Date Published: November 17, 2009
Publisher: Public Library of Science
Author(s): Philippe Georgel, Cécile Macquin, Seiamak Bahram, David M. Ojcius. http://doi.org/10.1371/journal.pone.0007803
Abstract: Toll-Like Receptors (TLR) are critical elements of the innate arm of the vertebrate immune system. They constitute a multigenic family of receptors which collectively bind a diverse array of – exogeneous as well as endogeneous – ligands. An exponential burst of knowledge has defined their biological role in fight against infections and generation/modulation of auto-immune disorders. Hence, they could at least be conceptually recognized – despite being structurally unrelated – as innate counterparts to Major Histocompatibility Complex (MHC) molecules – equally recognizing antigenic ligands (albeit structurally more homogeneous i.e., peptides), again derived from self and/or non-self sources – preeminent this time in adaptive immunity. Our great disparities in face of infections and/or susceptibility to auto-immune diseases have provoked an intense search for genetic explanations, in part satisfied by the extraordinary MHC allelic repertoire. An equally in-depth and systematic analysis of TLR diversity is lacking despite numerous independent reports of a growing number of SNPs within these loci. The work described here aims at providing a preliminary picture of the allelic repertoire – and not purely SNPs – of all 10 human TLR coding sequences (with exception of TLR3) within a single cohort of up to 100 individuals. It appears from our work that TLR are unequally polymorphic: TLR2 (DNA alleles: 7/protein alleles: 3), 4 (4/3), 7 (6/3), 8 (9/2) and 9 (8/3) being comparatively least diverse whereas TLR1 (11/10), 5 (14/12), 6 (10/8) and 10 (15/10) show a substantial number of alleles. In addition to allelic assignment of a large number of SNPs, 10 new polymorphic positions were hereby identified. Hence this work depicts a first overview of the diversity of almost all human TLR genes, a prelude for large-scale population genetics as well as genetic association studies.
Partial Text: In collaboration with other pattern recognition Receptors, Toll-like Receptors (TLRs) govern the innate arm of the immune system. The first member of this family, TOLL, was originally described through its function in the Drosophila embryonic development and was later found to be involved in the response against Gram positive bacteria and fungi within this organism . This discovery prompted the in silico identification of mammalian orthologues. A positional cloning approach of the Lpsd locus led to the identification of the mouse TLR4 and to its description as the LPS sensor which is essential for an efficient response against Gram-negative bacteria . These findings emphasized the evolutionary conservation of the TLRs and suggested that these molecules play a fundamental role in the early detection of pathogens and the appropriate subsequent innate immune response (reviewed in , ).
Susceptibility to infections is heritable, as demonstrated in a seminal study which examined and compared the cause of death among adoptees to that of their biological or adoptive parents . Although many immunodeficiencies have been mapped to a single locus (monogenic diseases, such as X-linked or autosomal severe combined immunodeficiencies), until recently, a role for TLR in susceptibility to infections had not firmly been assessed. Two genetic studies have demonstrated the involvement of downstream components used by the TLR pathway: IRAK4 mutations have been linked to increased infections by pyogenic bacteria  and UNC-96B defects in two children was associated to HSV-1 encephalitis . The same group has now identified a TLR3 mutation in humans and reports for the first time HSV-1 encephalitis in children carrying the mutated allele . This contrasts with the wealth of information which was produced with animal models in which experimental infections in mice have been instrumental in defining the TLR specificities. Despite high homologies between mouse and human genes, it is now clear that human genetic diversity (as opposed to genetic homogeneity in mouse models) and environmental conditions which may or may not bring in close proximity a pathogen and its host, render human genetic studies, in addition to the development of animal models, a required step to understand immune responses to natural infections. To contribute to this goal, we have resequenced TLR1–10 (but not TLR3 which was excluded from the analysis for technical reasons, see Materials and Methods) in up to one hundred unrelated individuals of Caucasian origin, and “scanned” all the synonymous and non-synonymous SNPs in the coding sequence. Because we focused our study to the coding sequences of TLRs, we concentrated our present analysis on non-synonymous SNPs, which can introduce modifications potentially affecting the function of the corresponding proteins. Nevertheless, we also recorded synonymous SNPs and compared them to those which are already described in public databases. From this effort, several points deserve to be highlighted. (i) First, the general picture emerging from the comparison of TLR1–10 allelic repertoire is that these molecules are not equally polymorphic. Clearly, TLR10, 5, 1 and 6 show the highest level of variation in our Caucasian sample with respectively 15, 12, 10 and 8 possibilities for the encoded proteins. Only 3 non-synonymous SNPs we detected for TLR4, whereas 29 are deposited in databanks. The limited number of alleles identified in our study likely reflects the lowest amount of samples which were analysed, compared to more extensive investigations , . On the contrary, besides this sub-group of highly variable TLRs, TLR 7, 8 and 9 appear less flexible. We detected only two known non-synonymous SNPs in TLR7 (out of 12 which are recorded), one in TLR8 (in addition to 3 known non-synonymous SNPs) and two in the TLR9 coding sequence (out of 16 already described). We speculate that selection pressure prevents mutations in these genes because the proteins encoded by these TLRs recognize nucleic-acid-based structures which are highly conserved and subjected to almost no possible variation in their size, charge or other physicochemical features. Conversely, evolutionary constrains have likely shaped TLR1, 4, 5 and 6 to render them able to detect a whole panel of pathogens, despite expression of highly variable molecules such as LPS and lipoproteins among different microorganisms. In this regard, TLR2 limited variation remains enigmatic, but the multiple alleles of the still orphan TLR10 suggests that this molecule, possibly associated with TLR1 or TLR2, likely participates in bacterial determinants sensing. (ii) Second, several new non-synonymous polymorphic positions were identified in this study (I57M, L443I, V542A and T565S in TLR1, S247T, S353G and M484I for TLR5; D428N in TLR8; R311Q in TLR9 and L59I in TLR10). All these positions are located in the ectodomain of the TLRs which contains the LRR and could therefore potentially affect ligand binding and/or dimer formation. Future studies, using TLR expression vectors transfected in culture cells, should clarify the effect of these mutations on the functionality of these receptors. In this respect, it is interesting to note that such studies designed to quantify the impact of specific SNPs in TLRs have been performed only in few instances . (iii) Third, our effort, combined with those of others, is important to obtain reliable sequence and SNP information, which ultimately, are validated by comparing data from multiple sources. For instance, we observed that the so-called S248N variant of TLR1 is present in 3 alleles (alleles 5, 6 and 7; Table S1) with a combined frequency of 70.24%. In agreement with others , our data indicate that Asn (N) residue at position 248 rather represents the normal form of TLR1 and its replacement by a Ser (S) identifies a variant form. (iv) Lastly, we wish to emphasize that our effort provided not only frequencies for individual SNPs in the TLR coding sequences, but also afforded an overall picture of the allelic repertoire for this important class of proteins. Resequencing is a useful approach to assemble SNPs at the level of a single gene and, as pointed by others , , is more powerful than single SNP analysis to analyse complex traits. Because recombination between SNPs may affect gene variability, our description of all the alleles in a single cohort provides valuable information regarding the heterogeneity of TLR genes in an ethnically-defined population. It should also contribute to clarify the effect of single SNPs in TLRs which were not convincingly replicated in several independent investigations (see  for a recent update on TLR polymorphisms linked to human disease).