Preprint Complex Genetics and Regulatory Drivers of Hypermobile Ehlers-Danlos Syndrome: Insights from [GWAS] Meta-analysis, 2025, Petrucci-Nelson et al

[forestglip]

Complex Genetics and Regulatory Drivers of Hypermobile Ehlers-Danlos Syndrome: Insights from Genome-Wide Association Study Meta-analysis

Spoiler: Authors

Taylor Petrucci-Nelson, Sacha Guilhaumou, View ORCID ProfileTakiy-Eddine BERRANDOU, Cortney Gensemer, Adrien Georges, Matthew Huff, Margaux-Alison Fustier, Josephine Henry, Asraa Esmael, Olivia Jaye, Ranan Phookan, Sarah Dooley, Kathryn Byerly, Brian Loizzi, Roman Fenner, Emma Mach, Amy Weintraub, Victoria Daylor, Julianna Weninger, Natalie Koren, Erika Bistran, Charlotte Griggs, Molly Griggs, Sydney Severance, Rebecca Byrd, Sunil Patel, Steven A. Kautz, Anne Maitland, View ORCID ProfileNabila Bouatia-Naji, Russell A. Norris

[Line breaks added]

Background
Hypermobile Ehlers Danlos syndrome (hEDS) is the most common subtype of EDS, a group of heritable connective tissue disorders. Clinically, hEDS is defined by generalized joint hypermobility and chronic musculoskeletal pain, but its impact extends beyond the musculoskeletal system.

Affected individuals frequently experience autonomic, gastrointestinal, immune, and neuropsychiatric involvement, highlighting both the multisystemic nature of the condition and challenges of diagnosis. In contrast to other EDS subtypes with defined genetic causes, the molecular basis of hEDS has remained elusive.

Methods
We conducted a genome wide association study (GWAS) of hEDS across three case controls studies, including 1,815 cases and 5,008 ancestry-matched controls. Fixed effects metaanalysis of 6.2 million variants was complemented with LDAK gene based association testing, transcriptome wide association studies, and integrative annotation across multiple tissues and cell types including eQTLs, enhancer marks and open chromatin accessibility profiles, supported by luciferase assays on one candidate variant.

LDscore genetic correlations were assessed between hEDS and 19 frequently reported comorbid conditions.

Results
Two loci reached genome wide significance, including a regulatory region near the atypical chemokine receptor 3 gene (ACKR3) on chromosome 2. Functional annotation supports ACKR3 risk alleles colocalize with eQTLs in tibial nerve, alter enhancer activity, and generate a de novo AHR transcription factor regulatory site, implicating neuroimmune and pain signaling pathways.

Gene-based and transcriptome wide analyses identified common variants in a locus containing multiple candidates, including SLC39A13, a zinc transporter critical for connective tissue development previously implicated in a rare form of EDS, and PSMC3, a gene involved in central nervous system development.

LDscore regression revealed significant genetic correlations between hEDS and joint hypermobility, myalgic encephalomyelitis/chronic fatigue syndrome, fibromyalgia, depression, anxiety, autism spectrum disorder, migraine, and gastrointestinal diseases.

Conclusions
These results establish the first evidence of common variant contributions to hEDS, supporting a complex, multisystem model involving neuroimmunestromal dysregulation. Our findings add novel indications to hEDS pathogenesis and provide solid foundations for future molecular definition and therapeutic discovery.

Web | PDF | Preprint: MedRxiv | Open Access

 

[forestglip]

From GeneCards:

ACKR3

UniProtKB/Swiss-Prot Summary for ACKR3 Gene

Atypical chemokine receptor that controls chemokine levels and localization via high-affinity chemokine binding that is uncoupled from classic ligand-driven signal transduction cascades, resulting instead in chemokine sequestration, degradation, or transcytosis. Also known as interceptor (internalizing receptor) or chemokine-scavenging receptor or chemokine decoy receptor.

Acts as a receptor for chemokines CXCL11 and CXCL12/SDF1 (PubMed:16107333, 19255243, 19380869, 20161793, 22300987). Chemokine binding does not activate G-protein-mediated signal transduction but instead induces beta-arrestin recruitment, leading to ligand internalization and activation of MAPK signaling pathway (PubMed:16940167, 18653785, 20018651).

Required for regulation of CXCR4 protein levels in migrating interneurons, thereby adapting their chemokine responsiveness (PubMed:16940167, 18653785). In glioma cells, transduces signals via MEK/ERK pathway, mediating resistance to apoptosis. Promotes cell growth and survival (PubMed:16940167, 20388803).

Not involved in cell migration, adhesion or proliferation of normal hematopoietic progenitors but activated by CXCL11 in malignant hemapoietic cells, leading to phosphorylation of ERK1/2 (MAPK3/MAPK1) and enhanced cell adhesion and migration (PubMed:17804806, 18653785, 19641136, 20887389).

Plays a regulatory role in CXCR4-mediated activation of cell surface integrins by CXCL12 (PubMed:18653785).

Required for heart valve development (PubMed:17804806). Regulates axon guidance in the oculomotor system through the regulation of CXCL12 levels (PubMed:31211835). ( ACKR3_HUMAN,P25106 )

(Microbial infection) Acts as a coreceptor with CXCR4 for a restricted number of HIV isolates. ( ACKR3_HUMAN,P25106 )

SLC39A13

UniProtKB/Swiss-Prot Summary for SLC39A13 Gene

Functions as a zinc transporter transporting Zn(2+) from the Golgi apparatus to the cytosol and thus influences the zinc level at least in areas of the cytosol (PubMed:21917916, 23213233). May regulate beige adipocyte differentiation (By similarity). ( S39AD_HUMAN,Q96H72 )

PSMC3

UniProtKB/Swiss-Prot Summary for PSMC3 Gene

Component of the 26S proteasome, a multiprotein complex involved in the ATP-dependent degradation of ubiquitinated proteins.

This complex plays a key role in the maintenance of protein homeostasis by removing misfolded or damaged proteins, which could impair cellular functions, and by removing proteins whose functions are no longer required. Therefore, the proteasome participates in numerous cellular processes, including cell cycle progression, apoptosis, or DNA damage repair.

PSMC3 belongs to the heterohexameric ring of AAA (ATPases associated with diverse cellular activities) proteins that unfolds ubiquitinated target proteins that are concurrently translocated into a proteolytic chamber and degraded into peptides. ( PRS6A_HUMAN,P17980 )

 

[forestglip]

Genetic correlations, from supplementary table 13:

As expected, hEDS shared a substantial proportion of its genetics with joint hypermobility (rg=0.41, P=8.0×10−4), its most reported clinical feature (Figure 4, Supplementary Tab 13).

The most significant genetic correlation was found with myalgic encephalomyelitis/chronic fatigue syndrome (rg=0.35, P=6.5×10-15,), a condition reported by ~20 to 30% of patients included in our meta-analysis (Supplementary Table 2).

We also observed a robust genetic correlation with chronic pain (rg=0.33, P = 5.16×10−12; reported by 73% to 91% patients), and to a less extent and significance with fibromyalgia (rg=0.21, P=0.043).

Consistent with the high burden of gastrointestinal symptoms reported in >80 % patients, hEDS genetically correlated notably with irritable bowel syndrome (rg=0.31, P=1.71×10−13), gastroesophageal reflux disease (rg=0.18, P=2.87×10−5), and nominally with gastroparesis (rg=0.25, P=0.019).

Beyond connective tissue and gastrointestinal track, hEDS showed significant shared genetic architecture with several neurological and psychiatric diseases including migraine (rg=0.30, P=1.9×10−8; reported by 50–68% of patients), major depressive disorder (rg=0.17, P=1.2×10−7; reported by 51 to 63% patients), nominally with autism spectrum disorder (rg=0.13, P=0.008, ~10% of patients) but not with anxiety (rg=0.19, P = 0.23), despite report by a large proportion of hEDS 64–76% of patients.

Of note, our current analyses do not support large and significant genetic correlations between hEDS and structural or vascular manifestations including mitral valve prolapse, tricuspid valve disease, pelvic organ prolapse, or specific hernia subtypes (ventral, umbilical, inguinal), despite their frequent clinical reporting11 (Supplementary Table 2)

 

[ME/CFS Science Blog]

The effect sizes are much larger (OR of 1.29 and 1.66) than in DecodeME.

 

[ME/CFS Science Blog]

None of the regions found here came close to genome-wide significance the DecodeME dataset.

Chromosome	rsID	p-value in DecodeME
2	rs2708184	0.637
8	rs16880769	0.0672

 

[EndME]

The data that they used for ME/CFS is indeed the DecodeME data (see table 7 of supplementary data). Looks like ME/CFS status was verified in the hEDS population by ticking a ME/CFS box in a questionnaire and for the ALL of us set is was simply using EHR data.

 

[EndME]

On S4ME it has been previously argued that:
1- it seems unplausible that self-reported hEDS would stand for much, contrary to ME/CFS
2- DecodeME data strongly suggests that self-reported symptoms and self-reported diagnosis of ME/CFS (verified by self-reported symptoms) identify a group of people suffering from a distinct pathology

This study seems to suggest that people with self-reported hEDS however look like they indeed have a pathology that is both somewhat distinct from other illnesses, whose risk genes carry a greater effect than those in ME/CFS do and who at same time share a genetic overlap with different conditions.

From a first glance it looks like this study would throw a spanner into one of those 2 arguments brought up on S4ME.

 

[Utsikt]

On S4ME it has been previously argued that it seems unplausible that self-reported hEDS would stand for much, contrary to ME/CFS, and at the same time it has been argued that DecodeME data strongly suggests that self-reported symptoms in ME/CFS identify a group of people suffering from a distinct pathology.

This study seems to suggest that people with self-reported hEDS however look like they indeed have a pathology that is both somewhat distinct from other illnesses, whose risk genes carry a greater effect than those in ME/CFS do and who at same time share a genetic overlap with different conditions.

From a first glance it looks like this study would throw a spanner into one of those 2 arguments brought up on S4ME.

DecodeME wasn’t self-reported ME/CFS, though. It was self-report of having received an ME/CFS diagnosis by an HCP, and fulfilling the ME/CFS criteria according to an additional questionnaire.

What was the inclusion criteria in this study?

 

[EndME]

DecodeME wasn’t self-reported ME/CFS, though. It was self-report of having received an ME/CFS diagnosis by an HCP, and fulfilling the ME/CFS criteria according to an additional questionnaire.

What was the inclusion criteria in this study?

I anticipated such a comment so I'd already edited my original comment (presumably whilst you made this comment). I don't think it really makes a difference to my assessment.

 

[ME/CFS Science Blog]

From a first glance it looks like this study would throw a spanner into one of those 2 arguments brought up on S4ME.

My own argument has been that the hEDS diagnosis likely captures a lot of people who do not have a connective tissue disorder but some other disease. Think the genetic data of this study is still consistent with that.

The two hits show strong effects but it isn't very clear what they mean. They do not directly point to known EDS genes. The authors also did not find significant genetic correlations between hEDS and structural or vascular manifestations including mitral valve prolapse, tricuspid valve disease, pelvic organ prolapse, or specific hernia subtypes. Even the genetic correlation with hypermobility itself (rg = 0.41) was relatively modest.

One of the more interesting findings, however, was the gene-based analysis where SLC39A13 showed up. It encodes a zinc transporter critical for connective tissue development. Mutations in this gene cause the recessive spondylocheirodysplastic form of EDS. Perhaps this points to (undiagnosed) EDS cases in the hEDS group.

 

[ME/CFS Science Blog]

DecodeME data strongly suggests that self-reported symptoms and self-reported diagnosis of ME/CFS (verified by self-reported symptoms) identify a group of people suffering from a distinct pathology

I am not sure this statement is correct by the way - regardless of the debate about hEDS in this thread. All human traits have heritability. My guess would be that if you take people with severe fatigue you might also find significant hits in a GWAS the size of DecodeME.

 

[Utsikt]

This study seems to suggest that people with self-reported hEDS (…) share a genetic overlap with different conditions.

Are you referring to the correlations here?

 

[EndME]

I am not sure this statement is correct by the way - regardless of the debate about hEDS in this thread. All human traits have heritability. My guess would be that if you take people with severe fatigue you might also find significant hits in a GWAS the size of DecodeME.

The argument that has been made on distinctness of ME/CFS has not been made on finding significance but rather on the genetic risk profile looking distinct to known things. I think something similar can be argued here.

 

[Yann04]

Do we have a GWAS of hyper mobility? Is there an argument to be made that the hEDS GWAS just picks out not necessarily disease related hypermobile genes?

 

[ME/CFS Science Blog]

The argument that has been made on distinctness of ME/CFS has not been made on finding significance but rather on the genetic risk profile looking dinstinct to other things.

Not sure I would agree with that either: some of the genetic correlations we calculated with LDSC were quite high.

 

[EndME]

Are you referring to the correlations here?

LD-score analysis.

 

[EndME]

Not sure I would agree with that either: some of the genetic correlations we calculated with LDSC were quite high.

You don't have to agree with me or not, but I think it's the consensus on S4ME, or at least such a sentence is currently part of a draft of a factsheet on ME/CFS for clinicans (see here) and I don't think anybody has disagreed with that statement on that thread.

 

[Utsikt]

LD-score analysis.

Don’t they just show sample overlaps?

If a substantial portion of the hEDS group had ME/CFS, isn’t a correlation just as expected?

What does that mean for hEDS?

 

[Utsikt]

You don't have to agree with me or not, but I think it's the consensus on S4ME, or at least such a sentence is currently part of a draft of a factsheet on ME/CFS for clinicans (see here) and I don't think anybody has disagreed with that statement on that thread.

This is the sentence:

Almost nothing is known about mechanisms involved in ME/CFS but recent genetic studies show that the syndrome picks out a specific biological process, or cluster of processes, with a profile of genetic risk factors not found in other conditions. (One gene may overlap for chronic pain.)

Is that also the case in this GWAS?

 

[EndME]

Is that also the case in this GWAS?

Is what exactly the case? The factsheet does not provide an exact argument for how this statement was made, so how do you want to analyse whether it is the case here? The lead variants picked out here seem to not have identified as significant in previous GWAS (at least according to a quick google search) but my understanding is that things aren't that simple.