Heme oxygenase-1 (HO-1) has been suggested as a cytoprotective gene during liver transplantation. Inducibility of HO-1 is modulated by a (GT)n polymorphism and a single nucleotide polymorphism (SNP) A(-413)T in the promoter. Both a short (GT)n allele and the A-allele have been associated with increased HO-1 promoter activity. In 308 liver transplantations, we assessed donor HO-1 genotype and correlated this with outcome variables. For (GT)n genotype, livers were divided into two classes: short alleles (<25 repeats; class S) and long alleles (≥25 repeats; class L). In a subset, hepatic messenger ribonucleic acid (mRNA) expression was correlated with genotypes. Graft survival at 1 year was significantly better for A-allele genotype compared to TT-genotype (84% vs. 63%, p = 0.004). Graft loss due to primary dysfunction (PDF) occurred more frequently in TT-genotype compared to A-receivers (p = 0.03). Recipients of a liver with TT-genotype had significantly higher serum transaminases after transplantation and hepatic HO-1 mRNA levels were significantly lower compared to the A-allele livers (p = 0.03). No differences were found for any outcome variable between class S and LL-variant of the (GT)n polymorphism. Haplotype analysis confirmed dominance of the A(-413)T SNP over the (GT)n polymorphism. In conclusion, HO-1 genotype is associated with outcome after liver transplantation. These findings suggest that HO-1 mediates graft survival after liver transplantation.
Orthotopic liver transplantation (OLT) is the best available treatment for patients with end-stage liver failure. It is well recognized that, during the transplant procedure, livers are exposed to various stressful stimuli such as ischemia and reperfusion injury. Heme oxygenase-1 (HO-1) has been shown to provide cytoprotection during liver ischemia and reperfusion. Moreover, it has been suggested to have an immune modulating effect (1). In various experimental OLT models, upregulation of HO-1 has been shown to protect livers from I/R injury and to improve graft survival (2,3). HO-1 catalyzes the oxidative detoxification of excess heme resulting in equimolar amounts of free iron (Fe2+), biliverdin and carbon monoxide. All products formed in this process possess potential beneficial effects in the transplant setting. CO has vasodilatating effects, thereby maintaining microvascular hepatic blood flow (4,5). Biliverdin and the subsequently formed bilirubin possess potent antioxidant effects (6–9). Free iron is highly reactive by itself, however, cellular Fe2+ released via heme degradation upregulates the expression of the Fe2+ sequestrating protein ferritin as well as that of an Fe2+ pump, thereby limiting the amount of free iron and preventing the generation of reactive oxygen species (10–12).
We previously studied the endogenous regulation of HO-1 during human liver transplantation and showed a dual role for HO-1, with either cytoprotection or increased cytotoxicity, depending on the initial level of overexpression (13). However, none of the clinical variables analyzed in this study could explain the variation in initial expression of HO-1 in the donor livers. We therefore decided to study the impact of genetic differences on HO-1 expression and outcome after OLT.
Expression of the HO-1 gene is modulated by two functional polymorphisms in the promoter: a (GT)n polymorphism and a single nucleotide polymorphism (SNP) (14–19). (GT)n is the most frequent of the simple repeats scattered throughout the human genome and many of these exhibit a length polymorphism (20). Most of these variable sites are not expected to have any functional effect, because they are located in intragenic regions and introns. However, the HO-1 (GT)n repeat resides in a regulatory sequence and a short (GT)n allele has been associated with enhanced transcriptional activity of the gene (14,17–19). In kidney transplantation the influence of HO-1 (GT)n polymorphism has recently been studied by Baan et al. (21) and Exner et al. (22), who found a positive correlation between a short GT repeat and graft function and survival after transplantation. In addition to the (GT)n polymorphism, the A(-413)T SNP has been identified as a functionally relevant variation of the HO-1 gene (15,16). Using a transient transfection assay of HO-1 promoter luciferase genes in bovine aortic endothelial cells Ono et al. have shown that the A-allele of this SNP is associated with a higher promoter activity than the T-allele. Interestingly, the A(-413)T SNP appeared in vitro to be more important for HO-1 promoter activity than the (GT)n polymorphism (15,16). Only limited work has been conducted evaluating the A(-413)T SNP in clinical research (15).
Based on the accumulating evidence that HO-1 is an important enzyme influencing graft survival after transplantation, we hypothesized that these two functionally relevant polymorphisms in the HO-1 promoter are associated with outcome after OLT. Therefore, we analyzed the two functional HO-1 promoter polymorphisms in donor genomic DNA in relation to outcome after human liver transplantation. Furthermore, we studied the functional relevance of these polymorphisms by measuring hepatic messenger ribonucleic acid (mRNA) expression.
Patients and Methods
Between January 1996 and January 2005, a total number of 465 consecutive OLTs were performed at the University Medical Center Groningen. After exclusion of children (<18 years), 320 transplants in 282 adult patients were included in this study. Of 308 donors (96%), cryopreserved splenocytes were available for the HO-1 genotyping. Median follow-up time for this cohort was 4 years and 6 months (range 20–81 months).
ABO blood group-identical or compatible grafts from brain-death donors with normal or near normal liver function tests were used for all patients. A standardized technique was used for implantation, as has been described previously (23,24). During the study period, immunosuppressive protocols were based on tacrolimus or cyclosporine A, either with or without azathioprine and a rapid taper of steroids. Biopsy-proven acute rejection was treated when clinically indicated, with a bolus of methylprednisolone on three consecutive days. Steroid-resistant rejections were treated either by conversion to tacrolimus in patients on cyclosporine A, or by giving five doses of antithymocyte globulin (4 mg/kg i.v.) on alternating days. Doppler ultrasound was performed routinely at postoperative days 1, 3 and 7 and on demand to rule out vascular or biliary complications or parenchymal lesions. Cholangiography via a biliary drain was routinely performed between postoperative day 10–14 and later on demand (i.e. for rising cholestatic parameters or dilatation of bile ducts on ultrasound). Tissue and data collection was performed according to the guidelines of the medical ethical committee of our institution and the Dutch Federation of Scientific Societies.
HO-1 genotype assessment
Genomic DNA was isolated from donor splenocytes using a commercial kit (Gentra Systems, Minneapolis, MN). The 5′-flanking region of the HO-1 gene containing the (GT)n polymorphism was amplified by polymerase chain reaction (PCR) using as forward primer 5′-CAG CTT TCT GGA ACC TTC TGG-3′ (sense), carrying a 6-FAM fluorescent label (Sigma, Malden, The Netherlands), and as reversed primer 5′-GAA ACA AAG TCT GGC CAT AG GAC-3′ (antisense). PCR and genotyping procedures were similar as described earlier (25). Sequence analysis of the amplification products of individuals homozygous for the 222 and 229 base-pairs alleles showed correspondence with GT numbers 26 and 29, respectively (results not shown). Allelic repeats were divided into two subclasses using a similar classification based on transfection studies as described previously (26). A short allele, with less than 25 GT repeats, was designated as class S, and long allele with 25 or more GT repeats (amplicons of 220 base-pairs and more), as class L (26). Recipients of class S allele liver transplants (homozygous SS and heterozygous SL) were compared with recipients of non-class S allele transplants (LL).
The SNP A(-413)T (rs2071746) was analyzed using the ABI7900HT TaqMan system (Applied Biosystems, Foster City, CA) with a probe/primer assay hCV15869717, developed by and purchased from Applied Biosytems (Assay-on-Demand). Recipients of at least one A-allele liver transplants (homozygous AA and heterozygous AT) were compared with recipients of heterozygous T-allele recipients (TT).
Collection of liver biopsies, RNA isolation and reverse-transcriptase polymerase chain reaction
In a subset of 38 patients we collected liver biopsies at the end of cold storage to compare HO-1 mRNA expression in the various genotype groups. RNA isolation and cDNA synthesis were performed as described before (13). cDNA levels of HO-1 and 18S were measured by real time polymerase chain reaction (RT-PCR) using the ABI PRISM 7900 HT Sequence detector (Applied Biosystems). Nucleotide sequences of Primers (Invitrogen, Paisley, Scotland) and Probes (Eurogentec, Herstal, Belgium) were designed using Primes Express software (PE Applied Biosystems). Probes were 5′ labeled by a 6-carboxy-fluoresceine (FAM) reporter and 3′ labeled with a 6-carboxy-tetra-methyl-rhodamin (TAMRA) quencher. RT-PCR data were analyzed using the comparative cycle threshold (CT) method. Briefly, the difference in cycle times, ΔCT, was determined as the difference between the tested gene and the reference RNA, 18S. We then obtained ΔΔCT by finding the difference compared to a control group of liver biopsies from patients undergoing an hemihepatectomy for colorectal metastasis. The fold induction (FI) was calculated as 2−ΔΔCT.
Clinical outcome parameters
Outcome parameters included serum concentrations of aspartate aminotransferase (AST) and alanine aminotransferase (ALT), as marker for ischemia/reperfusion injury after OLT, graft survival, incidence of acute rejection and causes of graft loss. Recipient data were obtained from a prospectively collected database. Donor data were extracted from the national and hospital's donor databases.
Acute rejection was suspected on the basis of daily liver function tests, fever and deterioration of the clinical condition and proven by needle biopsy of the liver. The degree of acute rejection was histologically graded according to the Banff classification (27). Only rejections within the first 3 months with grade II and III, or grade I with a clinical indication for treatment, were considered in this study. As individual causes of graft loss, five different etiologies were identified: (1) primary dysfunction (PDF), defined as either primary nonfunction (PNF) or graft loss due to initial poor function (IPF). PNF was defined as nonlife sustaining function of the liver requiring retransplantation or leading to death within 7 days after OLT. IPF was defined as early graft dysfunction characterized by serum AST levels >2000 U/L on any day between postoperative day 2–7 and a prothrombine time (PT) >16 sec (modified according to Ploeg et al. ), which was not explained by biliary or vascular complications; (2) hepatic artery thrombosis, which was always confirmed by doppler ultrasonography and or angiography; (3) nonanastomotic biliary strictures, as detected on imaging studies of the biliary tree and in the absence of arterial complications (29); (4) recurrent disease; and (5) nongraft related graft loss, including extrahepatic conditions that contributed to the loss of the donor liver, such as postoperative sepsis and multiorgan failure.
All data are reported as median and interquartile ranges (IQR) or number with percentage.
Collection of laboratory values from the central hospital database was conducted as follows. For every postoperative biochemical variable of each patient a time curve was constructed before further analysis. In case multiple measurements of a parameter were performed on 1 day, these values were averaged to a single value before further analysis. Likewise, in case laboratory values were missing on certain days, these values were interpolated. Extrapolations were not performed.
Groups were compared with the chi-square test or Mann-Whitney U-test, where appropriate. Biochemical variables were compared using the daily values, but also the total course during the first two postoperative weeks was compared by calculating the area under the curve (AUC), using the trapezium rule. Graft survival curves were calculated according to the Kaplan–Meier method and compared using the log-rank test. A two-tailed p-value of <0.05 was considered statistically significant. Statistical analyses were performed using SPSS version 12.0.2 (SPSS, Inc., Chicago, IL).
To study linkage disequilibrium (LD) between the two polymorphisms in the promoter of the HO-1 gene, the frequencies of the combined genotypes of the (GT)n polymorphism and A(-413)T were counted. LD is the occurrence of two or more polymorphism variants together on the chromosome more often than could be expected based on recombination possibilities, most likely due to their close locations, but it may also arise when the combination confers a selective advantage. A haplotype is a vector of polymorphisms. Haplotype frequencies were estimated from the genotype counts using the expectation-maximization algorithm (own software). LD is then determined from these haplotype frequencies by means of D′ and the correlation coefficient r2, which both range from 0 to 1 with 1 implying the strongest possible LD and 0 as no LD.
Distribution of HO-1 genotypes in the donor population
The allelic distribution of the (GT)n polymorphism in the HO-1 promoter of liver donors is given in Figure 1. The distribution of (GT)n is bimodal, with a peak at 22 repeats (22%) and the other at 29 repeats (45%). Forty-two (14%) patients received a liver from a donor homozygous for class S allele, 130 (42.2%) from a heterozygote (SL) and 136 (44.2%) from a donor homozygous for the class L allele.
With respect to the T(-413)A SNP, the distribution of the genotypes was as follows: 92 (30%) patients received a liver from an AA genotype donor, 153 (50%) from an AT genotype donor, and 61 (20%) from a TT genotype donor, (in two samples genotyping of A(-413)T failed).
There were no significant differences in donor and recipient characteristics of patients receiving a liver from a donor with a class S allele (SS or SL) or from a donor without a class S allele (LL). Also no significant differences were found between the group of patients who received a liver from a donor with an A-allele (AA or AT) and the group of patients who received a liver without an A-allele (TT genotype) (Table 1).
Table 1. Comparison of donor and recipient characteristics in relation to donor HO-1 genotype
A(-413) T SNP1
S-receiver (n = 172)
LL (n = 136)
A-receiver (n = 245)
TT (n = 61)
1SNP analysis failed for two donors.
2At time of donor procedure.
Continuous variables are presented as median and interquartile range, categorical variables as numbers with percentage.
AST = aspartate amino transferase; ALT = alanine amino transferase; gGT = gamma glutamyl transferase; AP = alkaline phosphatase; CIT = cold ischemia time, time between start cold perfusion in the donor and end of cold preservation of the liver graft; WIT = warm ischemia time, time between the end of cold ischemic preservation of the liver and portal vein perfusion in the recipient; LOS = length of stay; ICU = intensive care unit.
Donor age (years)
Total bilirubin (umol/L)
Dopamine use (n = 177)
Cause of death
Cerebral vascular accident
Recipient age (years)
LOS at ICU
Are the two HO-1 polymorphisms in linkage disequlibrium?
Haplotype frequencies were estimated from data in Table 2 and are presented in Table 3. The two most prevalent haplotypes were A(-413)_(GT)29 and (-413) T_(GT)22 indicating that the ‘favorable’ A-allele is in LD with the ‘unfavorable’ class L genotype. The LD measures D′ and r2 were 0.87 and 0.50, respectively, indicating strong LD between the two promoter polymorphisms. The class L genotype cosegregates at -413 more often than expected with the A-allele, the same holds for the combination of the class S genotype and T-allele at -413.
Table 2. Number of A(-413)T genotypes in each genotype of the (GT)n polymorphism
(GT)n repeat length polymorphism
Combinations occurring less then four times are not shown in this table.
Table 3. Estimated haplotype frequencies from Table 2 with the EM algorithm
(GT)n number of repeats repeats
Estimated haplotype frequency (%)
Haplotypes with a frequency of less than 1% are not shown.
Is there an association between HO-1 genotypes and mRNA expression?
In a subgroup of 38 livers, material was available to measure hepatic HO-1 mRNA expression. The FI of the HO-1 mRNA in liver biopsies, retrieved at the end of the cold storage period, was significantly higher in A-receiver genotype livers compared to the TT-genotype (p = 0.03) (Figure 2). No difference in mRNA expression was found for the recipients of class S or LL- livers.
Although these findings provide support for a functional relevance of the A(-413)T SNP and not for the (GT)n polymorphism, they do not demonstrate the dominance of one of these polymorphisms. Therefore we next studied HO-1 mRNA expression in the various haplotype combinations (Figure 3). Within the group of non-class S allele transplants (LL), HO-1 mRNA was higher in livers with an A allele (haplotype A-L, A-L) compared to livers with a T-allele (haplotype T-L, T-L). A similar comparison of different haplotypes within the group of class S livers was not possible due to the low frequency of the S-A, S-A haplotype (Table 3).
Are HO-1 polymorphisms associated with outcome after liver transplantation?
Survival: In the entire cohort of 308 transplants, overall actuarial graft survival rate at 1 and 5 years was 80% and 71%, respectively. Graft survival rates were significantly better in recipients of livers with at least one A-allele, compared to recipients of a TT genotype liver; log-rank p = 0.004 (Figure 4). In addition, within the group of livers with at least one A-allele, there were no differences between AA and AT genotypes. No differences were found between recipients of class S or LL- livers.
Ischemia/reperfusion injury: Postoperative serum levels of AST and ALT as a marker of ischemia/reperfusion injury, are presented in Figure 5 A and B. Recipients of a liver with TT genotype had significantly higher serum transaminase levels, as expressed by the AUC for the first 2 weeks (AST, p = 0.01; ALT, p = 0.009). There were no significant differences in serum AST or ALT in recipients of a class S liver, compared to recipients of a non-class S liver (LL).
Acute rejection: The overall incidence of clinically relevant acute rejection within the first 3 months after OLT was 34%. There were no statistically significant differences in the incidence of acute rejection between any of the genotypes. Moreover, no differences were found in the severity of rejection among the different genotype groups (Table 4).
Table 4. Incidence of acute rejection within the first 3 months after OLT in relation to donor HO-1 genotype
S-receiver (n = 172)
LL (n = 136)
A-receiver (n = 245)
TT (n = 61)
1SNP analysis failed for two donors.
2Grades of rejection according to the BANFF classification. For three patients no histological data were available.
Categorical variables as numbers with percentage.
Causes of graft loss: The number of grafts lost in patients receiving a liver with a S-allele was 50 (29%) and the number of grafts lost in patients receiving a liver with LL genotype was 37 (27%). The number of grafts lost in patients receiving a liver with an A-allele was 62 (25%) and the number of grafts lost in patients receiving a liver with TT genotype was 25 (41%) (p = 0.004). To find an explanation for the observed differences in overall graft survival in relation to the A(-413)T SNP, we next examined the individual causes of graft loss (Table 5). Primary graft dysfunction was a significantly more frequent cause of graft loss in the group of TT-genotype livers (10%) compared to livers with an A-allele (2%); odds ratio 3.73 (95% Confidence interval 1.02–13.60; p = 0.03). For the other most common causes of graft loss, including hepatic artery thrombosis, nonanastomotic biliary strictures, recurrent disease and nongraft related causes, no significant differences were found in the distribution among the different genotypes (Table 5).
Table 5. Causes of graft loss grouped by donor HO-1 genotype
S-receiver (n = 172)
LL (n = 136)
A-receiver (n = 245)
TT (n = 61)
1SNP analysis failed for two donors.
Hepatic artery thrombosis
Nonanastomotic bilary strictures
Not graft related
In this study, we have examined the relationship between two functionally relevant polymorphisms in the promoter of the HO-1 gene in the donor and postoperative outcome in a large cohort of 308 liver transplant recipients. There are three novel findings in this study. Firstly we observed significantly worse outcome in patients receiving a liver from a TT genotype donor, compared to recipients from donors with at least one A allele. Secondly, we have shown that the A(-413)T SNP and the (GT)n polymorphism are in LD with each other in this predominantly Caucasian population. Finally, we have shown, for the fist time in a human population, the differences in functional relevance of these two HO-1 promoter polymorphisms. Our association study of the various haplotypes and actual HO-1 mRNA expression suggests that the A(-413)T SNP is of greater functional relevance than the (GT)n polymorphism. No differences were found in any outcome parameter between the class S and the LL-receivers of the (GT)n polymorphism. The power of this study with an overall sample size of 308 subjects was greater than 80% to detect a difference of 13% in graft survival at the statistical significant lever of 5%.
An association with functional polymorphisms of the HO-1 gene and clinical outcome parameters has also been found in other pathological conditions, such as pulmonary emphysema, certain cardiovascular diseases and malignancies (14,19,30–35). With respect to transplantation, two groups have previously reported an association between HO-1 polymorphism in the donor and outcome after kidney transplantation (21,22). Baan et al. and Exner et al. have shown a positive correlation between the presence of a short (GT)n allele in the HO-1 promoter and a favorable outcome after kidney transplantation (21,22). Although our data and the two studies in kidney transplant recipients all point toward a critical role for HO-1 in maintaining graft function after solid organ transplantation, in detail the studies are different. The two studies in kidney transplantation revealed an association between the (GT)n polymorphism and outcome after transplantation, whereas we found an association with the A(-413)T SNP. Unfortunately, this SNP was not tested in the two previous studies in kidney transplant recipients. Moreover, a third large genetic association study between the (GT)n polymorphism and outcome after kidney transplantation did not provide evidence for a protective effect of class S alleles on kidney graft survival (36). The LD between the short (GT)n variant and the T-allele at -413 in the current study, in combination with the known dominant effect of the A(-413)T SNP in relation to the (GT)n polymorphism (16), could possibly explain the inconsistent results of studies in kidney transplant recipients focusing on the (GT)n polymorphism only. It could well be that HO-1 expression has actually been lower in kidney grafts with a short (GT)n allele. Unfortunately, tissue levels of HO-1 mRNA, as a marker of actual HO-1 gene expression, were not measured in the three studies in kidney transplantation.
In our study population, which mainly consisted of Caucasians, we found the A(-413)T SNP and the (GT)n polymorphism within the promoter of the HO-1 gene to be in LD. The two most frequent haplotypes were the A-allele at -413 in combination with a long (29) (GT)n allele (43.2%) and the T-allele at -413 in combination with a short (22) (GT)n allele (20.3%). This finding is in accordance with results from previous studies in Japanese populations (15,16). Theoretically, these combinations are counterproductive, as the A-allele at −413 and a short (GT)n allele are both associated with enhanced expression of HO-1, whereas the T-allele and a long (GT)n allele are associated with reduced HO-1 expression. Our data, however, consistently point toward a dominant effect of the A(−413)T SNP over the (GT)n polymorphism. Not only clinical outcome parameters but also hepatic HO-1 mRNA correlated with the A(−413)T SNP, but not with the (GT)n polymorphism. To our knowledge, this is the first study in humans suggesting an association between mRNA expression and the various HO-1 haplotypes. Similar observations have been made by Ono et al. who have studied the functional role of the A(−413)T SNP and the (GT)n polymorphism in an in vitro system of bovine aortic endothelial cells, using a luciferase reporter assay (15). These investigators suggested that, with respect to HO-1 promoter activity, the A(−413)T SNP is dominant over the (GT)n polymorphism. In a previous study, we have shown the HO-1 mRNA expression correlates well with protein expression in human livers (13).
The exact mechanisms explaining the clinical observations in this study are incompletely understood. Experimental studies have previously shown that upregulation of HO-1 protects liver grafts against ischemia/reperfusion injury and improves graft survival (2,3,37–39). Especially, steatotic livers, which are highly sensitive to ischemic injury, seem effectively protected against this type of injury by induction of HO-1 (37,40). The inferior outcomes of livers with the unfavorable TT genotype (associated with a reduced HO-1 promoter activity) in the current study supports these previous findings. The effect of HO-1 genotype on graft survival could, at least partly, be explained by a higher incidence of PDF in livers with a TT-genotype, compared to livers with an A-allele (p = 0.03). However, when the Kaplan–Meier curves are carefully observed, the lines started to separate from day 25 and were further divergent later after transplantation. In accordance with this, the differences in serum transaminases became more pronounced in the second week after transplantation. These observations suggest that the observed differences in graft survival are not only explained by differences in ischemia/reperfusion injury, but also result from other mechanisms. In fact, the absolute number of grafts lost due to PDF is relatively small, again suggesting that other factors have also contributed to the observed differences in graft survival. Apparently, the impact HO-1 is not limited to the early postoperative period. We speculate that other, possibly immune-mediated processes, could explain the more late effects of HO-1 on graft survival. Several studies have shown that HO-1 is a key enzyme in certain immune processes. Nevertheless, we observed no differences in the incidence or severity of acute rejection between the various genotypes. However, it would be of interest to study HO-1 mRNA expression in donor livers more long-term after transplantation and to see if differences in HO-1 expression persist. Unfortunately, we had no access to repeated biopsies during long-term follow-up after OLT. More studies on the mechanisms underlying the more long-term effects of HO-1 on graft survival will be needed.
In conclusion, in this large series of 308 liver transplant recipients, we found an association between donor HO-1 genotype and outcome after liver transplantation. Livers with at least one A-allele of the A(−413)T SNP had significantly better graft survival rate and a lower rate of PDF than livers with the TT genotype. In addition, our data indicate a functional dominance of the A(−413)T SNP over the (GT)n polymorphism. These data suggest that HO-1 is critically involved in maintaining graft function during and after liver transplantation.
We are grateful to Marcel Mulder for the HO-1 genotyping and Danielle Nijkamp for assistance in donor data collection. This study was funded by grants from the Dutch Federation of Scientific Research (ZonMW 920–03-309 to CIB and 907–00-043 to RJP).