Introduction

The LST-1 gene is located 15 kilobases (kb) upstream of the tumour necrosis factor alpha (TNF) locus (TNFSF2) and encodes a small protein that modulates immune responses and cellular morphogenesis and is constitutively expressed in leucocytes and dendritic cells.¹ Its gene product has been shown to have an inhibitory effect on lymphocyte proliferation.² Polymorphism has been described in intron 3 of the LST-1 gene, including two dinucleotide repeats termed TNFd and TNFe.³ It has been reported that variability in TNF production is associated with alleles at the TNFd locus. A strong correlation between inheritance of the TNFd3 allele and high TNF production by leucocytes in vitro was shown.⁴ The experiment suggested strong linkage disequilibrium between an as yet undescribed functional polymorphism in the TNF gene and the TNFd3 allele, despite a distance of some 8 kb between the TNFd locus and the TNF gene.

TNF region microsatellites have been typed in a number of disease association studies.^5,6,7 With the conventional size-based method of typing, there is an assumption that alleles of identical size are identical in sequence. Often this is not the case, due to the presence of variable base insertions/deletions either within or in close proximity to the microsatellite, termed homoplasy. Homoplasy has been reported within the TNFa microsatellite, which fundamentally affects the calculation of TNFa allelic size.^8,9 Thus a number of suballeles have now been described at the TNFa microsatellite.¹⁰

The presence of homoplasy in the TNFd microsatellite would have strong implications in disease association studies that have analysed this locus. TNF region polymorphism has been strongly implicated in severe acute graft-versus-host disease (aGVHD) and other bone marrow transplant (BMT)-related complications. Middleton et al¹¹ demonstrated TNFd3 homozygosity associated with grade III/IV aGVHD in HLA-matched sibling BMT. A second study also tested association of aGVHD with TNFd in a large cohort of sibling donor/recipient pairs.¹² Their genotype results correlated with acute and chronic GVHD and mortality, and in addition patients who were homozygous d3 had higher transplant-related mortality rates.

We analysed the TNFd microsatellite locus using DNA conformational approaches in order to determine the presence or absence of further suballeles. This paper demonstrates that induced heteroduplex generator (IHG) technology can successfully identify and genotype the TNFd microsatellite, detecting sequence variations, leading to the identification of TNFd1 and TNFd4 suballeles, a clear advance over existing typing methods. The TNFd1 suballele would otherwise have been typed as a TNFd3 allele using conventional typing approaches. Using the Arlequin software package we have also demonstrated strong linkage disequilibrium between the suballeles and other polymorphic loci in the TNF region. In addition we have retyped the same DNA sibling transplant panel, previously typed as TNFd3 homozygotes by Middleton et al,¹¹ for the TNFd1 suballele using IHG analysis.

Results and discussion

Identification of TNFd locus suballeles

Figure 2a shows a conventional size-based typing approach for the TNFd microsatellite locus using nondenaturing polyacrylamide gel electrophoresis (PAGE). Natural heteroduplexes are formed during each reannealing stage of the PCR reaction when two microsatellite alleles from a heterozygous individual cross-hybridize, forming a heteroduplex. In using the conventional typing procedure it was noticeable that samples with the same genotype gave different natural heteroduplex patterns (Figure 2a, arrows). This could only occur where there is a further level of polymorphism within or adjacent to the repeat sequence, a phenomenon termed homoplasy. It was noted that the variation appeared to be restricted to TNFd3- and d4-positive genotypes and we therefore decided to attempt to identify unequivocally the potential variant allele using DNA conformational approaches. An IHG reagent, designated IHG1, was synthesized which mimicked an undescribed TNFd allele (Figure 1). Using this reagent in combination with PAGE we were able to identify all existing TNFd alleles and genotypes. Using IHG1 we identified a novel d4-related allele that was not always identified using nondenaturing PAGE (Figure 2b, arrows). However, using IHG1, we were unable to discriminate clearly between the TNFd3 allele and its potential variant. Therefore, a second IHG was constructed (designated IHG2) with one extra GA repeat than present in IHG1 (see Figure 1). Using this reagent it was possible to identify clearly the presence of a previously undescribed variant allele designated TNFd1b (Figure 2c) in individuals previously genotyped as d3 homozygotes. IHG1 and IHG2 may therefore be used in combination to type accurately these two novel alleles, and all previously described alleles in the TNFd microsatellite locus.

Figure 2.

(a) Conventional nondenaturing PAGE TNFd microsatellite typing. Black and white arrows indicate natural heteroduplexes that indicate sequence variation not identified by conventional analysis of PCR product length. PCR primers from a previous study¹³ were used for amplification of the TNFd microsatellite and IHG1 and IHG2 (IHG and primer sequences shown in Figure 1). PCR amplifications were carried out in a total reaction volume of 20 l, containing 100 ng of genomic DNA or IHG, 0.5 U Taq polymerase (Advanced Biotechnologies), 0.5 M of each primer (forward and reverse), 1.5 mM MgCl₂, 200 m of each dNTP, 67 mM Tris-HCL (pH 8.8), 16 mM (NH₄)₂SO₄ and 0.01% Tween. The PCR was optimized on an MJ PTC-100 thermal cycler (GRI, Braintree, UK). The PCR parameters for both genomic DNA and IHGs were as follows: initial denaturation for 5 min at 95°C, followed by 32 cycles of 95°C for 1 min, 62°C for 30 s, 72°C for 1 min with a final extension at 72°C for 5 min. (b) TNFd microsatellite typing using induced heteroduplex analysis with IHG2. White arrows indicate induced heteroduplexes, which clearly discriminate between alleles 4 and 4b. The method of IHG analysis has been described previously:^14,15 cross-matching parameters were 95°C for 2 min, followed by an initial controlled cooling step to 65°C (ramping rate=1.0°C/s) and holding for 1 min. A second controlled cooling step was inserted to 45°C (ramping rate 0.03°C/s) and held for 1 min, with a final holding step at 4°C. Heteroduplexes were resolved on nondenaturing 15% Protogel^® polyacrylamide gels (National Diagnostics, Hull, UK) using a 'triple wide' minigel system (30 cm 8 cm; CBS Scientific Company, Del Mar, USA): gel constitution 37.5:1 (w/v acrylamide:bisacrylamide). Electrophoresis was carried out at 300 V for 90 min. Gels were post-stained in 1 Tris borate EDTA (TBE) buffer containing 0.5 g/ml ethidium bromide and examined using a 302 nm UV transilluminator. Images for analysis were acquired using an EDAS 120 system (Kodak). M, molecular weight marker (100 bp ladder). (c) TNFd microsatellite typing using induced heteroduplex analysis with IHG1. Induced heteroduplexes clearly discriminate between alleles 3 and 1b. Genotypes are shown for each lane.

Full figure and legend (217K)

Figure 1.

Nucleotide sequence alignment of TNFd5, d4, d3 and d1 alleles with IHG1 and IHG2. Forward and reverse primer annealing sites are shown underlined. Repeat regions 1, 2 and 3 are highlighted in bold typeface. The IHGs were synthesized as single oligonucleotides, using 0.2 mol membrane columns on a PerSeptive 8900 Nucleic Acid Synthesis System. After synthesis, deprotection and precipitation, the oligonucleotides were amplified by PCR using the primers as shown. Prior to analysis, a stepwise dilution of each IHG was carried out. The optimal range was between 10^-4 and 10^-6 dilution.

Full figure and legend (88K)

Figure 3 shows the nucleotide sequences of the two novel TNFd alleles. We have classified the variant allele previously typed erroneously as d3, as a suballele of TNFd1 and have designated it TNFd1b. This designation is based on the principle of TNFd allelic discrimination by sequence variation in the first of three GA repeat blocks. The TNFd1b allele has the same number of repeats as the TNFd1 allele in the first repeat block but has two extra GA repeats in the second repeat block. The situation with the TNFd4 suballele is more complex. Sequence variation involves the creation of two imperfect repeats via guanine to adenosine substitutions, one in the first GA repeat region and one in the third. We have designated this new suballele TNFd4b.

Figure 3.

Nucleotide sequence alignment of TNFd4, d4b, d3, d1 and d1b with IHG 1 and IHG 2. Alleles d1 and d1b differ by two GA repeats in repeat region 2. Alleles d4 and d4b differ by two imperfect repeats (GA >AA) in repeat region 1 and repeat region 3.

Full figure and legend (45K)

Allele frequencies of the two novel TNFd alleles

Using these new IHG-based typing methods, we genotyped a random cohort of 100 individuals from a BMT donor registry using both size-based and conformational approaches. Conventional TNFd allele frequencies did not differ significantly from previous studies.¹⁶ However, when IHG typing was incorporated, allele frequencies changed significantly (Table 1). The TNFd1b allele had an allele frequency of 0.155. This figure represents 28.4% of previously classified 'TNFd3' alleles. In our study, the TNFd1b allele is the third most frequent allele. The TNFd4b allele had an allele frequency of 0.04, 13.5% of previously classified 'TNFd4' alleles.

Table 1 - Allele frequencies in 100 Caucasian adult volunteer bone marrow donors from southwest England, showing a comparison of alleles detected by nondenaturing PAGE and IHG analysis.

Full table

Contribution of the two novel TNFd alleles to extended TNF haplotypes

In order to identify extended TNF region haplotypes that include the new TNFd alleles, we genotyped the same random cohort at a number of microsatellite and SNP loci. Maximum likelihood analysis (Arlequin 2.000¹⁷) was used to identify conserved haplotypes, and linkage disequilibrium analysis was used to analyse the association between loci. The TNFd1b allele was found to be significantly associated with a haplotype that includes the TNFa6 allele. There was significant linkage disequilibrium (P=4 10^-14) between the TNFd1b and TNFa6 alleles. The TNFd4b allele was strongly associated with a haplotype including the TNFa2, TNFd4b, TNF -1031C and -238A alleles (unpublished data). Linkage disequilibrium between the TNFd4b and TNF -238A allele was highly significant (P<0.000001).

Reanalysis of TNFd3 homozygotes in an identical sibling bone marrow transplant cohort

A previous study on the influence of cytokine polymorphism on HLA-identical sibling stem cell transplantation¹¹ reported an association with homozygosity of the d3 allele and increased risk of grade III/IV aGVHD. We revisited this study and genotyped the d3-homozygous samples using the TNFd IHG2 reagent, in order to identify any TNFd1b alleles. Of the 28 sib-transplant panel retyped for presence of the d1 suballele, two individuals were homozygous and nine individuals were heterozygous for this suballele. In all, 17 individuals were negative for this subtype.

Conclusions

In the current study, the TNFd1b allele has been shown to be the third most frequent allele in the control cohort. It comprises 28% of alleles that have been typed as TNFd3 in previous disease association studies. As the TNFd3 allele was by far the most frequent allele, this has obvious implications for those studies that have associated this allele with given conditions.^{11,18,19,20,21} We have demonstrated this by reanalysing TNFd3 homozygotes in an identical sibling BMT cohort.

The d3 allele has been associated with a number of extended TNF haplotypes. If the TNFa/TNFd microsatellite association is analysed, the d3 allele is associated with the a6, a7, a10 and a11 alleles. These are relatively common alleles in Western populations. We have shown that the TNFd1b and not the TNFd3 allele is strongly associated with the TNFa6-associated haplotype. This finding is important as a number of studies have associated this haplotype with disease association,^19,20 leading to the conclusion that the TNFd1b allele alone would be a strong marker of disease association in these studies. Further to this we have identified strong linkage disequilibrium between the TNFd4b allele and the TNF -238A allele. A number of studies have associated this allele with immune disease.²²

Our IHG-based approach for the analysis of the TNFd microsatellite permits identification of homoplasy at the TNFd locus. The identification of two new suballeles (TNFd1b and TNFd4b) has implications for disease association and expression studies that have up to now been unable to identify them. We are currently applying IHG technology to all TNF region microsatellites in order to identify further suballeles.

References

1. de Baey A, Fellerhoff B, Maier S, Martinozzi S, Weidle U, Weiss EH. Complex expression pattern of the TNF region gene LST1 through differential regulation, initiation, and alternative splicing. Genomics 1997; 45: 591–600. | PubMed | ChemPort |
2. Rollinger-Holzinger I, Eibl B, Pauly M et al. LST1: a gene with extensive alternative splicing and immunomodulatory function. J Immunol 2000; 15: 169–176.
3. Holzinger I, de Baey A, Messer G, Kick G, Zwierzina H, Weiss EH. Cloning and genomic characterisation of LST1: a new gene in the human TNF region. Immunogenetics 1995; 42: 315–322. | PubMed |
4. Turner DM, Grant SCD, Lamb WR et al. A genetic marker of high TNF- production in heart transplant recipients. Transplantation 1995; 60: 1113–1117. | PubMed | ISI | ChemPort |
5. Hajeer AH, Lear JT, Ollier WE et al. Preliminary evidence of an association of tumour necrosis factor microsatellites with increased risk of multiple basal carcinomas. Br J Dermatol 2000; 142: 441–445. | Article | PubMed | ISI | ChemPort |
6. Kunstmann E, Epplen C, Elitok E et al. Helicobacter pylori infection and polymorphisms in the tumor necrosis factor region. Electrophoresis 1999; 20: 1756–1761. | PubMed |
7. Mu H, Chen JJ, Jiang Y, King MC, Thomson G, Criswell LA. Tumor necrosis factor a microsatellite polymorphism is associated with rheumatoid arthritis susceptibility through an interaction with the HLA-DRB1 shared epitope. Arthritis Rheum 1999; 42: 438–442. | Article | PubMed |
8. Honchel R, McDonell S, Schaid DJ, Thibodeau SN. Tumor necrosis factor- allelic frequency and chromosome 6 allelic imbalance in patients with colorectal cancer. Cancer Res 1996; 56: 145–149. | PubMed |
9. Greenberg SJ, Fujihara K, Selkirk SM et al. Novel compound tetra-, dinucleotide microsatellite polymorphism in the tumor necrosis factor/lymphotoxin locus. Clin Diagn Lab Immunol 1997; 4: 79–84. | PubMed |
10. Van der Silk AR, Shing DC, Eerligh P, Giphart MJ. Subtyping for TNFa microsatellite sequence variation. Immunogenetics 2000; 52: 29–34. | Article | PubMed | ISI | ChemPort |
11. Middleton PG, Taylor PR, Jackson G, Proctor SJ, Dickinson AM. Cytokine gene polymorphisms associating with severe acute graft-versus-host disease in HLA-identical sibling transplants. Blood 1998; 92: 3943–3948. | PubMed | ISI | ChemPort |
12. Cavet J, Middleton PG, Segall M, Noreen H, Savies SM, Dickinson AM. Recipient tumor necrosis factor-alpha and interleukin-10 gene polymorphisms associate with early mortality and graft-versus-host disease severity in HLA-matched sibling bone marrow transplants. Blood 1999; 94: 3941–3946. | PubMed | ISI | ChemPort |
13. Udalova IA, Nedospasov SA, Webb GC, Chaplin DD, Turetskaya RL. Highly informative typing of the human TNF locus using six adjacent polymorphic markers. Genomics 1993; 16: 180–186. | Article | PubMed | ISI | ChemPort |
14. Wood N, Bidwell J. Genetic screening and testing by induced heteroduplex formation. Electrophoresis 1996; 17: 247–254. | PubMed |
15. Morse HR, Olomolaiye OO, Wood NA et al. Induced heteroduplex genotyping of TNF-alpha, IL-1 beta, IL-6 and IL-10 polymorphisms associated with transcriptional regulation. Cytokine 1999; 11: 789–795. | Article | PubMed | ChemPort |
16. McDonnell GM, Kirk CW, Middleton D et al. Genetic association studies of tumour necrosis factor a and b and tumour necrosis factor receptor 1 and 2 polymorphisms across the clinical spectrum of multiple sclerosis. J Neurol 1999; 246: 1051–1058. | Article | PubMed | ISI | ChemPort |
17. Schneider S, Roessli D, Excoffier L. Arlequin ver 2.000: software for population genetics data analysis. Genetics and Biometry Laboratory, University of Geneva, Switzerland.
18. Niizeki H, Naruse T, Hecker KH et al. Polymorphisms in the tumour necrosis factor (TNF) genes are associated with susceptibility to effects of ultraviolet-B radiation on induction of contact hypersensitivity. Tissue Antigens 2001; 58: 369–378. | PubMed |
19. Mulcahy B, Waldron-Lynch F, McDermott MF et al. Genetic variability in the tumor necrosis factor-lymphotoxin region influences susceptibility to rheumatoid arthritis. Am J Hum Genet 1996; 59: 676–683. | PubMed | ISI | ChemPort |
20. Hohler T, Grossmann S, Stradmann-Bellinghausan B et al. Differential association of polymorphisms in the TNFalpha region with psoriatic arthritis but not psoriasis. Ann Rheum Dis 2002; 61: 213–218. | Article | PubMed | ISI | ChemPort |
21. Turner DM, Pravica V, Sinnott PJ, Hutchinson IV. Polymorphism in the promoter region of the gene encoding human allograft inflammatory factor-1. Eur J Immunogenet 2001; 28: 449–450. | Article | PubMed | ISI | ChemPort |
22. Bidwell JL, Keen LJ, Gallagher G et al. Cytokine gene polymorphism in human disease: on-line databases. Genes Immun 1999; 1: 3–19. | Article | PubMed | ISI | ChemPort |

Acknowledgements

We would like to thank Dr. Nigel Wood for synthesis of the IHG reagents, and Doris Culpan for her help with the nucleotide sequencing.