Synaptraglutide
A theoretical, CNS-exposed tri-agonist of the GLP-1, GIP and glucagon receptors, designed to probe the metabolic–neurodegeneration axis ("type 3 diabetes"). It builds on validated incretin pharmacology and adds two engineering hypotheses — brain-penetrant delivery and biased, desensitization-resistant signaling — explicitly to interrogate why peripheral GLP-1 drugs engaged Alzheimer biomarkers yet did not slow clinical decline.
A fresh theoretical candidate publishes automatically every Monday.
How to read this dossier
Every substantive claim is tagged by evidence tier so the line between what is known and what is conjecture is never blurred:
Directly supported by published, peer-reviewed experimental or clinical evidence.
A reasonable extrapolation from validated, closely-related findings.
A novel design proposition inferred from related science; untested.
A possible future outcome with little or contradicting direct evidence.
Worked example — full methodology
Reference edition RDX-NM1 (Synaptraglutide), kept as a depth example of how each weekly brief is graded.
1 · The unmet challenge
Type 2 diabetes and obesity are associated with markedly increased risk of Alzheimer's disease and related dementias, and brain insulin resistance is a recognised feature of the Alzheimer brain — a link summarised by the "type 3 diabetes" framing. Established fact[1]
Incretin (GLP-1 receptor) agonists are neuroprotective in many preclinical models and modulate neuroinflammation. Established fact[2] Yet in the pivotal Phase 3 EVOKE / EVOKE+ trials (3,808 participants, early symptomatic Alzheimer's), oral semaglutide did not slow clinical progression versus placebo, even though it produced ~10% nominal reductions in CSF p-tau and neuroinflammation biomarkers. Established fact[3]
The gap between biomarker engagement and clinical benefit is plausibly explained by limited brain exposure of peripherally-restricted GLP-1 drugs and by single-pathway engagement. AI hypothesis The unmet need we target: a peptide that achieves verified, durable CNS receptor engagement across complementary metabolic and neurotrophic pathways — and a development plan that gates the neuro-program on proof of brain exposure rather than peripheral effect.
2 · Candidate profile
| Working name | Synaptraglutide (theoretical) |
|---|---|
| Scientific codename | RDX-NM1 |
| Modality | Unimolecular, lipidated multi-agonist peptide (~40–48 aa), once-weekly subcutaneous |
| Biological targets | GLP-1R (agonist) · GIPR (agonist) · GCGR / glucagon receptor (partial agonist) — all class B/secretin-family GPCRs Established |
| Delivery target (non-signaling) | Receptor-mediated transcytosis motif to raise blood–brain-barrier penetrance Hypothesis |
| Proposed mechanism | Gs→cAMP→PKA/CREB signaling with engineered bias toward sustained cAMP and reduced β-arrestin desensitisation Prediction + glucagon-driven energy expenditure |
| Primary intended use (research framing) | Probe of metabolic correction and CNS pathway engagement in metabolic–neurodegenerative disease |
| Status | Theoretical — never synthesised or tested |
3 · Proposed mechanism of action
Peripheral metabolic arm. GLP-1R and GIPR agonism enhances glucose-dependent insulin secretion, slows gastric emptying and reduces appetite; glucagon-receptor agonism raises energy expenditure and improves hepatic lipid handling. The combination is the validated basis of marketed and late-stage incretin drugs. Established fact[4]
Central arm (hypothesised). If meaningful CNS exposure is achieved, the same Gs/cAMP/CREB cascade is expected to drive neurotrophic (e.g., BDNF-linked) and anti-inflammatory signaling in neurons and glia, and GIPR engagement may modulate microglial/astrocytic activation. AI hypothesis Whether this translates to neuronal protection in humans is speculative — the EVOKE result is a direct caution that biomarker movement need not equal clinical benefit.[3]
4 · Engineering rationale
The design layers three established techniques and two novel hypotheses onto a validated tri-agonist scaffold.
Established engineering (high confidence)
① Protease-stabilised backbone — an α-aminoisobutyric acid (Aib) substitution at position 2 and selected non-natural residues resist DPP-4 and serum proteases; this is standard across the "-glutide" class. Established[4]
② C18 fatty-acid lipidation + albumin binding — extends half-life to support once-weekly dosing, as in semaglutide. Established[5]
③ Balanced tri-receptor potency tuning — relative GLP-1R/GIPR/GCGR activity is set to maximise metabolic benefit while limiting glucagon-driven glycaemic and cardiovascular liabilities, mirroring the empirical tuning behind triple agonists. Prediction[6]
Novel hypotheses (the "next-generation" bet)
④ Brain-penetrant delivery. A receptor-mediated transcytosis motif (e.g., a transferrin-receptor-binding or Angiopep-class shuttle sequence) is conjugated to raise CNS exposure beyond peripherally-restricted incretins, directly addressing the exposure limitation EVOKE exposed. AI hypothesis[7]
⑤ Signaling-biased GLP-1R agonism. Tuning toward sustained cAMP with reduced β-arrestin recruitment is hypothesised to lower receptor desensitisation and sustain neuronal signaling; biased GLP-1R agonism is an active, published research area, though its in-vivo neuro consequences are unproven. AI hypothesis[8]
5 · Expected benefits — with confidence
Confidence reflects how directly each expectation is supported by existing evidence for closely-related molecules. These are predictions about a theoretical compound, not claims.
Metabolic weight & glycaemic effect
Evidence-based prediction Tri-agonism of this class produces large weight loss and glycaemic improvement in humans (retatrutide ~24% at 48 wk in Phase 2; ~28.7% at 68 wk reported in Phase 3).[6] A well-tuned analogue would very likely share this profile. High but not certain because exact potency tuning of a novel sequence is unverified.
Measurable CNS receptor engagement (PET/CSF)
AI hypothesis Conjugated shuttle strategies have raised brain uptake of large biologics in models,[7] but reliably delivering a 40+ aa peptide agonist into human CNS at pharmacologically active levels is unproven. Moderate confidence the strategy can increase exposure; low confidence on the magnitude needed.
Engagement of Alzheimer-related biomarkers (p-tau, neuroinflammation)
Speculative Peripheral semaglutide already moved these biomarkers ~10% in EVOKE,[3] so greater CNS exposure might move them more — but this is inference, not evidence for this molecule.
Slowing of clinical cognitive decline in humans
Speculative The single most important caution in this dossier: EVOKE showed that biomarker engagement by a GLP-1 drug did not translate into clinical benefit in early Alzheimer's.[3] Any claim of cognitive benefit is unsupported and treated as a long-odds research question, not an expectation.
6 · Potential risks & limitations
Gastrointestinal intolerance. Nausea, vomiting and discontinuations are a dose-limiting class effect; in retatrutide Phase 3 the highest dose was effective but intolerable for a meaningful fraction of participants. Established[6]
Glucagon-axis effects. Glucagon agonism can raise glucose, heart rate and hepatic workload; potency balance is safety-critical. Established
Lean-mass loss with rapid weight reduction, and unknown effects of central agonism (appetite/aversion circuitry, mood, reward). Prediction / Speculative
Delivery-motif risks. Transferrin-receptor targeting can create peripheral "sink" effects, off-target uptake and immunogenicity; conjugates raise manufacturing and stability burdens. Hypothesis-stage risk[7]
Immunogenicity and long-term CNS safety of a brain-exposed multi-agonist are entirely uncharacterised. Speculative
7 · Development challenges
The central scientific challenge is proving CNS target engagement — not merely peripheral effect — because EVOKE demonstrated peripheral biomarker movement is insufficient. This demands a CNS exposure readout (radiolabelled PET ligand or CSF pharmacology) as an early go/no-go gate. Prediction Further challenges: separating central benefit from dose-limiting peripheral GI toxicity; manufacturing a stable conjugate at scale; controlling immunogenicity; and the weak translational track record of rodent neurodegeneration models. Established
8 · Competitive comparison
Against the real incretin landscape. Figures are published results for the marketed/clinical agents; the RDX-NM1 row is the theoretical target profile, not data.
| Agent | Receptors | Peak weight loss (reported) | CNS-exposure engineering | Neuro program | Status |
|---|---|---|---|---|---|
| Semaglutide | GLP-1R | ~15% (STEP) | None (peripheral) | EVOKE — did not slow AD | Marketed[5] |
| Tirzepatide | GLP-1R / GIPR | ~20–22% (SURMOUNT) | None | None | Marketed[9] |
| Retatrutide | GLP-1R / GIPR / GCGR | ~24% Ph2; ~28.7% Ph3 | None | None | Phase 3[6] |
| Cagrilintide | Amylin / calcitonin | ~10–15% (combo higher) | None | None | Phase 3 |
| RDX-NM1 (theoretical) | GLP-1R / GIPR / GCGR + delivery motif | Target: tri-agonist range | Yes (hypothesised) | CNS-gated by design | Theoretical |
The differentiating bet: none of the real agents were engineered for, or validated to achieve, CNS exposure — yet the neuro-metabolic hypothesis hinges on exactly that. RDX-NM1 exists to make CNS exposure an explicit, testable design goal rather than an afterthought.
9 · Confidence scorecard
10 · References & provenance
What is directly supported: incretin receptor biology and signaling; the metabolic efficacy of GLP-1/GIP/glucagon agonism; the EVOKE clinical result; the diabetes–dementia association. What is inferred: that greater CNS exposure or biased signaling would change clinical outcomes — these rest on related, not direct, evidence. Links resolve to the primary record or a live literature search.
- Insulin resistance / "type 3 diabetes" and Alzheimer's risk — review. PubMed search ↗
- GLP-1 receptor agonists, neuroprotection & neuroinflammation — reviews. PubMed search ↗
- EVOKE / EVOKE+ Phase 3, oral semaglutide in early Alzheimer's (did not meet primary endpoint; ~10% CSF biomarker reductions), reported Dec 2025. Coverage ↗
- Triple GLP-1/GIP/glucagon agonism for obesity — mechanism review. PMC ↗
- Semaglutide pharmacology & STEP weight-loss program. PubMed search ↗
- Retatrutide (triple agonist): Phase 2 (NEJM 2023) and Phase 3 TRIUMPH program (~28.7% weight loss reported). Lancet (TRANSCEND-T2D-1) ↗ · Phase 3 release ↗
- Receptor-mediated transcytosis / blood–brain-barrier shuttles for biologics (transferrin receptor, Angiopep). PubMed search ↗
- Biased agonism at the GLP-1 receptor (cAMP vs. β-arrestin). PubMed search ↗
- Tirzepatide (GLP-1/GIP) SURMOUNT obesity program. PubMed search ↗
11 · Future research roadmap
What would have to be true — and proven — before real-world development of anything like RDX-NM1 could responsibly be considered. Each phase gates the next.
Sequence & structure design
Model the tri-agonist scaffold and candidate shuttle conjugation sites; predict receptor docking, stability, and aggregation; shortlist sequences. Deliverable: ranked candidate set with predicted potency/selectivity.
Receptor & signaling validation
Measure cAMP and β-arrestin responses at GLP-1R/GIPR/GCGR; quantify potency, selectivity and bias factor; confirm protease resistance. Go/no-go: balanced agonism + measurable, intended bias.
Brain-penetrance proof (the critical gate)
Transwell BBB models, then rodent PK with CSF sampling and, ideally, a radiolabelled PET ligand. Hard gate: no neuro-program proceeds without evidence of CNS exposure at active concentrations — the explicit lesson from EVOKE.
Metabolic then neuro
Diet-induced-obesity rodents for metabolic effect; only then disease models for neuro endpoints, interpreted cautiously given poor model translation. Pre-register endpoints; report negatives.
Tolerability, conjugate safety, anti-drug antibodies
GLP/GI tolerability, glucagon-axis safety, peripheral-sink and immunogenicity of the delivery motif, lean-mass effects, and chronic CNS-exposure toxicology.
Human exposure before human efficacy
Confirm human CNS target engagement (CSF/PET) in early-phase studies before any cognition endpoint trial — so that a future EVOKE-style result, if it came, would at least answer a different question than the one already answered.