What an MTHFR AA Genotype Call Means

If a report hands you “MTHFR AA,” it almost always means rs1801133 with two copies of the minor allele, which is identical to what the older literature calls 677TT or homozygous C677T. The difference is bookkeeping. MTHFR sits on the minus strand of chromosome 1, and the classical C→T nomenclature is written in the orientation of the coding sequence. Forward genomic orientation is used by dbSNP, reference FASTA files, and most modern VCFs. In that orientation the same base change is written G→A, so AA equals TT equals homozygous 222Val. Likewise, GG equals CC equals homozygous reference. The protein consequence is p.Ala222Val either way. Nothing about your biology changed when the letters did.

That single point resolves most of the confusion people arrive with, but it also creates a real hazard: a strand-ambiguous call can be silently wrong, and a genotype you cannot verify is not worth reasoning about. The rest of this page covers how to check the call in raw data and what the reduced-activity claim rests on biochemically. It also covers why two SNPs are a poor summary of a gene with hundreds of coding variants and which measurements tell you something the genotype cannot.

Verifying the call in your own data

The two variants people ask about have stable coordinates. In GRCh38, rs1801133 is the forward-strand G>A change at chr1:11,796,321. It is also written as c.665C>T, p.Ala222Val, or the “C677T” of the older amino-acid numbering. Likewise, rs1801131 is the forward-strand T>G change at chr1:11,794,419. It is also written as c.1286A>C, p.Glu429Ala, or the “A1298C”. In GRCh37 the same sites are chr1:11,856,378 and chr1:11,854,476. If you have a genome, pull them by position rather than by rsID, because rsID annotation depends on whichever dbSNP build your pipeline used:

bcftools view -H -r chr1:11796321,chr1:11794419 sample.vcf.gz | cut -f1-5,10

Two checks matter more than the genotype string itself. First, confirm that REF matches the reference base at that position, which catches lifted-over or mis-normalized files:

bcftools norm -c ws -f GRCh38.primary_assembly.fa -Ou sample.vcf.gz | \
  bcftools view -H -r chr1:11796321,chr1:11794419

Second, look at depth and genotype quality in the FORMAT fields. A 1/1 call at DP 6 on a 30x genome is a coin flip. At these two sites on a well-covered short-read genome you should see DP in the 20s or higher and GQ at ceiling. If you sequenced at low coverage with imputation, treat the call as a posterior probability and check the imputation INFO or dosage field rather than the hard call.

Array data is where strand errors live. A 23andMe or AncestryDNA raw text file uses GRCh37 coordinates and reports rs1801133 on the plus strand, so the line reads something like rs1801133 1 11856378 AA. Because this site is an A/G polymorphism, it is not strand-ambiguous in the A/T or C/G sense, which is why the flip is recoverable by inspection. The site rs1801131 is reported as T/G on the plus strand, so a “GG” there is 1298CC. If you are running your own PLINK pipeline, set alleles from the reference rather than trusting the manifest: plink2 --vcf sample.vcf.gz --ref-from-fa --fa GRCh38.fa. Reserve --flip for lists you have independently confirmed.

What “reduced activity” rests on

Methylenetetrahydrofolate reductase converts 5,10-methylenetetrahydrofolate to 5-methyltetrahydrofolate, the methyl donor that remethylates homocysteine to methionine. The 222Val protein is thermolabile: it loses activity on mild heating and it binds its flavin adenine dinucleotide (FAD) cofactor less tightly, so the functional deficit depends on how much FAD and folate are around. In vitro, homozygous 222Val cells run at roughly a third of reference activity and heterozygotes around two thirds, and the enzyme is stabilized by 5-methyltetrahydrofolate binding. Resequencing and expression studies of MTHFR allozymes show the same pattern for many variants: reduced activity tracks with reduced protein stability rather than a broken active site ¹. That mechanism is why cofactor status matters more than genotype status.

The population framing also matters. The 677T allele is common, roughly a quarter to a third of chromosomes in European ancestries and higher in several Latin American populations, which puts homozygotes in the range of ten percent or more of many populations. A variant that common is a modifier whose measurable effect on plasma homocysteine appears mainly when folate intake is low and largely flattens at adequate folate status. Such a variant is not a disease. In populations with fortified grain, the average homocysteine difference between AA and GG is small.

Two SNPs are a bad summary of MTHFR

The bigger limitation of an “MTHFR AA” report is coverage. Orientation does not matter as much. MTHFR has hundreds of coding variants, and a deep mutational scan of the gene assayed missense effects across a range of folate and cofactor conditions, finding that many variants are conditionally damaging: their functional deficit appears at low folate and is rescued at high folate ². A separate survey of rare MTHFR coding variants in humans found that a meaningful fraction reduce enzyme activity in vitro and that activity is restored by supplementing the flavin cofactor, which is the same folate-remedial mechanism operating on rarer alleles than 677T ³. Targeted resequencing studies of the gene in clinical cohorts have similarly turned up variants outside the two famous positions ⁴.

The practical consequence for someone with sequence data: do not stop at two rsIDs. Extract the whole locus and look at every coding change.

bcftools view -r chr1:11,785,000-11,806,000 sample.vcf.gz -Oz -o mthfr.vcf.gz
vep -i mthfr.vcf.gz --cache --assembly GRCh38 --everything \
    --pick_allele_gene --transcript_filter "stable_id match ENST00000376592" \
    --tab -o mthfr.vep.tsv

Then read the output with the right priors. MODERATE impact missense calls in MTHFR are mostly common and benign. What deserves attention is a rare missense (gnomAD allele frequency below roughly 0.001) or a frameshift, stop-gain, or canonical splice call, especially in trans with 677T. Severe MTHFR deficiency is a recognized inborn error of metabolism caused by loss-of-function alleles, and it looks nothing like the common polymorphism. If your VCF shows a rare loss-of-function allele in this gene, that is a finding for a clinical geneticist and a confirmatory assay. A home spreadsheet is not the place for it.

Measure the pathway instead of inferring it

Genotype at rs1801133 explains a small share of variance in the thing you care about, which is whether one-carbon metabolism is running well. The pathway has direct readouts, and they are inexpensive. Plasma total homocysteine is the integrative marker of remethylation capacity. Serum folate reflects recent intake while red blood cell folate reflects roughly the prior three months. Vitamin B12 is the other cofactor for methionine synthase, and because serum B12 is a noisy measure, methylmalonic acid is the more specific functional indicator of B12 sufficiency. Riboflavin status matters specifically for 222Val carriers because riboflavin is the precursor of the FAD cofactor the variant enzyme binds weakly.

Our view is to order homocysteine, RBC folate, B12, and methylmalonic acid together, once, and then interpret the genotype in that context rather than the reverse. An AA genotype with homocysteine in the single digits and RBC folate comfortably mid-range tells you the variant is well compensated in your current diet. The same genotype with elevated homocysteine and low RBC folate is a nutritional question with a genetic modifier attached. The interpretation, including whether any supplementation is appropriate for you, belongs with a physician or a registered dietitian. Nutrient status genuinely modulates gene-level effects, which is the premise of nutrigenomic work, and also the reason single-gene diet prescriptions from a consumer report overreach the evidence ⁵. Repeating the measurements is what turns a static genotype into something you can act on with a clinician: one homocysteine value has substantial biological and analytical variability, and a trend over a year is far more informative.

Questions people also ask

Does MTHFR come from mom or dad? Both. MTHFR is autosomal, on chromosome 1, so you carry one copy from each parent. An AA (677TT) result means both parents transmitted the A allele, so each is at least a heterozygote. Phasing tools on your own genome cannot tell you which allele came from which parent without parental data.

Are eggs bad for MTHFR? No. Eggs are a dietary source of choline and betaine, which feed the alternative betaine-homocysteine methyltransferase route for remethylating homocysteine, a pathway that is independent of MTHFR. There is no mechanism by which egg intake is specifically harmful to 677T carriers.

What foods should people with MTHFR avoid? No food is contraindicated by this genotype. The common claim that carriers must avoid folic acid is not supported by the enzymology: folic acid is reduced to tetrahydrofolate upstream of the MTHFR step, and carriers metabolize it. Whether methylfolate offers any advantage over folic acid for you is a question for a clinician, and the answer depends on your measured folate and homocysteine. Your genotype alone does not settle it.

What is high when you have MTHFR? Plasma total homocysteine is the marker that can run high, and mainly when folate or B12 status is marginal. Many AA homozygotes with adequate folate have entirely normal homocysteine.

What are signs of having the MTHFR gene? Everyone has the MTHFR gene, and the common 677T variant has no reliable symptoms or physical signs. Fatigue, brain fog, and similar complaints attributed to it online have many causes and are not diagnostic of any genotype. Variant status is established by genotyping and its metabolic consequence by blood measurement. Symptoms do not establish either.

Footnotes

1. Yvette N. Martin, Oreste E. Salavaggione, Bruce W. Eckloff, et al. Human methylenetetrahydrofolate reductase pharmacogenomics: gene resequencing and functional genomics. Pharmacogenetics and Genomics, 2006. https://doi.org/10.1097/01.fpc.0000194423.20393.08 ↩

2. Jochen Weile, Nishka Kishore, Song Sun, et al. Shifting landscapes of human MTHFR missense-variant effects. The American Journal of Human Genetics, 2021. https://doi.org/10.1016/j.ajhg.2021.05.009 ↩

3. Nicholas J. Marini, Jennifer Gin, Janet Ziegle, et al. The prevalence of folate-remedial MTHFR enzyme variants in humans. Proceedings of the National Academy of Sciences, 2008. https://doi.org/10.1073/pnas.0802813105 ↩

4. Chiamaka N. Aneji, Hope Northrup, Kit Sing Au. Deep sequencing study of the MTHFR gene to identify variants associated with myelomeningocele. Birth Defects Research Part A: Clinical and Molecular Teratology, 2012. https://doi.org/10.1002/bdra.22884 ↩

5. Milan Kumar Lal, Eshita Sharma, Rahul Kumar Tiwari, et al. Nutrient-Mediated Perception and Signalling in Human Metabolism: A Perspective of Nutrigenomics. International Journal of Molecular Sciences, 2022. https://doi.org/10.3390/ijms231911305 ↩