GPCRs: Drug Mechanisms and Disease Modules

Overview

G protein-coupled receptors are the molecular target of a large share of approved drugs and the site of a well-defined set of Mendelian endocrinopathies. This project builds dismech mechanism modules around them, on two axes:

1. Drug-mechanism modules — one per major GPCR-targeting drug class, so a disease entry's treatment can conforms_to a shared receptor → transducer → effector chain instead of re-deriving it, following the pattern the antimicrobial set established (projects/ANTIMICROBIAL.md).
2. Disease modules — the recurring ways a GPCR or its transducer breaks: loss of function (hormone resistance), constitutive activation, and transducer defects.

The gene side is curated in parallel in ai-gene-review (projects/GPCR.md): receptor and transducer gene reviews, PANTHER family reviews, and GO pathway modules (Gs-cAMP-PKA, Gq-PLC-Ca2+, planned Gi/o, G12/13, arrestin). The receptor reviewed there is the molecular anchor for the module here.

Scope (what this is and is not)

• In scope: the pharmacological mechanism by which a GPCR drug changes a disease's pathophysiology, including on-target adverse effects that are themselves mechanisms (opioid respiratory depression, antipsychotic parkinsonism, 5-HT2B valvulopathy).
• Not in scope: a drug–indication database, dosing, or pharmacokinetics except where it gates mechanism. NCIT P302 already supplies indications (projects/NCIT_TREATMENT_INDICATIONS.md).
• Granularity: one module per receptor-and-direction (agonism vs antagonism), not per drug. Drugs sit on the conforming disease entries' treatments with therapeutic_agent bindings.

Existing modules that touch GPCRs

Read each before creating a neighbour, and extend rather than duplicate where the scope already fits.

Workstream A — Drug-mechanism modules

Candidate modules, ordered by how many existing disease entries would conform:

Each module follows the create-module skill: a trigger → receptor engagement → transducer → effector → tissue consequence chain, with the treatment target_mechanisms drug pattern, module_categories: [PHARMACOLOGY] (plus TOXICOLOGY where an adverse-effect arm is modelled), and receptor genes bound to HGNC.

Adverse-effect arms to model explicitly, because they are mechanisms in their own right: opioid respiratory depression (brainstem MOR), D2-blockade extrapyramidal effects and hyperprolactinemia, beta-blocker bronchospasm (beta2 off-selectivity), 5-HT2B agonist valvulopathy.

First pairing: mu-opioid receptor (OPRM1)

Chosen first (2026-10-15): opioid drugs appear in about 18 entries (Opioid_Use_Disorder, Acute_Opioid_Poisoning, Acute_Post-Surgical_Pain, Alcohol_Use_Disorder, ...) and in addiction_cycle and spinal_hsp90_opioid_enhancement, each describing the mechanism on its own.

• Mu-opioid receptor agonism (morphine, fentanyl, oxycodone; methadone and buprenorphine for OUD). Branches: analgesia (spinal and supraspinal), respiratory depression (brainstem; the overdose mechanism), constipation (enteric), reward/tolerance/dependence (link to addiction_cycle). Buprenorphine, a partial agonist, belongs here with its respiratory ceiling recorded on that branch rather than in a third module.
• Mu-opioid receptor antagonism (naloxone reversal, naltrexone relapse prevention, peripheral antagonists for opioid-induced constipation).
• Biased agonism (oliceridine) remains the open question below; this module pair is where it will need an answer.
• Kappa-opioid modules (nalfurafine; mixed agonist-antagonists) come later.

Workstream B — Disease modules

The loss/gain pair is deliberately symmetric: the same receptor (CASR, TSHR, LHCGR) appears on both sides, which is the test that the modules are mechanism-level rather than disease-level.

Plan

Phase 0 — scoping

• [ ] Census: which kb/disorders entries already name a GPCR drug in treatments (therapeutic_agent) and which GPCR genes appear in genetic:
• [ ] Agree tranche-1 receptor list with the ai-gene-review side
• [ ] Decide whether the drug-mechanism set becomes a kb/module_collections/ entry (GPCR drug mechanisms)

Phase 1 — first modules

• [x] mu-opioid receptor agonism (first module; see above) — mu_opioid_receptor_agonism (#14453)
• [x] mu-opioid receptor antagonism — mu_opioid_receptor_antagonism (#14456)
• [x] beta-adrenergic receptor blockade — beta_adrenergic_receptor_blockade (#15087)
• [ ] GLP-1 receptor agonism — glp1_receptor_agonism (#15252, in review)
• [ ] GPCR loss-of-function hormone resistance
• [ ] GPCR constitutive activation

Phase 2 — conformers and remaining modules

• [ ] Add conforms_to on the treatment-linked pathophysiology nodes of the candidate conformers above
• [x] The four opioid-treated entries, in #14973: Opioid_Use_Disorder and Acute_Opioid_Poisoning against both mu-opioid modules, Acute_Post-Surgical_Pain against agonism, Alcohol_Use_Disorder against antagonism
• [x] Three beta-blocker-treated entries, in #15087: Heart_Failure, Essential_Hypertension and Atrial_Fibrillation against beta_adrenergic_receptor_blockade. Migraine and Glaucoma are deliberately not wired — see that module's scope notes
• [x] Four antipsychotic-treated entries, in #15299: Schizophrenia, Bipolar_Disorder, Cannabis_Hyperemesis_Syndrome and Subacute_Delirium against d2_receptor_antagonism. The four span three indications — psychosis/mania, antiemesis and agitation — which is the point: one receptor action, three reasons to prescribe. Schizophrenia keeps its dopamine-dysregulation target alongside the receptor node, because D2 blockade acts downstream of that presynaptic abnormality
• [x] Three angiotensin-blocker-treated entries, in #15271: Essential_Hypertension, Heart_Failure and IgA_Nephropathy against at1_receptor_blockade. Essential_Hypertension's ACE inhibitor is pointed at its own renin-angiotensin-aldosterone node instead, because an ACE inhibitor never occupies the receptor — see that module's scope notes
• [x] Two incretin-treated entries, in #15252: Type_2_Diabetes_Mellitus and Obesity against glp1_receptor_agonism, both at the receptor step. The module names no dominant weight-loss arm, so conforming Obesity against one would assert more than the evidence carries
• [x] Three dopamine-agonist-treated entries, in #15350: Parkinsons_Disease, Restless_Legs_Syndrome and Peripartum_Cardiomyopathy against dopamine_receptor_agonism. All three anchor without a downstream edge; the first two at the trigger, and Peripartum_Cardiomyopathy at the module's lactotroph node instead, because bromocriptine is given there for prolactin suppression rather than for striatal signalling
• [x] Two triptan-treated entries, in #15470: Migraine and Migraine_with_Aura against serotonin_1b_1d_receptor_agonism, both at the trigger. Familial_Hemiplegic_Migraine is deliberately not wired — see that module's scope notes
• [x] Two CGRP-antagonist-treated entries, in #15624: Migraine, through its named gepant treatment, and pseudotumor_cerebri, through its erenumab treatment. The second is not a migraine entry on purpose, so the module is indication-independent and both the small-molecule and the antibody route are exercised. Migraine's anti-CGRP antibody treatment is left unwired because it bundles a receptor-directed agent with two peptide-directed ones
• [x] Two antihistamine-treated entries, in #15686: Urticaria and Alpha-gal_Syndrome against h1_receptor_antagonism, both at the trigger. The second is not an urticaria entry on purpose, so the module is indication-independent. Four further entries (Mast_Cell_Activation_Syndrome, Maculopapular_Cutaneous_Mastocytosis, Systemic_Mastocytosis and Monoclonal_Mast_Cell_Activation_Syndrome) were recorded here as blocked by bundling an H1 and an H2 agent into one treatment. That premise was wrong, and reading the four settled it: a treatment that binds its components as separate agents can carry one target_mechanisms link per receptor and assert nothing false, so bundling blocks nothing. Two of the four already bound both classes separately; the other two had agents missing rather than conflated. #15778 wired all four to h1_receptor_antagonism with no split, adding the missing agent bindings
• [x] One H2-blocker-treated entry, in #15689: Gastroesophageal_Reflux_Disease against h2_receptor_antagonism, at the trigger. Its existing H2 antagonist treatment was wired rather than a new one added
• [x] Three beta2-agonist-treated entries, in #15706: Asthma (both its pre-existing short- and long-acting treatments wired to one node, since the receptor is the same and only the route and duration differ), Thunderstorm_Asthma and Formaldehyde_Poisoning against beta2_adrenergic_receptor_agonism, all at the trigger. The third is a toxicological entry rather than an airway disease, so the module is not tied to asthma; its link carries receptor-level evidence only, because that treatment's own notes record that no formaldehyde-specific efficacy publication exists. Three further entries treat with a beta2 agonist and were not wired at the time, each blocked by one treatment covering more than one drug class. Two of the three were since unblocked in #15777 and are now conformers; the third should not be. Chronic_Obstructive_Pulmonary_Disease bound the generic NCIT:C319 Bronchodilator agent in a single Bronchodilators treatment, so wiring it would have asserted beta2 action for a LAMA. #15777 split that treatment into a long-acting beta-2 agonist and a long-acting muscarinic antagonist, keeping all four of its citations on both halves with per-class explanations, and wired the beta2 half. Cough_Variant_Asthma turned out not to be bundled at all: its treatment is beta2-only in its own description, and the generic agent was an under-specific binding rather than a bundle, so #15777 refined the agent and wired it with no split. Sulfur_Mustard_Poisoning is a third shape and is the one that stays out: its salmeterol is bound at agent level (CHEBI:9011), but inside Macrolide and Inhaled Corticosteroid Therapy for Bronchiolitis Obliterans alongside azithromycin and fluticasone propionate, and that treatment's own target_mechanisms says the combination is used for anti-inflammatory rather than bronchodilator action. Splitting it would assert a bronchodilator mechanism the entry explicitly disclaims, so it should be left alone
• [x] Four antihistamine-treated mastocytosis entries, in #15778: Mast_Cell_Activation_Syndrome, Systemic_Mastocytosis, Maculopapular_Cutaneous_Mastocytosis and Monoclonal_Mast_Cell_Activation_Syndrome against h1_receptor_antagonism, all at the trigger. Maculopapular_Cutaneous_Mastocytosis' node is the only one bound to the dermis, since its entry is cutaneous-only; the other three carry no location. Two of the four needed agent bindings added, and the anti-leukotriene half of Monoclonal_Mast_Cell_Activation_Syndrome's treatment is recorded in its notes as deliberately unbound -- CHEBI:49159 leukotriene antagonist is a CHEBI role term and so sits outside the ChemicalEntityTerm enum, and no NCIT class term for the drug class was found
• [x] One muscarinic-antagonist-treated entry, in #15795: Chronic_Obstructive_Pulmonary_Disease against muscarinic_receptor_antagonism, at the trigger. It is the second receptor occupancy node in that entry, alongside the beta2 one from #15777, which is the point of keeping the two modules separate
• [x] Five calcimimetic-treated entries, in #15925: CKD-Mineral_Bone_Disorder, Familial_Hypocalciuric_Hypercalcemia_1 and _2, Parathyroid_Gland_Carcinoma and Parathyroid_Hyperplasia against casr_positive_allosteric_modulation, all at the trigger. The five span four different reasons parathyroid hormone is excessive -- reduced renal calcium and vitamin D handling, an inactivating CASR variant, a GNA11 signalling defect, and autonomous malignant or hyperplastic tissue -- which is what makes this module indication-independent rather than a CKD module, and the point is carried by a source rather than assumed: the cited case report states that cinacalcet suppresses parathyroid hormone in parathyroid carcinoma in the same way as in uraemic hyperplasia. Two further entries name a calcimimetic and are not wired. Multiple_Endocrine_Neoplasia_Type_1 has no calcimimetic treatment to wire, only a mention inside a parathyroidectomy treatment that binds no agent of its own. Autosomal_Recessive_Hypophosphatemic_Rickets_1 records a single adjunctive course in one adult that was stopped because higher doses were not tolerated, and its own description says it does not establish routine efficacy, so wiring it would present an abandoned single-patient trial as a therapeutic mechanism. The autosomal dominant hypocalcaemia entries are recorded in the module as never-conformers: they are caused by activating receptor variants, so a calcimimetic is the wrong direction of action and those entries name it as something to avoid
• [x] Three P2Y12-antagonist-treated entries, in #15928: Coronary_Artery_Disease, Carotid_Stenosis and CADASIL_Type_1 against p2y12_receptor_antagonism, all at the trigger. Those three are the only entries whose treatment binds clopidogrel or ticagrelor as therapeutic_agent. Six others name a P2Y12 antagonist in a treatment context and are not wired, because none of them binds one as an agent, so a target_mechanisms link would assert a receptor action the entry does not name. They fail that test in four different ways, which is why the worklist has to be read per entry rather than as one search: Peripheral_Artery_Disease binds the generic NCIT:C1327 Antiplatelet Agent, which covers aspirin as well; Carotid_Artery_Occlusion names clopidogrel in its treatment prose but binds only acetylsalicylic acid; Grange_syndrome, Kawasaki_Disease and Renal_Artery_Obstruction bind no agent at all on the relevant treatment; and Secondary_Erythromelalgia names ticlopidine only as one of the platelet inhibitors that failed to relieve its pain, so it is not treated with the class at all. Only the first of those is the generic-binding shape that the Bronchodilator refinements in #15777 fixed. Five further entries name the class or one of its drugs but are not candidates: Alexander_Disease (clopidogrel as a mouse-model probe), Atypical_Hemolytic_Uremic_Syndrome (clopidogrel as a precipitant), CYCS-Related_Thrombocytopenia (whose own notes say a carrier's clopidogrel is not evidence that antiplatelet drugs treat the disease), and Platelet-type_Bleeding_Disorder_8 and Platelet-type_Bleeding_Disorder_16, where P2RY12 is the causal gene and loss of the receptor is the disease rather than the therapy. Note that a grep for drug names alone returns thirteen entries, not fourteen: Platelet-type_Bleeding_Disorder_16 names only "P2Y12 inhibitors" as a class, so searching for p2y12 as well as the drug names is what makes the two counts agree
• [x] Nine nodes across the three migraine entries, in #15934: Migraine at four, Familial_Hemiplegic_Migraine at three and Migraine_with_Aura at two, against migraine_neurovascular_mechanism. This one adds no nodes at all — every node it conforms was already in the entry, which is what makes it the cheapest kind of conformance and the clearest demonstration of what conformance is for: the three entries had modelled the same chain under different names, and nothing could compare them until the module existed
• [ ] Remaining Workstream A modules
• [x] D2 receptor antagonism — d2_receptor_antagonism (#15299)
• [x] AT1 receptor blockade — at1_receptor_blockade (#15271)
• [x] dopamine receptor agonism — dopamine_receptor_agonism (#15350)
• [x] 5-HT1B/1D agonism — serotonin_1b_1d_receptor_agonism (#15470)
• [x] CGRP receptor antagonism — cgrp_receptor_antagonism (#15624)
• [x] H1 receptor antagonism — h1_receptor_antagonism (#15686)
• [x] H2 receptor antagonism — h2_receptor_antagonism (#15689)
• [x] beta2 adrenergic receptor agonism — beta2_adrenergic_receptor_agonism (#15706)
• [x] muscarinic receptor antagonism (airway) -- muscarinic_receptor_antagonism (#15795)
• [x] CaSR positive allosteric modulation -- casr_positive_allosteric_modulation (#15925)
• [ ] S1P receptor modulation — deferred behind P2Y12: only Multiple_Sclerosis treats with an S1P modulator. Of the four other entries mentioning S1P, two are mouse models (Lafora disease, glaucoma), one uses a research-tool antagonist by rat knockdown, and one lists fingolimod only in prose about MS immunotherapies
• [x] P2Y12 receptor antagonism — p2y12_receptor_antagonism (#15928)
• [ ] V2 receptor antagonism
• [ ] MC4R agonism
• [x] migraine neurovascular mechanism — migraine_neurovascular_mechanism (#15934). A disease module, not a drug one, factoring the cortical spreading depolarization → trigeminovascular activation → central sensitization chain out of Migraine, Migraine_with_Aura and Familial_Hemiplegic_Migraine, which each re-derived it under different node namings. Added from the resolution of the CGRP open question below. It carries no treatments block: the migraine pharmacology is already in serotonin_1b_1d_receptor_agonism and cgrp_receptor_antagonism, and the conformers that use those drugs reach them through their own therapeutic_vulnerability nodes. Migraine conforms to both, Migraine_with_Aura to the 5-HT1B/1D module only, and Familial_Hemiplegic_Migraine to neither, because triptans and ergots are contraindicated there for their vasoconstrictive effects. That last case is a second reason the shared module carries no drug arm: it would attach a therapy to a conformer that must avoid it
• [ ] Gs alpha defect module

Decisions

• Agonism and antagonism of the same receptor are separate modules (decided 2026-10-15), because their downstream consequence chains differ.
• One receptor family plus one direction of pharmacological action makes one module (decided 2026-10-25, generalizing the rule above). A module is never defined by an indication: conformers come from whatever diseases happen to be treated that way, not the reverse.
• 5-HT1B/1D agonism and CGRP receptor antagonism are two modules, not one (decided 2026-10-25, resolving the open question below). Three reasons, in increasing order of weight. By the rule above, they are different receptors in opposite directions, so bundling them would bundle on shared indication. The triptan arm carries a vasoconstriction profile that the gepant arm exists specifically to avoid, so one module could not hold both adverse chains without contradicting itself. And the pharmacology review the CGRP module is grounded on (PMID:34076887, Garelja, Walker and Hay, Br J Pharmacol 2022) draws the line itself: it separates the drugs that bind the CGRP peptide from the receptor antagonists, and subdivides the latter into small molecules and one antibody antagonist. That third reason also settles where erenumab goes, which the first two left open, and it scopes cgrp_receptor_antagonism to receptor antagonism with the peptide-directed antibodies out.

The question's premise was a false choice, and the answer is both: a migraine neurovascular mechanism module is a genuinely distinct third module, not a substitute for either drug module, and is now in the Workstream A queue above. - H1 and H2 antagonism are two modules, and so are airway beta2 agonism and muscarinic antagonism (decided 2026-10-26). Both were single rows in the Workstream A table above, and both are now split, by the rule two entries up. H1 and H2 are different receptors on different cells with disjoint effector chains -- cutaneous vasodilatation, extravasation and itch against parietal-cell acid secretion -- and disjoint conformers; they share only the word antihistamine. The airway pair are different receptors in opposite directions, Gs-coupled relaxation against Gq blockade, sharing only that the two drug classes are often inhaled together.

The queue table bundled on shared indication twice, which is the failure the rule above names, so a bundled row there should be read as a candidate for splitting rather than as a decision already taken.

• A module pull request may split another entry's bundled treatment (decided 2026-10-27 by doing it in #15777 and having it reviewed, rather than in advance). The open question asked whether this is a modelling fix or a clinical-content change belonging to whoever curated the entry. The split that settled it kept every agent and every citation and only separated two mechanisms that were asserted as one, carrying all four of the treatment's references onto both halves with per-class explanations; nothing about which drugs the entry recommends changed. Removing or adding an agent remains the original curator's call (see the ranitidine item under Deferred follow-ups) — the line is between restating what an entry already says and changing what it says.

Two corrections to the question's own framing came out of the same work, and both are recorded above: bundling two drug classes into one treatment does not block conformance when the agents are bound separately, so five of the six entries the question listed as blocked -- the four mastocytosis entries and Cough_Variant_Asthma -- were never bundling problems at all; and the sixth, Chronic_Obstructive_Pulmonary_Disease, was blocked by a generic agent term covering more than one class rather than by bundling as such, which is what actually blocks conformance.

Deferred follow-ups

Items raised in review and deliberately not done at the time, recorded here so they are not lost. None blocks anything.

• Join the H1 anticholinergic node to the pathograph. Anticholinergic Effects of First-Generation Antihistamines in h1_receptor_antagonism has no edges at all, which is deliberate: its effects are conventionally muscarinic, no cited abstract says so, and an edge from H1 occupancy would assert that blocking H1 causes thirst and tachycardia. The #15686 reviewer suggested a target_mechanisms link from the first-generation treatment instead, which asserts only that the agents cause them and rests on the sentence already cited. Not pushed because it would have reset an approval for a rendering improvement.
• Drop ranitidine from Gastroesophageal_Reflux_Disease's H2 treatment, or decide to keep it. That treatment binds ranitidine while conforming to a module that leaves it unbound, because NDMA contamination led to recalls and registration suspensions (PMID:35696227). The entry records the caveat and cites that source, but the roster is unchanged: removing an agent is a clinical-content decision and was another curator's call. Whoever takes it should also delete the trailing clause "Whether to drop it from the roster is a clinical-content decision and has not been made here", which describes a pull request rather than the treatment.
• ~~Split the four entries that bundle an H1 and an H2 antihistamine into one treatment.~~ Done in #15778, and the premise did not survive contact with the four entries: no split was needed. A treatment binding its components as separate agents can carry one target_mechanisms link per receptor and assert nothing false, so bundling two classes into one treatment does not block conformance by itself — only a generic agent does, which is what actually blocked Chronic_Obstructive_Pulmonary_Disease's bronchodilator treatment. What the four needed instead:
  • Mast_Cell_Activation_Syndrome — H1/H2 Antihistamines, already binding cetirizine and famotidine separately. Wired as it stood.
  • Systemic_Mastocytosis — Anti-Mediator Symptomatic Therapy, already binding an H1 and an H2 class term plus cromoglicate and adrenaline separately. Wired as it stood; the H1 link names the H1 agent and says nothing about the other three.
  • Maculopapular_Cutaneous_Mastocytosis — H1 and H2 Antihistamines, bound only the H1 class term. Gained NCIT:C29702.
  • Monoclonal_Mast_Cell_Activation_Syndrome — Anti-mediator Therapy, bound no agent at all. Gained NCIT:C29578, NCIT:C29702 and NCIT:C29714; still the hardest of the four, and the one a conformance pass would have silently skipped.
• Correct h2_receptor_antagonism's notes, which state the superseded premise as fact. The note says three entries bundle an H1 and an H2 antihistamine and that "None can be routed to this module's occupancy node, or to the H1 module's, without asserting one receptor's action for the other receptor's drug". Both halves are wrong: there are four, and #15778 routed all four to the H1 module without asserting anything of the kind. h1_receptor_antagonism's equivalent passage was corrected in the same pull request; the two were worded differently, so fixing one did not find the other. A kb/ change, so it could not ride this projects/-only pull request.

Open questions

• How should biased agonism (G protein vs arrestin, e.g. oliceridine) be represented — a node within the agonism module, or a hypothesis group?
• Should the peptide-directed anti-CGRP antibodies (fremanezumab, galcanezumab, eptinezumab) get a ligand-sequestration module of their own? They reduce signalling through the same receptor one step upstream, so such a module would share cgrp_receptor_antagonism's effector chain — which is a reason to decide it deliberately rather than fold them in. Migraine's anti-CGRP antibody treatment bundles one receptor-directed and two peptide-directed agents and wants splitting either way.
• Is tissue a third scoping axis? The scoping rule under Decisions is one receptor family plus one direction of action. Beta2 agonism does not fit it: thirteen entries treat with a beta2 agonist, for three mechanistically disjoint purposes that share one trigger and one transducer and diverge only at the effector — airway smooth muscle relaxation (Asthma, Thunderstorm_Asthma, Cough_Variant_Asthma, Chronic_Obstructive_Pulmonary_Disease, Formaldehyde_Poisoning, Sulfur_Mustard_Poisoning); enhancing neuromuscular transmission (Congenital_Myasthenic_Syndrome and its _6 and _7 entries, ALG2-Congenital_Disorder_of_Glycosylation, Combined_Oxidative_Phosphorylation_Deficiency_58, RYR1-Related_Myopathy); and driving potassium intracellularly (Hyperkalemic_Periodic_Paralysis). beta2_adrenergic_receptor_agonism is scoped to the airway on this table's own wording plus the tissue-scoped precedent of cardiac_camp_pka_calcium_handling and spinal_hsp90_opioid_enhancement, and that scope does not settle the question. Either the other two chains become their own modules, or they become further effector arms off a shared trigger — in which case one module would serve about ten conformers instead of three. That figure is three already wired, plus the six unwired entries whose beta2 agonist sits in a treatment of its own, plus ALG2-Congenital_Disorder_of_Glycosylation, where albuterol shares a treatment with ephedrine — a borderline case, since ephedrine is itself a sympathomimetic with beta activity rather than a different class. It assumes the three entries blocked by genuine cross-class bundling stay out; unbundling all three would make it thirteen. MC4R and V2 have the same shape, so this is worth deciding before either.

muscarinic_receptor_antagonism (#15795) put numbers on the same question, and they are starker. A census taken while writing it found antimuscarinic drugs used across the corpus for four mechanistically disjoint purposes sharing that module's trigger and diverging at the effector: airway bronchodilation, 3 entries; detrusor overactivity and neurogenic bladder, 11; striatal cholinergic blockade in dystonia and related movement disorders, chiefly trihexyphenidyl, 20; and reversal of a cholinergic crisis with atropine, 10. The module is scoped to the airway because that is what this page's queue row named — which means it is scoped to the smallest of the four arms, on the wording of a table rather than on the biology, and the other 41 entries have no module. If tissue is a scoping axis, that is three more muscarinic modules to write; if it is not, this one should be widened into a shared trigger with four effector arms. Either way the current state is an artefact, and the module's notes say so.

A fourth group names a beta2 agonist without treating with one: several cardiac entries cite beta-adrenergic stimulation as an arrhythmia provocateur or a thing to avoid. Those are not conformer candidates in any scoping, and the count above excludes them.

The three blocked airway entries are declared in this page's frontmatter, so they link from the prose above. The neuromuscular and potassium-shift entries deliberately are not, pending this decision: declaring them would put seven more entries in the project's disease list on the strength of a scoping question nobody has answered yet. Add them if the answer is one shared trigger. - Should there be a module for the H2-mediated symptoms antihistamines treat? Raised by #15778. The four mastocytosis entries take an H2 antagonist for flushing, gastric hypersecretion and hypotension in mast cell mediator release, and h2_receptor_antagonism cannot hold that: it is scoped to the parietal-cell acid-secretion chain, which is one of those symptoms and not the others. So the H2 half of those four treatments is unwired for a reason unrelated to the bundling premise above — it is module scope, and the remedy is either a second H2 module scoped to mediator-release symptoms or a widening of the existing one's effector arms. Both options cross the one-receptor-one-direction rule in the same way the tissue question above does, which is a reason to decide them together.