Patients with myalgic encephalomyelitis have too much of a protein in their blood and too little on their cells. Same protein. Same patients. That isn't a contradiction — it's a cleavage event. And the protein they're losing controls two independent arms of innate immune suppression that no one realized were connected until last year.
The Paradox
In July 2025, Moreau, Fluge, and Mella published data from two independent cohorts — 249 ME patients with 63 controls in Canada, replicated in 141 ME patients in Norway. They measured a GPI-anchored membrane protein called SMPDL3B. The findings:
Plasma SMPDL3B: significantly elevated in ME patients. Correlated with symptom severity.
Membrane-bound SMPDL3B: reduced on monocytes.
PLCXD1 expression: elevated — the gene encoding the PI-PLC enzyme that cleaves GPI-anchored proteins from membranes.
The cell is making more SMPDL3B. An enzyme is cutting it off faster than the cell can retain it. The protein accumulates in the blood because the membrane can't hold it.
Estrogen upregulates SMPDL3B expression, which may explain why female ME patients show higher plasma levels and greater severity. The body's attempt to compensate — produce more of the protein — amplifies the paradox: more production, more substrate for cleavage, more loss.
Two Functions, Ten Years Apart
SMPDL3B was studied in two separate literatures that never cited each other.
In 2015, Heinz and colleagues showed that SMPDL3B maintains specific ceramide species in cell membranes. Knock it out, and those ceramides deplete. Membrane order drops. Toll-like receptors become hyper-responsive. The inflammatory program activates. Add the ceramides back, and the phenotype reverses. The causal direction is clear: membrane lipid composition controls innate immune signaling, not the reverse.
In late 2025, Wang and colleagues published in Immunity that SMPDL3B is also a cGAMP hydrolase. cGAMP is the second messenger that activates the cGAS-STING pathway — a separate arm of innate immunity that detects cytoplasmic DNA and triggers interferon responses. SMPDL3B degrades cGAMP, dampening STING signaling. Mice lacking SMPDL3B had elevated cGAMP, heightened immune activation, and reduced viral loads.
One protein. Two brakes on two independent innate immune pathways. Discovered a decade apart by groups working in different fields — membrane biophysics and innate immunity signaling — on the same molecule.
What Cleavage Does
Wang's paper revealed something Moreau's couldn't: the released SMPDL3B retains its cGAMP hydrolase activity outside the cell. PI-PLC cuts the GPI anchor, and the protein floats free — but it keeps working. It degrades extracellular cGAMP, which means it can dampen STING signaling in neighboring cells that never lost their own SMPDL3B.
One cleavage event, divergent consequences. The source cell loses both brakes — TLR goes hyper-responsive from ceramide depletion, STING goes unbraked from cGAMP accumulation. The neighboring cells gain an extra cGAMP scavenger they didn't ask for. The immune landscape around the affected monocyte becomes incoherent: one cell screaming, its neighbors silenced.
This hasn't been demonstrated in vivo in ME/CFS or Long COVID patients. It's inferred from Wang's biochemistry. But the architecture is worth stating because it may explain something otherwise puzzling: why immune profiles in post-viral disease are heterogeneous even within a single tissue. The cells that lost their SMPDL3B are hyper-activated. The cells bathed in the released protein may be suppressed. Same tissue, opposite states.
The Bridge to Long COVID
The Moreau data is from ME/CFS, not Long COVID. But the bridge is short. Novak's 2026 comparison found near-identical autonomic failure profiles in LC and ME/CFS (95% and 89%, respectively). The Kumar LC-Mo monocyte — the reprogrammed monocyte subpopulation that defines Long COVID's immune signature — is TLR-hyper-responsive and pro-inflammatory. That is exactly the phenotype Heinz showed SMPDL3B loss produces.
The hypothesis writes itself: the monocyte that becomes LC-Mo may be the monocyte that lost its SMPDL3B. Not proven — no one has measured SMPDL3B on LC-Mo cells specifically. But the phenotypic match between SMPDL3B-deficient monocytes (Heinz) and LC-Mo monocytes (Kumar) is precise enough to warrant the test.
Wang's upstream trigger adds another connection. SMPDL3B is induced and stabilized by viral membrane perturbation — a direct response to infection. SARS-CoV-2 disrupts membranes. The initial induction of SMPDL3B may be a normal acute-phase brake. The pathology begins when PLCXD1 stays elevated and keeps cleaving the protein off after the acute infection resolves. The brake was deployed. Then the brake was removed. What drove it remains the open question.
The Fix That Exists
Moreau's team tested DPP4 inhibitors — saxagliptin and vildagliptin — for their ability to inhibit PI-PLC activity independently of their primary DPP4 function. Both restored membrane-bound SMPDL3B in vitro. Both reduced the soluble (cleaved) form. The mechanism is direct: block the enzyme that cuts the protein off, and the protein stays on the membrane.
These are generic diabetes drugs. Saxagliptin costs roughly $10/month. Vildagliptin is available across Europe. The Open Medicine Foundation has announced plans to test low-dose saxagliptin in ME/CFS through their CTN-Lite decentralized trial program — the first clinical trial directly targeting the SMPDL3B-PI-PLC axis.
No equivalent trial is registered for Long COVID. Despite the phenotypic overlap, the LC field and the ME/CFS field are running separate therapeutic pipelines on overlapping biology. The RECOVER program has baricitinib (JAK inhibition — downstream of TLR and STING), LDN (opioid receptor modulation), and semaglutide (GLP-1, anti-inflammatory). None targets the membrane itself.
What I Don't Know
What drives chronic PLCXD1 upregulation. Wang showed that viral membrane stress induces SMPDL3B expression. Moreau showed that PI-PLC expression is elevated in ME/CFS. But what keeps PLCXD1 turned on months or years after the initial infection? PLCXD1 belongs to a poorly characterized phospholipase C family — the literature is thin. This is the critical mechanistic gap.
Whether this operates identically in Long COVID. The Moreau data is from ME/CFS cohorts. The Novak phenotypic overlap and the Kumar LC-Mo match suggest it likely does, but SMPDL3B has not been measured on monocytes from Long COVID patients. The experiment is straightforward. It hasn't been done.
Whether CD55 modifies the picture. CD55 — another GPI-anchored protein on the same monocyte subset — is upregulated in severe COVID, suppressing type I interferon-stimulated genes. That's the opposite direction from SMPDL3B loss. Two GPI-anchored proteins, one going up, one coming off, on the same cell. The net effect is unclear.
Whether the renal connection matters. A follow-up study found subclinical kidney dysfunction in ME patients that tracks with SMPDL3B levels — consistent with the protein's known role in podocyte membrane integrity. Three metabolic endophenotypes emerged from SMPDL3B stratification. If membrane dysfunction extends beyond monocytes to podocytes, the therapeutic implications broaden.
The convergence of Wang's immunology (Immunity, 2025) and Moreau's clinical data (J Translational Medicine, 2025) wasn't planned. Two groups studying the same protein for different reasons arrived at the same membrane from different directions. Ten years separated their key findings. The field that studies what SMPDL3B does to lipids didn't read the field that studies what it does to cGAMP. The result is a decade of lost insight into how one protein coordinates two arms of innate immune suppression — and a trial of cheap generic drugs to put it back.