Association of the haem oxygenase-1 gene with inflammatory bowel disease

a CHILD Research Center, Department of Pediatrics, The Pennsylvania State University, College of Medicine, Hershey, PA, United States of America b Department of Public Health Science, The Pennsylvania State University, College of Medicine, Hershey, PA, United States of America c Department of Surgery, The Pennsylvania State University, College of Medicine, Hershey, PA United States of America d Department of Obstetrics and Gynecology, The Pennsylvania State University, College of Medicine, Hershey, PA, United States of America e Department of Radiology, Perelman School of Medicine, University of Pennsylvania, Philadelphia, PA, United States of America.


Introduction
Haem oxygenase 1 (HO-1) catalyses the degradation of haem to biliverdin and carbon monoxide.HO-1 plays an important role in regulating intestinal homeostasis through anti-apoptosis and angiogenesis in cancer [1].HO-1 also has anti-inflammatory properties and thus it has been suggested that it may have a role in the recovery from inflammatory bowel disease (IBD).However, in previous genetic association studies, HO-1 single nucleotide poly-morphisms (SNPs) have not been shown to associate with adult IBD [2,3] and paediatric Crohn's disease [2] based on the (GT)n dinucleotide repeat within the HO-1 promoter region, or adult IBD based on the SNP rs2071746 upstream of the HO-1 gene [4].Furthermore, in our case-control study (submitted for publication) we did not observe association of the HO-1 rs2071746 with paediatric IBD, which is consistent with the available literature.The HO-1 gene is known to be regulated by the interleukin-10 / signal transducer and activator of transcription 3 (IL-10/STAT3) signalling pathway [5, 6,].Of interest, IL-10 and STAT3 SNPs have been shown to associate with IBD [7][8][9][10].Recently, mutations in IL-10 and IL-10 receptors (IL-10RA and IL-10RB) were shown to be associated with paediatric IBD [7,[11][12][13][14], and our recent case-control study (submitted) showed association of IL-10 SNP rs3024496 with paediatric IBD (p = 0.022), as well as an epistatic interaction (dominant-additive p = 0.0018) of IL-10RB (rs2834167) and HO-1 (rs2071746).In the present family based case-trio study, we hypothesised that HO-1 SNPs would be associated with paediatric IBD.We also studied HO-1 gene expression in B lymphocyte cell lines from paediatric IBD patients carrying IBD-associated gene variants and their response to the inflammatory inducers, lipopolysaccharide (LPS) and tumour necrosis factor alpha (TNFα), as well as HO-1 gene expression in diseased and nondiseased tissues from adult IBD patients.

Study samples
For the case-trio study, 52 trio families (51 triad and 1 dyad families) were studied.Genomic DNA samples were obtained from the Crohn's and Colitis Foundation of America (CCFA) DNA Databank.CCFA genomic DNAs were collected from samples originating from: the University of North Carolina at Chapel Hill; University of Chicago, IL; Cedars-Sinai Hospital, CA; Massachusetts General Hospital; University of Pittsburgh, PA; and Mount Sinai Hos-pital, NY.Informed written consent was obtained from parents or guardians of the patients.The genomic DNA samples were stored at ˗80 o C until use.The study protocol was approved by the ethics committee of Penn State University College of Medicine Institutional Review Board (IRB HY98-057) [15].

Diseased tissues from adult IBD patients
Intestinal tissues were obtained from 17 Crohn's disease and 10 ulcerative colitis patients undergoing surgery at the Penn State Hershey Medical Center.All use of human tissues was approved by the Human Subjects Protection Office of The Pennsylvania State University College of Medicine, and was undertaken with the understanding and written consent of each subject.Macroscopically normal areas of intestine and areas of intestine with obvious disease were classified by a pathologist [15].Nondiseased tissue samples were excised from noninflamed, macroscopically normal tissue that was adjacent to, but a minimum of 5 cm away from, the diseased tissue.The intestinal tissues were immediately submerged in RNAlater Solution (Ambon, CA, USA) and stored at 4°C overnight.Tissues were stored frozen at ˗70°C until total RNA extraction.

SNP selection criteria for IL-10, IL-10RA, IL-10RB, STAT3 and HO-1
Seven SNPs from these five genes were studied, including: rs1800872 (C-592>A), rs3024498 and rs3024496 from IL-10 [16]; rs3135932 from IL-10RA [11]; rs2834167 from IL-10RB [11,16]; rs744166 from STAT3; and rs2071746 from HO-1 [4,17].The criteria for SNP selection and a summary of these SNPs, including genetic variation, chromosomal position, gene location, and disease implication were described in our previous paper [18].Briefly, the criteria for SNP selection were based on: (a) the potential relevance of these SNPs in the function and the regulation of genes, which have been associated with IBD and other diseases, and/or have a known functional role in inflammatory processes; (b) the gene location, either within the coding region that changes the encoded amino acid, or at 5′ upstream or 3′ untranslated region (UTR) potentially affecting RNA transcription, RNA stability, or protein translation; and (c) being polymorphic in the study samples as tested in our preliminary study and having minor allele frequency (MAF) information in existing databases.

B cell line culture and treatment with LPS and TNFα
B cell lines were established from blood samples from IBD patients by means of infection with Epstein-Barr virus (EBV) [15].Informed parental or guardian written consent documentation was obtained for all patient samples received from the Pennsylvania State University IBD Biobank and the Hershey Medical Centre.Cell lines were cultured in RPMI-1640 medium supplemented with 10% foetal bovine serum and incubated at 37°C in an atmosphere of 5% CO 2 [15].The cells were treated with LPS (1 μg/μl) and TNFα (0.1 μg/μl) for 24 hours.The cells were collected and cell pellets were stored at ˗80 o C until used for RNA isolation.All the experiments were repeated three times.

Genotype analysis
The genotype of all seven SNPs was determined with polymerase chain-reaction (PCR)-based RFLP/cRFLP as described previously (RFLP: restriction fragment length polymorphism) [19].The methods for PCR amplification, gel electrophoresis, and genotype score were as previously described [18].In brief, 100 ng DNAs were used for PCR in a 30 μl reaction volume.The PCR cycling profile was 95 o C for 2 min, 5 cycles at 95 o C for 30 sec, 50 o C for 1 min, and 72 o C for 1 min, then 30 cycles at 95 o C for 30 sec, 58 o C for 1 min, and 72 o C for 1 min, followed by a final extension step at 72 o C for 4 min.PCR products were digested with an appropriate restriction enzyme, separated with polyacrylamide gel electrophoresis (8%), and the genotypes scored according to the gel pattern of digested PCR products.

Case-trio analysis
Suppose the case-trio study involves n triad families (each containing one affected offspring and two parents, the parents either affected or unaffected), m dyad families (each containing one affected offspring and one parent, either father or mother, with the parent affected or unaffected), and s affected singleton monads (each containing only one affected offspring).At each SNP location, suppose there are two alleles A and a, with allele frequencies p and q, respectively.Let be the relative risk parameters, R = p 2 ψ 2 + 2pqψ 1 + q 2 .For the triad families, table 1 summarises the conditional probabilities of parental genotype and offspring genotype given that the offspring is affected.The likelihood function corresponding to the triad families is given by: logL Triad = [ 4n 1 + 3n 2 + 3n 3 + 2n 4 + 2n 5 + 2n 6 + 2n 7 + n 8 + n 9 ] logp + [ n 2 + n 3 + 2n 4 + 2n 5 + 2n 6 + 2n 7 + 3n 8 + 3n 9 + 4n 10] logq For the dyad families, table 2 summarises the conditional probabilities of parental genotype and offspring genotype given that the offspring is affected.The likelihood function corresponding to the dyad families is given by: logL For the singleton monads, suppose the number of genotype AA, Aa, aa are s 1 , s 2 , s 3 , respectively.The conditional probabilities of the genotypes given IBD affection are Original article prob(AA|disease) = p 2 ψ 2 / R, prob(Aa|disease) = 2pqψ 1 / R, and prob(aa|disease) = q 2 / R.
The likelihood function corresponding to the singleton monads is given by: The final likelihood function is the summation of two parts corresponding to the triad families and the dyad families, respectively: logL = logL Triad + logL dyad + logL monad (4) Using the likelihood ratio test, we can test the association between the genotype at each SNP location and the risk of IBD, with the following 5 types of hypotheses tested against the null hypothesis H 0 : ψ 1 = 1 and ψ 2 = 1: 1. Unrestricted model (no restriction on the risk parameters ψ 1 and ψ 2 )

Reverse transciptase PCR
Total RNA was isolated from B cells and intestine tissues with a Qiagen RNA isolation Mini kit.First strand copy DNA was synthesised with Invitrogen Superscript III Kit, according to the manufacturer's manual.The reverse transcriptase (RT)-PCR primers for HO-1 are given in table 3. PCR amplification was carried out at 95 o C for 2 min, 30 cycles at 95 o C for 30 sec, 58 o C for 1 min, and 72 o C for 1 min, followed by a final extension step at 72 o C for 4 min.PCR products were separated by 1.5% agarose gel electrophoresis and photographed under ultraviolet light.The glyceraldehyde 3-phosphate dehydrogenase (GAPDH) gene was used as a reference gene, and the gene relative mRNA expression level of the gene under study was noted as the ratio of the RT-PCR products of each target gene to GAPDH [20].
Table 1: Conditional probabilities of parental genotype and offspring genotype given that offspring is affected (for the triad families).

Parental genotype
Offspring genotype Conditional probability Number of obs.
Table 2: Conditional probabilities of parental genotype and the offspring genotype given that the offspring is affected (for the dyad families).

Parental genotype Offspring genotype Conditional probability Number of obs.
AA

Gel image analysis with Quantity One
Digital gel images were converted to TIF format and then analysed using Bio-Rad Quantity One software v4.5.0.The trace quantity was normalised against the reference gene, GAPDH.The ratio of HO-1 to GAPDH was used as relative gene expression level of HO-1 for the study.

Statistical analysis
Pearson's chi-squared tests with two degrees of freedom for allelic association were performed using Haploview software to analyse genetic patterns of linkage disequilibrium.We also calculated the genotype-based odds ratio (OR) using Fisher's contingency tables and tested association similarly to analyse genetic association.The difference was considered as significant when p <0.05.
We used a likelihood approach to analyse the genetic association with paediatric IBD using data from a case-trio study [21].As shown in table 4, HO-1 rs2071746 was highly associated with paediatric IBD in different models: p = 0.0009 in an unrestricted model, p = 0.0005 in a recessive model, p = 0.0010 in a multiplicative model, and p = 0.0002 in an additive model.IL10 rs1800872 was also found to be associated with paediatric IBD (p = 0.0277 in a dominant model).
The other genes, IL10-RA, IL-10RB, and STAT3, did not show a significant association with paediatric IBD, although we observed a likelihood ratio of 3.57 in an unrestricted model and 3.21 in a dominant model for IL-10RB, and a likelihood ratio of 3.82 in an unrestricted model, a likelihood ratio of 3.77 in a multiplicative model and a likelihood ratio of 3.70 in an additive model for STAT3.A full list of the maximum likelihood estimations of allele frequencies and risk parameters, and likelihood ratio test statistics and p-values for all models 1-5 are shown in table 4.

HO-1 gene expression in B cell lines from paediatric IBD patients
We aimed to study the potential function of disease-associated alleles in HO-1 gene expression and their effect on cellular response to inflammation stimulated by LPS or TNFα compared with untreated cells.Four paediatric IBD B cell lines (genotypes of the five studied genes indicated in figure 1) were treated with LPS (1 μg/μl) or TNFα (0.1 μg/μl).The results are shown in fig 1.

Effect of HO-1 rs207146 on HO-1 gene expression
Of the four cell lines, only the 777-27-1 cell line carried a homologous wild type HO-1 genotype, the other three cell lines were heterozygous for rs207146.Compared with the HO-1 expression in control cells, the relative HO-1 expression varied among cell lines, with 777-27-1 expression be- ing at a middle level, much lower than the 888-58-1 cell line.Clearly the HO-1 rs2071746 has no significant effect on HO-1 gene expression in B cells.

IL10 and IL10RB variants and HO-1 gene expression
The highest HO-1 gene expression level was observed for the cell line 888-58-1, the only cell line without any mutation in IL-10 and IL-10RB (both homozygous for the wild type allele).In contrast, the cell lines 95-09-1 (heterozygous at IL10 rs1800872, homozygous at both IL10 rs3024496 and IL10RB rs2834167; 888-118-1 (heterozygous at both IL10 rs1800872 and IL10RB rs2834167); and 777-27-1 (heterozygous at both IL10 rs3024496 and IL10RB rs2834167) exhibited much lower HO-1 expression than the cell line 888-58-1 (fig 1).These results support our speculation that IL10 and IL10RB may have a functional role in the regulation of HO-1 gene expression, although the number of the cell lines analysed is small to reach a firm conclusion.

Response of mutated alleles in IL-10, IL-10RB and HO-1 to LPS and TNFα stimulation
To study the regulation of HO-1 expression in response to inflammation via the IL10/STAT3 pathway, B cells were treated with LPS and TNFα.LPS can induce inflammation, and TNFα can mediate inflammation via its primary role in the regulation of immune cells.
Overall changes in HO-1 expression level in response to LPS and TNFα were not significant (fig.1).A trend of an increase of HO-1 gene expression was observed in the 888-58-1 B cell line that has no mutated alleles in IL-10 and IL-10RB; however in the other three cell lines with mutations in IL-10 and/or IL-10RB, HO-1 expression was reduced or showed no clear change in the presence of LPS  and TNFα.Although changes in the expression of the wild type for HO-1 and IL-10 cell line were small, the cell lines with mutation in IL-10 and IL-10RB had a negative or no effect on HO-1 expression in response to inflammation induced by LPS and TNFα.

HO-1 expression in diseased intestinal tissues from adult IBD patients
We further studied HO-1 expression and its association with pro-or anti-inflammatory activity in IBD affected intestinal tissues.As Crohn's disease can occur in both the small intestine and large intestine, but ulcerative colitis affects only the large intestine, the tissues selected from the 27 adult IBD patients included three IBD subtypes: 17 Crohn's disease tissues of which 12 were located in the small intestine and 5 in the large intestine, and 10 ulcerative colitis tissues located in the large intestine.We studied HO-1 expression and its association with inflammatory activity by comparing diseased tissues to nondiseased (noninflamed) surgical tissue from the same patients.To examine differences in gene expression and inflammation between Crohn's disease and ulcerative colitis we compared 12 Crohn's disease tissues from both the small and large intestine with 10 ulcerative colitis tissues from the large intestine.To examine gene expression between disease locations, we compared 12 Crohn's disease tissues from the small intestine with 5 Crohn's disease tissues from the large intestine.The results are shown in fig. 2.
The HO-1 expression level varied greatly among the 27 IBD patients.The variation may be due to many different factors, such as disease severity, genetic variation, clinical treatment, drug therapy, etc.To study the anti-inflammatory function of the IL-10 pathway via HO regulation, we compared the HO-1 fold-change mRNA expression of diseased with nondiseased tissues.The results are given in fig.
In the 10 patients with ulcerative colitis (in the large intestine), HO-1 expression was downregulated in 7 patients (70%, varying from 0.23-to 0.81-fold), and upregulated in only 2 (20%, 1.7-1 and 2.02-fold, respectively); in one there was no change (10%, 0.99-fold) (fig.2B and table 5).This similarity between Crohn's disease and ulcerative colitis suggests a common regulation of HO-1 expression in the human large intestine.In 15 patients with IBD affecting the large intestine, HO-1 was downregulated in 10 (67%) and upregulated in 3 (20%); there was no clear change in 2 (13%).The differential expression of HO-1 in this group was similar to that in ulcerative colitis, not Crohn's disease, raising a question of tissue-specific regulation of HO-1 expression in the small intestine and large intestine responses to intestinal inflammation.

Discussion
Association of HO-1 gene with paediatric IBD IBD is a chronic and recurrent intestinal inflammatory disorder.HO-1 has been shown to be a protective and anti-inflammatory protein.A role of the HO-1 gene in IBD has been demonstrated in an IBD animal model [22,23].However, a genetic association of HO-1 with adult IBD and paediatric IBD has not been observed.In various association studies [2], in which HO-1 (GT)n variants were used as markers, no association was found with IBD (Crohn's disease n = 179, ulcerative colitis n = 110, control n = 56) [3], with Crohn's disease (n = 504, controls n = 370) [2], or with paediatric Crohn's disease (n = 119 [2]).A genetic association study with another HO-1 SNP rs2071746 upstream of the gene also found no association with adult IBD (Danish cohort 336 Crohn's, 498 ulcerative colitis, 779 controls) [4].Moreover, in a case-control study we did not observe an association of HO-1 rs2071746 with paediatric IBD.However, in the present case-trio study we showed an association of HO-1 with paediatric IBD.Utilisation of casetrio data from triad and dyad families benefits genetic association studies by reducing both false negative and false positive rates [21].The integration of additional genotypic information from the parent(s) has the potential to increase the statistical power.Also, with the genotypes of the parent(s) included in the likelihood model, the analysis result is less affected by population stratification and more robust in terms of false positive rates.Though our case-trio study does not contain singleton monads, the likelihood approach is general enough to be applied to a traditional case-control study where only the genotypic information of the child is available.

HO-1 gene expression and function in intestinal inflammation
HO-1 and its enzymatic product carbon monoxide regulate both acute and chronic inflammation and ameliorate intestinal inflammation by promoting bacterial clearance.An early investigation of HO-1 expression in clinical and experimental colitis demonstrated a scattered expression of HO-1 in the epithelium of severely inflamed colonic mucosa of patients with IBD compared with control specimens from patients with, for example, diverticulitis, suggesting dysregulated expression in IBD.Therefore, a protective role of HO-1 was suggested, if it is induced before the onset of inflammation [24].HO-1 is a protein with a wide range of anti-inflammatory and immune regulatory functions, and with significant protective roles in various T cell-mediated diseases.Cellular localisation of HO-1 gene expression indicated that in patients with active ulcerative colitis, HO-1 staining is pos-itive in mononuclear cells in the submucosa of the colon but negligible in the epithelial cells [25].
In dextran sulphate sodium (DSS)-induced colitis mice, DSS markedly induced HO-1 expression in the colon epithelium.Inhibition of HO-1 expression aggravated the colitis.The anti-inflammatory mechanism involved the induction of regulatory T cells, IL-17 and apoptosis [21,23,26].In DSS-induced mouse colitis, the anti-inflammatory effects of grapefruit-derived nanovesicles were mediated by upregulation of HO-1 and inhibition of IL-1β and TN-Fα in intestinal macrophages [27].In a 2, 4, 6 trinitrobenzene sulphonic acid (TNSB )-induced colitis animal model, significantly increased colonic HO-1 expression was observed in comparison with the control group.Octreotide treatment increased HO-1 expression, but decreased expression of nuclear factor kappa-light-chain-enhancer of activated B-cells (NFκB) [22].In biopsy specimens from 18 ulcerative colitis patients and 13 colon cancer patients, expression of HO-1 mRNA and protein was significantly increased in the colonic mucosa of patients with active ulcerative colitis compared with normal mucosa [25].These results collectively indicate that there may be an induction of HO-1 in the colon of ulcerative colitis patients In the present study, we observed a downregulation of HO-1 expression in diseased intestinal tissues compared to adjacent normal tissues from 9 out of 17 Crohn's disease patients (53%) and 7 out of 10 ulcerative colitis patients (70%).Evidence indicated that HO-1 / carbon monoxide pathway ameliorates intestinal inflammation [1].Downregulation of HO-1 expression will probably weaken endogenous defensives, shifting the pro/anti-inflammation balance towards inflammation, potentially resulting in tissue injury in IBD.
The observations of an increased HO-1 expression in IBD by Kokurs et al. [25] as well as an upregulation and down regulation of HO-1 expression in different individuals in this study, suggest that the activation and maintenance of HO-1 expression is controlled by many factors, including genetic factors (as discussed above), oxidative stress, inflammatory processes and disease stage.A further study of the underlying mechanisms, on clinical samples, is necessary.

Impact of IL-10/STAT3 pathway on HO expression
The IL-10/STAT3 pathway is involved in regulation of inflammation.In this pathway, IL-10 signalling transduction occurs through binding of IL-10 to its receptors ILl10RA and ILl10RB to form a complex, with downstream molecules janus kinase-1 (Jak1) and tyrosine kinase 2 (Tyk2) activating STAT3 to regulate HO-1 transcription.Previous studies indicated that I-10 and Tyk2 SNPs were genetically associated with IBD, and mutations in IL-10RA and IL-10RB have been found to be associated with paediatric and very early onset IBD.In this study, we showed that SNPs rs1800872 of IL-10 and rs2071746 of HO-1 are associated with paediatric IBD.In our previous study we also observed a significant epistatic interaction between IL-10RB gene in the IL-10/STAT3 signalling pathway and target HO-1 gene (table 5).These findings indicate that the IL-10/STAT3 pathway plays a role in anti-inflammation in paediatric IBD.These studies help in the discovery of genetic contributors to intestinal inflammation and provide specific targets in personalised medicine.

Induction of HO-1 gene expression by proinflammatory LPS and TNFα
Although HO-1 possesses anti-inflammatory properties, we did not, in this study, observe a significant effect of proinflammatory LPS and TNFα on HO-1 expression in human B cells.This may be due to the small number of samples studied, and/or related to the patients' treatment (information that was not available).A further investigation of HO-1 expression in response to LPS and TNFα treatment, with a larger number of samples with clinical treatment information, is needed for understanding the role of HO-1 in the control of inflammation in IBD.Dorresteijn et al. [28] showed in a recent study that LPS downregulates HO-1 expression in primary human mononuclear blood cells.They further indicated that LPSdependent HO-1 regulation is cell type-and species-specific.In that study, human B cells (as opposed to mouse B cells), unlike mononuclear blood cells, lacked expression of TLR4 and were not activated by treatment with LPS.Furthermore, the B lymphocyte cell lines used in the present study were transformed by Epstein Barr virus and were not primary B cells.These issues may contribute to the observed differences between the results of the present study and those in the previous literature [28].
HO-1 as a potential target for therapy HO-1 has anti-inflammatory properties and has been postulated to have a role in the amelioration of IBD.This was demonstrated with grapefruit-derived nanovesicles (GDNs) that can induce HO-1 activity and produce endogenous antioxidant and anti-inflammatory moieties such as bilirubin and carbon monoxide in DSS-induced mouse colitis [29].This indicates that HO-1 may act as an immune modulator molecule, attenuating inflammatory response in IBD.

Conclusion
Our previous results and present case-trio study suggest: that genes IL-10 and HO-1 are involved in paediatric IBD; that gene-gene interaction of IL-10RB with HO-

Table 4 :
Parameter estimation and hypothesis testing summary.

Table 5 :
Ho-1 gene expression in diseased tissues from adult patients with inflammatory bowel disease.
Swiss Medical Weekly • PDF of the online version • www.smw.chPublished under the copyright license "Attribution -Non-Commercial -No Derivatives 4.0".No commercial reuse without permission.See http://emh.ch/en/services/permissions.html.
1 is involved in paediatric IBD; and that mutations in IL10 and IL10RB reduced HO-1 expression.These results indicate that the IL-10/STAT3 pathway and the IL-10 receptor IL10RB may regulate HO-1 activity, which in turn may lead to protective and/or anti-inflammatory effects in IBD therapy.statement This study was supported by Children Miracle Network (CMN) Research grant (ZL, Children Miracle Network Grant 132698).Study samples were provided by Crohn's and Colitis Foundation of American (CCFA) and the Pennsylvania State University Hershey Medical Center and Hershey Biobank.