COMPOUND DEEP DIVES · PEPTIDE SCIENCE 101
Conventional wisdom says peptides can’t survive the gut — and for most of the field’s history, that was a reasonable working assumption. Peptide bonds are exactly what gastric pepsin and intestinal proteases evolved to cleave. Throw a structurally intact research compound into pH 1.5–3.5 gastric acid, then expose whatever survives to trypsin, chymotrypsin, and elastase in the small intestine, and you’d expect little to reach systemic circulation. That expectation shaped twenty years of injection-first peptide research design.
The preclinical literature from 2020–2026, however, tells a more complicated story. A convergence of structural chemistry, nanotechnology, permeation enhancer science, and computational modelling has produced oral peptide formulations achieving measurable, reproducible systemic bioavailability in non-human primate and dog models. Not 80% bioavailability — nobody is claiming that. But 4–10%, delivered with pharmacokinetic variability under 30%, in some cases producing pharmacodynamic responses equivalent to subcutaneous injection. For the research community tracking compounds in our Research Compound Catalogue, that shift matters. It changes what’s possible to study, how protocols are designed, and which delivery questions are worth asking.
The framing has shifted too. A 2026 review in Frontiers in Drug Delivery put it plainly: the core challenge is no longer whether oral peptide delivery is feasible — it’s how to optimise formulation strategy for each compound’s specific physicochemical profile (Khalid et al., 2026, PMID: 41953894). That reframing is not hype. It’s a precise description of where the literature currently sits. What follows is our read of the key mechanisms, the strongest preclinical data, and — critically — what the data still cannot tell us.
The gastrointestinal barrier presents four distinct, layered obstacles to orally administered peptides. A 2026 review in Materials Today Bio catalogued these as: the harsh luminal pH environment (gastric pH 1.5–3.5); enzymatic proteolytic degradation by pepsin, trypsin, chymotrypsin, and elastase; the intestinal mucus layer (10–700 µm thick depending on GI region), which physically traps large, charged, or hydrophilic molecules; and the intestinal epithelial barrier itself, where tight junction proteins (occludin, claudin family) restrict paracellular passage of macromolecules (Wang X et al., 2026, PMID: 41585434).
Each barrier has been studied in isolation and in combination. The dominant preclinical models in the literature reviewed here include rat jejunal closed-loop preparations, minipig duodenal models, dog pharmacokinetic studies, and cynomolgus monkey whole-animal bioavailability studies. In vitro models include standard Caco-2 monolayer assays (measuring transepithelial transport), mucus-producing Caco-2/HT29-MTX co-culture systems, and simulated gastric fluid (SGF) and simulated intestinal fluid (SIF) incubation assays.
Compounds studied range from well-characterised GLP-1 receptor agonists — semaglutide, exenatide, and liraglutide — to engineered dual agonist peptides (GIP/GLP-1 coagonist, LY series), lipidated GLP-1 analogues (MEDI7219), and insulin as a model macromolecule for paracellular transport studies. These are not obscure choices: GLP-1 class peptides represent the highest-value proof-of-concept category because oral semaglutide (Rybelsus) is the only oral peptide drug to have reached market approval, giving researchers a validated comparator bioavailability benchmark.
Formulation strategies studied include: sodium caprate (C10) and SNAC (sodium N-[8-(2-hydroxybenzoyl)amino caprylate]) as permeation enhancers; solid lipid nanoparticles (SLNs) and nanostructured lipid carriers (NLCs); cyclodextrin-based nanoparticle complexes; pH-responsive enteric coatings; mucoadhesive polymer systems (chitosan, alginate, Carbopol); and lipidation and cyclisation as structural chemical modifications. A complementary computational thread — machine learning prediction of GI stability from amino acid sequence — was also reviewed (Wang F et al., 2023, PMID: 36709014).
This post draws primarily on reviews and primary data published 2020–2026, with minimum 8 PubMed-indexed citations. Our Research Notes section documents related formulation science relevant to the compounds we carry.
Before evaluating solutions, it helps to understand the scale of what formulation science is working against. In unprotected form, the majority of peptide research compounds are degraded before reaching the jejunum — the primary absorption site for most orally administered molecules. Gastric transit time averages 1–4 hours; enzymatic exposure begins within minutes of ingestion. Mucus layer viscosity varies by GI region, with the highest-viscosity regions corresponding to peak absorptive surface area. Tight junctions restrict paracellular molecular passage to roughly 1 nm in diameter — a hard physical ceiling for intact peptides above approximately 500 Da.
The strategies that have moved the needle on bioavailability address these barriers in parallel, not sequentially. A 2020 systematic review in Pharmacology & Therapeutics established the key principle: combining protease inhibitors with permeation enhancers consistently produces additive or synergistic bioavailability improvements compared to either approach in isolation, across multiple peptide classes (Yamamoto et al., 2020, PMID: 32201316). That synergy principle underlies every high-performing formulation in the 2023–2026 data.
The most mechanistically characterised permeation enhancers in the current literature are SNAC (the delivery agent in oral semaglutide) and sodium caprate (C10). Both work by transiently reducing tight junction protein expression in gastric and intestinal epithelium — but their mechanisms diverge in important ways.
In cynomolgus monkey models, Tran et al. (2024, PMID: 38070657) evaluated SNAC and C10 erodible tablets (300 mg) for delivery of a GIP/GLP-1 dual agonist peptide engineered for superior pepsin stability. Key findings:
SNAC’s mechanism was confirmed histologically: reduced tight junction protein levels in gastric epithelial tissue, consistent with simultaneous paracellular and transcellular permeation. The same research group’s earlier minipig and rat data (Tran et al., 2023, PMID: 36592951) added mechanistic depth: confocal live imaging in rat intestine demonstrated rapid paracellular transport of a model macromolecule (fluorescein dextran FD4) following C10 administration, while minipig duodenal tissue showed C10-mediated reduction in tight junction protein expression and increased membrane fluidity. Absolute bioavailability in the minipig model reached ~2% relative bioavailability for the GLP-1/GIP coagonist.
Crucially, both studies documented that PE efficacy is not uniform across peptides — it is governed by the individual peptide’s physicochemical properties. Aggregation state, proteolytic stability, charge distribution, and lipophilicity all modulate how much benefit a given PE confers. This is not a detail — it fundamentally means PE data from one peptide class does not generalise to another without compound-specific optimisation studies.
The most scalable high-performance formulation strategy in the current data is lipid-based nanoparticle encapsulation. Solid lipid nanoparticles (SLNs) and nanostructured lipid carriers (NLCs) physically shield peptide cargo within a lipid matrix, protecting against both enzymatic degradation and chemical denaturation while enabling transcellular uptake via clathrin-mediated endocytosis and membrane fusion pathways.
A 2024 review in Advanced Pharmaceutical Bulletin documented the key advantages of SLN/NLC systems: particle size typically 50–400 nm, high drug encapsulation efficiency, biodegradable and biocompatible lipid matrices, reduced first-pass hepatic metabolism via lymphatic uptake, and mucosal adhesion properties that extend GI contact time (Mehrdadi, 2024, PMID: 38585451).
The most precise in vitro dataset comes from Birro et al. (2025, PMID: 40354907), using microfluidic-manufactured DOTAP/DSPE-PEG2kDa SLNs loaded with exenatide:
The PEGylation finding is a useful nuance: non-PEGylated SLNs perform better at the epithelial membrane, while PEGylated SLNs penetrate mucus more effectively. Optimal formulation design requires choosing which barrier is rate-limiting for a given peptide — an argument for compound-specific rather than universal SLN formulation.
Cyclodextrin-based nanoparticles offer an alternative encapsulation platform. In rat models, Presas et al. (2021, PMID: 33152374) demonstrated that cationic amphiphilic cyclodextrin nanoparticles (101 nm, -35 mV zeta potential, 5.0% peptide loading of liraglutide) maintained colloidal stability in intestinal biorelevant media for up to 4 hours, protected liraglutide from proteolytic degradation throughout that window, and produced a blood glucose pharmacological response in rats comparable to subcutaneous free liraglutide solution. Matching injectable pharmacodynamic response via an intestinal route, in a rat model, is the strongest preclinical validation available short of human pharmacokinetic data.
The highest absolute oral bioavailability figure in the current non-rodent preclinical dataset comes from AstraZeneca’s MEDI7219 programme. MEDI7219 is a stabilised, lipidated GLP-1 agonist designed from the ground up for oral delivery — not a retrofit of an existing injectable compound. Tyagi et al. (2023, PMID: 37896196) reported the following in dog pharmacokinetic models:
The bioadhesive sustained-release formulation outperformed the immediate-release enteric-coated tablet on both bioavailability and variability metrics — suggesting that extended mucosal contact time is a material advantage, not just an incremental improvement. This is the mechanistic rationale for mucoadhesive polymer systems: Carbopol, chitosan, and alginate bioadhesives documented in Baral & Choi’s 2025 Pharmaceutics review extend GI residence time at the epithelial surface, directly increasing total absorption opportunity (Baral KC & Choi KY, 2025, PMID: 40284395).
Table 1: Key Preclinical Oral Bioavailability Data — Permeation Enhancer and Nanoparticle Strategies
| Compound | Study Type | Key Outcome | Citation |
|---|---|---|---|
| GIP/GLP-1 dual agonist + SNAC tablet | Cynomolgus monkey, in vivo | 4.2% absolute oral bioavailability; ~4× higher than semaglutide SNAC (1.2%) | Tran et al., 2024, PMID: 38070657 |
| GIP/GLP-1 dual agonist + C10 tablet | Cynomolgus monkey, in vivo | 5.7% absolute oral bioavailability; 100% tablet erosion within 60 min | Tran et al., 2024, PMID: 38070657 |
| GLP-1/GIP coagonist + C10 | Rat / minipig, in vivo | ~2% relative bioavailability in minipig; C10 confirmed to reduce tight junction proteins | Tran et al., 2023, PMID: 36592951 |
| MEDI7219 (lipidated GLP-1 agonist) + Na CDC/PG, Carbopol tablet | Dog, in vivo | 10.1% absolute oral bioavailability; 26% CV% variability | Tyagi et al., 2023, PMID: 37896196 |
| Liraglutide + cyclodextrin NPs (101 nm) | Rat, intestinal administration | Proteolytic protection for 4 hr; blood glucose response comparable to SC injection | Presas et al., 2021, PMID: 33152374 |
| Exenatide + microfluidic SLNs (~120 nm) | In vitro (Caco-2, Caco-2/HT29-MTX) | 94.2% encapsulation; 2× epithelial transport vs unformulated peptide; full proteolytic protection | Birro et al., 2025, PMID: 40354907 |
The emerging computational layer in oral peptide delivery science deserves attention. Wang et al. (2023, PMID: 36709014) trained machine learning models on 109 peptide incubation datasets in simulated gastric fluid (SGF) and simulated intestinal fluid (SIF), achieving 75.1% classification accuracy for gastric stability prediction (k-Nearest Neighbor model) and 69.3% for intestinal stability prediction (XGBoost) from amino acid sequence alone. Feature importance analysis identified three primary determinants of GI stability: peptide lipophilicity, structural rigidity, and molecular size.
This is the first publicly available computational tool for pre-screening oral peptide candidates before physical formulation work begins — a meaningful reduction in early-stage research cost and timeline, even at 75% accuracy.
The structural modification findings from Khalid et al. (2026, PMID: 41953894) add context: cyclisation (which increases structural rigidity), lipidation (which increases lipophilicity and membrane permeability), and metabolic stabilisation have collectively enabled therapeutic-level efficacy at single-digit oral bioavailability figures. The ML model and the formulation chemistry tell the same story from different angles — rigidity and lipophilicity are the variables to engineer for.
Table 2: Formulation Strategy Comparison — Mechanism and Preclinical Status
| Formulation Strategy | Primary Barrier Addressed | Key Mechanism | Preclinical Status |
|---|---|---|---|
| SNAC permeation enhancer | Epithelial tight junctions | Reduces occludin/claudin expression; paracellular + transcellular | Non-human primate; 1 marketed product (semaglutide) |
| Sodium caprate (C10) | Epithelial tight junctions + membrane | Reduces tight junction proteins; increases membrane fluidity | Rat, minipig, non-human primate |
| Solid lipid nanoparticles (SLNs) | Enzymatic degradation + mucus + epithelial barrier | Lipid matrix encapsulation; transcellular endocytosis; lymphatic uptake | In vitro (Caco-2); limited in vivo |
| Cyclodextrin nanoparticles | Enzymatic degradation + epithelial barrier | Encapsulation; cationic surface-mediated mucosal adhesion | Rat, intestinal administration |
| Carbopol bioadhesive tablets | Mucus + GI transit time | Extended epithelial contact duration; sustained release | Dog (MEDI7219 programme) |
| Lipidation + enteric coating | Gastric acid + enzymatic degradation + membrane permeability | pH-responsive protection; fatty acid-mediated membrane permeation | Dog, non-human primate |
| Enzyme inhibitor + PE combination | Enzymatic degradation + epithelial barrier | Additive/synergistic protection across both barriers | Multiple species; review-level evidence |
For the compounds in our Metabolic Compounds range, and particularly the GLP-1 class research compounds including GLP-1 and Retatrutide, these formulation advances represent the foundational science that makes non-injectable research protocols increasingly viable in academic and independent settings.
The preclinical data reviewed here is the most encouraging in the field’s history for oral peptide delivery. It is also, overwhelmingly, not human data. That distinction is not pedantic — it is the central methodological limitation of the entire evidence base, and it should govern how any of this research is interpreted.
Limitation 1: Animal model GI physiology does not replicate human GI physiology.
Rat, minipig, dog, and cynomolgus monkey models each offer different approximations of human GI conditions. Gastric volume, transit time, enzyme expression levels, mucus composition, and tight junction protein density all differ across species — and all differ from human values in ways that materially affect bioavailability extrapolation. A 5.7% absolute bioavailability result in cynomolgus monkeys (Tran et al., 2024) is a meaningful proof-of-concept. It is not a prediction of human bioavailability. The translation gap between non-human primate and human GI pharmacokinetics for macromolecules remains poorly characterised in the peer-reviewed literature.
Limitation 2: Even best-in-class formulations achieve low absolute bioavailability by small-molecule standards.
The 10.1% figure from the MEDI7219 dog study (Tyagi et al., 2023) is a high-water mark for the current literature. Standard small-molecule oral drugs typically achieve 30–80%+ absolute bioavailability. Whether single-digit or low-double-digit oral bioavailability for a research peptide produces sufficient systemic exposure for the intended research purpose depends entirely on compound-specific potency — a calculation that cannot be generalised. For potent receptor agonists active at picomolar concentrations, 5% bioavailability may be entirely adequate. For compounds requiring sustained high-level systemic exposure, it may not be.
Limitation 3: Randomised controlled trial data is almost entirely absent.
With the exception of oral semaglutide (Rybelsus), no oral peptide formulation technology in this review has completed phase III RCT evaluation. The evidence base for SNAC beyond semaglutide, for C10, for SLN platforms, for cyclodextrin nanoparticles, and for lipidated peptide oral formulations is uniformly preclinical or early-phase. The jump from a compelling cynomolgus monkey bioavailability study to validated human pharmacokinetics has historically been where promising formulation science has failed.
Limitation 4: Permeation enhancer safety with chronic dosing is uncharacterised.
SNAC and C10 both work by transiently disrupting epithelial tight junction integrity — reducing occludin and claudin expression in gastric and intestinal epithelium. The Tran et al. (2024) data confirmed this histologically. What those studies do not characterise: the long-term effects of repeated tight junction disruption on mucosal barrier function, intestinal microbiome composition, or systemic inflammatory markers in chronic-use research protocols. This is not a theoretical concern — it is a gap in the literature that warrants attention before any chronic oral PE administration protocol is designed.
Limitation 5: In vitro to in vivo correlation (IVIVC) is weak for this class.
Caco-2 and Caco-2/HT29-MTX models are the standard in vitro screen for epithelial permeability and mucus penetration, respectively. They do not replicate human intestinal peristalsis, mucus turnover rate, immune cell interactions, or the dynamic luminal environment of a fed versus fasted state. Birro et al. (2025)’s impressive exenatide SLN transport data in Caco-2 co-cultures may or may not translate to meaningful in vivo bioavailability — the IVIVC for nanoparticle-formulated peptides has not been systematically validated.
Limitation 6: Machine learning GI stability models are trained on limited data.
The Wang et al. (2023) ML models were trained on 109 peptide incubation datasets. For reference, drug-likeness prediction models in small-molecule pharma typically require thousands to tens of thousands of compounds for robust generalisation. A 109-compound training set is likely to produce significant generalisation error when applied to structurally novel peptide candidates outside the training distribution. The 75.1% accuracy figure is a starting point, not a reliable prediction engine for arbitrary peptide sequences.
Limitation 7: PE efficacy is compound-specific, not class-universal.
Both the Tran et al. (2023, 2024) datasets explicitly document that C10 and SNAC produce substantially different bioavailability enhancements depending on the individual peptide’s aggregation state, proteolytic stability, and physicochemical properties. This means data from GLP-1 class PE studies cannot be extrapolated to predict PE efficacy for structurally distinct compounds such as BPC-157, TB-500, or GHK-Cu without compound-specific optimisation work. Anyone claiming a universal PE-based oral delivery solution for diverse peptide classes is not reading the same literature we are.
These limitations are worth keeping in full view when interpreting the trajectory of oral peptide delivery research. The direction is clearly positive. The gap between “preclinical proof-of-concept” and “validated human pharmacokinetics” remains large for most compounds in this space.
The research question has genuinely changed. Oral peptide delivery was, for most of the field’s history, studied as a theoretical possibility with negligible practical traction. The 2020–2026 data closes that chapter. In non-human primate and dog models, multiple formulation strategies — SNAC erodible tablets, C10 permeation enhancement, lipidated peptide bioadhesive systems, SLN encapsulation, and cyclodextrin nanoparticle platforms — have produced measurable, reproducible systemic bioavailability for GLP-1 class and related peptides. The MEDI7219 dog data at 10.1% with 26% CV is a legitimate landmark result for the field.
What this means for research protocol design: the physicochemical properties of the specific compound remain the primary variable. Lipophilicity, structural rigidity, molecular size, and proteolytic stability govern both the degree to which any formulation strategy will be effective and which strategy is the right starting point. There is no universal oral peptide formulation. There are increasingly well-characterised formulation-compound matching principles that the literature is beginning to make systematic.
For the compounds in our Longevity Compounds and Recovery Compounds catalogues — including Epithalon, CJC-1295, and Tesamorelin — the formulation science remains earlier-stage than the GLP-1 class. But the mechanistic principles being established for GLP-1 delivery are not GLP-1-specific. They apply to any peptide research compound for which oral delivery would be a meaningful experimental variable. Tracking this literature is worth the time. We’ll continue to do so on our Research Notes page.
Khalid et al. (2026). Navigating the complexity of oral peptide delivery: challenges and strategies to enhance oral bioavailability. Frontiers in Drug Delivery. PMID: 41953894.
Wang X et al. (2026). Strategies for overcoming multiple barriers of oral administration of protein and peptide therapeutics. Materials Today Bio. PMID: 41585434.
Baral KC & Choi KY (2025). Barriers and Strategies for Oral Peptide and Protein Therapeutics Delivery: Update on Clinical Advances. Pharmaceutics. PMID: 40284395.
Birro BA et al. (2025). Unlocking the potential of microfluidic assisted formulation of exenatide-loaded solid lipid nanoparticles. International Journal of Pharmaceutics. PMID: 40354907.
Chavda VP & Balar PC (2025). Oral delivery of protein and peptide therapeutics. Progress in Molecular Biology and Translational Science. PMID: 40122651.
Tran H et al. (2024). Development and evaluation of C10 and SNAC erodible tablets for gastric delivery of a GIP/GLP1 peptide in monkeys. International Journal of Pharmaceutics. PMID: 38070657.
Mehrdadi S (2024). Lipid-Based Nanoparticles as Oral Drug Delivery Systems: Overcoming Poor Gastrointestinal Absorption and Enhancing Bioavailability of Peptide and Protein Therapeutics. Advanced Pharmaceutical Bulletin. PMID: 38585451.
Parida P et al. (2024). Current Advancements on Oral Protein and Peptide Drug Delivery Approaches to Bioavailability: Extensive Review on Patents. Recent Advances in Drug Delivery and Formulation. PMID: 39356096.
Wang X et al. (2024). Obstacles, research progress, and prospects of oral delivery of bioactive peptides: a comprehensive review. Frontiers in Nutrition. PMID: 39610876.
Wang F et al. (2023). Advancing oral delivery of biologics: Machine learning predicts peptide stability in the gastrointestinal tract. International Journal of Pharmaceutics. PMID: 36709014.
Tran H et al. (2023). In Vivo Mechanism of Action of Sodium Caprate for Improving the Intestinal Absorption of a GLP1/GIP Coagonist Peptide. Molecular Pharmaceutics. PMID: 36592951.
Tyagi P et al. (2023). Systems Biology and Peptide Engineering to Overcome Absorption Barriers for Oral Peptide Delivery: Dosage Form Optimization Case Study Preceding Clinical Translation. Pharmaceutics. PMID: 37896196.
Presas E et al. (2021). Pre-Clinical Evaluation of a Modified Cyclodextrin-Based Nanoparticle for Intestinal Delivery of Liraglutide. Journal of Pharmaceutical Sciences. PMID: 33152374.
Yamamoto A et al. (2020). Approaches to improve intestinal and transmucosal absorption of peptide and protein drugs. Pharmacology & Therapeutics. PMID: 32201316.
biohacker.team sources all research compounds with HPLC purity verification and third-party certificate of analysis (COA) documentation available on request. Our formulation science reading list — including the oral delivery literature reviewed in this post — is maintained on our Research Notes page. If you’re evaluating compounds from our Cognitive Compounds or Aesthetics Compounds ranges and have questions about delivery format considerations in a research context, our team is reachable via the contact page. We don’t claim to have resolved the oral bioavailability problem for every compound we carry — the literature hasn’t resolved it yet either — but we track the data that’s moving the field forward and we share what we find.
For research use only. Not for human consumption. Not intended to diagnose, treat, cure, or prevent any disease or condition. All compounds are sold strictly for in-vitro and animal research purposes. Not approved for human use.