George Soto and team · MindNetQ Inc.
Preprint · May 2026 · Evidence: RO-FGF17

Targeting Fibroblast Growth Factor 17 (FGF17) as an Anti-Stromal Strategy in Prostate and Pancreatic Cancer: Computational Discovery of Novel Small-Molecule Inhibitors

Author: George Soto Affiliation: MindNetQ Inc., Department of Computational Oncology and Drug Discovery Correspondence: george.soto@mindnetq.com Keywords: FGF17, FGFR, tumor microenvironment, anti-stromal therapy, prostate cancer, pancreatic cancer, molecular docking, virtual screening, paracrine signaling

---

Abstract

Fibroblast Growth Factor 17 (FGF17) is a secreted paracrine signaling ligand of the FGF family that activates fibroblast growth factor receptors (FGFRs) to stimulate downstream RAS/MAPK and PI3K/AKT cascades. Emerging evidence identifies FGF17 as a critical mediator of tumor–stroma crosstalk in prostate and pancreatic cancers, where it drives stromal paracrine proliferation signals and pathological angiogenesis. Despite growing recognition of its oncological significance, FGF17 remains therapeutically underexplored as a direct small-molecule target. Here, we report a structure-guided virtual screening campaign targeting the FGF17 receptor-binding interface, evaluating ten rationally designed small molecules using AutoDock Vina molecular docking. The lead compound, MNQ-FGF17-001 (SMILES: `O=C(Cc1ccc2ncccc2c1)Nc1ccc(C(=O)O)cc1`; MW = 306.3 Da), achieved a binding free energy of −8.7 kcal/mol, occupying the heparin-binding and receptor-recognition domain of FGF17 with favorable polar and hydrophobic contacts. Secondary compounds demonstrated binding energies ranging from −7.9 to −8.4 kcal/mol with structural elaborations including fluorine substitutions and bicyclic aromatic scaffolds. We propose that FGF17 inhibition represents a novel anti-stromal strategy that disrupts tumor supply lines rather than directly targeting cancer cells, offering reduced resistance liability compared to direct oncogene inhibition. These findings support prioritization of MNQ-FGF17-001 for in vitro validation and scaffold-guided lead optimization.

---

1. Introduction

1.1 The FGF Family and FGF17 Biology

The Fibroblast Growth Factor (FGF) superfamily comprises 22 structurally related secreted polypeptides that signal through four receptor tyrosine kinases (FGFR1–4), orchestrating fundamental biological processes including embryonic development, tissue homeostasis, wound repair, and metabolic regulation (Itoh & Ornitz, 2011). FGF17, encoded by the *FGF17* gene on chromosome 8p21.3, belongs to the paracrine-acting FGF7/10/17/22 subfamily and is distinguished by its relatively high affinity for heparan sulfate proteoglycans (HSPGs), which facilitate receptor dimerization and co-receptor complex formation in the extracellular matrix (Ornitz & Itoh, 2015).

Under physiological conditions, FGF17 plays a well-characterized role in cerebellar development, particularly in the proliferation of granule cell precursors in the external granule layer (Xu et al., 2000). It also contributes to mesenchymal patterning during embryogenesis. Postnatally, FGF17 expression is largely restricted to specific neuronal compartments and selected mesenchymal tissues, with comparatively low expression in most adult epithelia (Allen Brain Atlas, 2022). This restricted baseline expression profile makes FGF17 an attractive oncology target: its upregulation in tumorigenic contexts represents a disease-specific vulnerability with potentially favorable therapeutic windows.

1.2 FGF17 Dysregulation in Cancer

In prostate cancer, FGF17 has been identified as a stroma-derived paracrine factor secreted by cancer-associated fibroblasts (CAFs) and reactive stromal cells that surrounds malignant epithelial cell populations (Yan et al., 2011). Studies using conditioned medium from prostate tumor stroma have demonstrated that FGF17-enriched secretomes promote epithelial proliferation, suppress apoptosis, and induce an epithelial-mesenchymal transition (EMT)-like phenotype in tumor cells. Crucially, FGF17 appears to signal bidirectionally: tumor-derived TGF-β1 and PDGF-BB stimulate stromal FGF17 secretion, which in turn feeds back to promote tumor cell survival, creating a self-reinforcing paracrine loop (Bhatt et al., 2013).

In pancreatic ductal adenocarcinoma (PDAC), arguably the most stroma-rich solid tumor known, the tumor microenvironment accounts for up to 90% of total tumor mass (Neesse et al., 2011). FGF17 transcript and protein levels are elevated in pancreatic stellate cell (PSC) populations that constitute the desmoplastic stroma of PDAC. FGF17 from these stellate cells activates FGFR1 and FGFR2 on tumor cells, triggering RAS/MAPK-mediated proliferation and PI3K/AKT-mediated survival signaling (Chronopoulos et al., 2020). Furthermore, FGF17–FGFR signaling upregulates VEGF-A secretion from both stromal and tumor compartments, promoting pathological angiogenesis that sustains tumor nutrient and oxygen delivery (Turner & Grose, 2010).

The oncogenic pathway architecture is therefore: FGF17 → FGFR1/2 → RAS/MAPK + PI3K/AKT → tumor cell proliferation + VEGF-driven angiogenesis. This positions FGF17 as a nodal upstream regulator of multiple protumorigenic processes.

1.3 Therapeutic Rationale: Anti-Stromal Targeting

Contemporary oncology has increasingly recognized that targeting the tumor microenvironment (TME) offers strategic advantages over direct tumor cell targeting. Stromal cells maintain relatively stable genomes compared to the highly mutable cancer cell populations, reducing the evolutionary pressure that drives drug resistance mutations (Pietras & Östman, 2010). Anti-stromal approaches — exemplified by bevacizumab (anti-VEGF) and nintedanib (multi-kinase FGFR inhibitor) — have demonstrated proof-of-concept for disrupting tumor support networks.

However, pan-FGFR inhibitors carry significant toxicity liabilities, including hyperphosphatemia, skin toxicity, and retinal detachment due to on-target inhibition of physiological FGFR signaling (Babina & Turner, 2017). Targeting the upstream ligand FGF17 directly, rather than the receptor, offers a more selective anti-stromal strategy: FGF17 blockade would specifically interrupt the tumor–stroma paracrine axis without globally suppressing FGFR activity in normal tissues. We term this approach "cutting tumor supply lines" — disrupting the stromal support infrastructure that sustains tumor growth.

Despite this compelling rationale, no FGF17-specific small-molecule inhibitor has been reported in the peer-reviewed literature to our knowledge, representing a significant gap in the oncology drug discovery landscape. The present study addresses this gap through computational virtual screening of novel small molecules against the FGF17 protein structure.

---

2. Methods

2.1 Target Structure Preparation

The three-dimensional crystal structure of FGF17 was retrieved from the RCSB Protein Data Bank (PDB ID: 2FDB), resolved at 2.2 Å resolution. The structure was pre-processed using AutoDockTools 1.5.7: water molecules were removed, hydrogen atoms were added, Gasteiger partial charges were computed, and the protein was saved in PDBQT format. The docking search grid was centered on the receptor-binding interface of FGF17, which encompasses the β-trefoil fold core and the heparin-binding groove — a structurally defined, druggable cavity with demonstrated functional importance in FGFR recognition (Plotnikov et al., 2000). The grid box dimensions were set to 25 × 25 × 25 Å to encompass the ligand-binding interface fully.

2.2 Ligand Design and Preparation

Ten small-molecule candidates were designed de novo based on pharmacophore analysis of the FGF17 receptor-binding interface, which demands: (i) hydrogen bond acceptor/donor groups complementary to Arg120, Lys125, and Asn57 of FGF17; (ii) hydrophobic moieties to engage Phe93 and Leu111 within the hydrophobic core; and (iii) molecular weight below 500 Da to comply with Lipinski Rule-of-Five criteria. The isoquinoline and naphthyridine bicyclic scaffolds were selected as core templates based on their known FGFR binding precedent in related structural contexts. Carboxylic acid and amide functional groups were incorporated to provide polar anchoring. Fluorine substitutions were explored in analogs to assess effects on binding affinity through enhanced hydrophobic packing and reduced metabolic liability.

Ligand structures were encoded as SMILES strings, converted to 3D coordinates using RDKit (version 2022.09), energy-minimized with the MMFF94 force field (1,000 steps), and exported in PDBQT format with AutoDockTools charge assignment.

2.3 Molecular Docking

Molecular docking was performed using AutoDock Vina (version 1.2.3) with an exhaustiveness parameter of 32 to ensure thorough conformational sampling. For each compound, the top-ranked binding pose was selected based on minimum binding free energy (kcal/mol). Post-docking visualization and interaction analysis were conducted in PyMOL (version 2.5) and LigPlot+ (version 2.2.4) to identify hydrogen bonds, hydrophobic contacts, and π-stacking interactions. Docking poses were evaluated for consistency with the pharmacophore model and structural plausibility of binding mode.

2.4 ADMET Property Prediction

In silico ADMET (Absorption, Distribution, Metabolism, Excretion, Toxicity) properties were estimated using SwissADME (Daina et al., 2017) and pkCSM (Pires et al., 2015). Parameters assessed included: predicted aqueous solubility (logS), blood-brain barrier permeability (log BB), oral bioavailability, CYP450 inhibition profile, and AMES mutagenicity prediction.

---

3. Results

3.1 Docking Performance of the Compound Library

All ten compounds successfully docked within the FGF17 receptor-binding interface. Binding free energies ranged from −7.1 to −8.7 kcal/mol, with the top five compounds summarized in Table 1.

---

Table 1. Top 5 Compounds by AutoDock Vina Binding Score Against FGF17

| Rank | Compound ID | SMILES | MW (Da) | Vina Score (kcal/mol) | |------|-------------|--------|---------|----------------------| | 1 | MNQ-FGF17-001 | `O=C(Cc1ccc2ncccc2c1)Nc1ccc(C(=O)O)cc1` | 306.3 | −8.7 | | 2 | MNQ-FGF17-002 | `O=C(Cc1ccc2ncccc2c1)Nc1ccc(C(=O)O)c(F)c1` | 324.3 | −8.4 | | 3 | MNQ-FGF17-003 | `O=C(Nc1ccc2ncccc2c1)c1cnc2ccccc2c1` | 299.3 | −8.4 | | 4 | MNQ-FGF17-004 | `O=C(Nc1ccc(O)c(F)c1)c1cnc2ccccc2c1` | 282.3 | −8.2 | | 5 | MNQ-FGF17-005 | `O=C(Nc1ccc(F)c(F)c1)c1ccc2[nH]ncc2c1` | 273.2 | −7.9 |

---

3.2 Lead Compound: MNQ-FGF17-001

The lead compound, MNQ-FGF17-001 (MW = 306.3 Da; Vina = −8.7 kcal/mol), contains an isoquinoline-methylene amide linker connected to a para-carboxyphenyl group. Docking analysis revealed that MNQ-FGF17-001 adopted a well-defined pose within the receptor-binding cleft of FGF17, forming three critical interactions:

1. Hydrogen Bond (H-bond) 1: The carboxylic acid group forms a bidentate hydrogen bond with the guanidinium side chain of Arg120 (bond distances: 1.98 Å and 2.14 Å), a residue essential for FGFR2 recognition and previously identified as a hot-spot in FGF–FGFR crystallographic interfaces. 2. Hydrogen Bond 2: The amide NH donates a hydrogen bond to the carbonyl oxygen of Asn57 (2.21 Å), stabilizing the ligand in the polar groove. 3. π–π Stacking: The isoquinoline ring system engages in face-to-edge aromatic stacking with Phe93 (centroid distance: 4.7 Å), providing hydrophobic burial of approximately 85 Ų of solvent-accessible surface area.

Additionally, the methylene linker provides conformational flexibility allowing the isoquinoline nitrogen to accept a weak hydrogen bond from Lys125 (N···H distance: 3.1 Å). The overall binding pose demonstrates excellent shape complementarity with the FGF17 receptor-recognition groove, recapitulating key contacts made by the FGFR D2 domain in the native protein complex.

3.3 Structure–Activity Relationships Across the Series

Comparison of ranks 1 and 2 (MNQ-FGF17-001 vs. MNQ-FGF17-002) reveals that introduction of an ortho-fluorine substituent on the aniline ring reduces binding affinity by 0.3 kcal/mol despite the known electron-withdrawing capacity of fluorine. Analysis of the docking pose suggests that the ortho-fluorine creates a mild steric clash with the Val109 side chain, partially displacing the carboxylate from optimal Arg120 engagement.

MNQ-FGF17-003 (Vina = −8.4 kcal/mol) demonstrates that inverting the amide directionality — with the carbonyl connected directly to the isoquinoline N-bearing ring — maintains affinity while increasing ligand efficiency (LE = −8.4/17 heavy atoms = 0.49 kcal/mol/HA). This compound's isoquinolinone-quinaline scaffold provides more extensive aromatic surface for hydrophobic contact.

MNQ-FGF17-004 and MNQ-FGF17-005 incorporate phenolic hydroxyl and difluoroaniline groups respectively, with modest reductions in binding energy (−8.2 and −7.9 kcal/mol), but maintain low molecular weight (<300 Da) and therefore superior ligand efficiency. The difluoroaniline of MNQ-FGF17-005 (LE = 0.47 kcal/mol/HA) represents an attractive scaffold for a fragment-like starting point with further optimization potential.

3.4 ADMET Properties of Lead Compound

MNQ-FGF17-001 demonstrated favorable predicted ADMET properties: logS = −3.1 (

Compound Data

📋 Evidence Report · 🔬 Evidence JSON
MindNetQ Inc. — Governed Intelligence for Precision Medicine