Can’t believe in a TCRA association (at the moment)

Nature genetics published recently an association paper of an of an autism researcher researcher writing here on narcolepsy

Using genome-wide association (GWA) in Caucasians with replication in three ethnic groups, we found association between narcolepsy and polymorphisms in the TRA@ (T-cell receptor alpha) locus, with highest significance at rs1154155 (average allelic odds ratio 1.69, genotypic odds ratios 1.94 and 2.55, P < 10-21, 1,830 cases, 2,164 controls). This is the first documented genetic involvement of the TRA@ locus, encoding the major receptor for HLA-peptide presentation, in any disease.

I am always cautious of these “first ever” claims. Indeed, I can not follow the conclusions of this expert writing his first T cell report here. So I wrote a letter and sent it to NG, basically saying

The study by Hallmayer et al. [1] reports a moderate association of TCRα SNPs with narcolepsia. Given the largely frustrating attempts to find any consistent TCR disease association [2], there may be some questionmarks given the evolutionary emergence of the huge repertoire that is being encoded in this genomic region [3].
First, commercial available SNP panels with a rather uneven coverage of the human genome may not be the optimal method for analysing such a complicated genomic region containing redundant sequence. That may be the reason why more adequate methods like CDR3 spectratyping [4] are available for the genetic analysis of the TCRα.
Second, there is obvious population stratification in the TCRα SNPs (as seen in tables 2 and 3) which is also expected by previous work on the TCRß [5]. The results of the reported Seldin panel does not exclude stratification in this particular region. It would be interesting to see the detailed geographical and ethnical history of all samples including some more informative gene marker like LCT-13910C/T that varies strongly among Caucasians ([6] and unpublished own observations).
Third, as being acknowledged by the authors, the associated SNPs fall in an area of somatic recombination [7] that affects under normal circumstances about one quarter of the genotyped cells. This ratio may vary considerably in a disease like narcolepsia where T lymphocytes can be increased or decreased [8]. In other words it is unclear if the study shows the “first documented genetic involvement of the TRA@ locus” or just some concommittant lymphocytosis; if it shows any inherited disease specific antigen selection mechanism or just some reactive inflammation; truly a TCR disease bias or just an artifact. The authors don’t report the age distribution of their individuals [9] although this might also have confounded the results with an age-associated decline in the T cell repertoire [10] and a life-long persistence of some T cell subsets [11].
Previous CNV scans at least excluded the 14q11.2 region from the analysis where genotyping results may appear as single copy deletions (and rarely as duplications As there is also an “imprint of CNV on SNP genotypes” [12] an increase in SNP allele counts may be mimicked as well.
Any signal induced by lymphocytosis would need to include a segment 3′ to the last constant region exon (the position of rs1263646) with decreasing likelihood from J1 to J61 (the other two reported SNPs are next to J9 and J13). Major signals from V segments are unlikely due to broken up LD in oligo- or polycolonal T cell expansion. Interestingly, there is also a signal at the TCRß receptor; the diversity of TCRα selected as partner for a given VDJ-ß rearrangement is low [13] and may be the reason why a TCRα association is being easier to be detected.
Any biased allele count could be further recognized by disturbing Hardy Weinberg equilibrium (HWE) in cases only. Table 3 results indeed indicates that in the two collections that replicate their findings, the case genotypes do not follow HWE (Asian cases p exact=0.02, Caucasian cases p exact=0.04).
In summary, before speculating about the impact for over 100 other HLA-associated disorders, it may be interesting to know if the present association can be replicated in non-recombining somatic cells.
1. Hallmayer J, Faraco J, Lin L, Hesselson S, Winkelmann J, et al. (2009) Narcolepsy is strongly associated with the T-cell receptor alpha locus. Nat Genet.
2. Turner SJ, Doherty PC, McCluskey J, Rossjohn J (2006) Structural determinants of T-cell receptor bias in immunity. Nat Rev Immunol 6: 883-894.
3. Marchalonis JJ, Kaveri S, Lacroix-Desmazes S, Kazatchkine MD (2002) Natural recognition repertoire and the evolutionary emergence of the combinatorial immune system. FASEB J 16: 842-848.
4. Seitz S, Schneider CK, Malotka J, Nong X, Engel AG, et al. (2006) Reconstitution of paired T cell receptor alpha- and beta-chains from microdissected single cells of human inflammatory tissues. Proc Natl Acad Sci U S A 103: 12057-12062.
5. Subrahmanyan L, Eberle MA, Clark AG, Kruglyak L, Nickerson DA (2001) Sequence variation and linkage disequilibrium in the human T-cell receptor beta (TCRB) locus. Am J Hum Genet 69: 381-395.
6. Smith GD, Lawlor DA, Timpson NJ, Baban J, Kiessling M, et al. (2009) Lactase persistence-related genetic variant: population substructure and health outcomes. Eur J Hum Genet 17: 357-367.
7. Spits H (2002) Development of alphabeta T cells in the human thymus. Nat Rev Immunol 2: 760-772.
8. Coelho FM, Pradella-Hallinan M, Alves GR, Bittencourt LR, Pedrazzoli Neto M, et al. (2007) A study of T CD4, CD8 and B lymphocytes in narcoleptic patients. Arq Neuropsiquiatr 65: 423-427.
9. Hall JG (2003) A clinician’s plea. Nat Genet 33: 440-442.
10. Yager EJ, Ahmed M, Lanzer K, Randall TD, Woodland DL, et al. (2008) Age-associated decline in T cell repertoire diversity leads to holes in the repertoire and impaired immunity to influenza virus. J Exp Med 205: 711-723.
11. Nikolich-Zugich J (2008) Ageing and life-long maintenance of T-cell subsets in the face of latent persistent infections. Nat Rev Immunol 8: 512-522.
12. Redon R, Ishikawa S, Fitch KR, Feuk L, Perry GH, et al. (2006) Global variation in copy number in the human genome. Nature 444: 444-454.
13. Hamrouni A, Aublin A, Guillaume P, Maryanski JL (2003) T cell receptor gene rearrangement lineage analysis reveals clues for the origin of highly restricted antigen-specific repertoires. J Exp Med 197: 601-614.

The response of the journal (“Nature Genetics publishes the very highest quality research in genetics” with the publication guideline “Provides strong evidence for its conclusions”) were some boilerplates

Thank you for submitting your Correspondence commenting on the Hallmayer et al. paper recently published in Nature Genetics. We have carefully evaluated the points raised in your Correspondence, and we regret that we will not be sending your manuscript for peer review. Manuscripts submitted to Nature Genetics are evaluated for their potential general interest to our readership before a decision is made as to whether they should be sent for peer review, and papers may fall short of consideration for publication on a number of editorial grounds. In this case, we believe that the specific caveats you raise are unlikely to account for the association of these specific SNPs in the TRA@ locus with narcolepsy risk, particularly since the association of these correlated SNPs with narcolepsy was observed consistently in the discovery samples and in independent replication samples drawn from diverse ancestral populations, i.e. we find it difficult to envision that the observed specificity for these SNPs and consistency of replication could arise as an artifact of somatic rearrangements at the TRA@ locus or underlying population structure.
I am sorry that I cannot provide you with a more positive response on this occasion and hope you will think of us when preparing manuscripts in the future.

signed by an expert in “classical mouse mutants Splotch and Loop-tail to study the genetics of neural tube closure” and “the developmental mechanisms underlying vertebrate left-right axis specification”.
So I wrote another email

yes, I know that NG does not believe in an artifact by somatic recombination – otherwise this paper would not have been published.I am, however, 100% sure that the overall signal intensity at the TCRA locus goes down in current GWA experiments. This can be easily confirmed with the raw data and will convince you that lymphocyte counts matter. Calling of a genotype at the TCRA locus in a individual is a complex mixture of somatic + recombined DNA template, further complicated by variability on a population level. The used birdseed algorithm may be particular sensitive to cnvs
So it is not unexpected, that you have an (unnoticed!) deviation from Hardy Weinberg in cases – textbooks provide many explanations for that but in my hands it is usually a genotyping error. This view is also supported by genotyping failure of rs12587781 in table 1.
My guess is (80% probability) that an oligoclonal expansion is responsible for the observed association […]
I provide a good theoretical explanation in my letter why just the reported SNPs are being affected although I do not really understand why the increase in the allele frequency is uniform across the collection.
In any case, replication in another sample does not help you to identify a spurious result (please note that the association can NOT even be replicated in all collections). Replication is not an adequate method to account for confounding.
Journals like NG should care about the scientific integrity of what they publish; if the prepublication process fails for whatever reasons, post publication comes into place (as far as I see there is no T cell expert on the Hallmayer paper)…

Response: None (4+ weeks later):
So I am asking here: Does anyone has empirical data to show the proposed association of oligoclonal expansion and SNP allele frequency at the constant region? The rather identical replication in the original paper is astonishing, I agree, but given the fact that the original data have not been published, we can believe it or not.
Response: None (1 year later):
By writing another paper on “Common variants in P2RY11 are associated with narcolepsy” the authorsdon’t repeat their TCRA claim. What a strange science policy …