Night owl genes: what CRY1, PER3 and 351 chronotype loci can tell you

Chronotype is about half inherited

Chronotype is usually measured by asking. The Morningness-Eveningness Questionnaire scores the times people would choose for sleep, work and exercise; large genetic studies use a single question, whether you are a morning or an evening person^{[3, 2]}. Age, sex and the light people live in account for a large part of the variation, and genes for part of it^[2].

Twins give the first estimate of how large that genetic part is. In 1981 the Older Finnish Twin Cohort answered a four-step question on diurnal type: 2,836 identical and 5,917 non-identical same-sex pairs. Genetic effects accounted for 49.7 percent of the differences between people, with the rest put down to environments the twins did not share. Most of the genetic share, 38.0 percent, came from dominance, the way the two copies of a gene combine, and 11.7 percent from additive effects^[1]. Other twin studies range from 37 to 54 percent^[3].

351 regions from 697,828 people

Earlier genome-wide studies, of up to 128,286 people, had found 24 chronotype loci. In 2019 Jones and colleagues combined 449,734 UK Biobank participants with 248,098 research participants of 23andMe, all of European ancestry, and found 351^[2]. UK Biobank asked one question with five answers, and the spread of answers is the trait the study was measuring.

Among the strongest signals were genes of the body clock itself: PER2, PER3, FBXL3, and for the first time PER1, CRY1 and ARNTL. The loci were also enriched for genes active in the retina, the hypothalamus and the pituitary, and two of them, RGS16 and INADL, act in the retinal cells that report light to the brain's master clock^[2]. Part of being a night owl may be how strongly the eye passes on daylight.

Self-report can mislead, so the study checked the loci against wrist activity monitors worn by 85,760 participants. The variants moved the timing of sleep and showed little link with its length or quality. The 5 percent of people carrying the most morning-linked variants had the middle of their five least active hours at 03:06, against 03:32 for the 5 percent carrying the fewest: 25.1 minutes earlier on average^[2].

Counting only the common variants that genotyping chips capture, chronotype heritability in UK Biobank was 13.7 percent, at the lower end of earlier estimates of 12 to 21 percent^[2]. The gap between that and the twin figure is usual for complex traits; our guide to polygenic scores in plain words explains where it comes from.

What a chronotype score explains

A polygenic score adds up a person's morning- or evening-linked variants, each weighted by its effect. Maukonen and colleagues built one from 313 of the 351 lead variants and tested it in 8,433 Finns of the FINRISK 2007 and 2012 surveys, who had answered both the single question and six items of the full questionnaire. A higher score went with being an evening type on both measures. It explained 1.39 percent of the questionnaire score and 1.89 percent of the single-question answer^{[3, 5]}.

None of the 313 variants on its own was linked to chronotype in the Finnish sample, although most pointed the same way as in the large study^[3]. That is what a score of tiny effects looks like in 8,000 people: the signal exists only in the sum.

CRY1 and PER3: the family variants

The night-owl gene of the headlines is a change in CRY1, one of the brakes of the molecular clock. Patke and colleagues studied a 46-year-old woman with delayed sleep phase disorder. In the laboratory her evening melatonin rise came at 2:32 am, where 8 to 10 pm is typical, and during 14 days without clocks or daylight her sleep-wake rhythm ran on a 24.5-hour cycle, against 24.2 hours in a control subject^[6].

She carried c.1657+3A>C (rs184039278), a change at a splice site that makes cells skip exon 11, 72 letters long, and build CRY1 without 24 of its amino acids. The shortened protein bound the clock's activators more tightly and lengthened the cycle in cell experiments. Late sleep ran through her family in a dominant pattern, and in six unrelated Turkish families of 70 people, 38 carriers reported disturbed sleep, late or broken, while their non-carrier relatives did not^[6]. Some of the same researchers later linked the variant to ADHD, starting from families selected for having both ADHD and insomnia^[7].

The early-bird counterpart is a pair of changes in PER3, P415A and H417R, inherited together. They were found in people with familial advanced sleep phase who also scored higher on questionnaires about low and seasonal mood. Mice carrying the human version shifted their sleep-wake cycle when days were kept short^[8].

Families come to attention because their members are extreme, so a variant can look stronger there than in carriers found at random. In 2022 Weedon and colleagues looked up the reported sleep variants in up to 191,929 people with sleep data and sequence data, from UK Biobank, two Finnish cohorts and a US study. In UK Biobank none was linked to a recorded sleep disorder or to sleep medication. Only CRY1 and PER3 moved sleep timing at all^[4].

Both effects are small. The PER3 carriers' higher odds of calling themselves definite morning people, 1.36, passed genome-wide significance; the CRY1 difference in sleep midpoint just missed the usual 0.05 threshold. In the same study, 64 people carrying one of 50 different PER2 variants that cut the protein short had a sleep midpoint about 57 minutes earlier, the one large effect the population search turned up^[4].

MEIS1: restless legs, broken sleep and a little eveningness

Restless legs syndrome is an urge to move the legs, with unpleasant sensations, that comes on at rest and is worse in the evening and at night; it affects up to 10 percent of people of European ancestry^[9]. The first genome-wide study of it, in 2007, found variants in MEIS1, a developmental gene, among three regions that each raised risk by more than 50 percent^[10].

A 2017 meta-analysis of 15,126 cases and 95,725 controls, replicated in 30,770 cases and 286,913 controls, confirmed MEIS1 as the strongest genetic factor. Its lead variant, rs113851554 in intron 8, carried an odds ratio of 1.92 per copy^[9]. The same allele turns up in insomnia studies with odds ratios of 1.14 to 1.27, and in the 2019 chronotype study, where its other allele went with slightly higher odds of being a morning person, 1.06^[11]. Part of the insomnia signal may be people kept awake by their legs.

Finland: a rare CRY1, a common PER3

The three variants are spread very unevenly. In gnomAD's genome data, the PER3 pair is about five times as common in Finns as in other Europeans, and the CRY1 variant about 16 times rarer, with 3 copies in 10,626 Finnish chromosomes. MEIS1's restless-legs allele is also more common in Finland^{[12, 13, 14]}.

Common in Finland did not mean visible in Finns. In the FINRISK and Health 2000 to 2011 cohorts, 22.8 percent of the 149 PER3 carriers called themselves definite morning people, against 22.4 percent of non-carriers. Fewer than five CRY1 carriers were found, and none of them answered definitely evening^[4]. Both samples were small.

Light matters too. One test of latitude comes from Brazil, where 12,884 people in one time zone, from the equator to 32 degrees south, became more evening-type with distance from the equator^[15]. That study stopped far short of Finland's 60 to 70 degrees north, and none of the studies cited here tests latitude within Finland. Among 18,087 Finnish men in four surveys, 9.5 percent were evening types after age standardisation, and morning types were less common in the 2000s than two decades earlier^[16].

One older candidate still circulates. A CLOCK variant, 3111T/C (rs1801260), was linked to eveningness in candidate-gene studies that were small and inconsistent^[3], but the GWAS Catalog lists no genome-wide association for it with any trait^[17].

What the Aimosti report shows

Chronotype appears in three places in the report. The Traits section, in the well-supported tier, reads rs12927162 near CASC16, a locus found in the 23andMe study of 89,283 people and again in the 2019 study, and labels the genotype a morning, intermediate or evening tendency^[18]. Below the credibility firewall, the emerging scores include the 311-variant chronotype score, which ships 311 of PGS000336's 313 variants and ranks your file against a European reference^[5]. A higher percentile means a tendency towards evening, and the card states that it explains roughly 1 to 2 percent of chronotype.

The Frontier, in the same exploratory tier, holds the single-gene cards: CRY1 (labelled mixed evidence), PER3 and MEIS1 (replicated, small effect) and CLOCK 3111T/C (contested). Two short-sleeper cards and a caffeine-and-sleep card sit in the same group; the caffeine side is in our guide to caffeine and CYP1A2. Whether a card can be read depends on the file.

No card turns a genotype into advice about sleep or a sleep diagnosis, and none of them says anything about mood: the PER3 card notes that its family's mood scores have not been seen in the general population. Where a marker could not be read, the report says so instead of showing the common genotype as a result. Why that distinction matters is in what a missing VCF record means.