Stabilized cCMP and cUMP analogues as orthogonal immunomodulators to the cGAS-STING axis

Initial intuition

Tal et al. (Cell 2021) characterized the Pycsar system—a family of bacterial cyclases that synthesize cyclic pyrimidines (cCMP, cUMP) in response to phage infection, alongside the cyclic purines of CBASS. Human biology currently knows of no canonical equivalent for cyclic pyrimidine signals: our innate systems (cGAS-STING, RIG-I, MDA5) are hardwired to cGAMP and 2’3’-cGAMP exclusively.

This absence of a eukaryotic ortholog is doubly interesting:

1. Blank chemical space — no human pharmacological probe exploiting cCMP/cUMP exists. This is typically the kind of niche where Bactaegion in silico hits can generate a significant first publication.
2. Potential TLR activation — early literature on extracellular nucleotides (DAMP, PAMPs) suggests that TLR3 (dsRNA) and TLR7/8 (ssRNA) could be sensitive to structured pyrimidine motifs. Activation by stable cCMP/cUMP is not excluded.

Hypothesis

A library of stabilized cCMP and cUMP analogues, designed to resist human phosphodiesterases (PDEs):

• Phosphorothioate substitutions (S-cCMP, S-cUMP)
• 2’-O-methyl on the ribose
• Modified bases (5-methyl-C, pseudouridine)
• Lipophilic conjugation for cellular internalization (short alkyl chains, ProTide prodrugs)

should contain at least one selective activator of at least one human TLR pathway, without significant overlap with the cGAS-STING axis. This selectivity would open up:

1. Orthogonal vaccine adjuvants (broad-spectrum utility)
2. Immunomodulators for patients with deficient interferon response (chronic HCV, certain autoimmunities)
3. Research tools to dissect human innate subnetworks

In silico plan

1. Pycsar structures: retrieve AlphaFold structures for the identified Pycsar cyclases (P0DV24, P0DV28, P0DV40). Characterize the catalytic pocket and product specificity (cCMP vs cUMP vs hybrids). Compare to CBASS CD-NTases for chemical differentiation.

2. RDKit enumeration: ~150 cyclic pyrimidine analogues. Lipinski filter + estimated pharmacokinetic properties (LogP 1-3, PSA < 140, HBD < 5).

3. Docking on human receptors:

   • TLR3 (PDB 3CIY) — dsRNA site
   • TLR7 (PDB 5GMH) — ssRNA / small ligand site
   • TLR8 (PDB 5WYZ) — ssRNA site
   • Human cGAS (PDB 6BV8) — selectivity negative control

   Target: Vina score < −7 on ≥1 TLR AND > −4 on cGAS (selectivity).

4. Short MDs on the top 20 hits, verification of binding mode stability over 100 ns.

Limitations and risks

• The “TLRs sensitive to cyclic pyrimidines” hypothesis is speculative: current literature mostly indicates TLR recognition by linear RNA or single-stranded DNA, not by individual cyclic nucleotides. If all hits miss the TLRs, we fall back on an adjacent CBASS antagonist project—not bad, but outside the Pycsar thesis.
• Human PDEs (PDE1-11) are diverse and some hydrolyze a priori stable analogues. The ADMET screen must include an estimation of the simulated plasma half-life (DEREK, ADMET-AI software).
• The chemical synthesis of modified cCMP/cUMP is non-trivial (the 2’,3’ cycles are sensitive). A competent partner in nucleotide chemistry will be necessary in the wet phase (DNDi-like, M4K Pharma, or academic CRO).

Links to the Bactaegion scope

V1 Family Pycsar (high translational priority, small molecule modality). Seminal paper Tal 2021 (Cell, DOI 10.1016/j.cell.2021.07.011). Strategy aligned with the family sheet: model product-effector chemistry, screen for more stable cCMP/cUMP analogues. Annotated catalog + virtual library.

This lead is complementary to CBASS mimetics: it explicitly aims to avoid cGAS-STING overlap to settle into an immune niche not yet served pharmacologically.