rs1906252 — POU3F2

POU3F2 6q16.1 — A Regulatory Locus for Cortical Neuron Development and Cognitive Ability

One of the most consistently replicated loci in the genetics of human cognition sits on chromosome 6q16.1. The rs1906252 variant is located within a large uncharacterised non-coding RNA at chr6:98,102,413 (GRCh38), approximately 730 kilobases upstream of POU3F2¹¹ POU3F2
POU class 3 homeobox 2, also known as BRN2 — a transcription factor expressed almost exclusively in the brain that controls the differentiation of upper-layer cortical projection neurons. The gap in linear distance is not unusual for regulatory biology: enhancers and regulatory elements routinely act across hundreds of kilobases via chromatin looping, and the 6q16.1 region harbours several candidate regulatory elements including the brain-expressed microRNA MIR2113²² MIR2113
microRNA 2113, located ~78 kb from rs1906252, which may modulate POU3F2 expression post-transcriptionally.

POU3F2 (BRN2) is essential for the development of cortical layer II–IV projection neurons — the neurons that underlie associative, executive, and higher-order cognitive functions. Mouse knockout experiments demonstrate that loss of POU3F2 produces severe cortical architectural defects, loss of upper-layer neurons, and impaired cognition in memory and learning tasks. The observation that common genetic variation at this locus predicts cognitive performance in humans suggests that the A allele fine-tunes POU3F2 expression during development in a way that produces marginally more efficient cortical organisation.

This SNP is related to rs9320913, another POU3F2 6q16.1 locus variant (~34 kb away) that was the original lead SNP at this locus from the Rietveld 2013 Science educational attainment GWAS. rs1906252 and rs9320913 likely tag overlapping functional elements in this regulatory region.

The Mechanism

The A allele at rs1906252 is the minor allele globally (gnomAD v4: ~39%), though it approaches 50% in European populations (~49%), suggesting drift or mild positive selection in that ancestry group. Each copy of the A allele is associated with a small linear increase in general cognitive ability (g) — the direction is consistent across the allele dose series CC → AC → AA.

The molecular mechanism is not resolved. The variant sits within a 1.2-megabase lncRNA with no characterised function, in a region containing MIR2113. Functional genomics data from adult brain tissue show active regulatory chromatin in this region, consistent with an enhancer role, but no published reporter assay has directly tested the allelic effect of rs1906252 on POU3F2 or MIR2113 expression. The causal variant at this locus may be rs1906252 itself or a nearby variant in strong LD.

The Evidence

Trampush et al. 2015³³ Trampush et al. 2015
American Journal of Medical Genetics B moved this locus beyond genome-wide significance for direct cognitive ability — not merely educational attainment proxy phenotype — by combining the COGENT cognitive GWAS cohort with educational attainment summary statistics in 68,159 individuals, yielding P = 1.65×10⁻⁹ for g. Crucially, the direction was unambiguous: each A allele linearly increased g scores.

Davies et al. 2016⁴⁴ Davies et al. 2016
Molecular Psychiatry confirmed the region in UK Biobank (N = 112,151), identifying 6q16.1 among 20 genome-wide significant regions for cognitive function. Lee et al. 2018⁵⁵ Lee et al. 2018
Nature Genetics — the largest educational attainment GWAS to date, N = 1.1 million — included 1,271 genome-wide significant loci, with pathway analysis pointing to brain-expressed genes in neurodevelopmental pathways.

Hashizume et al. 2018⁶⁶ Hashizume et al. 2018
Genes Brain Behav provided the functional bridge: POU3F2 knockout mice exhibit impaired object recognition, spatial memory, and reduced adult hippocampal neurogenesis, establishing POU3F2 as a critical regulator of memory and learning in the postnatal brain — making it biologically plausible that variation in POU3F2 regulatory elements shapes cognitive differences in humans.

The effect size per A allele is approximately 0.03–0.05 standard deviations in g — small in individual terms but highly significant at population scale. AA homozygotes are not measurably smarter than CC homozygotes in everyday functioning; the signal emerges only in adequately powered population studies.

Practical Actions

For AC and AA carriers, no intervention changes the genotype — the benefit is already present. The practical value is understanding that this locus reflects POU3F2-dependent cortical circuitry, which requires adequate neurotrophic and nutritional support to fully express its potential. Choline is the rate-limiting precursor for cortical acetylcholine neurotransmitter synthesis and phosphatidylcholine membrane production; DHA is the primary structural omega-3 in cortical neuron membranes. Ensuring adequate supply of both specifically supports the POU3F2-specified upper-layer cortical neurons that this locus implicates.

For CC carriers, no pharmacological intervention modifies this variant, but the absence of the A allele simply means this particular locus contributes less to cognitive reserve. The same nutritional considerations apply — arguably more so, since CC individuals have less genetic headroom from this particular pathway.

Interactions

rs1906252 at 6q16.1 and rs9320913 at the same locus likely tag overlapping regulatory elements upstream of POU3F2; individuals with risk alleles at both may have a compounded reduction in POU3F2 pathway tone, but this has not been formally studied. At the pathway level, POU3F2-dependent cortical neurons are the downstream targets of BDNF signalling (rs6265, Val66Met) and dopaminergic modulation (COMT rs4680); individuals carrying risk alleles at those SNPs alongside CC at rs1906252 would have compounded reductions in cortical circuit efficiency, though no published study has modelled the combined genetic effect.