A common TCN1 loss-of-function variant is associated with lower vitamin B₁₂ concentration in African Americans

TO THE EDITOR:

Vitamin B₁₂ (VB₁₂), or cobalamin, is a water-soluble vitamin. It serves as an enzyme cofactor in the 1-carbon metabolism pathway, which plays a key role in a wide range of biological processes, including red blood cell (RBC) formation, DNA synthesis, and myelination of the central nervous system.^1-3  VB₁₂ deficiency is clinically associated with megaloblastic anemia and neurodegenerative disorders and is also linked to cardiovascular diseases, which is thought to be mediated via hyperhomocysteinemia.^1,3  Uptake of VB₁₂ from the diet in the gastrointestinal tract depends on intrinsic factor (encoded by GIF).^3,4  VB₁₂ in blood is bound to haptocorrin (HC; encoded by TCN1) and transcobalamin (encoded by TCN2) and circulates as holohaptocorrin (holoHC) and holotranscobalamin, respectively.^3,5,6 

Although VB₁₂ deficiency often results from poor gastrointestinal absorption,⁷  genetic factors also contribute to phenotypic variation. Previous genome-wide association studies (GWASs), primarily focusing on populations of European and Asian ancestry, have identified 13 loci (FUT2, CUBN, TCN1, MUT, MS4A3, FUT6, PRELID2, CLYBL, CD320, TCN2, ABCD4, MMAA, and MMACHC) associated with VB₁₂.^8-15  GWASs in Chinese¹¹  and Indians¹⁴  identified VB₁₂-associated shared genetic variants across ancestry groups, as well as population-specific variants. VB₁₂ and holoHC concentrations are higher in blacks than in whites,¹⁶  yet no GWAS has examined the genetic determinants of VB₁₂ specifically in African Americans (AAs). In the current study, we used whole-genome sequencing (WGS) from the National Heart, Lung, and Blood Institute (NHLBI) Trans-Omics for Precision Medicine (TOPMed) Program to assess genetic variants associated with VB₁₂ in an unselected population-based sample of AAs from the Jackson Heart Study (JHS).¹⁷ 

The JHS is a population-based longitudinal study based in Jackson, Mississippi^17,18 ; additional information is reported in supplemental Methods (available on the Blood Web site). The JHS was approved by the Institutional Review Board of the University of Mississippi Medical Center, and participants provided written informed consent. At the baseline examination, VB₁₂ was measured in plasma in a subset of 1851 JHS participants using a homogenous enzyme immunoassay system and Hitachi 911 equipment (Roche Diagnostics, Indianapolis, IN).¹⁹  A total of 3406 JHS participants underwent 30× coverage WGS through the TOPMed project at the Northwest Genome Center at the University of Washington, and genotype calling was performed by the Informatics Resource Center at the University of Michigan, as previously described.²⁰  After quality control (supplemental Methods), a total of 29 665 030 variants (minor allele frequency > 0.001) and 1280 participants (20.6-91.6 years of age, 65.5% female, VB₁₂ = 639.9 ± 289.3 pg/mL) were included in the discovery analysis (phs000964.v2.p1 for sequencing data and phs000286.v5.p1 for VB₁₂ data on the Database of Genotypes and Phenotypes; https://www.ncbi.nlm.nih.gov/gap).

Association of each single variant with rank-based inverse-normal–transformed VB₁₂ concentration was tested in JHS participants adjusted for age, sex, and the first 10 principal components (PCs) of genetic ancestry using a linear mixed model approach to account for familial relationships, as implemented in EPACTS 3.2.6 on the University of Michigan ENCORE server (https://encore.sph.umich.edu). Genome-wide significance was predefined as P < 1.69 × 10⁻⁹, or 0.05/29 665 030 variants tested; 33 variants across 2 genomic regions (TCN1 and FUT2) reached genome-wide significance (P < 1.69 × 10⁻⁹) for VB₁₂ (supplemental Figure 1).

The index variant rs34530014 on chromosome 11 is a 1-bp deletion of TCN1 (minor allele frequency = 0.036; P = 6.48 × 10⁻¹⁵; Table 1; Figure 1A) associated with lower VB₁₂. It is a loss-of-function (LoF) frameshift mutation (p.Val58Cysfs) prevalent only among African ancestry populations, originally reported in 2 Afro-Caribbean pedigrees with HC deficiency.²¹  Replication of the novel LoF variant TCN1-rs34530014 was performed using genomic data from 2 AA electronic medical record–linked biobanks: the Biobank of Vanderbilt University (BioVU)²²  and BioMe of Mount Sinai²³  (supplemental Methods). Because TCN1-rs34530014 was not directly genotyped or imputed in BioVU or BioMe, we used its linkage disequilibrium (LD) proxy rs11822978 (r² = 0.98 in JHS) and confirmed its association with lower VB₁₂ in 3924 additional AAs from BioVU and BioMe (Table 1). TCN1-rs34530014 is distinct from other VB₁₂-associated TCN1 variants (including 2 missense variants, rs34324219 and rs34528912) identified in Europeans and Indians.^9,12-14  Of the TCN1 variants previously associated with lower VB₁₂, only rs34324219 was nominally associated with lower VB₁₂ in the JHS (P = .009; supplemental Table 1). Together, these results demonstrate an ancestry-specific role for rs34530014 in lower VB₁₂ concentration in African ancestry populations and substantial allelic heterogeneity in the genetic architecture of VB₁₂ at the TCN1 locus.

The identification of this naturally occurring LoF mutation prevalent in AAs allowed us to further investigate additional hematologic and clinical consequences of mild HC deficiency, beyond the association with lower VB₁₂ concentration. The association of rs34530014 with additional quantitative traits, including RBC indices and homocysteine, was assessed in the JHS and the Women’s Health Initiative SNP Health Association Resource²⁴  (supplemental Methods). An elevated concentration of methylmalonate is a more sensitive and specific marker for VB₁₂ deficiency^1,3 ; unfortunately, it was not available in our data sets. In addition, a phenome-wide association study (PheWAS) of TCN1-rs11822978 was performed in BioVU and BioMe. A total of 340 traits was scanned for association using logistic regression adjusted for age, sex, and the first 10 PCs, including metabolic, neurological, and hematologic outcomes and disorders. Individual study association results were combined using fixed-effect inverse-variance–weighted meta-analysis, and P values were corrected for multiple testing. Although marginal associations (P < .05) were observed for mean corpuscular hemoglobin and mean corpuscular volume, as well as diagnosis codes Vitamin Deficiency and Other Deficiency Anemia, none of these reached the Bonferroni-corrected threshold (P > .006, or 0.05/8 traits for RBC indices, supplemental Table 2; and P > 1.47 × 10⁻⁴, or 0.05/340 traits for PheWAS, supplemental Table 3).

In contrast to mutations of TCN2, which result in transcobalamin II deficiency (OMIM #275350) with severe clinical manifestations,²⁵  the clinical impact of HC deficiency is poorly defined. The distinct consequences of TCN1 and TCN2 mutations are consistent with the fact that holotranscobalamin is the bioactive form of VB₁₂ that is taken up by all tissues, whereas the biological role of holoHC is unclear.⁶  Despite the relatively large overall sample, the small number of homozygotes (4 in JHS) limited our ability to accurately assess the recessive effect of the TCN1-rs34530014 on VB₁₂ and related traits and disorders. Therefore, further investigation, including very large phenotypic data sets of African ancestry individuals, will be needed to draw firm conclusions about the clinical impact of HC deficiency.

The index single nucleotide polymorphism at the other genome-wide significant association peak in the JHS on chromosome 19 was rs507766 (located in the 3′ untranslated region of FUT2; P = 5.51 × 10⁻¹¹) (Figure 1B). This noncoding variant is in only moderate LD (r² = 0.55 in JHS) with the common secretor variant FUT2-rs601338 (P = 1.49 × 10⁻⁷ for association with VB₁₂ in JHS), which was previously identified in Europeans and Indians.^8,9,14  Thus, we found evidence for 2 functional alleles involving FUT2: 1 with an additive genetic effect (rs507766) and 1 with a dominant effect (rs601338) on VB₁₂ (supplemental Table 4). We confirmed 4 previously identified loci (MUT, MMACHC, FUT6, and CD320; supplemental Table 1) at P < .05 with consistent association directions of effects.

We report a novel association between the African ancestry–specific LoF variant TCN1-rs34530014 and VB₁₂ concentration and an additional FUT2 variant rs507766 associated with VB₁₂ specific to AAs, highlighting the ancestry heterogeneity of genetic factors that influence VB₁₂. Another population-specific VB₁₂-associated LoF variant, CLYBL-rs41281112 (p.Arg259Ter), is found in ∼3% of apparently healthy non-African individuals (but is rare in Africans).¹¹  Further elucidation of the genetics and role of HC-bound VB₁₂ may lead to improved diagnosis and differentiation of true bioavailable VB₁₂ deficiency from total VB₁₂ deficiency.