AUTHORS
Konstantin Helmsauer, Joseph Palmeri2, Jacob Gutierrez3, Thahmina Ali4, Benan Nalbant5, Caleb Lareau6
AFFILIATIONS
1 Memorial Sloan Kettering Cancer Center, New York, New York, United States
2 Memorial Sloan Kettering Cancer Center, New York, New York, United States
3 Memorial Sloan Kettering Cancer Center, New York, New York, United States
4 Memorial Sloan Kettering Cancer Center, New York, New York, United States
5 Memorial Sloan Kettering Cancer Center, New York, New York, United States
6 Memorial Sloan Kettering Cancer Center, New York, New York, United States
ABSTRACT
The thymus is the primary site of T cell development. Thymus function declines across the life course, is highly variable between individuals and has recently been linked to health outcomes in adults. Yet, we have little understanding of the biological processes driving thymic involution and the extent to which these processes are heritable. These fundamental questions underlying thymic biology remain as thymic output is typically quantified with dedicated assays that poorly scale to population-level assessments. Namely, thymus function is quantified by qPCR of T cell receptor excision circles (TRECs): circular DNA generated during TCRD excision and VJ recombination at the TCRA locus. Once T cells proliferate, TRECs are diluted and therefore act as a proxy for recent thymic emigrant (RTE) T cells in blood. Here, we show that TRECs can be recovered from standard blood whole-genome sequencing (WGS) by identifying discordant or split reads that span the TREC circle junctions.
We first asked whether we could detect TREC reads in blood WGS. Using the TenK10K WGS cohort, TREC reads were indeed detectable in 916 of 1,925 donors (47.6%), at 0.06±0.003 TREC reads per diploid genome (Mean±SE). In matched scRNA-seq data, TRECs correlated specifically with the proportion of naive T cells (Kendall’s τ=0.25, P<2.2x10-16). To establish that TREC reads report on thymic output, we performed WGS on FACS-sorted RTE CD4 T cells vs. memory CD4 T cells from N=3 donors. TREC reads were two orders of magnitude higher in RTE (8.84 (7.29–11.50), mean (range), reads per diploid genome) than in memory cells (0.05 (0.02–0.07)). The specific TREC junction measured by qPCR in the clinic was the single most frequent junction detected. However, it accounted for only 0.5% of TREC reads; the remainder arose from a diverse set of 912 distinct V–J combinations. WGS therefore captures a substantially broader TREC repertoire than the single junction assayed by qPCR. To assess specificity, we applied the same pipeline to WGS data from immortalized B-lymphoblastoid cell lines of diverse genetic ancestries from the 1000 Genomes project. We recovered only a single TREC read across all N=2,504 donors, indicating that neither sequencing artifacts nor genotype diversity is a source of false-positive signal.
Having validated the approach, we quantified TREC content from blood WGS for N=365,928 donors in the NIH All of Us biobank. 61.5% of donors had detectable TRECs. Thymic output declined steeply with age: 96.3% of donors aged 18–25 were TREC-positive (0.343±0.0015 reads per diploid genome) compared with 20.5% of donors over 80 (0.014±0.0003). Women carried 1.67-fold more TRECs than men (0.135±0.0003 vs. 0.081±0.0003). A phenome-wide association study identified associations of TREC not only with sepsis but also with e.g. diabetes type II and obesity, highlighting the interplay between metabolic disease and thymus function.
We then performed a genome-wide association study of thymus function in All of Us. Cross-ancestry meta-analysis identified 80 independent significant loci (10,059 variants with P<5x10-8). Within the EUR ancestry group alone, we identified 49 loci (6771 variants with P<5x10-8). SNP heritability in this group was h²=0.044±0.007. We observed concordant hits for TREC positivity and for TREC content among positive donors (31/49 loci), indicating that the association is not driven by the detection threshold. TREC positivity was genetically separable from overall T cell abundance (estimated from WGS using ImmuneLENS). Although the two traits were moderately genetically correlated (rg=0.421±0.031), 32/49 loci remained genome-wide significant after conditioning on T cell fraction using mtCOJO, consistent with most effects not being mediated by circulating T cell proportions. Associated loci recovered genes with established roles in thymus development (EYA2, HOX cluster), thymic epithelial cells (both LTBR and its ligands LTA/LTB), or developing T cells (BCL2L11; TERT). Beyond these, our analysis uncovered many more genes with uncharacterized roles in thymopoiesis, including TNFRSF19, MEOX1, GRHL2, RBBP8, or FAM53B.
Together, these results establish thymus function as a quantitative trait that can be recovered from standard blood WGS. Applied to a single biobank, this revealed the genetic architecture of human thymopoiesis, recapitulating genes with established roles in thymic biology while nominating new candidate regulators of thymus function.
DISCLOSURES
Konstantin Helmsauer: No relevant conflicts of interest to declare.