clustering was performed using the R mva package with the Hclust function (32,33). Details can be found at http://142.150.56.35/~LiverArrayProject/home.html.

GO database interrogations (the ‘mineGO’ analysis)

We used the R packages GO and AnnBuilder from the Bioconductor Open Source and Open Development Software Project (34) to develop an automated method for interrogating the Gene Ontology (GO) database, similar to the EASE method described by Hosack et al. (22). Full details of our method can be found at http://142.150.56.35/~LiverArrayProject/home.html. In brief, as in the EASE method we have divided the GO functional categories into levels (moving broadly from less-specialized to more-specialized functions). Each clone on the array is annotated across a number of GO categories, such that there will be a discrete number of genes in each category (at a given GO level) depending on the total number of genes represented on a specific microarray. We then compare the distribution of genes in a given GO category as expected by chance alone (specific to the microarray in use) to the distribution of genes in the same category in the list of genes determined to be significant (statistically and by fold increase/decrease) in our experimental conditions. A Fisher exact test is used to calculate significance, and ‘enrichment’ of a given GO category is taken to mean that the functional category (at a given GO level) has significantly more genes present than would be expected by chance alone, to a p-value of <0.05. The programing code for our method can be freely downloaded from the above Web site.

Results

Living donor liver graft reperfusion is associated with a pro-inflammatory HIRI

As noted above, the process of liver preservation and reperfusion is well-documented to be pro-inflammatory in both experimental and clinical studies. Our initial studies were aimed at quantifying the degree and timeline of inflammatory and ischemia/reperfusion changes in the LDLT operation. As shown in Table 2, the donor right lobe grafts were generally subjected to 1.5 h of cold ischemia, 1 h of warm ischemia, 1 h of perfused ischemia (after the portal venous anastomosis but prior to the hepatic artery anastomosis) and a further 2 h until the completion of the bile duct anastomosis. Thus, overall, the PHA biopsy takes into account roughly 3 h of post-reperfusion effect (portal venous and hepatic artery). In comparison, the DD liver biopsies reflect the local effects of almost 23 h of braindeath.

Table 2. Patient demographics
 LDLTDDHCVHBVPBC
  1. N/A: not applicable; Data = mean ± SD.

n2410461831
Donor age (years)37.5 ± 1235.5 ± 1849.9 ± 840 ± 1253.8 ± 9
Donor sex (%male)509073780
Recipient age (years)54.7 ± 846.2 ± 14.1N/AN/AN/A
Recipient sex (%male)7162N/AN/AN/A
Hours braindeathN/A22.6 ± 10.8N/AN/AN/A
Hepatic fibrosis002.2 ± 1.042.4 ± 1.041.36 ± 0.76
Disease activity001.4 ± 0.51.6 ± 0.5N/A
CIT (hrs)1:47 ± 2.69.18 ± 3.25N/AN/AN/A
WIT (hrs)0:54 ± 10:58 ± 0:31N/AN/AN/A
HIT (hrs)0:58 ± 1.51:02 ± 0:20N/AN/AN/A
Artery – final biopsy (hrs)0:58 ± 1.5N/AN/AN/AN/A

As seen in Figure 1, based on histologic criteria of inflammation and organ injury, the right hepatic lobe remains essentially unchanged in response to surgical manipulation and the response to the systemic stresses of surgery (the PRECLAMP phase). However, after reperfusion, the PHA graft consistently demonstrated evidence of neutrophilic infiltration. Tissue glutathione levels decreased only after reperfusion, consistent with a HIRI in the later phase of the LDLT operation (Figure 1E). DD liver tissue was essentially normal by histologic criteria (Figure 1D).

Figure 1.

LDLT is associated with a pro-inflammatory HIRI. Representative H&E sections from routine histological examinations of the liver biopsies on which gene expression profiling was subsequently performed, 20× magnification. (A) OPENING; (B) PRECLAMP; (C) PHA (the indicated region is magnified, demonstrating the typical, mild neutrophilic infiltration scattered through the sections of PHA liver biopsies); (D) deceased donor (DD) liver tissue; (E) Glutathione assays on LDLT liver biopsies. Data are presented as mean SEM, n = 6 per group. Units are arbitrary, normalized to the OPENING values for each patient. *p < 0.05, ANOVA.

Surgical stress and transplantation injury alter hepatic gene expression

Both surgical stress (PRECLAMP) and transplantation injury (PHA) effect alterations in hepatic gene expression, as shown in Tables 3 and 4. In PHA liver tissue, the expression of a total of 129 genes was increased by at least 2-fold, and the expression of 106 was decreased by at least 2-fold, when compared to OPENING tissue gene expression (p < 0.01). Interestingly, there was only a single gene (C9orf3) whose expression was altered in the PRECLAMP phase and not also altered in PHA liver tissue (data not shown); as such, the data presented in Tables 3 and 4 can be viewed as a time course of gene expression over the course of the LDLT operation. To ensure that our microarray measurements could be confirmed using other methods, we used real-time PCR to compare gene expression in opening, PRECLAMP and PHA liver biopsies. We made the assumption that the genes whose levels were most up-regulated or down-regulated in PHA liver tissue would be the ones most likely to be altered in response to acute liver stress in general. We therefore performed real-time PCR on the genes most up-regulated, and most down-regulated (by fold changes relative to baseline OPENING values), in PHA tissue, as outlined in Tables 3 and 4. We arbitrarily aimed to validate at least 25 genes, deeming this a reasonably large subset for functional analyses. In Figure 2, we present the 25 genes, all successfully validated—15 increased and 10 decreased in response to transplantation injury. Overall, our validation rate exceeded 70% (data not shown).

Table 3. Genes up-regulated in response to acute hepatic stress
CloneIDAccNumUnigeneGene NameSymbolPRECLAMP/OPENSD p (pre vs open) PHA/OPENSD p (pha vs open)
  1. Genes up-regulated in the PHA liver biopsies by at least 2-fold, p < 0.01 vs OPENING tissue. The behavior of the same genes in PRECLAMP samples is also presented; PRECLAMP genes which increased by at least 2-fold (relative to OPENING) are highlighted in red. Data are presented as the mean of 25 PHA samples, 10 PRECLAMP samples, in comparison to 20 OPENING samples. Stats: ANOVA; p-values correspond to the column at left.

137360R38114Hs.351316transmembrane 4 superfamily member 1TM4SF16.02359272.100.000115.30657641.780.0001
488548 AA047110 Hs.293464 pre-B-cell colony enhancing factor 1 PBEF1 4.94980151.680.00018.05454081.460.0001
222041H83378NACyclin-dependent kinase inhibitor 1A (p21, Cip1)CDKN1A4.93972761.650.00016.85261481.610.0001
428077 AA002237 Hs.252189 syndecan 4 (amphiglycan, ryudocan) SDC4 5.76550372.080.00016.8112261.620.0001
140447R66021Hs.169531DEAD (Asp-Glu-Ala-Asp) box polypeptide 21DDX213.89198011.640.00016.54083611.510.0001
262808N28322NAThrombospondin 1THBS14.16198012.190.00015.77241471.790.0001
146483 R78823 NA Transglutaminase 2 (C polypeptide) TGM2 2.41309221.560.00024.7357441.590.0001
418264W92662Hs.349611protein kinase C, alphaPRKCA2.66898721.760.00014.64806921.740.0001
501851AA128002Hs.79741hypothetical protein FLJ10116FLJ101162.87776831.910.00024.42246681.390.0001
291191W03376Hs.364345nicotinamide N-methyltransferaseNNMT3.70997051.530.00014.41218441.240.0001
51771H23321NAESTEST3.39501552.130.00034.3342441.400.0001
52027 H24115 Hs.512708 transglutaminase 2 (C polypeptide) TGM2 2.28070971.560.00044.14926521.640.0001
143439R74579Hs.356061microtubule-associated protein 1 light chain 3 betaMAP1LC3B1.89055261.270.00013.84325131.470.0001
4552577BG337015Hs.2128dual specificity phosphatase 5DUSP51.7473031.360.00023.83649851.630.0001
24838T80560NAExostosin 1 (heparan sulfate elongation)EXT13.73037482.590.00343.8266961.770.0001
130269R21221Hs.414795plasminogen activator inhibitor type 1SERPINE12.60894471.680.00023.80089621.520.0001
5578122BM468475Hs.101651hypothetical protein LOC90133LOC901332.36687781.500.00013.65431841.410.0001
52419H25214NAChromosome 9 open reading frame 61C9orf612.43390141.760.00023.65289451.530.0001
274305H49219Hs.28578muscleblind-like 1MBNL12.07494821.610.00173.61377161.610.0001
489747AA101987Hs.418291interleukin 10 receptor, betaIL10RB2.29051531.800.00173.51618451.530.0001
120559T95199Hs.438737metallothionein 1F (functional)MT1F2.36260551.450.00023.51350461.310.0001
382658AA069485Hs.433307branched chain keto acid dehydrogenase E1BCKDHA2.31699121.600.00023.45094511.310.0001
502897 AA128572 Hs.252189 syndecan 4 (amphiglycan, ryudocan) SDC4 2.70939831.460.00013.40921051.290.0001
49029H14887Hs.75678FBJ murine osteosarcoma viral oncogene homolog BFOSB2.95481032.930.02323.39991582.280.0001
222188H83491NAStaphylococcal nuclease domain containing 1SND12.44059361.330.00013.25418961.320.0001
470647AA032031Hs.133916hypothetical protein LOC152485LOC1524852.36287851.560.00013.22356131.350.0001
142944R71124Hs.10784chromosome 6 open reading frame 37C6orf371.69915451.450.00113.19646451.530.0001
3839322 BE738657 Hs.293464 pre-B-cell colony enhancing factor 1 PBEF1 2.36913841.380.00013.17097481.440.0001
417706W89053NALOC125150LOC1251502.07031921.410.00013.14596781.390.0001
128812R10165Hs.127767similar to CG14894-PALOC1305021.89064771.460.00063.14545341.360.0001
284614 N71848 Hs.372571 muscleblind-like 2 MBNL2 2.8966991.610.00013.10426031.350.0001
240148H82698Hs.298119myeloid/lymphoid or mixed-lineage leukemia 3MLL31.85839951.500.00033.08167121.500.0001
143842R75975Hs.303649chemokine (C-C motif) ligand 2CCL23.49062311.540.00013.02198551.390.0001
136040R35451Hs.435735solute carrier family 5SLC5A61.66515011.510.00132.98039031.480.0001
21815T65624Hs.298654dual specificity phosphatase 6DUSP62.16987691.670.00142.97907121.770.0001
51362H22698Hs.258561general transcription factor IIBGTF2B1.94521031.400.00022.91049711.500.0001
46042H09007Hs.75199protein phosphatase 2, regulatory subunit B (B56)PPP2R5B1.71959661.350.00042.86113451.250.0001
236109H53723Hs.313544nucleosteminNS2.31817831.530.00012.85998851.310.0001
152802R50467Hs.76422phospholipase A2, group IIAPLA2G2A1.89709843.150.45772.77494582.380.0004
205813H58269NADiaphanous homolog 2 (Drosophila)DIAPH21.62289511.410.00182.76164271.370.0001
111840T84868Hs.90232ProSAPiP1 proteinProSAPiP12.06090161.330.00012.72266581.220.0001
44590H06595Hs.380138chimerin (chimaerin) 1CHN11.72433191.440.00092.70393181.340.0001
110051T85234NATriple functional domain (PTPRF interacting)TRIO1.97992971.810.00912.6995871.400.0001
5088039BI262416Hs.166975splicing factor, arginine/serine-rich 5SFRS51.75764511.410.00042.69202611.530.0001
429914AA033787Hs.167641G protein-coupled receptor 108GPR1081.92570631.680.00062.68401381.390.0001
2157607AI478910Hs.439190chromosome 14 open reading frame 31C14orf312.46559481.320.00012.68040051.290.0001
324024W46504Hs.115285dihydrolipoamide S-acetyltransferaseDLAT1.7199661.470.00152.67925391.660.0001
175090H39162Hs.409230lysophosphatidic acid acyltransferase, alphaAGPAT12.035791.760.00242.67304081.480.0001
4660714BG469305Hs.406013keratin 18KRT181.64150461.320.00022.66823391.320.0001
273065N33087Hs.443793myosin regulatory light chain interacting proteinMYLIP1.71589761.710.02072.66309041.570.0001
486381AA043760Hs.258561general transcription factor IIBGTF2B1.706491.580.00832.66215711.400.0001
23590T77476Hs.9691guanine nucleotide binding protein, alpha 13GNA131.62265971.310.00012.62463171.480.0001
2619620AW131518Hs.9075serine/threonine kinase 17a (apoptosis-inducing)STK17A1.78102181.260.00012.6072061.270.0001
5533389BM470481Hs.211556ELOVL family member 6ELOVL62.0753811.600.00042.59835441.520.0001
428336AA005419Hs.356494hypothetical protein LOC283680LOC2836801.65097421.430.00122.59395431.200.0001
190666H38594Hs.129882interphotoreceptor matrix proteoglycan 1IMPG12.00122811.610.00012.57487071.330.0001
241629H91613Hs.418241metallothionein 2AMT2A2.05095761.370.00012.56639541.280.0001
1721285AI151284Hs.376617MICAL-like 2FLJ234711.61931441.350.00082.5482971.230.0001
5481755BM915559Hs.76244spermidine synthaseSRM1.88067981.490.00042.54451181.510.0001
239744H80591Hs.356321nuclear factor of activated T-cellsNFATC22.28800531.360.00012.49702361.270.0001
178627H49329Hs.433759barrier to autointegration factor 1BANF11.63004951.270.00012.49528891.330.0001
501994AA125917Hs.66interleukin 1 receptor-like 1IL1RL11.42488231.450.01762.48807491.450.0001
47950H12136NADedicator of cytokinesis 1DOCK11.8790611.440.00012.47518151.350.0001
26104R12535NAChimerin (chimaerin) 1CHN11.79514511.440.00082.44907681.380.0001
155826R72253Hs.323445chromosome 9 open reading frame 75C9orf751.59826621.310.00032.44025631.200.0001
5836756BQ008175Hs.159118adenosylmethionine decarboxylase 1AMD11.82100171.450.00122.43058761.370.0001
304962W38982Hs.315562glutamate-cysteine ligase, modifier subunitGCLM2.10813611.520.00022.42693511.400.0001
133085R26344Hs.10784chromosome 6 open reading frame 37C6orf371.60500241.390.00122.41040621.380.0001
132691R26028Hs.28988glutaredoxin (thioltransferase)GLRX2.07754421.570.00022.4066561.400.0001
178494H46498Hs.516393cripto, FRL-1, cryptic family 1CFC11.400131.510.04832.40182451.440.0001
117640T89963NAProtein phosphatase 1, regulatory subunit 12APPP1R12A1.7532261.320.00012.37017591.260.0001
5787700BQ049778Hs.130946egl nine homolog 1 (C. elegans)EGLN11.43750411.370.00382.35228841.500.0001
502153AA126675Hs.65377hypothetical protein BC009489LOC929791.75360741.750.02672.34489341.370.0001
304820W38872Hs.159118adenosylmethionine decarboxylase 1AMD11.78098211.300.00012.33778541.350.0001
139818R62373NAKIAA0669 gene productKIAA06691.8694861.330.00012.33599691.480.0001
5725681BM804942Hs.166254likely ortholog of rat vacuole membrane protein 1VMP11.88900741.460.00012.32458951.320.0001
321665W33115Hs.78465v-jun sarcoma virus 17 oncogene homolog (avian)JUN1.65925891.220.00012.29296481.530.0001
26582R14003NAELAV (Hu antigen C)ELAVL31.7535171.420.00032.29239971.300.0001
293298N92095NAESTEST2.12956971.300.00012.28642631.710.0001
120036T94968NAESTEST1.87164011.500.00012.28582511.290.0001
174706H30357Hs.36992ATPase, H+/K+ exchanging, alpha polypeptideATP4A2.29563191.740.00062.28241071.490.0001
4800687BG696089Hs.13014ADP-ribosylation factor GTPase activating protein 3ARFGAP31.68590391.370.00012.27315641.370.0001
290242N92185Hs.1119nuclear receptor subfamily 4, group A, member 1NR4A11.57045981.270.00012.27283831.210.0001
238558 H64803 NA Muscleblind-like 2 MBNL2 1.95920911.410.00032.26146051.460.0001
5228392BI835258Hs.433345RNA, U22 small nucleolarRNU221.35595941.400.02882.25205451.450.0001
151515H03751Hs.353519hypothetical protein MGC26963MGC269631.37625481.230.00012.24560391.350.0001
428542AA005393Hs.51299NADH dehydrogenase (ubiquinone) flavoprotein 2NDUFV21.55795551.580.03062.24397281.880.0002
265267N27681Hs.274402heat shock 70kDa protein 1AHSPA1A1.63445581.430.00122.23775891.300.0001
211943H66745NADynamin binding proteinDNMBP1.41715461.240.00052.23444671.250.0001
123640R02721Hs.19413S100 calcium binding protein A12 (calgranulin C)S100A121.75946541.230.00012.2338171.420.0001
267804N34210Hs.32352zinc finger protein 395ZNF3951.60548631.410.00112.22209961.450.0001
5731537BM805846Hs.177766ADP-ribosyltransferase (NAD+ADPRT1.4868431.340.00122.2217921.360.0001
4345490BF792909Hs.18271golgi phosphoprotein 3 (coat-protein)GOLPH31.85404121.520.00052.2201851.500.0001
67071T70439NAInterleukin 22 receptor, alpha 1IL22RA12.04577551.920.01722.21967681.480.0001
325645W52609Hs.28988glutaredoxin (thioltransferase)GLRX2.14656331.530.00012.21901281.400.0001
149932H01247Hs.154057matrix metalloproteinase 19MMP191.74741861.370.00012.19270821.200.0001
3866474 BE777770 Hs.181243 activating transcription factor 4 ATF4 1.82323681.470.00022.18670421.330.0001
47400H10627NASignal-induced proliferation-associated 1 like 2SIPA1L21.48348911.270.00012.17795381.470.0001
162333H27657Hs.369441G protein-coupled receptor kinase interactor 1GIT11.68216411.320.00012.17365661.260.0001
415343W92173Hs.380774DEAD (Asp-Glu-Ala-Asp) box polypeptide 3, X-linkedDDX3X1.7444021.360.00012.17167951.430.0001
186430H44309Hs.111024solute carrier family 25 (mitochondrial carrierSLC25A11.68288931.570.00382.15861471.450.0001
45088H08650Hs.130946egl nine homolog 1 (C. elegans)EGLN11.32452321.350.02292.15837511.520.0001
2406400AI830609Hs.414362cytochrome b5 reductase b5R.2CYB5R21.88168081.590.00232.15598431.300.0001
231663H93176NA6-phosphofructo-2-kinase/fructose-2,6-biphosphatase 3PFKFB31.84260491.420.00012.145711.320.0001
338437 W51822 Hs.181243 activating transcription factor 4 ATF4 1.92750591.640.00272.14454351.580.0001
491644AA150254Hs.371003placenta-specific 8PLAC80.96192780.96NA2.14094891.910.0000
470635AA031376Hs.252820vascular endothelial growth factor-related proteinPGF1.8160021.510.00022.13856431.390.0001
142606R70894Hs.161999TGFB-induced factor (TALE family homeobox)TGIF1.51265941.230.00012.13270371.330.0001
272210N44809Hs.6838ras homolog gene family, member EARHE1.91146931.590.00202.13037291.440.0001
148427H12368Hs.417402mucolipin 1MCOLN11.44119361.300.00132.12557741.190.0001
416094W85966Hs.19333hypothetical protein FLJ10349FLJ103491.66820721.410.00042.12510391.320.0001
4723271BG567139Hs.7917likely ortholog of mouse hypoxia induced gene 1HIG11.58565161.420.00222.12056561.430.0001
36682R25310NACUG triplet repeat, RNA binding protein 2CUGBP21.72143621.390.00212.11965131.630.0001
5549800BM552637Hs.279806DEAD (Asp-Glu-Ala-Asp) box polypeptide 5DDX51.73245741.250.00022.11905411.460.0001
278740N98254NARNA terminal phosphate cyclase-like 1RNAC1.50591161.320.00102.103191.370.0001
5533275BM460992Hs.91773protein phosphatase 2, catalytic subunit, alphaPPP2CA1.79892041.390.00012.08971141.260.0001
128159R10412Hs.170472translocated promoter region (to activated MET)TPR1.28980941.310.02382.07992161.240.0001
290758W01296Hs.11360disrupted in renal carcinoma 2DIRC21.97571891.470.00012.07882191.200.0001
40038R53445NAESTEST1.55965671.340.00032.07522241.250.0001
25402R11719Hs.15243HOM-TES-103 tumor antigen-likeHOM-TES-1031.49933831.210.00012.07366361.250.0001
143920R76842NAhypothetical protein FLJ23153FLJ231531.15288171.340.22512.07342341.840.0002
173265H21130Hs.183800Ran GTPase activating protein 1RANGAP11.57726681.300.00022.07277611.270.0001
133471R27271Hs.443271programmed cell death 1 ligand 1PDCD1LG11.83326561.610.01192.07198111.370.0001
489961AA115453Hs.68257general transcription factor IIF, polypeptide 1GTF2F11.75741761.560.00262.06412931.350.0001
153322R47760NATU3A proteinTU3A1.74054141.360.00022.05481831.400.0001
5421424BM006748Hs.433455enolase 1, (alpha)ENO11.38226161.290.00252.04676881.340.0001
144944R78657NAChemokine (C-X-C motif) receptor 4CXCR41.6673691.300.00022.04494761.340.0001
415021W93055Hs.1908proteoglycan 1, secretory granulePRG11.85690591.380.00012.04051921.310.0001
305406W23477Hs.191320hypothetical protein FLJ32731FLJ327311.39180621.370.01762.03653811.330.0001
45932H09565Hs.25001Tyr 3-monooxygenase/Trp 5-monooxygenase activationYWHAG1.46016861.320.00242.02440731.370.0001
5752052BM920804Hs.124940Rho family GTPase 1RND11.82532771.430.00012.01943851.480.0001
266794N31391Hs.197081A kinase (PRKA) anchor protein (gravin) 12AKAP121.36994521.230.00152.00891021.260.0001
277423N47738Hs.21355doublecortin and CaM kinase-like 1DCAMKL11.83270211.630.00692.00373741.360.0001
342089W61046Hs.203772FSHD region gene 1FRG11.36198761.410.00002.00197432.200.3432
Table 4. Genes down-regulated in response to acute and chronic hepatic stress
CloneIDAccNumUnigeneGene NameSymbolPRECLAMP/OPENSD p (pre vs open) PHA/OPENSD p (pha vs open)
  1. A comparison of the genes down-regulated by at least 2-fold, p < 0.01, in 25 PHA liver biopsies when compared to the baseline gene expression in 20 OPENING liver tissue. Stats: ANOVA; p-values correspond to the column at left. The behavior of the same genes in 10 PRECLAMP liver tissue biopsies is presented also; genes that have decreased by at least 2-fold relative to the OPENING liver tissue are highlighted in green.

241037H91396Hs.284712bile acid Coenzyme ABAAT0.31526540.190.00010.14388690.100.0001
294160W01986NAESTEST0.39416460.270.00010.21386510.100.0001
196784R92830NAESTEST0.31008480.290.00010.23196240.120.0001
240729H90986Hs.38085similar to RIKEN cDNA 0610008P16 geneMGC159370.36966580.290.00060.23423140.200.0001
4767327BG618308Hs.75431fibrinogen, gamma polypeptideFGG0.2282250.150.03990.23474070.140.0267
416102W85970Hs.443161dihydropyrimidinaseDPYS0.37616560.170.00010.25318760.050.0001
2728822AW296131Hs.193228alanine-glyoxylate aminotransferase 2AGXT20.29870540.160.00010.25630050.140.0001
309685W30810NALeucine rich repeat (in FLII) interacting protein 2LRRFIP20.49689120.210.00010.29505450.090.0001
124365R00992Hs.145384glycine-N-acyltransferaseGLYAT0.42145380.200.00030.29638380.170.0001
767598AA418235Hs.10119hypothetical protein FLJ14957FLJ149570.5139870.380.00070.29856250.150.0001
41563R52864Hs.22003solute carrier family 6, member 1SLC6A10.42816510.240.00040.30172760.120.0001
123730R01281Hs.411942src family associated phosphoprotein 1SCAP10.50160320.260.00110.3169190.130.0001
5798249BM928510Hs.427202transthyretin (prealbumin, amyloidosis type I)TTR0.26130310.090.00060.32172640.270.0007
23778T77325Hs.458291polycystic kidney disease 2 (autosomal dominant)PKD20.42805460.120.00010.33367350.050.0001
37647R60672Hs.289319hypothetical protein LOC284323LOC2843230.4851230.100.00010.33498380.090.0001
230477H81127Hs.50732protein kinase, AMP-activated, beta 2 non-catalytic subunitPRKAB20.50065460.210.00010.33656920.080.0001
44477H07071Hs.109225vascular cell adhesion molecule 1VCAM10.4805690.130.00010.33746810.150.0001
2882043AW475074Hs.42151histamine N-methyltransferaseHNMT0.67809110.480.00890.34305660.130.0001
416118W85981Hs.448993cytochrome P450, family 4, subfamily A, polypeptide 11CYP4A110.41689590.120.00010.34821120.050.0001
301565W17011Hs.438311angiotensin II receptor-like 1AGTRL10.47486760.170.00020.34919320.150.0001
48267H12153Hs.407520chimerin (chimaerin) 2CHN20.56343680.190.00010.35168520.150.0001
505530AA147026NAESTEST0.63429020.210.00100.35628090.100.0001
246246N77085Hs.1282complement component 6C60.47861370.360.00370.35692840.240.0001
232656H73495NAESTEST0.43740490.140.00010.36138770.060.0001
2319582AI708697Hs.407138queuine tRNA-ribosyltransferase domain containing 1QTRTD10.62456020.320.00140.36260970.110.0001
40175R53688Hs.173933nuclear factor I/ANFIA0.59868190.140.00010.3662870.130.0001
37458R35197Hs.119583-hydroxysteroid epimeraseRODH0.44474640.260.00020.36697110.170.0001
291075W00403Hs.369424ectonucleoside triphosphate diphosphohydrolase 7ENTPD70.60972530.270.00080.37448030.100.0001
4730866BG621623Hs.449631hemoglobin, gamma AHBG10.40014410.160.01810.37664890.130.0015
47396H10439Hs.458301family with sequence similarity 35, member AFAM35A0.58652340.290.00010.37758020.110.0001
5760037BM923901Hs.78996proliferating cell nuclear antigenPCNA0.4185450.170.00070.38067830.160.0001
504278AA149501Hs.440934arginase, liverARG10.37602030.230.00030.3815360.190.0001
112779T85956Hs.15456PDZ domain containing 1PDZK10.51891130.170.00010.38161730.110.0001
123919R01518NAComplement component 5C50.66364490.510.01480.38250650.180.0001
42639R61092Hs.512235inositol 1,4,5-triphosphate receptor, type 2ITPR20.60327580.190.00010.38390670.130.0001
206255H58539NAESTEST0.62012170.240.00030.38514360.140.0001
66329T66826NAESTEST0.57948730.480.00990.38543380.190.0001
491272AA114890Hs.1501syndecan 2 (cell surface-associated, fibroglycan)SDC20.49155320.200.00030.38692070.140.0001
145554R78119Hs.308019hypothetical protein FLJ10980FLJ109800.61919250.350.00380.3874480.080.0001
2250186AI620916Hs.302772SH3 domain binding glutamic acid-rich protein like 2SH3BGRL20.56099090.160.00010.3891010.170.0001
282007N54238NAAdrenergic, alpha-1A-, receptorADRA1A0.70985870.350.07970.39204250.130.0001
136037R35457Hs.24395chemokine (C-X-C motif) ligand 14CXCL140.65881290.480.02940.39288970.180.0001
230515H75839NAESTEST0.39490270.070.00180.39332920.140.0003
246571N73259Hs.429896alcohol dehydrogenase 6 (class V)ADH60.55095580.120.00010.39335070.140.0001
4803331BG698179NAInositol 1,4,5-triphosphate receptor, type 2ITPR20.61237040.270.00350.39490570.080.0001
209783H65614Hs.300774fibrinogen, B beta polypeptideFGB0.38938070.110.03200.39855460.100.0175
2130477AI494175Hs.278483histone 1, H4jHIST1H4J0.39537980.120.00380.40344050.090.0003
267340N33796Hs.283749ribonuclease, RNase A family, 4RNASE40.43400510.110.04560.40684940.120.0106
274227H49804Hs.75431fibrinogen, gamma polypeptideFGG0.42665130.340.02670.40820770.200.0066
49270H16644Hs.408142hypothetical protein KIAA1109KIAA11090.65483210.100.00010.40848090.080.0001
26094R12356Hs.82163monoamine oxidase BMAOB0.50240560.300.01220.40977930.240.0001
289678N77031NAHypothetical protein FLJ31842FLJ318420.53093520.230.00050.41090490.160.0001
415553W78787Hs.1281complement component 5C50.69733840.540.02930.41454260.200.0001
277481N48742Hs.20242hypothetical protein FLJ12788FLJ127880.5096710.120.00010.4146970.060.0001
214553H71209Hs.75599antithrombinSERPINC10.47115580.060.00010.41526250.070.0001
240356H78000Hs.81934acyl-Coenzyme A dehydrogenase, short/branched chainACADSB0.66619550.660.01510.41595890.160.0001
202383H53170Hs.10119hypothetical protein FLJ14957FLJ149570.41570240.140.00180.41903260.110.0001
182870H45103Hs.373838attractinATRN0.65046340.150.00030.42055540.060.0001
138917R62862Hs.437922v-myc myelocytomatosis viral oncogene homolog 1MYCL10.51970750.210.00010.42814490.060.0001
5926321BQ064577Hs.418167albuminALB0.47522760.210.00830.43352170.240.0003
502706AA126200NAInterleukin 1 receptor-like 1IL1RL10.64886950.210.00070.43550310.130.0001
22501T74081Hs.287749sterol-C5-desaturaseSC5DL0.7352130.210.01490.43626530.100.0001
38642R51129Hs.150595cytochrome P450, family 26, subfamily A, polypeptide 1CYP26A10.86568710.710.63380.43755930.180.0001
324242W47459Hs.458301family with sequence similarity 35, member AFAM35A0.59371650.360.00160.4376740.090.0001
240647H78062Hs.1084413-hydroxyanthranilate 3,4-dioxygenaseHAAO0.46067040.160.00020.43776290.230.0001
126513R06795Hs.424907hypothetical protein MGC35366MGC353660.53949410.350.01030.43995730.210.0001
193984R83870Hs.408106dopa decarboxylase (aromatic L-amino acid decarboxylase)DDC0.52604880.420.00450.4404140.320.0001
51052T74211NAESTEST0.59648080.290.16780.44041690.110.0001
172114H20335NAESTEST0.66710570.210.00200.44727330.200.0001
233786H66389Hs.38069complement component 8, beta polypeptideC8B0.6147170.230.00240.44744830.200.0001
278772N66548Hs.119597stearoyl-CoA desaturase (delta-9-desaturase)SCD0.50368920.140.00040.44795040.150.0001
416050W85921NAESTEST0.71263680.290.00750.45136350.110.0001
295420W04478NAPlasminogen-likePLGL0.64550510.420.26940.45267710.320.0026
127724R09709NAHypothetical protein FLJ20605FLJ206050.59995620.280.00430.45372880.150.0001
203251H54740Hs.284712bile acid Coenzyme ABAAT0.61291460.220.00480.45511570.130.0001
380548AA053851Hs.84522hypothetical protein FLJ31842FLJ318420.69109250.240.00160.45534250.110.0001
2518851AI879705Hs.34871zinc finger homeobox 1bZFHX1B0.60046660.210.00020.45580470.100.0001
231461H92252NAESTEST0.64561680.220.00170.45822910.110.0001
48659H14709Hs.116122pantothenate kinase 1PANK10.76483620.300.05810.45878890.210.0001
35076R25019NAESTEST0.64302330.250.02710.46085510.070.0001
416126W85985Hs.311346cytidine monophosphate N-acetylneuraminic acid synthetaseCMAS0.72373450.520.00850.46215880.080.0001
281588N51199Hs.195799microsomal triglyceride transfer protein (88kDa)MTP0.85793350.460.19440.46455080.190.0001
120288T97103NAPre-B-cell leukemia transcription factor 1PBX10.44283490.050.00660.46554990.080.0033
240300H89783Hs.159628antitrypsinSERPINA40.59813960.250.00990.46562020.240.0001
126392R06513Hs.134035CD5 antigen-like (scavenger receptor cysteine rich family)CD5L0.49413670.290.01040.46691910.280.0003
34286R19706Hs.408177pleckstrin and Sec7 domain containing 3PSD31.03581980.590.85950.46809340.130.0001
46166H09425Hs.152096cytochrome P450, family 2, subfamily J, polypeptide 2CYP2J20.5900390.150.00030.46891620.130.0001
243600N49822Hs.446309glutathione S-transferase A1GSTA10.40460320.260.00020.46908250.320.0001
201383R98623Hs.284712bile acid Coenzyme ABAAT0.52333110.250.00140.46922030.160.0001
138632R63418NAESTEST0.79424130.410.08070.46932430.150.0001
108703T72733Hs.335599hypothetical protein FLJ38348FLJ383480.60841590.100.00010.47039820.110.0001
123843R00659Hs.301819zinc finger protein 146ZNF1460.45423730.090.80010.47175990.080.9820
147217R80536NAKeratin associated protein 4–7KRTAP4–70.75607380.430.00980.47437620.060.0001
5726059BM805170Hs.178499HSPC063 proteinHSPC0630.74282610.170.00110.47594040.100.0001
26610R14034Hs.180946similar to RIKEN cDNA 2900024C23na0.71490530.160.00160.47614510.120.0001
48670H15967Hs.79101cyclin G1CCNG10.79674490.500.06640.47619460.190.0001
24064T78664Hs.244318death-associated protein kinase 1DAPK10.56315920.130.00010.48034290.060.0001
149156R82466Hs.373980IQ motif containing GTPase activating protein 2IQGAP20.7438530.380.01770.48254790.180.0001
243344N48101NAESTEST0.52429410.330.00170.48473540.250.0001
132630R25980NAESTEST0.63263910.380.00980.48479920.180.0001
415268W92014Hs.429896alcohol dehydrogenase 6 (class V)ADH60.60174210.380.01730.48764060.180.0001
247977N77712NAESTEST0.71490170.600.03370.49072250.240.0001
5798666BM928536Hs.438599sperm specific antigen 2SSFA20.7011660.260.00430.49243740.140.0001
305944W20035NAHypothetical protein LOC339005LOC3390050.75919630.400.01780.49308710.120.0001
45991H08515Hs.432780hypothetical protein MGC29956MGC299560.61099630.110.00010.49361720.090.0001
26053R11947Hs.354740K large conductance Ca-activated channel, subfamily MKCNMA10.76072730.200.03450.4936940.100.0001
120089T95249Hs.80876flavin containing monooxygenase 3FMO30.62051220.280.00290.49423930.270.0001
36576R25303NAMGC10940MGC109400.65091530.140.00030.49731250.130.0001
5109720BI257812Hs.49349beta-site APP-cleaving enzyme 1BACE10.66309630.200.00060.49793040.080.0001
141235R66548Hs.413175collagen, type XVIII, alpha 1COL18A10.66967460.310.00870.49815650.120.0001
152861R50648Hs.267442hypothetical protein FLJ14103FLJ141030.68332140.210.00060.49863830.090.0001
Figure 2.

Figure 2.

Real-time PCR verification of genes in Table 2 . (A). Real-time PCR was performed on the aRNA amplified from 5 OPENING samples, 5 PRECLAMP samples, and 6–15 PHA samples for the indicated genes. (B) Real-time PCR was performed on the unamplified RNA isolated from 8 OPENING, 8 PRECLAMP and 8 PHA samples. In each case, data are presented as mean ± SD in arbitrary units relative to β-actin. Stats: ANOVA; *p < 0.05 vs OPENING, **p < 0.01 vs OPENING, ***p < 0.001 vs OPENING, δp < 0.05 vs PRECLAMP. NB: PRECLAMP data for SCAP1, PRKAB2, and HNMT are not presented due to technical irregularities with the assay for these samples.

Figure 2.

Figure 2.

Real-time PCR verification of genes in Table 2 . (A). Real-time PCR was performed on the aRNA amplified from 5 OPENING samples, 5 PRECLAMP samples, and 6–15 PHA samples for the indicated genes. (B) Real-time PCR was performed on the unamplified RNA isolated from 8 OPENING, 8 PRECLAMP and 8 PHA samples. In each case, data are presented as mean ± SD in arbitrary units relative to β-actin. Stats: ANOVA; *p < 0.05 vs OPENING, **p < 0.01 vs OPENING, ***p < 0.001 vs OPENING, δp < 0.05 vs PRECLAMP. NB: PRECLAMP data for SCAP1, PRKAB2, and HNMT are not presented due to technical irregularities with the assay for these samples.

Our microarray data are based on aRNA amplified from the RNA extracted from patient samples. Although we have found that the amplification step yields results very close to unamplified RNA (see Materials and Methods), we performed a separate series of real-time verifications for a total of nine genes (six predicted to increase, three predicted to decrease in response to transplantation injury), using the unamplified RNA originally extracted from liver tissues. As shown in Figure 2B, our predictions remain correct for both unamplified and amplified RNA. The data from the real-time PCR validation studies using both unamplified and amplified sample RNA confirm an impression from the microarray data: alterations of gene expression in PHA liver tissue are quantitatively different from, but qualitatively similar to, gene expression in PRECLAMP liver tissue. That is, a gene with markedly increased expression in PHA liver tissue tended to have increased expression in PRECLAMP tissue also, though to a lesser extent. We conclude that the addition of HIRI and preservation injury exaggerates a profile already established by surgical stress.

Altered gene expression in response to surgical stress and transplantation injury is similar to that found in response to systemic stress, but distinct from chronic liver disease

Having defined and verified the altered expression of a subset of genes with implications for acute liver injury, we were interested to determine whether this subset defined a common response to acute liver stress or reflected generic liver injury. To answer this question, we compared gene expression of the 25 genes in the subset to expression in another setting of acute liver stress (the response to the systemic stress of braindeath) and to several chronic liver diseases (PBC, HCV and HBV). We queried the microarray gene signatures from our UHN Liver Disease Microarray Project (24) for the expression of these 25 genes in these conditions. DD liver biopsies were taken as the first step of the organ retrieval process so as to minimize surgical stress per se. As shown in the unsupervised hierarchical cluster analysis presented in Figure 3A,B, there was a generally consistent change in the levels of expression of the 25 genes in all three of DD, PRECLAMP and PHA liver biopsies, with a different pattern of expression when compared to the levels of expression in chronic liver disease. Importantly, there are differences in gene expression between any individual chronic liver disease state and between DD and PHA acute liver stress also (data not shown). Similar clusters are observed when all of the genes identified as being altered over the course of the LDLT experiment (Tables 3 and 4) are used in a hierarchical cluster analysis (data not shown). Our data demonstrate that the genes most affected by the acute liver stress of LDLT define a subset of genes that are similarly affected in response to braindeath, but not in response to chronic liver disease. Thus, the subset of verified genes may have implications for acute liver injury as a whole, as may many of the additional genes presented in Tables 3 and 4.

Figure 3.

Figure 3.

Hierarchical cluster analysis comparing acute and chronic liver stress. (A) Liver biopsy samples were clustered using the 25 genes verified in the real-time PCR analysis, comparing the levels of expression of these 25 genes in 25 PHA samples, 20 OPENING samples, 10 PRECLAMP samples, 10 DD samples, 18 HBV samples, 48 HCV samples and 31 PBC samples. Each gene in this figure is compared to the mean level of expression in control liver tissue (OPENING samples); red = increased and green = decreased relative to expression in OPENING liver tissue. (B) Details of the cluster analysis. Note that there are two large clusters; one comprised samples from patients with chronic liver stress (HCV, HBV, PBC), and the other comprised samples from patients with acute liver stress (DD, LDLT). Details of the clustering algorithm are found in Materials and Methods.

Figure 3.

Figure 3.

Hierarchical cluster analysis comparing acute and chronic liver stress. (A) Liver biopsy samples were clustered using the 25 genes verified in the real-time PCR analysis, comparing the levels of expression of these 25 genes in 25 PHA samples, 20 OPENING samples, 10 PRECLAMP samples, 10 DD samples, 18 HBV samples, 48 HCV samples and 31 PBC samples. Each gene in this figure is compared to the mean level of expression in control liver tissue (OPENING samples); red = increased and green = decreased relative to expression in OPENING liver tissue. (B) Details of the cluster analysis. Note that there are two large clusters; one comprised samples from patients with chronic liver stress (HCV, HBV, PBC), and the other comprised samples from patients with acute liver stress (DD, LDLT). Details of the clustering algorithm are found in Materials and Methods.

Functional analysis of the verified genes

In order to gain insight into the molecular pathways targeted by transplantation injury, we performed a bioinformatic functional analysis of the genes altered after graft reperfusion. In deference to the ‘noise’ of microarray analyses and in an effort to apply as impartial an analysis as possible, we applied two separate approaches. First, we used a commercially available software package to interrogate the function of the 25 genes verified by quantitative PCR. PathwayAssist (Stratagene) links genes based on their published functions; however, as with other packages its ability to link genes is entirely dependent on those that are currently annotated. Of the 25 gene subset (15 increased and 10 decreased), 9 were functionally annotated within this package (VCAM1, MAP1LC3B, SDC4, CDKN1, MT1F, SCAP1, THBS1, TGM2, FOSB). As shown in Figure 4, these genes were linked by functions with clear relevance to liver injury: cellular remodeling, focal adhesion, apoptosis and regeneration. Second, we developed a method—MineGO—to interrogate the GO database (35) that allows us to compare the distribution—by function—of all genes found on the array to the distribution of genes identified in our array analysis. This method is similar to the EASE method developed by Hosack et al. (22) and allows us to identify which general functions are ‘enriched’ (see Materials and Methods) in our array analysis. Taking a p-value of <0.05, we found that for various GO levels our microarray results were enriched for genes with functions relevance to acute liver injury, such as inflammatory response, programed cell death, response to pathogens and healing (results for GO level 5 are presented in Table 5). As seen in Table 5, our method produces results very similar to the EASE analysis. In both cases we chose a GO level that describes reasonably specific functions. The identity of the genes in each functional category is found in Table 6. These data argue that both the 25 gene subset and the whole microarray data set have good biological plausibility.

Figure 4.

Functional analysis of the verified genes altered after reperfusion of the LDLT right lobe graft. Validated genes annotated in the PathwayAssist software package (VCAM1, MAP1LC3B, SDC4, CDKN1, MT1F, SCAP1, THBS1, TGM2 and FOSB) were found to be linked by a number of biological functions, as shown. Links to other genes are indicated by the arrows; genes closely linked to the verified genes are also presented.

Table 5. GO Database functional analysis of the entire microarray data set
GO biological process (level 5)MineGO Fisher testEASE scoreEASE Fisher test
  1. A comparison of the results from an interrogation of the GO database using both the method described in Materials and Methods (MineGO) and the EASE package. Level 5 GO functional categories describe reasonably specific functions and are presented in this table. EASE calculates both a probability score (the EASE score) and a Fisher exact test: both are presented.

carboxylic acid metabolism0.009340.00030.0001
lipid biosynthesis0.045340.01540.00405
coenzyme metabolism0.024490.120.0328
fatty acid metabolism0.033210.06410.0176
inflammatory response0.035160.08920.0274
induction of programmed cell death0.044520.1410.0414
response to bacteria0.000640.0020.0001
wound healing0.00001not foundnot found
protein localization0.00047not foundnot found
response to pathogenic bacteria0.000900.00090.0003
blood vessel development0.020170.08340.0118
response to pathogen0.00120not foundnot found
biogenic amine metabolism0.004470.00780.0007
amino acid derivative metabolism0.006280.01150.0012
amino acid derivative biosynthesis0.007930.02280.0014
cytolysis0.002570.00860.0003
Table 6. Gene identities in enriched GO functional categories
GO category level 5 (mineGO analysis)Gene Symbols 
  1. Gene symbols for the functional categories enriched in our interrogation of the GO database are presented—these categories correspond to those in Tables 3 and 4.

carboxylic acid metabolismAGPAT1GCLMME2PPP2CABAATGLYATPRKAB2
 CYP4A11ARG1ACADSBMGC35366DDCSCDCYP2J2
lipid biosynthesisAGPAT1CYB5R2PRKAB2HSD17B6SCDCMASPBX1
coenzyme metabolismPBEF1MLL3DLATGCLMGLYATPANK1 
fatty acid metabolismGLYATPRKAB2CYP4A11ACADSBSCDCYP2J2 
inflammatory responseIL10RBCCL2S100A12C5CXCL14ATRN 
induction of programmed cell deathCDKN1APRKCASTK17APPP2CADAPK1 
response to bacteriaCCL2S100A12C6C5C8B 
wound healingSERPINE1IL10RBFGGSERPINC1 
protein localizationARFGAP3RND3TPRAKAP12 
response to pathogenic bacteriaCCL2C6C5C8B 
blood vessel developmentMMP19PGFFGGANG 
response to pathogenCCL2C6C5C8B 
biogenic amine metabolismSRMAMD1TTRDDC 
amino acid derivative metabolismSRMAMD1TTRDDC 
amino acid derivative biosynthesisSRMAMD1DDC 
cytolysisC6C5C8B 

Discussion

In this study, we present gene expression profiling data which identifies the genes most influenced by the early (1–3 h) response to transplantation injury, and which illustrates a divergent liver response to acute and chronic stress. A subset of 25 genes was verified by real-time PCR. This 25 gene subset defines a molecular signature that is common to two forms of acute liver stress (braindeath, LDLT) but not seen in chronic liver stress (HCV, HBV, PBC). The functions of the 25 genes are linked functionally in pathways that are clearly relevant to acute liver stress, and the entire microarray data set is enriched by genes that play biologically relevant roles in acute stress. Thus, this study identifies both a molecular signature for acute liver stress and provides a robust data set for genes with relevance to the process.

Many of the genes verified in the 25 gene subset provide insight into the molecular pathogenesis of acute liver stress. PBEF1 is a novel inflammatory cytokine that is anti-apoptotic in neutrophils (36). PBEF is also up-regulated in activated lymphocytes (37), has been identified as a biomarker of acute lung injury (38) and may play an important role in glucose homeostasis (39). To our knowledge our work is the first to link this fascinating gene, whose functions have clear implications for organ-specific responses to acute stresses, with acute liver injury. SDC4, in combination with antithrombin, inhibits monocyte and lymphocyte chemotaxis in response to IL-8 and Rantes and disrupts endotoxin-induced adherence of neutrophils to human umbilical endothelial cells in vitro (40,41). The protein-crosslinking enzyme TGM2 is protective in experimental carbon tetrachloride (CCl4)-induced rodent acute liver injury, possibly by accelerating the phagocytosis of apoptotic cells (42,43). TGM2 protein levels are increased in the early stages of human hepatic fibrosis, and decreased in the later stages (42); this biphasic association may support the experimental suggestion of a protective effect, but also illustrates the fact that the protein serves multiple functions (44). Nicotinamide N-methyltransferase (NNMT), which catalyzes the N-methylation of nicotinamide and other pyridines, is one of only a few genes increased in the fulminant hepatitis induced in rodents by administration of heat-killed Propionibacterium acnes (45). Our work is the first to associate NNMT with acute liver injury in humans, though how increases in NNMT influence the immediate response to IRI is unclear. On the other hand, FosB is one component of the transcription factor AP-1, which has been extensively investigated in the response to inflammatory stimuli and is redox-sensitive (46,47). P21/Cip has been demonstrated previously to be increased in chronic hepatitis and has broad implications for the response to liver injury in general. The sensescence-related, anti-proliferative p21/Cip1 protein may play an important role in regulating hepatocyte cell cycle progression in liver regeneration, contributes to hepatocyte necrosis experimentally in the acute liver injury induced by CCl4 and is associated with human hepatitis (48,49). p21/Cip1 has also been demonstrated to be increased in experimental ischemia/reperfusion in vitro, though our work is the first to link p21 with this mechanism clinically (50). Taken as a whole, this work establishes an intriguing link between a defined subset of genes and the early inflammatory micro-environment associated with clinical transplantation injury.

In order to determine the specificity of this 25 gene subset, we compared the expression levels of these genes in two conditions of acute liver stress (LDLT, braindeath) and three forms of chronic liver stress (HCV, HBV, PBC). We have had excellent success in verifying our microarray data using real-time PCR, with verification rates exceeding 70%. With high quality gene expression data it is possible to compare multiple disease states and derive a set of genes whose expression varies more in one state than another. Interestingly, we found that the levels of these 25 genes were consistently altered in acute liver stress, but largely unaltered by chronic liver stress (Figure 3). These results argue for the specificity of our findings and suggest that the 25 genes constitute a molecular signature for acute liver stress. The fact that they are linked in biologically relevant pathways, and that our array data set is enriched for genes that play important roles in acute organ injury, argues that our study has identified a subset of genes with important functions in acute liver injury.

Although infiltrating immune cells and the consequent changes in cellular composition may influence the distinction between chronic and acute liver gene expression, changes in cellular composition were unlikely to play a large role in the gene expression determined over the course of the LDLT operation. There were small numbers of infiltrating inflammatory cells in the PHA liver biopsies (Figure 1), but the fact that gene expression changes began in response to surgical stress and were exaggerated following IRI suggests that the bulk of the gene expression effect was determined by cells already resident in the liver during the response to surgical manipulation. In addition, we required a stringent consistency in identifying genes of interest, to the extent that gene expression signals from inconsistently present inflammatory cells were ‘washed out.’ Taken together our data argue that acute liver stress is funneled along common effector mechanisms that result in a definable subset of gene expression changes.

Recent work has demonstrated a pivotal role for TLR4 in HIRI in a rodent model (90 min warm ischemia, 6 h reperfusion), and suggested that the IP-10 chemokine might be important for inflammatory cell recruitment (51). Two of the annotated genes in our microarray data set are linked at the TLR pathway: PI3R1—down, Jun—up. Relaxing our criteria to genes that are increased or decreased by 1.5-fold relative to opening, with p < 0.01, we found four genes involved in the TLR pathway (PI3R1, Jun, LBP and CCL3). These results are consistent with a role for TLR in clinical acute hepatic stress responses. We did not find increased IP-10 mRNA in acute hepatic stress, though it increases consistently in chronic HCV; we have confirmed these results by real-time PCR (data not shown). It is possible that a longer period of reperfusion is required before IP-10 is up-regulated in clinical transplantation injury.

In this study, we also present an automated method for examining the functions of genes identified in any microarray analysis. Our method in outline is quite similar to the EASE method developed by Hosack et al. (22), and both methods produce similar results (Table 5). Our results therefore act as an independent verification of the EASE method. The programing code for our method is available for download and manipulation at http://142.150.56.35/~LiverArrayProject/home.html. In addition, the genes identified in this paper are linked directly to the NCBI database on our server.

In summary, our data suggest that there may be a relatively small set of genes that participate in many forms of acute liver injury. These genes may define the acute hepatic ‘danger’ signal (1). Our results are biologically plausible, as demonstrated by our functional interrogation of this large data set, and by the identities of the genes verified by quantitative PCR. Further studies are warranted to determine the generalizability of our findings and explore their potential therapeutic implications, particularly in the setting of liver transplantation.

Acknowledgments

The authors are grateful to the surgeons of the Multi-Organ Transplant Program at the University of Toronto for their participation in this study, in particular Drs. Mark Cattral and Paul Greig. Jean Parodo provided invaluable assistance with tissue glutathione assays. The authors also thank Catalina Coltescu for her help with maintaining our HCV clinical database. This work was partially funded by an unrestricted educational grant provided by Fujisawa Canada, and by an educational grant from the Physicians Services Incorporated Foundation.

Ancillary