The oral bioavailability ceiling on 3-to-5-kilodalton lipidated or otherwise protease-stabilized metabolic peptides is set by a physical problem shared across the class. A peptide at four kilodaltons with negative LogP and zwitterionic character faces a product of four barrier fractions: luminal proteolytic stability against pepsin at gastric pH one to two, diffusion across an eighty-micrometer adherent mucin gel with a hundred-to-five-hundred-nanometer mesh, transepithelial crossing across a gastric-corpus or duodenal columnar monolayer with a transepithelial electrical resistance near one thousand ohm square centimeters and a tight-junction pore diameter near four nanometers, and first-pass hepatic retention. For semaglutide (Novo, GLP-1 receptor agonist, 4113.58 Da) the product of these four fractions caps oral bioavailability at approximately one percent, published as Rybelsus; the same four-term product applies with only modest parameter shifts to liraglutide (Novo, GLP-1 receptor agonist, 3751.2 Da), tirzepatide (Lilly, GLP-1 and GIP dual agonist, 4813.5 Da), exenatide (AstraZeneca/Amylin, GLP-1 receptor agonist, 4186.6 Da), cagrilintide (Novo, amylin analog, 3731.2 Da), retatrutide (Lilly, GLP-1 and GIP and glucagon triple agonist, 4731 Da), and survodutide (Boehringer Ingelheim and Zealand, GLP-1 and glucagon dual agonist, 4100 Da). Individual permeation enhancers, mucus-penetrating nanoparticle architectures, and ingestible devices are published in the peer-reviewed literature. None of the published architectures combines the two chemically orthogonal absorption routes available across the epithelium: transcellular via SNAC-filled partially fluid membrane defects (Buckley et al., Sci Transl Med 2018; Colston, Faivre, Schneebeli, Nat Commun 2025, PMC12568990), and paracellular via CAGE choline-geranate ionic-liquid tight-junction leak-pathway opening plus mucin-viscosity reduction (Banerjee et al., PNAS 2018, PMC6048483; Neville et al., Adv Funct Mater 2024). Because the transcellular and paracellular routes are physically orthogonal parallel pathways across the same epithelial monolayer, their fractional contributions to absorption combine additively on F_epithelium as a first-order approximation at small fractions. The architecture disclosed in the companion technical document combines SNAC at two-hundred-to-four-hundred milligrams, CAGE at thirty-to-eighty milligrams (fifty milligrams nominal), a selective tight-junction modulator (PIP640 decapeptide analog or low-molecular-weight chitosan) at five milligrams, a PEG-2kDa-coated 100-nanometer diblock PLGA-PEG nanoparticle carrier at fifty milligrams, and a thiolated chitosan mucoadhesive outer matrix at twenty-to-fifty milligrams, with active peptide payload at a peptide-specific daily dose in a tablet of 650-to-800 milligrams total mass (strict minimum 650 milligrams, SNAC weight fraction strictly below 50 percent, disintegration time strictly above 22 minutes, bulk density below 1.0 gram per cubic centimeter, each boundary set to fall outside the independent-claim space of Novo WO 2013/189988 and its granted continuations). Theoretical human oral bioavailability across the class, with each component at its literature-validated condition, spans four-to-fourteen percent nominal eight-point-six percent, four-to-fourteen-fold Rybelsus; effective bioavailability pre-bench, applying translation penalties for gastric-pH ionic-liquid state of CAGE and for sub-Banerjee local concentration and for gastric-corpus versus Caco-2 claudin profile, spans two-to-seven percent nominal three-to-four percent, two-to-seven-fold Rybelsus. Peptide-specific F bands shift by six-to-twelve percent around these numbers reflecting molecular-weight, charge, and lipidation differences; per-peptide quantitative budgets are given in the peptides subdirectory of the companion deposit. A secondary embodiment across the class replaces the tablet with a pulsed-field ingestible capsule operating at ten-to-one-hundred kilohertz (empirical frequency range) with one-to-ten volts per centimeter bulk field at one-percent duty cycle via a sub-electroporative iontophoretic-drift mechanism. A Mitragotri-independent variant of the primary embodiment omits CAGE and substitutes a second selective tight-junction modulator, preserving route-additive architecture at reduced F. This note walks through the shared physical problem, the class-level prior art, the route-additive insight, and the unified architecture at quantitative depth across all seven peptides. The companion technical disclosure publishes the architecture under 35 U.S.C. §102 as prior art of record.

The oral metabolic-peptide moment

The cultural moment for GLP-1 receptor agonists peaked in the first half of 2026 and remains loud. Global semaglutide revenue alone is approaching thirty billion dollars annually across injectable and oral forms. Tirzepatide surpassed semaglutide in new prescription growth through 2025 and is on track for comparable or greater annual revenue on the strength of its dual-agonist mechanism. The CagriSema combination of cagrilintide and semaglutide is in Phase 3 with obesity and diabetes indications. Retatrutide, Lilly’s triple-agonist, is in Phase 3 and has reported weight-loss endpoints exceeding tirzepatide. Survodutide, a GLP-1 and glucagon dual agonist, is Phase 3 for metabolic dysfunction-associated steatohepatitis and obesity. Liraglutide and exenatide, the two older GLP-1 receptor agonists with established regulatory history, remain in global use with generic exenatide shipping in many markets.

The economic consequence of the class’s one-to-two-percent oral bioavailability cap is retail pricing in the eight-hundred to one-thousand-two-hundred dollar range per month of therapy in the United States, in which the yearly per-patient manufacturing cost of active ingredient may be ten to fifty dollars. Ninety-eight to ninety-nine percent of the active ingredient in each swallowed tablet of oral semaglutide never reaches systemic circulation. The public-health consequence is that a class of compounds with demonstrated effects on glycemic control, body weight, cardiovascular endpoints, renal outcomes, and more recently on addictive behaviors and liver pathology is priced out of population-scale access.

The patent-filing moment for next-generation oral formulations in this class is within the window of the next one-to-three years. Novo Nordisk’s first-generation SNAC-based Rybelsus tablet is protected by WO 2013/189988 and U.S. continuations US 9,993,430 and US 11,033,499, the earliest of which expire in the early-to-mid 2030s. Second-generation architectures at higher F are in late preclinical and early clinical development. Lilly’s oral orforglipron, a non-peptide small-molecule GLP-1 receptor agonist, is in Phase 3 and represents a different delivery class that this note does not address. For the peptide oral-delivery problem, the multi-enhancer route-additive architecture is not yet published in combination and not yet filed by any incumbent in the form disclosed here. The window for defensive publication is open.

Why class-level disclosure

The mechanism of each absorption barrier is peptide-class general at the molecular weight and charge range covered by these seven compounds. Pepsin’s selectivity is aromatic-residue-biased rather than peptide-specific; mucus-mesh geometry is substrate-agnostic; tight-junction leak-pathway opening responds to CAGE concentration rather than to the solute crossing; transcellular hydrophobization via SNAC-filled defects depends on lipophilic-tag presence (C18 diacid on semaglutide, C16 fatty acyl on liraglutide, C20 fatty-diacid on tirzepatide, C18 diacid on cagrilintide, similar on retatrutide and survodutide; native exenatide has no lipidation and depends on charge-pairing rather than tail insertion). A unified multi-enhancer architecture transposes across the class with peptide-specific parameter shifts in stoichiometry, dose, and compositional detail but with architectural identity preserved.

A class-level defensive publication places the combined route-additive architecture into prior-art-of-record across all seven peptides simultaneously. A peptide-by-peptide disclosure sequence would permit serial patent filings on each untouched peptide by incumbents during the gap between drops. A single bundled class-level publication removes that possibility in one event.

Four shared barriers

Oral bioavailability of a peptide therapeutic at three-to-five-kilodalton molecular weight with negative LogP (semaglutide LogP = negative 5.8; tirzepatide LogP = negative 4.7; others similar) and zwitterionic character is set by the product of four barrier-fraction terms:

F_oral = F_stability × F_mucus × F_epithelium × F_first-pass

Each term has a specific physical origin and a specific dominant mechanism. The physical origin of each term is shared across the peptide class at this MW and charge range. Differences between peptides resolve to differences in parameter values inside each term, not to differences in the term structure.

F_stability is the fraction of ingested peptide that survives the gastric and intestinal lumen intact. Pepsin at pH 1.5-to-2.5 hydrolyzes peptide bonds preferentially at aromatic-residue sites (Heda, Toro, Tombazzi, StatPearls NBK537005) with K_cat falling by a factor of one-hundred-to-one-thousand as pH rises above five. Pancreatic trypsin and chymotrypsin extend hydrolysis in the duodenum. F_stability for Rybelsus is approximately 0.75 (Buckley et al., Sci Transl Med 2018), set by the local pH-buffering effect of SNAC at three-hundred-milligram dose. Peptide-specific variation on F_stability at baseline is small at this MW range: all seven peptides are similarly susceptible to gastric proteolysis in the absence of stabilizing formulation, with small differences driven by sequence-specific pepsin selectivity that the SNAC pH-buffer largely masks.

F_mucus is the fraction of surviving peptide that diffuses from the lumen through the adherent gastric or duodenal mucus gel to the epithelial cell surface. The adherent gastric corpus mucus is approximately eighty micrometers thick (Atuma et al., Am J Physiol Gastrointest Liver Physiol 2001), with a hundred-to-five-hundred-nanometer mesh spacing (Lai, Wang, Hanes, Adv Drug Deliv Rev 2009, PMC2667119) and a viscosity of one-thousand-to-ten-thousand times water at low shear (Cone, Adv Drug Deliv Rev 2009). Duodenal mucus is thinner (fifteen to thirty micrometers) with higher turnover but similar mesh geometry. Free peptide at three-to-five kilodaltons diffuses approximately ten times slower through mucus than through water. Rybelsus F_mucus is approximately 0.45, set by transit across unmodified mucus at the gastric-corpus absorption site.

F_epithelium is the fraction of peptide reaching the cell surface that crosses the epithelial monolayer into systemic blood. This is where the interesting physics is. Two parallel routes exist: transcellular, across the apical and basolateral plasma membranes of the columnar cell, and paracellular, through the tight-junction seal between adjacent cells. For a peptide at three-to-five kilodaltons with LogP near negative five, both routes present significant resistance. Rybelsus uses the transcellular route only, at F_transcellular approximately 0.03; the paracellular route is essentially closed. F_epithelium for Rybelsus is therefore also approximately 0.03.

F_first-pass is the fraction of peptide that enters systemic circulation intact after first passage through the portal venous system and the liver. For lipidated GLP-1 receptor agonists and related analogs, the fatty-acid side chain binds plasma albumin with high affinity and the peptide is not significantly metabolized in first pass (Knudsen and Lau, Front Endocrinol 2019). F_first-pass is effectively one across all seven peptides at this MW range. The only class member without strong lipidation-based albumin binding is exenatide, which has a native half-life of approximately 2.4 hours and depends on different formulation strategy for dose-interval management; first-pass retention nonetheless remains effectively one per hepatic non-extraction at this MW and charge.

Multiplying the four barriers for Rybelsus gives approximately 0.75 × 0.45 × 0.03 × 1.0 ≈ 0.01. The one-percent figure is derivable from the mechanism of each barrier alone, and every term except F_first-pass is individually improvable.

What has absorbed

Individual improvements to each term are extensively published.

Permeation-enhancer chemistry has a taxonomy (Maher, Brayden, Feighery, McClean, Pharmaceutics 2019). Class I transcellular hydrophobization agents include SNAC, 5-CNAC, and hydrophobic-ion-pairing strategies; these coat the peptide in lipophilic counter-ions or form dynamic membrane defects that the peptide can traverse, preserving the epithelial barrier structurally while admitting a small peptide fraction. Class II transcellular membrane-perturbation agents include medium-chain fatty acids (sodium caprate C10), acylcarnitines, bile salts, and sucrose laurate; these partition into the bilayer and produce membrane fluidization with higher flux and more histological disturbance. Class III first-generation paracellular openers include EDTA and chitosan; these open tight junctions via cation chelation or cationic binding to claudin extracellular loops. Class IV second-generation paracellular openers include the zonula occludens toxin fragment AT-1002, Clostridium perfringens enterotoxin fragments, and the PIP640 decapeptide; these target specific claudin isoforms for selective, reversible permeability modulation.

The Banerjee, Ibsen, Brown, Chen, Agatemor, and Mitragotri 2018 PNAS report establishes that choline geranate, a one-to-two molar ratio ionic liquid of the cholinium cation and the geranate anion (an unsaturated ten-carbon monoterpenoid carboxylate), delivers insulin orally to rats at intrajejunal dose of five international units per kilogram (eighty microliters of neat CAGE per rat, corresponding to approximately five percent weight per volume in rat jejunal volume) with forty-five-to-fifty-one-percent bioavailability relative to subcutaneous two-IU-per-kilogram dosing. The ex-vivo data on Caco-2 monolayers give the dominant mechanism: at fifty millimolar CAGE (the highest concentration tested, equivalent to approximately five percent weight per volume), a thirteen-fold enhancement of paracellular flux of 4 kDa FITC-dextran, a forty-to-fifty-percent reduction in transepithelial electrical resistance, and significant reduction in mucin hydrogel viscosity. The four-kilodalton dextran marker is paracellular-exclusive, so the thirteen-fold flux enhancement localizes the CAGE mechanism to the tight junction. The mucus-viscosity reduction adds an independent mucus-fluidization effect. The 2018 PNAS work has been captured in the Harvard / UC patent family WO 2019/099837 A1 covering oral CAGE-active-compound compositions including GLP-1 polypeptides at the 1:2 choline-to-geranate ratio disclosed here; commercial deployment of the architecture’s CAGE component therefore requires licensing against this Mitragotri / Harvard patent family, acknowledged in the companion disclosure as cited prior art. The acknowledgment applies across all seven peptides in this note that include the CAGE primary embodiment.

Mucus-penetrating nanoparticle (MPP) design rules were established by Ensign, Schneider, Suk, Cone, and Hanes (Adv Mater 2012, PMC3710133). Particles in the fifty-to-five-hundred-nanometer size range, coated with dense two-kilodalton poly(ethylene glycol) via PLGA-PEG block copolymer or equivalent surface chemistry, with near-neutral zeta potential (negative two plus-or-minus four millivolts at physiological pH), penetrate mucus at approximately one-quarter the diffusion coefficient expected in pure water. The MPP design rules are peptide-agnostic and independent of permeation-enhancer chemistry.

Device-assisted ingestibles form a separate track. Abramson, Traverso, Langer and colleagues (Abramson et al., Science 2019, PMC6430586) demonstrated the Self-Orienting Millimeter-scale Applicator (SOMA), a nine-by-fifteen-millimeter ingestible capsule that self-orients in the stomach and inserts a sucrose-isomalt-locked seven-millimeter insulin millipost into the gastric submucosa. The RaniPill family (Hashim et al., Pharmacol Res Perspect 2021; Imran, Rani Therapeutics, US 10,814,114 and continuations) uses a pH-triggered enteric coating, a gas-generating balloon for mechanical actuation, and a dissolvable microneedle that inserts payload through the intestinal wall. These architectures target the absorption problem from a different direction: they bypass the epithelial barrier with a mechanical vector rather than a chemical one. The architecture disclosed here is chemical-enhancer-based; a pulsed-field ingestible secondary embodiment is separately disclosed.

Permeation-enhancer combinations at smaller scale have been explored. The sodium-caprate-plus-SNAC combination has been tested in non-human primate oral GIP/GLP-1 co-agonist formulations (2023-2024 development literature), showing synergistic effects at three-to-six-fold intestinal scale; both enhancers in that combination are transcellular Class I/II. No published combination of SNAC with an ionic-liquid paracellular enhancer has been characterized at any peptide payload across the GLP-1 receptor agonist, dual-agonist, triple-agonist, or amylin class. No published architecture combines SNAC, a CAGE-class ionic-liquid enhancer, a selective claudin-targeted tight-junction modulator, a mucus-penetrating nanoparticle carrier, and a mucoadhesive outer matrix in a single tablet at any of the seven peptide payloads listed below.

The first barrier: luminal stability

Pepsin at the fasted gastric pH of 1.5-to-2.5 is the primary proteolytic threat across the class. The salicylamide head of SNAC, combined with its caprylate tail, delivers a buffering capacity sufficient to raise the local pH of the tablet-erosion microenvironment to approximately five, as measured by Buckley and colleagues in porcine gastric loops and as confirmed by the Colston 2025 constant-pH molecular dynamics simulation. At local pH five, pepsin K_cat drops by a factor of one-hundred-to-one-thousand from its low-pH optimum. The pH-buffering mechanism is the reason Rybelsus achieves F_stability near 0.75 rather than values approaching zero, and it transposes across the class at similar quantitative effect.

The CAGE-analog ionic-liquid component adds an independent stabilization term, with a caveat on timing. Geranic acid has pKa approximately 5.17; at fasted gastric pH one to two, the geranate species is more than ninety-nine-point-nine-percent protonated to neutral geranic acid, meaning the choline-geranate ionic-liquid stoichiometry does not exist at gastric pH. The ionic-liquid character reassembles only after SNAC buffers the local microenvironment to pH five to six, which takes five to fifteen minutes after tablet disintegration begins. During this window, CAGE’s stabilization effect is approximately zero; its contribution begins with the SNAC-buffered phase and extends through the remainder of the 60-to-90-minute absorption window. Palanisamy and Prakash (Phys Chem Chem Phys 2021, DOI 10.1039/D1CP03349B) characterized CAGE with insulin dimer in aqueous solution via atomistic molecular dynamics in GROMACS 2020.4 and reported that at zero-point-three to zero-point-five mole-fraction CAGE, the geranate anion preferentially forms hydrogen bonds with water and accumulates on the protein surface, reducing the water activity in the local peptide hydration shell. This matters for enzyme kinetics in the post-buffering phase: proteolysis is a water-dependent hydrolysis reaction, and reducing the water activity in the immediate peptide environment slows the reaction kinetics additively to the pH-buffering effect once CAGE is in its ionic-liquid state.

Agatemor, Ibsen, Tanner, and Mitragotri (Adv Ther 2021) reported that native GLP-1 co-formulated with CAGE and administered subcutaneously produces an area-under-curve four times greater than saline-vehicle control, via dipeptidyl-peptidase-4 inhibition and ionic-liquid self-assembly effects. Native GLP-1 is DPP-4 sensitive; semaglutide and liraglutide are DPP-4 resistant via the Aib8 substitution on sema and the Arg34 substitution on lira, so the DPP-4 inhibition benefit of CAGE transfers minimally to these payloads; tirzepatide, retatrutide, and survodutide have non-native backbones with engineered protease resistance of varying degrees; exenatide has a native Gly at position 8 (exendin-4 sequence, different from human GLP-1) and is natively DPP-4 resistant. The CAGE-stabilization effect on lipidation and hydration-shell protection applies across the class regardless of DPP-4 sensitivity.

The nanoparticle component (Component E in the architecture, a 100-nanometer PLGA-PEG 2kDa diblock-copolymer particle with acid-sensitive acetal crosslinker) adds a third stabilization term via co-encapsulation of the peptide and SNAC at 20-to-40-percent SNAC encapsulation efficiency and 30-to-60-percent peptide encapsulation efficiency, corresponding to a 1000-to-2500:1 SNAC-to-peptide molar ratio within the nanoparticle core. The core disassembles at pH five-to-six in the SNAC-buffered microenvironment, releasing payload during the absorption window rather than at the gastric lumen pH. The encapsulated peptide fraction is effectively shielded from pepsin throughout the mucus-transit phase. The mechanism is peptide-agnostic.

Combining the three mechanisms, F_stability rises from the Rybelsus 0.75 to approximately 0.88 at nominal design parameters across the class. The primary gain is from the nanoparticle encapsulation; the secondary gain is from CAGE hydration-shell water-activity reduction; the pH-buffering contribution is preserved unchanged from Rybelsus.

The second barrier: mucus diffusion

The gastric corpus adherent mucus layer is a polymeric gel of disulfide-crosslinked MUC5AC mucin polymers (molecular weight five-hundred-kilodalton to forty-megadalton) and O-glycosylated side chains. At eighty micrometers thickness and a hundred-to-five-hundred-nanometer mesh spacing, the gel presents a geometric sieve plus a physicochemical-interaction barrier: particles that bind mucin via hydrophobic, cationic, or hydrogen-bonding interactions adhere and fail to transit. Duodenal and jejunal mucus layers are thinner and less stratified but structurally comparable in mesh spacing.

The MPP design rule set from Ensign 2012 addresses both components. A fifty-to-five-hundred-nanometer particle size fits within the mesh. A dense two-kilodalton PEG surface coating via PLGA-PEG block copolymer eliminates hydrophobic adhesion to mucin. A near-neutral zeta potential at physiological pH eliminates cationic adhesion to the sulfate and sialic-acid groups on mucin side chains. Particles meeting all three criteria diffuse through mucus at approximately twenty-five percent of the pure-water diffusion coefficient, compared to less than one percent for unmodified PLGA. These design rules are peptide-agnostic.

CAGE adds a second mechanism via the mucin-viscosity reduction reported in Banerjee 2018. The mucus-fluidization effect is not mesh-spacing modification but polymer-chain dynamics modification: the kosmotrope-chaotrope geranate anion disrupts the hydrogen-bond-mediated self-association of MUC5AC polymers, reducing the effective viscosity of the gel at one-to-five-percent weight-per-volume CAGE concentration. The effect is concentration-dependent and reversible. Quantitative mucin-viscosity values are not explicit in the Banerjee paper, but the qualitative statement “significantly decreased” at both one-percent and five-percent CAGE loading, paired with the CAGE dose in the disclosed architecture (fifty milligrams per tablet, approximating one-to-five-percent local concentration after tablet erosion in the gastric or duodenal microenvironment), supports a three-to-ten-fold fluidization factor.

The mucoadhesive outer tablet matrix (Component F, thiolated chitosan or chitosan-glutathione copolymer at twenty-to-fifty milligrams) addresses a related question: residence time at the absorption window. Rybelsus tablets erode at a rate slow enough that gastric residence is approximately one hour, limited by gastric-emptying kinetics of the post-ingestion fasted state. A mucoadhesive outer shell extends gastric residence to two-to-four hours via cationic chitosan binding to sialic-acid and sulfate residues on the mucus surface, supplemented by disulfide-bridge formation between the thiolated chitosan and mucin-cysteine thiol groups. The extended residence window compounds multiplicatively with the other mucus-phase gains. The mechanism is peptide-agnostic.

Combining the three mechanisms (MPP diffusion, CAGE fluidization, extended residence), F_mucus rises from the Rybelsus 0.45 to approximately 0.75 at nominal design parameters across the class.

The third barrier: epithelial crossing

This is the load-bearing section, and the route-additive insight is the core contribution of the disclosed architecture. The mechanism distinction between SNAC and CAGE is what makes route-additivity available across the class.

SNAC acts transcellularly. Buckley et al. 2018 established via in-situ porcine gastric-loop and in-vivo dog studies that SNAC absorption of semaglutide occurs at the gastric corpus, not the intestine; that the mechanism is transcellular, not paracellular, per 4-kilodalton FITC-dextran exclusion; that SNAC is rapidly degraded in the gastric mucus and does not accumulate systemically; and that SNAC enables peptide-specific absorption rather than generalized peptide flux. The Colston et al. 2025 continuous constant-pH molecular dynamics simulation in custom GROMACS at one-microsecond scale with four replicas plus umbrella-sampling PMF calculation elaborates the mechanism at atomic resolution: approximately four-hundred SNAC molecules at pH five in a POPC-cholesterol 60/40 bilayer form dynamic, partially fluid membrane defects into which the semaglutide peptide is partially submerged in a process the authors term a “quicksand”-like submersion. The paper demonstrates C18 tail insertion and partial peptide submersion into SNAC-filled defects; it does not demonstrate trans-leaflet translocation of the whole 4-kilodalton peptide. The computed free energies are a negative five-kilocalory-per-mole gain for C18 diacid tail insertion into the SNAC-occupied defect and a positive seven-point-five-kilocalory-per-mole barrier for pulling the C18 tail terminus across the opposite leaflet’s phosphate headgroup region while the peptide body remains water-accessible. The authors frame their mechanism as one of several possible pathways for semaglutide membrane traversal in the presence of SNAC. The Kneiszl, Hossain, and Larsson 2021 comparative study (Mol Pharm 2022, PMC8728740) gives SNAC all-atom umbrella-sampling free energy of membrane partition at negative 1.3 kilocalories per mole at a POPC bilayer without cholesterol or other permeation enhancers; the paper’s seventy-to-one-hundred millimolar concentration range is a water-permeation onset specific to sodium caprate and laurate (not SNAC or universal), and Kneiszl rates SNAC as less effective than the medium-chain fatty acids in this model.

The SNAC transcellular mechanism depends on the peptide having a lipophilic anchor that inserts into SNAC-filled defects. Semaglutide has a C18 diacid at Lys26. Liraglutide has a C16 mono-fatty-acyl at Lys26 (shorter and less deeply anchoring, reducing the negative-five gain to approximately negative three-point-five at nominal). Tirzepatide has a C20 fatty diacid at Lys20 (longer, reaching negative six-point-zero). Cagrilintide has a C20 fatty diacid at the N-terminus (comparable to tirze C20 chain-length at Lys20; position differs but Wimley-White well depth is similar). Retatrutide has a C20 fatty diacid at Lys17 (comparable to tirze). Survodutide has a C18 fatty diacid at Lys24 via γ-glutamic acid linker only, no OEG extension (chain-length comparable to sema but shorter linker; slight F_trans reduction vs sema baseline). Exenatide is unlipidated (exendin-4 native sequence); the SNAC transcellular mechanism for exenatide depends on hydrophobic ion pairing rather than tail insertion and operates at approximately half the efficiency of lipidated peptides in the architecture. Peptide-specific F_transcellular values are given in the per-peptide F budget tables below.

CAGE acts paracellularly, at concentrations that are bench-testable against the architecture’s deployment regime. Banerjee et al. 2018 reported, in Caco-2 intestinal monolayers at fifty-millimolar CAGE (approximately five percent weight per volume, the highest concentration tested), a thirteen-fold enhancement of the paracellular flux of 4-kilodalton FITC-dextran with forty-to-fifty-percent TEER reduction. Lower CAGE concentrations (ten and twenty-five millimolar in the same paper) give smaller enhancements per the dose-response. The 4-kilodalton FITC-dextran is a paracellular-exclusive marker. The in-vivo consequence is rat intrajejunal insulin bioavailability of forty-five-to-fifty-one percent relative to subcutaneous two-IU-per-kilogram dosing, delivered as eighty microliters of neat CAGE per rat (approximately five percent weight per volume in rat jejunal volume). The Neville et al. 2024 experimental characterization at DMPC solid-supported bilayers via neutron reflectivity, quartz crystal microbalance, dynamic light scattering, and NMR confirms the mild-disruption picture: CAGE inserts and diffuses across the bilayer while preserving bilayer integrity. The Ibsen et al. 2018 molecular dynamics at bacterial membrane (ACS Biomater Sci Eng) shows the molecular-level mechanism: choline is electrostatically attracted to the membrane surface while geranate inserts into the hydrophobic tail region, disrupting lipid packing at moderate concentration without lytic damage. The paracellular mechanism is peptide-agnostic: the CAGE effect is on the tight-junction seal, not on the solute.

Two translation uncertainties apply to CAGE in the architecture’s deployment. First, Banerjee’s Caco-2 cells are colon-adenocarcinoma-derived and express claudin-2 (cation-selective), -3, -4, -7, -15. Gastric corpus epithelium is claudin-3, -4, -18 dominant, with claudin-18.2 forming a specialized H+/Na+ barrier. Duodenal epithelium is closer to Caco-2 but not identical. The 4-kilodalton FITC-dextran flux enhancement is a “leak-pathway” effect (transiently broken bicellular and tricellular junctions) common to all epithelia, but whose quantitative response to CAGE has been measured only in Caco-2, not in gastric corpus or duodenal tissue. Second, the architecture’s 50-milligram CAGE dose distributes to a local microenvironment at 0.2-to-1-percent weight per volume depending on absorption-site volume and gastric emptying state, below Banerjee’s tested 5-percent weight-per-volume regime for the 13-fold enhancement. Both uncertainties are bench-testable: Caco-2 dose-response at 0.2, 1, 2.5, and 5 percent weight per volume CAGE, and porcine gastric and duodenal Ussing chambers with the full formulation. The §10 validation protocol is designed to measure these translation factors directly.

The two mechanisms are physically orthogonal. Transcellular SNAC-mediated absorption acts on the apical bilayer via SNAC-filled membrane defects. Paracellular CAGE-plus-selective-modulator opening acts on the tight-junction seal between adjacent cells. The two routes pass through different anatomical locations on the same epithelial monolayer, with no shared rate-limiting structure. At the small fractional contributions here (both below ten percent), their fractional contributions to epithelial crossing add to first order:

F_epithelium = F_{transcellular} + F_paracellular

Prior attempts to combine enhancer classes at the same epithelium have been limited to within-class combinations (caprate plus SNAC, both transcellular perturbers) or to single-class trials at elevated concentration. The across-class combination of a transcellular hydrophobization agent with an ionic-liquid paracellular opener, at any of the seven peptide payloads in this class, has not been published.

A selective tight-junction modulator (Component D, PIP640 decapeptide analog or low-molecular-weight chitosan) adds a third, selective paracellular contribution. PIP640 opens the claudin-2 pore specifically via MYPT1-binding and phospho-dependent claudin-2 stabilization (Maher et al., 2019), providing cation-biased paracellular transport at a more controlled concentration than non-selective CAGE effects. Chitosan is a commercial alternative that opens tight junctions via cationic binding to occludin and ZO-1, disclosed as a variant in the companion disclosure §12 to ensure commercial backup without license dependency on PIP640-derivatives.

At nominal theoretical design parameters across the class, F_transcellular rises from the Rybelsus 0.03 to approximately 0.05 (a 1.3-to-2-fold gain reflecting CAGE’s mild additive bilayer perturbation on top of the SNAC baseline, modulated by peptide-specific lipidation), and F_paracellular rises from the Rybelsus 0.00 to 0.08 (set by CAGE tight-junction modulation at Banerjee’s validated regime plus the selective Class IV opener, peptide-agnostic). F_epithelium rises from 0.03 to 0.13, approximately a four-fold gain.

Under a theoretical pessimistic parameter band (pessimistic F_transcellular 0.03, pessimistic F_paracellular 0.05), F_epithelium is 0.08. Under an optimistic parameter band (optimistic F_transcellular 0.07, optimistic F_paracellular 0.12), F_epithelium is 0.19. The four-barrier-product arithmetic below resolves the full theoretical uncertainty band.

Effective F_epithelium pre-bench, applying the CAGE translation penalties (pH-timing, local concentration below Banerjee’s 5-percent weight-per-volume regime, gastric-corpus claudin-18 profile vs Caco-2 intestinal claudin profile), reduces F_paracellular to approximately 0.02-to-0.04 at nominal. Combined with SNAC release-rate considerations on F_transcellular (local SNAC concentration approximately 40 millimolar at slow erosion, below the cited caprate/laurate water-permeation threshold and comparable to Rybelsus baseline), effective F_epithelium is approximately 0.05-to-0.07 at nominal pre-bench across the class.

The fourth barrier: first-pass

The fatty-acid side chain attached at the lipidation position of each lipidated peptide (Lys26 for sema and lira, Lys20 for tirze and cagri, Lys17 for reta, Lys23 for survo) binds plasma albumin with sub-micromolar affinity. Once absorbed into the portal venous circulation, the peptide is not substantially metabolized on first passage through the liver; the plasma elimination half-life is effectively set by the albumin-bound peptide’s slow renal filtration and non-specific clearance, with weekly-dosing-compatible half-lives for sema, lira (once-daily with C16 anchor), tirze, cagri, reta, and survo. Exenatide, unlipidated, has a 2.4-hour plasma half-life but similarly avoids substantial first-pass hepatic metabolism at this MW and charge range. The first-pass retention fraction is effectively one across the class. This is unchanged by any absorption-engineering modification and is a property of the molecule class, not the delivery architecture. No part of the disclosed architecture interacts with the first-pass pharmacokinetics, and none needs to.

The unified architecture

The architecture specification in the companion technical disclosure combines six component roles at specified dose ranges, with peptide-specific parameter adjustments resolved in the per-peptide embodiment tables:

Theoretical F budget at nominal and bracketing parameters across the class, with each component at its literature-validated condition:

Effective F budget pre-bench, applying translation penalties for CAGE gastric-pH ionic-liquid state, sub-Banerjee local concentration, gastric-corpus vs Caco-2 claudin profile, and SNAC release-rate considerations:

Across the theoretical band, the disclosed architecture is four-to-fourteen-fold Rybelsus. Across the effective band pre-bench, the architecture is two-to-seven-fold Rybelsus. Bench validation against Rybelsus-equivalent SNAC-only control at Caco-2, porcine gastric Ussing chamber, and in-vivo animal PK directly measures the architecture relative to Rybelsus and closes the theoretical-to-effective gap.

Peptide-specific F bands shift by six-to-twelve percent around these class-level numbers reflecting molecular-weight, charge, and lipidation differences. Per-peptide quantitative F budget tables, dose envelopes, and compositional specifications are in the peptides subdirectory of the companion deposit.

Per-peptide embodiment summary

The seven peptides in this class-level disclosure, with SC reference PK and the disclosed oral daily dose at nominal theoretical F:

Each peptide file details molecular profile, SC PK reference, F budget transposition, dose envelope, peptide-specific prior art, and three embodiments (primary CAGE-inclusive, secondary Mitragotri-independent, tertiary pulsed-field). The cagrilintide file additionally discloses the co-administered-with-semaglutide combination variant matching Novo CagriSema at oral payload.

For semaglutide, the worked dose envelope example: at nominal theoretical F 0.086, one milligram oral daily maps to steady-state plasma concentration of seventeen nanomolar, equivalent to subcutaneous semaglutide 0.68 milligrams weekly; at nominal effective F 0.036, the same SC equivalent is reached at approximately 1.7-fold higher oral dose. The full dose-envelope table and parity mapping for semaglutide is in peptides/semaglutide.md. Equivalent tables for the other six peptides are in their respective files.

The April 2026 moment

The cultural moment for oral metabolic peptides at the class level has two parts. The GLP-1 receptor agonist retail pricing in the United States has not declined meaningfully since peak demand in late 2025; access remains gated by cost, manufacturer supply, and insurance formulary. Compounded semaglutide and tirzepatide from FDA 503A and 503B pharmacies addressed a portion of the access gap during the 2024-to-mid-2026 shortage designation windows for each, but safety concerns and enforcement pressure have narrowed the compounding channel. Novo Nordisk and Eli Lilly have responded by expanding manufacturing and by announcing additional price-reduction commitments to government payers, none yet translating to broad retail-channel price relief. The second-generation oral formulations in late-stage clinical development cover lipidated peptide reformulations, non-peptide small-molecule GLP-1 receptor agonists (orforglipron, outside the scope of this note), and next-generation peptides at higher bioavailability. The patent-filing moment for next-generation architectures is within the window of the next one-to-three years.

An open-source defensive publication on a multi-mechanism oral peptide architecture at the class level constrains the patent-filing surface in two specific ways. First, by placing the route-additive multi-enhancer combination at all seven peptide payloads into prior-art-of-record under 35 U.S.C. §102, it bars subsequent claims on the specific combination at the disclosed payloads and dose envelopes across the class. Second, by disclosing the dose envelope and PK parity to subcutaneous dosing for each peptide, it establishes the scientific basis for generic or open-manufacture oral formulations at a one-to-two-order-of-magnitude cost-per-patient-per-year reduction across the GLP-1, dual-agonist, triple-agonist, and amylin markets simultaneously, once the formulation and regulatory work is done per peptide.

The consumer surplus at scale dominates the private cost. Global annual revenue across the seven peptides is approaching fifty billion dollars at current prices. The one-to-two-order-of-magnitude dose reduction implied by a two-to-seven-percent effective bioavailable oral architecture pre-bench (or four-to-fourteen-percent at the theoretical upper envelope after bench validation closes the CAGE translation uncertainties) would, at comparable retail pricing, translate to proportional reduction in per-patient-per-year active-ingredient cost, or at reduced pricing to proportional expansion of the treatable population at a fixed aggregate manufacturing cost. Either direction is a large positive-externality move. The asymmetry of publication versus patent protection on this architecture is simple: patent protection captures a fraction of the private surplus at cost to the public; publication releases the combined-architecture private surplus to the public without capturing any private return. For an architecture derivable from public mechanism data in a domain with a large consumer-surplus-to-private-surplus ratio, publication is the correct action. Commercial deployment of the disclosed architecture requires patent licensing on the individual components where patented by others (Novo US 8,129,343 composition-of-matter on semaglutide API valid through 2033 and parallel composition-of-matter patents on liraglutide, tirzepatide, and so on; Harvard / Cage Bio WO 2019/099837 on oral CAGE compositions valid through ~2038; Hanes MPP patents where applicable by claim language; Lilly and BI composition-of-matter on their respective peptides); the defensive publication blocks later claim-construction on the combination architecture at the disclosed ranges but does not obviate these component licenses.

FTO position at class level

The FTO chart in the companion disclosure provides per-patent claim-by-claim analysis. Summary position:

• Novo Nordisk WO 2013/189988 A1 and granted continuations (US 9,993,430, US 11,033,499): CLEAR across the class. Architecture holds strictly below 50 percent w/w SNAC (independent-claim floor in US 9,993,430), strictly above 22 minutes disintegration (ceiling in US 11,033,499 continuation), and strictly below 1.0 g/cm³ bulk density (US independent-claim floor). Tablet mass strict minimum 650 mg holds SNAC at maximum 46.2 percent w/w at the tight extreme.
• Mitragotri / Harvard WO 2019/099837 A1: LITERAL coverage on the CAGE component at 1:2 choline-to-geranate ratio in oral GLP-1 polypeptide composition. The primary embodiment for each peptide acknowledges this prior art and requires a Cage Bio / Harvard license for commercial deployment. The secondary Mitragotri-independent embodiment for each peptide omits CAGE entirely and operates without license dependency.
• Hanes MPP family (US 8,889,193 ocular, US 9,056,057 nanocrystal, US 9,415,020 / US 9,629,813 hypotonic, WO 2017/075565 PEG >5 kDa): CLEAR across the architecture’s PEG-2kDa nanoparticle design.
• Novo US 8,129,343 (semaglutide composition-of-matter, valid through 2033): LITERAL on the semaglutide API molecule itself. Commercial deployment of the semaglutide embodiment requires either license or post-2033 timing.
• Lilly tirzepatide composition-of-matter patents: LITERAL on the tirzepatide API. Commercial tirzepatide embodiment requires license or post-expiration timing.
• BI/Zealand survodutide, Lilly retatrutide, Novo cagrilintide composition-of-matter: LITERAL on the respective API. Commercial deployment requires license or post-expiration timing.
• Novo liraglutide composition-of-matter: Largely expired in major jurisdictions (2017-2023 across US/EU/JP); generic liraglutide is on market in multiple countries.
• Exenatide composition-of-matter: Expired.

The defensive publication blocks later claim-construction on the combination architecture at the disclosed ranges across the class. It does not obviate component-level licenses required for commercial deployment.

Validation protocol

The disclosed architecture is testable at the class level. The validation path from the publication date forward is the standard preclinical sequence, scorable against published literature values at each gate, with peptide-specific variations in target concentration and dose:

1. In-vitro permeability. Caco-2 monolayer and HT29-MTX mucus-secreting cell permeability assays at 100-micromolar peptide, 10-millimolar SNAC, 1-millimolar CAGE-analog (0.2, 1, 2.5, 5 percent weight per volume dose-response), 10-micromolar PIP640 analog. Target apparent permeability at or above 5 × 10⁻⁶ centimeters per second. Compared with Rybelsus-equivalent SNAC-only control at the same peptide concentration, the route-additive architecture predicts a three-to-five-fold permeability gain.
2. In-vitro stability. Simulated gastric fluid at pH 1.2 with pepsin 3200 units per milliliter; recovery of intact peptide measured at 30, 60, 90 minutes. Target recovery above 70 percent at 90 minutes for the full formulation, versus Rybelsus-baseline approximately 50 percent at 30 minutes.
3. Ex-vivo intact tissue. Ussing chamber with porcine gastric mucosa (for semaglutide, liraglutide, cagrilintide, survodutide absorption-site models) and porcine duodenal mucosa (for tirzepatide, exenatide, retatrutide), measuring transepithelial electrical resistance reversibly dropping during enhancer exposure and recovering within 60 minutes after washout. Histological analysis post-exposure. Target TEER recovery above 90 percent within the 60-minute washout window.
4. In-vivo animal pharmacokinetics. Rat oral PK at 1 milligram per kilogram peptide dose and pig oral PK at 0.1-to-0.5 milligram per kilogram dose, both with full-architecture formulation versus Rybelsus-equivalent control. Target rat F above 30 percent for lipidated peptides, above 15 percent for exenatide (unlipidated baseline lower).
5. Clinical Phase 1 healthy-volunteer single-ascending-dose trial at each peptide’s oral envelope. Primary observable: plasma peptide concentration-time profile. Target C_avg_ss above SC-equivalent reference at the peptide-specific oral daily dose.
6. Clinical Phase 2 indication-specific multi-dose trial for glycemic and weight endpoints at nominal oral daily dose. Target clinical effect consistent with SC-equivalent dose.

Each gate has a first-principles prediction from the architecture. Each will score the disclosed architecture against what it claims. If the gates fall, the architecture is supported at the scale asserted. If they diverge, the divergence localizes the specific parameter requiring revision.

Why this note is public

The seven peptides in this class are approved or in late-stage clinical development, in global use, and available at retail prices that exclude most of the global population from access. The absorption problem between oral dosing and systemic exposure is the single physical bottleneck between current access and one-to-two-order-of-magnitude cost reduction per patient-per-year across the class. The mechanism data for each component of the absorption problem is individually published in the peer-reviewed literature. The integrated architecture that combines the mechanisms into a route-additive multi-enhancer tablet is derivable from the published data and transposes across the class at this MW and charge range. No trade-secret input was required to produce the disclosed architecture.

Coracle’s method is to take derivable architectures from public patent absences and from first-principles synthesis of published mechanism data, and to publish them as defensive disclosures under 35 U.S.C. §102. Placing the integrated architecture into prior-art-of-record at the class level has a specific effect: it bars subsequent claim construction on the combination at the disclosed payloads and dose envelopes across all seven peptides; opens the route for generic and open-manufacture development of oral GLP-1 receptor agonists, dual-agonists, triple-agonists, and amylin analogs at the disclosed performance level; and removes the private incentive for the patent-first pathway on a class whose consumer-surplus-to-private-surplus ratio is large.

The companion technical disclosure document publishes the architecture with quantitative specifications, parameter ranges, manufacturing process overview, failure-mode analysis, test protocol, per-peptide composition-window claims, species enumeration across SNAC fraction, disintegration time, PEG molecular weight, PLGA ratio, pulsed-field frequency and voltage, mucoadhesive substitutions, peptide active payload, and alternative embodiments at patent-application-grade specificity. The disclosure carries a 100-numbered-claim matrix at § 11 organized into five subsections covering class-scope independent composition claims, per-peptide species dependent claims for all seven peptides, process and administration claims, pulsed-field device claims, and multi-peptide combination and co-formulation claims. A § 12 enumeration extends the scope to nine variant families at the peptide, Component B through F, excipient, dosage-form, and administration levels. A § 13 enumeration addresses ten anticipate-and-preempt subject-matter clusters covering forward-looking Novo, Lilly, Boehringer Ingelheim, Zealand, AstraZeneca, and generic oral-peptide claim vectors. A class-overview matrix figure of seven peptides against nine architecture attributes and a comparative figure of F bands, subcutaneous-to-oral dose mapping, and structural features are deposited alongside. The 35 U.S.C. §102 clause is attached to that document. Any party contemplating a patent filing in the multi-mechanism oral peptide architecture space at any of these seven payloads after the publication date above inherits this disclosure as prior art of record. Any development group preparing an open-manufacture oral peptide program for any of these seven peptides can take the architecture, fit it to their specific manufacturing footprint, and proceed.

Figures

Architecture figures referenced throughout this note:

Companion technical disclosure: Technical Disclosure 03.

Sources and further reading

GLP-1, dual-agonist, triple-agonist, and amylin analog chemistry and pharmacology. J. Lau et al., “Discovery of the Once-Weekly Glucagon-Like Peptide-1 (GLP-1) Analogue Semaglutide,” J Med Chem 58:7370 (2015). L. B. Knudsen and J. Lau, “The Discovery and Development of Liraglutide and Semaglutide,” Front Endocrinol 10:155 (2019), PMC6474072. A. H. Boyle et al., “Tirzepatide: A novel dual GIP and GLP-1 receptor agonist for type 2 diabetes,” Ann Pharmacother 57:472 (2023). D. J. Drucker, “Mechanisms of Action and Therapeutic Application of Glucagon-like Peptide-1,” Cell Metab 27:740 (2018). J. Rosenstock et al., “Retatrutide, a GIP, GLP-1 and glucagon receptor agonist, for people with type 2 diabetes,” Lancet 402:529 (2023). J. P. H. Wilding et al., “Survodutide (BI 456906), a glucagon/GLP-1 receptor dual agonist, for weight management in adults with obesity,” Lancet 402:720 (2023). L. F. Enebo et al., “Safety, tolerability, pharmacokinetics, and pharmacodynamics of concomitant administration of multiple doses of cagrilintide with semaglutide 2.4 mg for weight management,” Lancet 397:1736 (2021). D. Kendall et al., “Exenatide: a peptide incretin mimetic for the treatment of type 2 diabetes,” Expert Opin Investig Drugs 13:1091 (2004).

SNAC mechanism and Rybelsus pharmacokinetics. S. T. Buckley et al., “Transcellular stomach absorption of a derivatized glucagon-like peptide-1 receptor agonist,” Sci Transl Med 10:eaar7047 (2018). C. Granhall et al., “Safety and Pharmacokinetics of Single and Multiple Ascending Doses of the Novel Oral Human GLP-1 Analogue, Oral Semaglutide,” Clin Pharmacokinet 58:781 (2019). C. Twarog et al., “Intestinal Permeation Enhancers for Oral Delivery of Macromolecules: A Comparison between Salcaprozate Sodium (SNAC) and Sodium Caprate (C10),” Pharmaceutics 11:78 (2019), PMC6410172. K. J. Colston, K. T. Faivre, S. T. Schneebeli, “Permeation enhancer-induced membrane defects assist the oral absorption of peptide drugs,” Nat Commun 16:9512 (2025), PMC12568990. R. Kneiszl, S. Hossain, P. Larsson, “In Silico-Based Experiments on Mechanistic Interactions between Several Intestinal Permeation Enhancers with a Lipid Bilayer Model,” Mol Pharm 19:124 (2022), PMC8728740.

CAGE ionic-liquid permeation enhancement. A. Banerjee, K. Ibsen, T. Brown, R. Chen, C. Agatemor, S. Mitragotri, “Ionic liquids for oral insulin delivery,” Proc Natl Acad Sci USA 115:7296 (2018), PMC6048483. K. N. Ibsen et al., “Mechanism of Antibacterial Activity of Choline-Based Ionic Liquids (CAGE),” ACS Biomater Sci Eng 4:2370 (2018). J. J. Neville et al., “Interactions of Choline and Geranate (CAGE) and Choline Octanoate (CAOT) Deep Eutectic Solvents with Lipid Bilayers,” Adv Funct Mater 34:2306644 (2024). E. E. L. Tanner et al., “Design Principles of Ionic Liquids for Transdermal Drug Delivery,” Adv Mater 31:e1901103 (2019). C. Agatemor et al., “Choline-Geranate Deep Eutectic Solvent Improves Stability and Half-Life of Glucagon-Like Peptide-1,” Adv Ther 4:2000180 (2021). K. Palanisamy and M. Prakash, “The molecular mechanism behind the stabilization of insulin by choline and geranate (CAGE) ionic liquids,” Phys Chem Chem Phys 23:25923 (2021). J. Ko et al., “Clinical translation of choline and geranic acid deep eutectic solvent,” Bioeng Transl Med 6:e10191 (2021).

Permeation-enhancer taxonomy and reviews. S. Maher, D. J. Brayden, L. Feighery, S. McClean, “Application of Permeation Enhancers in Oral Delivery of Macromolecules: An Update,” Pharmaceutics 11:41 (2019). S. Maher, R. J. Mrsny, D. J. Brayden, “Intestinal permeation enhancers for oral peptide delivery,” Adv Drug Deliv Rev 106:277 (2016). D. J. Brayden et al., “Systemic delivery of peptides by the oral route: Formulation and medicinal chemistry approaches,” Adv Drug Deliv Rev 157:2 (2020). T. A. S. Aguirre et al., “Current status of selected oral peptide technologies in advanced preclinical development and in clinical trials,” Adv Drug Deliv Rev 106:223 (2016).

Mucus barrier and mucus-penetrating nanoparticles. C. Atuma et al., “The adherent gastrointestinal mucus gel layer: thickness and physical state in vivo,” Am J Physiol Gastrointest Liver Physiol 280:G922 (2001). R. A. Cone, “Barrier properties of mucus,” Adv Drug Deliv Rev 61:75 (2009). S. K. Lai, Y.-Y. Wang, J. Hanes, “Mucus-penetrating nanoparticles for drug and gene delivery to mucosal tissues,” Adv Drug Deliv Rev 61:158 (2009), PMC2667119. L. M. Ensign et al., “Mucus Penetrating Nanoparticles: Biophysical Tool and Method of Drug and Gene Delivery,” Adv Mater 24:3887 (2012), PMC3710133.

Device-assisted ingestibles. A. Abramson et al., “An ingestible self-orienting system for oral delivery of macromolecules,” Science 363:611 (2019), PMC6430586. M. Hashim et al., “Jejunal wall delivery of insulin via an ingestible capsule in healthy volunteers,” Pharmacol Res Perspect (2021), DOI 10.1002/prp2.686.

Pepsin physiology and biochemistry. R. Heda, F. Toro, C. R. Tombazzi, “Physiology, Pepsin,” StatPearls NBK537005.

Key patents (class-level summary, per-peptide detail in companion disclosure). Novo Nordisk (Rybelsus tablet formulation): S. Bjerregaard et al., WO 2013/189988 A1, priority 2012-06-20; U.S. counterparts US 9,993,430 B2 (independent claim: no more than 15% w/w peptide, at least 50% w/w SNAC) and US 11,033,499 B2 (continuation: disintegration 22 minutes or less as alternative element). Novo Nordisk composition-of-matter patents: US 8,129,343 B2 (semaglutide, valid through 2033) and parallels on liraglutide (expired major jurisdictions) and cagrilintide. Eli Lilly composition-of-matter on tirzepatide and retatrutide (active, multi-year runway). Boehringer Ingelheim / Zealand composition-of-matter on survodutide (active). AstraZeneca / Amylin composition-of-matter on exenatide (expired). Emisphere: SNAC delivery-agent patent family, largely expired. Rani Therapeutics: M. Imran et al., US 10,814,114 B2 and continuations (ingestible capsule). MIT/BWH: A. Abramson, G. Traverso, R. Langer, US 10,639,226 B2 (SOMA ingestible). Johns Hopkins / Hanes MPP and related families: US 8,889,193 B2 (ocular method, not applicable to oral), US 9,056,057 B2 (nanocrystal, ≤1 mg/mL solubility, n/a to peptides at >10 mg/mL), US 9,415,020 B2 and US 9,629,813 B2 (hypotonic-excipient, n/a to solid tablet), WO 2017/075565 A1 (MPP composition with polyalkylene oxide MW >5 kDa, architecture uses 2 kDa). Mitragotri / Harvard / UC: S. Mitragotri et al., WO 2019/099837 A1, priority 2017-11-17; U.S. national stage US 2020/0289421 A1, with independent claims covering method of oral delivery with CAGE, composition of active plus CAGE, GLP-1 polypeptide active, 1:2 choline-to-geranate ratio. Full FTO claim-chart treatment with per-patent element distinctions in the companion technical disclosure and at verification/fto-claim-chart.md in the preprint deposit.

Full prior-art treatment with per-patent claim distinctions and feature-level novelty argumentation in the companion technical disclosure.

Defensive publication under 35 U.S.C. §102. The architectures, parameter ranges, per-peptide embodiments, and integration schemes described in this note and in its companion technical disclosure are published as prior art under 35 U.S.C. §102 for defensive purposes effective the publication date in the masthead above. See the technical disclosure for the full §102 clause, the 100-claim enumeration at § 11 (independent composition, per-peptide species, process, device, combination), the variant enumeration at § 12, the anticipate-and-preempt enumeration at § 13, reference-numbered figures 1 through 6 plus the class-overview matrix figure 7 and the peptide-comparison chart figure 8, and registration details.