The Birth of ICOTYDE (2): From Peptide 1185 to PN-235, Why Continue to Modify It?
Peptide 1185 is already stable and highly engineered, so why continue to modify a small number of residues? This article uses real patent SAR to distinguish experimental results from structural interpretation.
The Birth of ICOTYDE · Part 2
Previous article: The Birth of ICOTYDE (1): How an Oral Cyclic Peptide Starts with Gastrointestinal Stability
Already Stable—Why Keep Modifying It?
Peptide 1185, discussed in the previous article, was already not a natural peptide waiting to be “drugged.” It has a Pen–Pen disulfide macrocycle, capped N/C termini, as well as F(4-2ae), 2-Nal, THP, and Lys(Ac); public data also show that it can remain stable in simulated gastrointestinal environments. The real question, therefore, is not “why use non-natural amino acids,” but rather: why continue to modify a few residues in an already highly engineered IL-23R cyclic peptide that can tolerate the gastrointestinal environment?
When it is compared side by side with PN-235—later called JNJ-77242113, JNJ-2113, and ultimately named icotrokinra—the changes are concentrated in three places: Trp→7-MeTrp, Gln→Lys(Ac), and the exocyclic tail Lys(Ac)–Asn–Asn→Glu–Asn–3-Pal–Sarc. The shared Pen–Asn–Thr and Pen–F(4-2ae)–2-Nal–THP indicate that the program did not start over from scratch, but continued searching around an already usable binding scaffold.
Figure 1. Complete structural difference map between Peptide 1185 and icotrokinra. Orange indicates regions that were substituted or reconstructed; the uncolored portions are the core scaffold shared by both molecules. Structures were reconstructed from public sequences using RDKit.
Change 1: Trp → 7-MeTrp Is a Small but Measurable Improvement
7-MeTrp differs from Trp only by an additional methyl group at the 7-position of the indole ring. Structurally, it changes the steric bulk at the edge of the aromatic side chain, the hydrophobic surface, and local conformational preferences; however, “looking reasonable” is not the same as experimental proof.
WO2021146441A1 provides a rare same-background comparison. The final sequence, SEQ ID NO.104, had an IC50 of 0.022 nM in the DB-cell IL-23-induced pSTAT3 HTRF assay; SEQ ID NO.158, in which only 7-MeTrp was changed back to Trp, had an IC50 of 0.0491 nM. In the PBMC pSTAT3 assay from the same patent, the two values were 0.00515 and 0.011 nM, respectively. Both sets of readouts point to an approximately two-fold difference.
This supports a moderate improvement in potency conferred by the 7-methyl group in this sequence context, but it does not support larger causal conclusions such as “methylation solved oral absorption.” The patent also discloses many modes of aromatic-ring substitution; the eventual retention of 7-MeTrp looks more like convergence after multiparameter screening than an answer that was obvious at a glance.
Change 2: Gln → Lys(Ac) Is Not Charge Neutralization
This step is the easiest to describe incorrectly. The side-chain terminus of Gln is an amide; after the lysine ε-amino group in Lys(Ac) is acetylated, its terminus is likewise a neutral amide. Therefore, this is not “turning a positively charged Lys neutral,” and it certainly should not be written as Gln→Lys(Ac) improving activity through charge neutralization. The obvious structural changes are: side-chain elongation, increased steric bulk, and relocation of the terminal amide to a new spatial position.
A clean matched pair in the patent also gives an important negative result: SEQ ID NO.108 differs from the final SEQ ID NO.104 only at this position—the former has Gln, and the latter has Lys(Ac)—and both are 0.022 nM in the same DB-cell pSTAT3 HTRF assay. In other words, the public data do not show that Lys(Ac) is superior to Gln in this functional activity assay.
It may still affect properties such as conformation, solubility, stability, selectivity, PK, or manufacturing that are not deconvoluted in this set of numbers, and it may show advantages only in other residue combinations. But these are medicinal chemistry interpretations, not conclusions that have already been proven by this SAR pair. The fact that the final candidate retained Lys(Ac) cannot be rewritten retroactively as “because it must be more potent.”
Change 3: Tail Reconstruction, with Sar as the Key Point
The exocyclic tail of Peptide 1185 is Lys(Ac)–Asn–Asn–NH₂, whereas the final structure becomes Glu–Asn–3-Pal–Sarc–NH₂. The patent shows that the researchers extensively adjusted this tail segment. The terminal Sarc (usually abbreviated as Sar) retained after multiple rounds of optimization provides a relatively clear chemical clue.
Sar is sarcosine, i.e., N-methylglycine; it makes the peptide bond between 3-Pal and Sar an N-methyl amide without an N–H. N-methylation changes local conformation and protease recognition of the peptide bond. In a model oligopeptide study, Kaminker et al. observed an approximately six-fold increase in elastase proteolysis half-life after replacing Gly with Sar. This very small backbone modification likely added another layer of protection to the stability of the final molecule.
From 2 nM to 7.1 pM
The publicly disclosed IL-23R binding IC50 for Peptide 1185 is 2 nM. By the time of JNJ-77242113, the final icotrokinra, the KD measured in an SPR experiment at 37°C had reached 7.1 ± 2.5 pM. 2 nM is equivalent to 2000 pM, so the two public readouts differ numerically by about 280-fold. Advancing from the nanomolar range to the single-digit picomolar range is precisely why researchers continued optimizing even after the scaffold was already stable.
Therefore, the transition from Peptide 1185 to PN-235 was not simply “adding more non-natural amino acids.” 7-MeTrp brought a moderate improvement in activity, Gln→Lys(Ac) adjusted the side-chain length and the position of the terminal amide, and the N-methyl group in the tail Sar provided new protection for protease stability. Through these small and continuous adjustments, the researchers gradually pushed a stable cyclic peptide scaffold toward the final candidate.
At this point, medicinal chemistry had pushed a gastrointestinally stable IL-23R scaffold to picomolar binding and cellular activity. But the contradiction also became sharper: icotrokinra is still a highly polar peptide of nearly 1900 Da. Did a molecule this large truly cross the intestinal wall? If its absolute oral bioavailability is very low, why can it treat psoriasis in the skin? These are the questions the next article will answer.
Next article: The Birth of ICOTYDE (3): 1898 Da—Why Can It Still Be Oral?