Although the etiology of primary sclerosing cholangitis (PSC) is unknown, it is most often referred to as an “autoimmune” liver disease. Genetically “complex” PSC has strong associations with the human major histocompatibility complex (MHC) on chromosome 6p21.3.1-6 The major susceptibility and resistance alleles/haplotypes for PSC are listed in Table 1.
|DRB1 P9 charge||A||B||Cw||DRB1||DQA1||DQB1|
The article by Hov et al.7 in this issue of HEPATOLOGY is the latest and largest study on human leukocyte antigens (HLA) in PSC. It goes beyond all previous studies by using three-dimensional modeling to explore the effect of key residues on the DR molecule in terms of disease risk. Genome-wide association studies have identified this region as having strong genetic associations with a range of different diseases, including PSC,1 primary biliary cirrhosis,8 and drug-induced liver injury.9, 10 In all of these cases, the MHC has been shown to be the most significant susceptibility determinant with the highest risk value. However, the key word in each of these studies is “risk”. Unlike Mendelian diseases, genetically “complex” diseases do not have a simple pattern of inheritance, the risk alleles are usually frequent in the healthy population, and the inheritance of a specific allele or group of alleles on a specific chromosomal segment (i.e., haplotype) is neither necessary nor sufficient for the disease to occur.
Considering all the points above, one may ask the question “why study genetics in complex disease?” The answer can be found within the Human Genome Project, which made a number of “promises” in relation to common disease, which can be listed under three broad headings: (1) to improve disease diagnosis, (2) to increase our understanding of disease pathogenesis, and (3) to develop new strategies for patient management and treatment including the development of novel therapies. The study by Hov et al.7 is intended to contribute to the overall understanding of the pathogenesis of PSC.
By translating specific HLA associations into amino acid sequences, the first step in this direction can be made. This approach to HLA-encoded disease risk was first published by Todd, Bell, and McDevitt in 1987,11 who mapped susceptibility for insulin-dependent diabetes to specific amino acid sequences of the HLA-DQβ polypeptide. This changed the way in which HLA associations were perceived. No longer were they seen as unexplainable genetic anomalies; it was now possible to put these associations into a functional context. Subsequent advances in polymerase chain reaction–based genotyping, the publication of the crystal structures for the MHC class II molecule,12 and the development of more sophisticated computer-based technologies for predictive modeling13 have completely revolutionized our approach to HLA in disease, and these new technologies have been widely applied. This can be seen with varying levels of sophistication in relation to “autoimmune” liver disease14-16 as well as nonliver diseases.17 The present study7 of the electrostatic modification of the HLA-DR molecule in PSC is the latest study to take this approach, and furthermore, it is one of many studies from this same group that have sought to define MHC-encoded susceptibility to PSC.1, 7, 18
Amino acid sequence variants for HLA-DRB1 were investigated in 356 patients with PSC from a single center. The basic principle is not a novel one (see above, Todd et al.11), but the techniques applied are up-to-date and this is the first study to consider all possible variants of HLA-DRB1 in PSC. Clearly aware of the previous studies, Hov et al.7 state “a consistent peptide-binding motif for the class II molecules associated with PSC has not been defined, and no attempts have been made to model how specific amino acids affect the structure and the electrostatic properties of the peptide-binding groove.” This statement is correct and forms the rationale for their study. The earlier studies of Farrant et al.,5 Olerup et al.,4 and Donaldson and Norris6 were all limited in scope. Farrant et al.5 proposed that susceptibility and resistance to PSC may be determined by the amino acid at position 38 of the second expressed DRB gene. In particular, they noted that the risk haplotypes encode the amino acid leucine at position 38, whereas the protective haplotypes encode alanine at position 38. Olerup was unable to affirm this association in the 1995 article,4 and although Donaldson and Norris6 confirmed the association in their extension of the series by Farrant et al.,5 the association with this amino acid is not confirmed in the present study nor in any of the studies from Sweden or Norway. Interestingly, the Donaldson and Norris6 review found that DRβ71 and DRβ86 were not major determinants of susceptibility to PSC. This observation contrasts with the situation in type-1 autoimmune hepatitis14, 15 and with the present study of PSC where DRβ86 makes a significant contribution to susceptibility and where DRβ71 may have a subsidiary role. One novel element of the study of Donaldson and Norris6 was to consider the DQB1 genes, and their report includes associations with proline at DQβ55 and with phenylalanine at DQβ87. Despite these observations and even more compelling evidence for a very strong role played by DQB1 alleles in a range of autoimmune diseases, DQB1 was not considered in the present study.17
In this new study of 356 “Scandinavian” patients with PSC, we see for the first time a complete analysis of the physiochemical and structural characteristics of the peptide-binding groove being compared, rather than a simple analysis of amino acid sequences. The study is restricted to HLA-DRB1, but it gives us a clear picture of the electrostatic potential around the three-dimensional structures of the different HLA molecules encoded by the different risk alleles. The study indicates that residues at positions 37 and 86 are the primary residues for disease risk. Other residue positions were found to have some influence, including residues 71 and 74, both of which have been identified in autoimmune hepatitis as major risk residues.14, 15
The highest and lowest risks of PSC were observed for carriers of asparagine (Asn37) (odds ratio = 5.7) and tyrosine (Tyr37) (odds ratio = 0.25). What does all this mean? When we consider the function of the MHC, we need to remember that the specificity of the peptide-binding groove is governed by the structural and chemical properties of a series of nine pockets in the binding groove. These are pockets, numbered P1 to P9 accommodate amino acid side chains of the antigenic peptide (Fig. 1). Risk alleles that encode asparagine at DRβ-37 (on risk haplotypes 1 and 2; Table 1) form P9 pockets with similar structural architecture and with a consistently positive electrostatic potential. These risk alleles are thought to encode molecules that present a restricted peptide repertoire. In comparison, protective alleles that encode tyrosine at DRβ-37 form P9 pockets with consistently negative electrostatic potential. Because HLA molecules are promiscuous and there is competition for binding of antigenic peptides by newly synthesized HLA molecules, any restriction resulting from this genetic variation can determine which antigenic peptides are preferentially bound and presented to the T cell receptor—a key step in the formation of the “immune synapse”. The fact that this is a competitive process in some ways accounts for the loose (and therefore complex) relationship between HLA and disease risk.
The authors of this article leave the reader with no doubt. Susceptibility to PSC is largely determined by DRβ-37, with some help from DRβ-86 and maybe DRβ-71 and DRβ-74. However, they also suggest that this is not the whole story.
There are differences between published series. The reasons for this are complex and are recognized by the authors.7 When considering studies performed between 1992 and 2011, we are not comparing like with like. There have been major advances in the methods used, which explains some but not all of the variation reported. It is true to say earlier studies were limited. However, even the present study made assumptions, particularly with regard to the potential role of paralogous DRB genes, and there are exclusions (HLA-DQB1 for example). Also, there is still some dispute over the secondary association with risk haplotype 2,3 which does not exist in the Scandinavian patients, but is present in the United Kingdom and has a significant effect on analysis of the UK series.5, 6
Finally, we have the ancestral 8.1 haplotype to consider (risk haplotype 1). HLA 8.1 raises two points for consideration. First, the patient population presented has a very large number of 8.1 homozygotes, much larger than would be expected, and it does not appear to be due to population homogeneity. Second, recent genome-wide association studies1 indicate a strong role for the HLA class I region in PSC, and earlier association studies suggested roles for HLA-C,18HLA-B,19 and MICA (MHC class I polypeptide-related sequence A).19, 20 Considering the involvement of these genes in activation of lymphocytes that are common in the liver, such as natural killer cells, natural killer T cells, and γδ T cells, any future studies of HLA will need to considered this region if we are to fully unravel the immunopathology of PSC.
This article marks a major step forward in PSC. Although there is still much work to be done, it presents a good model, particularly if it can be used to evaluate any future experimental studies of antigen presentation in this disease, as the authors suggest.