Ehlers-Danlos syndrome: genes and variants
Explore variant evidence for Ehlers-Danlos syndrome across 11 analyzed proteins (COL3A1, COL1A2, TGFBR2, TNXB, TGFBR1 and 6 more). Linked ClinVar records include 207 pathogenic or likely pathogenic variants, 808 variants of uncertain significance and 180 with conflicting classifications.
Coverage includes proteins already analyzed in CATVariant, not every gene involved in this condition. Database links are associations, not an assessment of clinical gene–disease validity. Computable evidence prioritizes variants for expert review and does not reclassify them. Source labels are pooled across this disease family.
Data updated 2026-11-05. Automated aggregation, not a clinical review date.
Download variant evidence (CSV)
Genes linked to Ehlers-Danlos syndrome
COL3A1: Collagen alpha-1(III) chain
Its type III collagen fibrils provide tensile support in arteries, bowel, uterus, skin, and other distensible connective tissues. Pathogenic variants cause vascular Ehlers-Danlos syndrome with marked arterial and hollow-organ fragility.
196 ClinVar pathogenic / likely pathogenic and 812 uncertain variants in COL3A1 have source records linked to Ehlers-Danlos syndrome. Association strength is not clinical gene validity.
COL1A2: Collagen alpha-2(I) chain
It contributes one of the three chains of type I collagen, providing tensile strength to bone, skin, tendon, blood vessels, and other connective tissues. Pathogenic variants can cause osteogenesis imperfecta, Ehlers-Danlos phenotypes, and related connective-tissue disorders.
4 ClinVar pathogenic / likely pathogenic and 13 uncertain variants in COL1A2 have source records linked to Ehlers-Danlos syndrome. Association strength is not clinical gene validity.
TGFBR2: TGF-beta receptor type-2
It binds TGF-beta ligands and activates TGFBR1 to initiate canonical and noncanonical signaling. Germline pathogenic variants cause Loeys-Dietz syndrome type 2, while somatic loss can remove growth-suppressive TGF-beta responses in cancer.
4 ClinVar pathogenic / likely pathogenic and 4 uncertain variants in TGFBR2 have source records linked to Ehlers-Danlos syndrome. Association strength is not clinical gene validity.
TNXB: Tenascin-X
Tenascin-X is an extracellular-matrix glycoprotein that supports collagen organization and tissue elasticity. Biallelic variants can cause classical-like Ehlers-Danlos syndrome.
1 ClinVar pathogenic / likely pathogenic and 66 uncertain variants in TNXB have source records linked to Ehlers-Danlos syndrome. Association strength is not clinical gene validity.
TGFBR1: TGF-beta receptor type-1
After activation by the ligand-bound receptor complex, it phosphorylates SMAD2 and SMAD3 to propagate TGF-beta signals. Germline pathogenic variants cause Loeys-Dietz syndrome type 1 with arterial aneurysm and dissection and variable craniofacial or skeletal features.
1 ClinVar pathogenic / likely pathogenic and 8 uncertain variants in TGFBR1 have source records linked to Ehlers-Danlos syndrome. Association strength is not clinical gene validity.
SMAD3: SMAD family member 3
It carries TGF-beta receptor signals into the nucleus to control extracellular-matrix, differentiation, and growth programs. Heterozygous loss-of-function variants cause Loeys-Dietz syndrome type 3 with arterial aneurysm and dissection risk and often early osteoarthritis.
1 ClinVar pathogenic / likely pathogenic and 7 uncertain variants in SMAD3 have source records linked to Ehlers-Danlos syndrome. Association strength is not clinical gene validity.
COL5A1: Collagen alpha-1(V) chain
It helps nucleate and regulate type I collagen fibril assembly, controlling fibril diameter and connective-tissue architecture. Haploinsufficiency or structural variants are a major cause of classical Ehlers-Danlos syndrome, with skin hyperextensibility, atrophic scarring, and joint hypermobility.
0 ClinVar pathogenic / likely pathogenic and 25 uncertain variants in COL5A1 have source records linked to Ehlers-Danlos syndrome. Association strength is not clinical gene validity.
COL5A2: Collagen alpha-2(V) chain
It partners with COL5A1-derived chains to regulate collagen fibril formation in skin, tendons, and other connective tissues. Pathogenic variants can cause classical Ehlers-Danlos syndrome with tissue fragility, hyperextensible skin, and joint hypermobility.
0 ClinVar pathogenic / likely pathogenic and 16 uncertain variants in COL5A2 have source records linked to Ehlers-Danlos syndrome. Association strength is not clinical gene validity.
COL1A1: Collagen alpha-1(I) chain
The alpha-1 chain of type I collagen, the main fibrillar collagen in connective tissue, bone, and skin. Together with its partner chain, it forms strong extracellular fibers, and COL1A1 variants are associated with osteogenesis imperfecta and several Ehlers-Danlos syndromes.
0 ClinVar pathogenic / likely pathogenic and 25 uncertain variants in COL1A1 have source records linked to Ehlers-Danlos syndrome. Association strength is not clinical gene validity.
ATP7A: Copper-transporting ATPase 1
It delivers copper to secretory-pathway enzymes and exports excess copper from cells, making it essential for systemic copper distribution. Loss-of-function variants cause Menkes disease or occipital horn syndrome, while some hypomorphic alleles produce distal motor neuropathy.
0 ClinVar pathogenic / likely pathogenic and 6 uncertain variants in ATP7A have source records linked to Ehlers-Danlos syndrome. Association strength is not clinical gene validity.
TGFB2: Transforming growth factor beta-2 proprotein
Its secreted signaling regulates extracellular matrix, cell differentiation, proliferation, and cardiovascular development. Haploinsufficiency causes a Loeys-Dietz-spectrum connective-tissue disorder with increased risk of thoracic aortic aneurysm and dissection.
0 ClinVar pathogenic / likely pathogenic and 4 uncertain variants in TGFB2 have source records linked to Ehlers-Danlos syndrome. Association strength is not clinical gene validity.
Weakly linked (only a few uncertain records): FBN2, COL6A1, RECQL4 and SMARCA2.
ClinVar pathogenic and likely pathogenic variants linked to Ehlers-Danlos syndrome
| Variant | Position | Protein part | Clinical label |
|---|---|---|---|
| COL3A1 G252C | 252 | Triple-helical region | Pathogenic / likely pathogenic (★★) |
| COL3A1 G360D | 360 | Triple-helical region | Pathogenic / likely pathogenic (★★) |
| COL3A1 G588S | 588 | Triple-helical region | Pathogenic / likely pathogenic (★★) |
| COL3A1 G744C | 744 | Triple-helical region | Pathogenic / likely pathogenic (★★) |
| COL3A1 G756E | 756 | Triple-helical region | Pathogenic / likely pathogenic (★★) |
| COL3A1 G897S | 897 | Triple-helical region | Pathogenic / likely pathogenic (★★) |
| COL3A1 G177V | 177 | Triple-helical region | Pathogenic / likely pathogenic (★★) |
| COL3A1 G252S | 252 | Triple-helical region | Pathogenic / likely pathogenic (★★) |
| COL3A1 G276S | 276 | Triple-helical region | Pathogenic / likely pathogenic (★★) |
| COL3A1 G318D | 318 | Triple-helical region | Pathogenic / likely pathogenic (★★) |
| COL3A1 G414D | 414 | Triple-helical region | Pathogenic / likely pathogenic (★★) |
| COL3A1 G414V | 414 | Triple-helical region | Pathogenic / likely pathogenic (★★) |
| COL3A1 G435D | 435 | Triple-helical region | Pathogenic / likely pathogenic (★★) |
| COL3A1 G435S | 435 | Triple-helical region | Pathogenic / likely pathogenic (★★) |
| COL3A1 G441R | 441 | Triple-helical region | Pathogenic / likely pathogenic (★★) |
| COL3A1 G570D | 570 | Triple-helical region | Pathogenic / likely pathogenic (★★) |
| COL3A1 G570S | 570 | Triple-helical region | Pathogenic / likely pathogenic (★★) |
| COL3A1 G588V | 588 | Triple-helical region | Pathogenic / likely pathogenic (★★) |
| COL3A1 G609E | 609 | Triple-helical region | Pathogenic / likely pathogenic (★★) |
| COL3A1 G612S | 612 | Triple-helical region | Pathogenic / likely pathogenic (★★) |
| COL3A1 G612A | 612 | Triple-helical region | Pathogenic / likely pathogenic (★★) |
| COL3A1 G984R | 984 | Triple-helical region | Pathogenic / likely pathogenic (★★) |
| COL3A1 G1068C | 1068 | Triple-helical region | Pathogenic / likely pathogenic (★★) |
| COL1A2 G613R | 613 | Pathogenic / likely pathogenic (★★) | |
| COL1A2 G901S | 901 | Pathogenic / likely pathogenic (★★) | |
| COL1A2 G1084C | 1084 | Pathogenic / likely pathogenic (★★) | |
| COL3A1 G213D | 213 | Triple-helical region | Pathogenic / likely pathogenic (★★) |
| COL3A1 G342R | 342 | Triple-helical region | Pathogenic / likely pathogenic (★★) |
| COL3A1 G378D | 378 | Triple-helical region | Pathogenic / likely pathogenic (★★) |
| COL3A1 G396D | 396 | Triple-helical region | Pathogenic / likely pathogenic (★★) |
| COL3A1 G405R | 405 | Triple-helical region | Pathogenic / likely pathogenic (★★) |
| COL3A1 G666S | 666 | Triple-helical region | Pathogenic / likely pathogenic (★★) |
| COL3A1 E682K | 682 | Triple-helical region | Pathogenic / likely pathogenic (★★) |
| COL3A1 G819D | 819 | Triple-helical region | Pathogenic / likely pathogenic (★★) |
| COL3A1 G822S | 822 | Triple-helical region | Pathogenic / likely pathogenic (★★) |
| COL3A1 G876A | 876 | Triple-helical region | Pathogenic / likely pathogenic (★★) |
| COL1A2 G913S | 913 | Pathogenic / likely pathogenic (★★) | |
| COL3A1 G177A | 177 | Triple-helical region | Pathogenic / likely pathogenic (★★) |
| COL3A1 E241K | 241 | Triple-helical region | Pathogenic / likely pathogenic (★★) |
| COL3A1 G273E | 273 | Triple-helical region | Pathogenic / likely pathogenic (★★) |
| COL3A1 G432D | 432 | Triple-helical region | Pathogenic / likely pathogenic (★★) |
| COL3A1 G663S | 663 | Triple-helical region | Pathogenic / likely pathogenic (★★) |
| COL3A1 G684E | 684 | Triple-helical region | Pathogenic / likely pathogenic (★★) |
| COL3A1 G708D | 708 | Triple-helical region | Pathogenic / likely pathogenic (★★) |
| COL3A1 G750R | 750 | Triple-helical region | Pathogenic / likely pathogenic (★★) |
| COL3A1 G834R | 834 | Triple-helical region | Pathogenic / likely pathogenic (★★) |
| COL3A1 G891R | 891 | Triple-helical region | Pathogenic / likely pathogenic (★★) |
| COL3A1 G918R | 918 | Triple-helical region | Pathogenic / likely pathogenic (★★) |
| COL3A1 G1038R | 1038 | Triple-helical region | Pathogenic / likely pathogenic (★★) |
| COL3A1 G1065E | 1065 | Triple-helical region | Pathogenic / likely pathogenic (★★) |
| COL3A1 G1029C | 1029 | Triple-helical region | Pathogenic / likely pathogenic (★★) |
| SMAD3 R287W | 287 | MH2 | Pathogenic / likely pathogenic (★★) |
| TGFBR1 K232N | 232 | Protein kinase | Pathogenic / likely pathogenic (★★) |
| COL3A1 G195R | 195 | Triple-helical region | Pathogenic / likely pathogenic (★★) |
| COL3A1 G258A | 258 | Triple-helical region | Pathogenic / likely pathogenic (★★) |
| COL3A1 G294R | 294 | Triple-helical region | Pathogenic / likely pathogenic (★★) |
| COL3A1 G423A | 423 | Triple-helical region | Pathogenic / likely pathogenic (★★) |
| COL3A1 E451K | 451 | Triple-helical region | Pathogenic / likely pathogenic (★★) |
| COL3A1 G456S | 456 | Triple-helical region | Pathogenic / likely pathogenic (★★) |
| COL3A1 G552R | 552 | Triple-helical region | Pathogenic / likely pathogenic (★★) |
Showing 60 of 207.
Uncertain variants prioritized for review in Ehlers-Danlos syndrome
| Variant | Position | Protein part | Clinical label | Evidence |
|---|---|---|---|---|
| COL3A1 K674E | 674 | Triple-helical region | Uncertain (★) | +6: 3 other pathogenic changes within 3 positions; K674N at the same position is pathogenic; not seen in the gnomAD population database; AlphaMissense 0.82 |
Which prediction tools work for Ehlers-Danlos syndrome
Observed separation of ClinVar pathogenic / likely pathogenic from benign / likely benign variants (AUROC × 100). This benchmark is not a clinical recommendation.
- AlphaGenome (regulatory): 100 out of 100 (learned from overlapping clinical labels, so this is optimistic)
- CATVariant: 98 out of 100 (learned from overlapping clinical labels, so this is optimistic)
- MutPred: 98 out of 100 (learned from overlapping clinical labels, so this is optimistic)
- REVEL: 98 out of 100 (learned from overlapping clinical labels, so this is optimistic)
- AlphaMissense: 97 out of 100
- ESM1b (LLR): 97 out of 100
- PolyPhen-2: 96 out of 100 (learned from overlapping clinical labels, so this is optimistic)
- MetaLR: 96 out of 100 (learned from overlapping clinical labels, so this is optimistic)
- EVE: 94 out of 100
- MetaSVM: 93 out of 100 (learned from overlapping clinical labels, so this is optimistic)
- CADD: 93 out of 100
- SIFT: 90 out of 100
- phyloP: 86 out of 100
- AlphaGenome (splicing): 78 out of 100 (learned from overlapping clinical labels, so this is optimistic)
- SpliceAI: 54 out of 100 (learned from overlapping clinical labels, so this is optimistic)
- Pangolin: 46 out of 100 (learned from overlapping clinical labels, so this is optimistic)
Same protein, different disease
- Familial thoracic aortic aneurysm and aortic dissection also has ClinVar records linked to COL3A1 variants; they fall in the same places as the Ehlers-Danlos syndrome variants (57 pathogenic / likely pathogenic).
- Osteogenesis imperfecta also has ClinVar records linked to COL1A2 variants; they fall mostly in different places as the Ehlers-Danlos syndrome variants (332 pathogenic / likely pathogenic).
- Ehlers-Danlos syndrome, classic also has ClinVar records linked to COL1A2 variants; they fall mostly in different places as the Ehlers-Danlos syndrome variants (265 pathogenic / likely pathogenic).
- Osteogenesis imperfecta with normal sclerae, dominant form also has ClinVar records linked to COL1A2 variants; they fall mostly in different places as the Ehlers-Danlos syndrome variants (51 pathogenic / likely pathogenic).
- Osteogenesis imperfecta, perinatal lethal also has ClinVar records linked to COL1A2 variants; they fall mostly in different places as the Ehlers-Danlos syndrome variants (42 pathogenic / likely pathogenic).
- Combined osteogenesis imperfecta and Ehlers-Danlos syndrome also has ClinVar records linked to COL1A2 variants; they fall mostly in different places as the Ehlers-Danlos syndrome variants (11 pathogenic / likely pathogenic).
- Familial thoracic aortic aneurysm and aortic dissection also has ClinVar records linked to TGFBR2 variants; they fall mostly in different places as the Ehlers-Danlos syndrome variants (78 pathogenic / likely pathogenic).
- Loeys-Dietz syndrome also has ClinVar records linked to TGFBR2 variants; they fall mostly in different places as the Ehlers-Danlos syndrome variants (49 pathogenic / likely pathogenic).
- Marfan syndrome also has ClinVar records linked to TGFBR2 variants; they fall mostly in different places as the Ehlers-Danlos syndrome variants (3 pathogenic / likely pathogenic).
- Ehlers-Danlos syndrome due to tenascin-X deficiency also has ClinVar records linked to TNXB variants; they fall mostly in different places as the Ehlers-Danlos syndrome variants (3 pathogenic / likely pathogenic).
- Loeys-Dietz syndrome also has ClinVar records linked to TGFBR1 variants; they fall mostly in different places as the Ehlers-Danlos syndrome variants (23 pathogenic / likely pathogenic).
- Familial thoracic aortic aneurysm and aortic dissection also has ClinVar records linked to TGFBR1 variants; they fall mostly in different places as the Ehlers-Danlos syndrome variants (23 pathogenic / likely pathogenic).
- Familial thoracic aortic aneurysm and aortic dissection also has ClinVar records linked to SMAD3 variants; they fall mostly in different places as the Ehlers-Danlos syndrome variants (29 pathogenic / likely pathogenic).
- Aneurysm-osteoarthritis syndrome also has ClinVar records linked to SMAD3 variants; they fall mostly in different places as the Ehlers-Danlos syndrome variants (18 pathogenic / likely pathogenic).
Diseases related to Ehlers-Danlos syndrome
- Familial thoracic aortic aneurysm and aortic dissection, also linked to COL3A1, COL5A1, COL5A2, SMAD3 and 3 more
- Connective tissue disease, also linked to COL1A1, COL1A2, COL3A1, COL5A1 and 2 more
- Ehlers-Danlos syndrome, classic, also linked to COL1A1, COL1A2, COL5A1 and COL5A2
- Loeys-Dietz syndrome, also linked to SMAD3, TGFB2, TGFBR1 and TGFBR2
- Osteogenesis imperfecta, also linked to COL1A1 and COL1A2
- Marfan syndrome, also linked to TGFBR1 and TGFBR2
- Osteogenesis imperfecta, perinatal lethal, also linked to COL1A1 and COL1A2
- Osteogenesis imperfecta with normal sclerae, dominant form, also linked to COL1A1 and COL1A2
- Ehlers-Danlos syndrome, arthrochalasia, also linked to COL1A1 and COL1A2
- Combined osteogenesis imperfecta and Ehlers-Danlos syndrome, also linked to COL1A1 and COL1A2
- Familial aortopathy, also linked to COL3A1 and TGFBR1
- Osteoporosis, also linked to COL1A1 and COL1A2
Frequently asked questions
Which genes have records linked to Ehlers-Danlos syndrome?
This view contains 11 analyzed proteins: COL3A1, COL1A2, TGFBR2, TNXB, TGFBR1 and 6 more. Links come from clinical records and association databases. They do not imply that every listed gene is a validated cause, and missing genes may not yet be analyzed.
What do the clinical classifications mean?
Linked records include 207 pathogenic or likely pathogenic variants, 808 variants of uncertain significance and 180 with conflicting classifications. Labels summarize source records; multi-condition records may not make a separate assertion for this disease. Check the original record and review status.
Does the evidence score change a VUS classification?
No. 1 VUS or conflicting variants reach the likely-pathogenic points range on the computable criteria available here. This is a research prioritization signal, not a clinical classification. Patient, family and other required evidence may be missing.
Can I download the variant evidence?
Download the CSV for all 1,398 variants in the selected disease scope, including clinical labels, review status, evidence criteria, predictor scores, functional measurements and population frequency where available.
About this data
Variant–disease links come from ClinVar, Open Targets and UniProt, pooled from eligible public CATVariant analyses of each human protein. Evidence scores use the ACMG/AMP Bayesian points scale with computable criteria only (position among known disease variants, rarity in gnomAD, calibrated predictors, deep mutational scanning); there is no family or patient data, so they prioritise variants for expert review and never classify them.
Download every variant as CSV · Browse all diseases · Methods · About the Center