The Birth of ICOTYDE (1): How an Oral Cyclic Peptide Begins with Intestinal Stability

The Birth of ICOTYDE · Part 1

What Was an Oral Cyclic Peptide Originally Intended to Solve?

In 2026, ICOTYDE (icotrokinra) became an FDA-approved oral IL-23 receptor (IL-23R) antagonist. It is a cyclic peptide composed of 13 residues, with a free-base molecular weight of approximately 1898 Da, yet it can be administered orally and exert systemic effects. If one looks only at the endpoint of the story, it is easy to understand the entire project as a technical effort focused on “how a macromolecule crosses the intestine.”

But tracing backward through the published patents reveals a somewhat counterintuitive starting point: the first priority in this program was not how to enable a large peptide to enter the bloodstream in substantial amounts, but how to prevent a peptide from being rapidly degraded in the gastrointestinal tract after oral administration.

These two questions are not the same. Even if a peptide has an opportunity to cross the intestinal epithelium, it cannot become an oral drug if it rapidly falls apart in the presence of gastric fluid, intestinal fluid, and proteases. Conversely, if a molecule can remain intact in the intestinal lumen or intestinal tissue, it may first become a starting point for a locally acting drug even when systemic exposure is low. Protagonist’s earliest IL-23R program was developed precisely along this logic.

The Target Had Been Validated, but the Molecular Format Remained a Challenge

IL-23 is an important cytokine in inflammatory diseases. It binds to a receptor complex composed of IL-23R and IL-12Rβ1, activates downstream signaling, and maintains cellular programs associated with multiple immune-inflammatory responses. Antibody drugs have already demonstrated that intervening in the IL-23 pathway can produce clear clinical effects. In other words, the main question facing the program was not “whether this target is worth pursuing,” but “whether it can be blocked using a molecular format suitable for oral administration.”

The contact between IL-23 and IL-23R is a protein–protein interaction. Such interfaces are usually broader and flatter than classical small-molecule pockets, with fewer deep cavities and defined anchoring points available for conventional small molecules. Antibodies can recognize them using a larger binding surface, but they must be injected; traditional small molecules are convenient for oral administration, but may not effectively cover this type of interface.

Cyclic peptides occupy a space between the two. They are larger than typical small molecules and can display a broader binding surface; cyclization can also reduce conformational freedom, allowing key side chains to face the target in a more stable manner. However, this size and structural advantage does not automatically confer oral properties. Peptide bonds may be cleaved by proteases, termini may also serve as entry points for degradation, and the chemical environment in the gastrointestinal tract continues to challenge the entire molecule.

The Program Beginning in 2014 Was First Directed Toward the Gastrointestinal Tract

The priority date of Protagonist’s early patent WO2016011208A1 can be traced back to July 2014. The patent title directly states “oral peptide inhibitors of IL-23 receptor,” and the molecules described were initially aimed primarily at inflammatory bowel disease. Their design objective had a clear GI-restricted character: to keep the compounds active in the gastrointestinal tract after oral administration, enable entry into intestinal tissue, and minimize exposure in the circulation.

This determined the order of early optimization. The team needed to obtain peptides capable of binding IL-23R, but binding activity was not the only readout. Candidates also had to undergo stability testing under conditions such as simulated gastric fluid (SGF), simulated intestinal fluid (SIF), and human intestinal fluid. A molecule that was strong in a binding assay but disappeared rapidly after entering intestinal fluid was not a qualified starting point.

Where Did the Peptide 1185 Sequence Come From?

Peptide 1185 was neither generated in a single computational step nor selected from a phage library in its final form. The early patents state that this series of IL-23R inhibitory peptides combined phage display with medicinal chemistry optimization. Phage display first searched large sequence libraries for cyclic peptide scaffolds capable of recognizing IL-23R. Researchers then brought those initial results back to the chemistry laboratory, changed residues one by one, and tested binding activity and stability.

Peptide 630 is one traceable early point in this route. Later research explicitly states that it originated from phage-display work against IL-23R. Its chemically modified sequence already contained the Pen–Asn–Thr–Trp–Gln–Pen core, followed outside the ring by F(4-OMe), 2-Nal, Aib, and Glu–Asn–Asn. In Peptide 1185, the Pen–Asn–Thr–Trp–Gln–Pen core was retained, while F(4-OMe), Aib, and Glu were replaced by F(4-2ae), THP, and Lys(Ac), respectively.

These changes show that Peptide 1185 was no longer a raw screening hit. Pen, 2-Nal, F(4-2ae), THP, and Lys(Ac) all had to be introduced by chemical synthesis, followed by repeated comparisons of neighboring analogues in IL-23R binding, cellular activity, and gastrointestinal stability assays. In other words, phage display provided the sequence scaffold and binding direction, while medicinal chemistry gradually transformed that starting point into Peptide 1185.

Peptide 1185: The Core Profile Had Already Emerged

In the subsequent patent WO2017011820A2, Peptide 1185 was explicitly listed as SEQ ID NO: 1185. Its simplified sequence is:

Ac–Pen–Asn–Thr–Trp–Gln–Pen–F(4-2ae)–2-Nal–THP–Lys(Ac)–Asn–Asn–NH₂

A disulfide bond is formed between the side chains of the two Pen residues, constraining the six residues near the N-terminus into a macrocycle. F(4-2ae) denotes phenylalanine bearing a 4-(2-aminoethoxy) substituent; 2-Nal is 2-naphthylalanine; THP is an unnatural residue based on a tetrahydropyran scaffold; and Lys(Ac) is side-chain-acetylated lysine. The N-terminus of the molecule is capped with an acetyl group, and the C-terminus is capped as an amide.

Figure 1. Peptide 1185: a representative compound disclosed in Protagonist’s early IL-23R cyclic peptide patent. The two Pen residues form a ring via a disulfide bond. This figure was redrawn using RDKit based on page 305 of the WO2017011820A2 patent (PDF page 307) and SEQ ID NO: 1185.

Peptide 1185 is noteworthy because it had already placed a Pen–Pen disulfide macrocycle, multiple unnatural or modified residues, and N-terminal acetylation and C-terminal amidation within the same molecule. It was still a peptide, but it had clearly moved beyond the chemical space usually covered by the 20 natural amino acids.

From the Early Profile to the Final Structure

After Peptide 1185, the IL-23R cyclic peptide program still went through multiple generations of medicinal chemistry. The quality review for FDA NDA 220149 confirmed that the final icotrokinra is a 13-residue Pen–Pen cyclic peptide, with a free-base molecular formula of C90H120N20O22S2 and a molecular weight of approximately 1898.17 Da. When placing the two side by side, it is sufficient here to note the clear scaffold continuity between the early and final structures; the specific residue substitutions and C-terminal redesign will be left for discussion in the next article.

Figure 2. Two-dimensional structure of icotrokinra. Redrawn using RDKit based on publicly available structural information from PubChem CID 162462321 and FDA NDA 220149, using the same core scaffold orientation as in Figure 1. This figure is used only to show the overall continuity between the early and final structures; detailed SAR is discussed in Part 2.

Why Did Peptide 1185 Already No Longer Resemble an Ordinary Peptide?

Pen vs Cys: adding steric hindrance next to the disulfide bond. The sulfur-containing side chain of Cys is –CH₂–SH; Pen, by contrast, has two additional methyl groups on the β-carbon adjacent to the sulfur atom. This β,β-dimethyl substitution is a fact that can be directly confirmed from the structure. The increased steric hindrance can restrict locally accessible conformations, alter the geometry near the disulfide bond, and may make it more difficult for proteases to accommodate adjacent peptide bonds, but these consequences cannot be written as inevitable outcomes based on structure alone. Patent Table E9 provides an informative approximate matched pair: SEQ ID NO.554 and 1028 have the same remaining sequence, with the main difference being that the first-position Pen is replaced by Cys; the former has a half-life category of 180–360 min in SIF, while the latter is <60 min. This supports the ability of Pen to contribute stability in this scaffold, but it is not Peptide 1185 itself, and it does not prove that the GI stability of the entire molecule is caused by Pen alone.

The disulfide between the two Pen residues also closes the six N-terminal residues into a macrocycle. Compared with a free linear peptide, ring closure reduces the conformational space that the backbone can sample, may lower the entropic cost of forming a binding-competent conformation, and at the same time changes the way proteases recognize and approach peptide bonds. A more accurate statement here is that “cyclization remodels the geometry of the entire substrate,” rather than that “a particular cleavage site disappears from this point onward.”

2-Nal vs Phe: expanding one benzene ring into a naphthalene ring. Phe provides a phenyl surface of fixed size; the 2-Nal in Peptide 1185 is 2-naphthylalanine, not 1-Nal. The additional fused benzene ring enlarges the hydrophobic and π-contact surface and also changes the shape of the side chain. For a broader protein–protein interaction interface, this may provide contact area or shape complementarity that natural Phe cannot cover; the patent does not prove that its individual role is to improve membrane permeation or oral absorption.

Phe / Tyr vs F(4-2ae): placing a polar terminus farther away. The patent explicitly defines F(4-2ae) as Phe[4-(2-aminoethoxy)]. Phe mainly provides an aromatic surface, while Tyr provides a shorter phenolic OH at the para position; F(4-2ae) retains the phenyl core, but uses –O–CH₂–CH₂–NH₂ to extend a polar, ionizable terminus farther away from the aromatic ring. What it demonstrates is not simply “more hydrophilic” or “more hydrophobic,” but rather that aromatic contacts, side-chain length, and the spatial position of polar groups can be tuned separately.

Lys vs Lys(Ac): retaining length while rewriting terminal charge. The –(CH₂)₄–NH₂ of ordinary Lys is mainly protonated at physiological pH; side-chain N-acetylation converts it into –(CH₂)₄–NHCOCH₃, a neutral amide. The side-chain reach is largely retained, but the strong basicity, net positive charge, and hydrogen-bond donor/acceptor pattern are changed, and the surrounding enzyme substrate-recognition environment may also be altered. The comparison here is between natural Lys and the Lys(Ac) already present in Peptide 1185, not a conflation of the Gln in Peptide 1185 with the Lys(Ac) in the final drug.

THP and terminal capping: natural amino acids have no fully equivalent reference. The patent describes THP as 4-amino-4-carboxy-tetrahydropyran: the amino and carboxyl groups are located on the same 4-position carbon of the oxygen-containing six-membered ring. It is not simply an elongated version of a natural side chain, so no one-to-one comparator is forcibly assigned; what is clearly visible is that the ring scaffold incorporates the α-carbon into a six-membered ring, reducing the type of free rotation seen in ordinary side chains. Peptide 1185 is also capped at the N-terminus with Ac– and amidated at the C-terminus as –NH₂, altering the charge and exopeptidase-recognition environment of the two free termini. Terminal capping is a classic stabilization strategy, but this article does not artificially separate its contribution from the other design elements.

Figure 3. Expanded chemical space from natural amino acids to Peptide 1185. Pen, 2-Nal, F(4-2ae), and Lys(Ac) respectively alter local spatial constraints, aromatic surface, polar-group positioning, and charge properties; THP has no fully equivalent natural amino acid reference. The figure presents structural facts and their reasonable medicinal chemistry implications, and does not indicate that the individual contribution of each modification has been proven by matched-pair experiments.

The significance of Peptide 1185 is therefore not merely that it “contains several unnatural amino acids.” Pen changes local steric constraints, 2-Nal expands the aromatic surface, F(4-2ae) combines aromatic contact with an extended polar function in the same side chain, and Lys(Ac) redefines charge and hydrogen bonding while retaining length. With the additional Pen–Pen cyclization and terminal capping at both ends, this peptide no longer presents the chemical and conformational pattern of an ordinary natural peptide when facing receptors and proteases.

Why did the gastrointestinal tract not digest it rapidly?

Table E22 in the patent family of WO2017011820A2 provides quantitative results for Peptide 1185 itself: its half-lives in SIF, SGF, and human intestinal fluid (HIF) were 33 h, 12 h, and 24 h, respectively. This is the direct experimental basis for the discussion of GI stability in this article, rather than a conclusion inferred backward from the final drug structure. The patent also reports that Peptide 1185 has an IC₅₀ of 2 nM in a human IL-23R competitive-binding ELISA, but stability and binding activity are two different readouts.

Pepsin does not necessarily complete cleavage simply because it encounters aromatic residues. The enzyme needs to recognize the surrounding sequence, bind the substrate, and position a specific peptide bond in a catalytically competent geometry. The macrocyclization of Peptide 1185, the sterically hindered Pen, the unnatural aromatic side chains, THP, and terminal capping collectively alter this recognition and positioning process. What the real data demonstrate is the stability phenotype of the entire molecule; apart from the approximate Pen/Cys matched pair mentioned above, the patent does not attribute the 33 h, 12 h, or 24 h values individually to any single residue.

But “surviving” is still not the same as “becoming a drug”

By the stage of Peptide 1185, the researchers already had a highly GI-stable IL-23R cyclic-peptide scaffold composed of multiple unnatural or modified residues, but it still was not the final drug. The truly critical subsequent question shifted from “how to make this peptide survive in the gastrointestinal tract” to “how to continue advancing IL-23R binding and overall developability to the level of a clinical candidate while retaining the stable scaffold.” Systemic exposure, permeability, dose, and food effects belong to the third article and will not be discussed in advance here.

Next article: The Birth of ICOTYDE (2): From Peptide 1185 to PN-235, Why Continue Modifying It?

The next article will follow the subsequent patents from Protagonist and Janssen to compare residue changes and patent SAR among Peptide 1185, genuinely representative intermediate structures, and the final candidate.

Scientific Sources

Protagonist Therapeutics, Inc. Oral Peptide Inhibitors of Interleukin-23 Receptor and Their Use to Treat Inflammatory Bowel Diseases. WO2016011208A1; Peptide Inhibitors of Interleukin-23 Receptor and Their Use to Treat Inflammatory Diseases. WO2017011820A2. Peptide 1185 is SEQ ID NO: 1185 in the latter.

Lay CS, Isidro-Llobet A, Kilpatrick LE, et al. Characterisation of IL-23 Receptor Antagonists and Disease Relevant Mutants Using Fluorescent Probes. Nature Communications. 2023;14:2882. https://doi.org/10.1038/s41467-023-38541-2

Related services: Custom Synthesis of Cyclic Peptides and Unnatural Amino Acid Peptides｜The Role of Unnatural Amino Acids in Peptide Drug Development