Selective suppression and biasing of chemokine receptors CCR9 and ACKR4 through targeting CCL25 with de novo miniproteins

This study demonstrates that de novo designed miniproteins targeting the chemokine CCL25 can effectively suppress or bias its signaling through receptors CCR9 and ACKR4, offering a novel therapeutic strategy for inflammatory diseases and new tools to study chemokine receptor activation.

The Gist

Imagine your body's immune system is a massive, bustling city. To keep the city safe, it needs to move its security guards (immune cells) to the right neighborhoods at the right time. These guards don't have GPS; instead, they follow "wanted posters" called chemokines. These posters tell the guards where to go.

One specific poster, called CCL25, is responsible for guiding guards to the gut (intestines). Usually, this is helpful. But sometimes, the city gets confused, and too many guards are sent to the gut, causing inflammation and diseases like Crohn's.

For years, scientists tried to stop this by trying to "jam the locks" on the guards' doors (the receptors) so they couldn't read the posters. But the locks are tricky, and the guards are stubborn. It's like trying to stop a thief by changing the lock on a door they can just pick anyway.

The New Strategy: Stealing the Poster

Instead of trying to jam the locks, this research team decided to try something smarter: They designed a tiny, custom-made "mask" to cover up the poster itself.

Here is how they did it, using a high-tech "AI architect":

1. The AI Architect (BindCraft): The scientists used a powerful computer program (powered by AI) to design brand-new, tiny proteins from scratch. Think of this like using a 3D printer to print a custom glove that fits perfectly over a specific shape.
2. The Target: They printed four different "gloves" (called miniproteins) designed to fit perfectly over the CCL25 poster.
3. The Result: When these tiny gloves snap onto the poster, the guards can no longer see the instructions. The guards stay put, and the inflammation stops.

The Two Types of Gloves

The team found that not all gloves work the same way. They discovered two very different types of "masks":

Type 1: The "Total Blocker" (The VUP25101, 107, and 112 gloves)
These gloves cover the part of the poster that the guards need to grab onto to get moving.

• What happens: The guards see the poster, but they can't hold it. They can't move, they can't signal, and they can't cause trouble.
• The Analogy: It's like putting a "Do Not Enter" sign over the map. The guards are confused and stop completely. This stops both the "movement" signal and the "cleanup" signal.

Type 2: The "Smart Filter" (The VUP25111 glove)
This one is the most fascinating. It covers a different part of the poster.

• What happens: The guards can still grab the poster and get the "move" signal (G-protein signaling). They still go to the gut. BUT, they don't get the "stop and clean up" signal (arrestin recruitment).
• The Analogy: Imagine a delivery driver who gets the address to drop off a package (movement) but doesn't get the instruction to call the police or file a report (inflammation/cleanup). The driver does the job, but without the chaotic side effects.
• Why this matters: This is a "biased" signal. It lets the body do the good work of moving cells without triggering the bad work of inflammation. It's like a surgeon's scalpel that cuts the tumor but leaves the healthy tissue untouched.

Why This is a Big Deal

1. New Tools for Old Problems: For a long time, scientists struggled to make drugs that work on these specific immune signals. This proves that using AI to design tiny, custom proteins is a winning strategy.
2. Precision Medicine: The "Smart Filter" (VUP25111) shows that we can be incredibly precise. We don't just have to turn the system "off"; we can tweak it to only turn off the bad parts while keeping the good parts running.
3. Future Hope: This gives hope for treating inflammatory bowel diseases and even stopping cancer from spreading to the gut, by using these tiny, custom-designed "masks" to trick the immune system into behaving.

In short: Instead of trying to break the lock on the door, the scientists built a custom cover for the map, effectively telling the immune guards to either "stop moving" or "move without causing a mess." It's a clever, high-tech way to calm down an angry immune system.

Technical Summary

1. Problem Statement

The chemokine signaling axis, specifically the interaction between the chemokine CCL25 and its receptors CCR9 (canonical) and ACKR4 (atypical), plays a critical role in immune cell migration and homeostasis. Dysregulation of this axis is implicated in inflammatory bowel diseases (IBD), Crohn's disease, and cancer metastasis.

• Therapeutic Challenge: Developing drugs targeting chemokine receptors (GPCRs) has historically been difficult, with few successful candidates reaching the clinic.
• Alternative Strategy: While most efforts target the receptor, natural regulatory mechanisms (e.g., tick evasins, atypical receptors) often target the chemokine ligand itself. However, chemokines are small, flexible proteins with low "ligandability," making them difficult to target with small molecules (NMR screens showed <1% hit rates with weak affinity).
• Goal: To develop high-affinity, specific modulators targeting CCL25 to inhibit or bias CCR9/ACKR4 signaling, leveraging recent advances in AI-driven de novo protein design.

2. Methodology

The study utilized a computational and experimental workflow centered on the BindCraft platform, an AI-driven inverse-design pipeline.

• Computational Design:

  • Target: A truncated form of human CCL25 (residues 1-77, CCL25_ΔCT) was used to focus designs on the core chemokine fold, excluding the long C-terminal extension.
  • Algorithm: BindCraft (based on AlphaFold2-Multimer) generated 75–125 amino acid miniprotein binders. No predefined binding hotspots were imposed, allowing the algorithm to explore the entire surface.
  • Filtering: 101 candidates passed initial filters. Designs were ranked by interface predicted TM-score (ipTM) and validated via Molecular Dynamics (MD) simulations (100 ns) and MM/GBSA binding free energy calculations.
  • Selection: Four miniproteins (VUP25101, VUP25107, VUP25111, VUP25112) were selected based on structural diversity, predicted binding energy, and interface characteristics.

• Experimental Validation:

  • Production: Miniproteins were expressed in E. coli, purified, and characterized.
  • Binding Affinity: Fluorescence Polarization (FP) assays measured binding to full-length CCL25.
  • Receptor Binding: Time-Resolved FRET (TR-FRET) assays assessed the ability of CCL25-miniprotein complexes to bind CCR9 and ACKR4.
  • Signaling Assays (BRET):
    • Arrestin Recruitment: Measured $\beta$-arrestin2 translocation.
    • G Protein Coupling: Measured recruitment of mini-G proteins (mGi) and downstream cAMP inhibition.
    • GRK Recruitment: Measured GPCR Kinase 3 (GRK3) translocation.
    • Internalization: Measured receptor endocytosis.
  • Cell Migration: Transwell assays using MOLT-4 lymphoblasts (CCR9+ cells) to quantify chemotaxis.
  • Structural Modeling: AlphaFold3 was used to model receptor:chemokine complexes and predict steric clashes with the miniproteins.

3. Key Contributions

• First De Novo Chemokine Modulators: Demonstrated the successful application of AI-driven de novo protein design to generate high-affinity binders for a difficult-to-target chemokine (CCL25).
• Mechanism of CRS1 Blocking: Identified that blocking the Chemokine Recognition Site 1 (CRS1)—the interaction between the chemokine body and the receptor N-terminus—is sufficient to completely suppress receptor activation for both canonical (CCR9) and atypical (ACKR4) receptors.
• Discovery of a Biased Agonist Complex: Uncovered a unique mechanism where a specific miniprotein (VUP25111) binds CCL25 in a way that creates a G protein-biased agonist. This complex activates G protein signaling but fails to recruit arrestins.
• Receptor Selectivity: Showed that the VUP25111:CCL25 complex inhibits CCR9 arrestin recruitment but does not inhibit ACKR4, highlighting the potential for receptor-selective modulation by targeting the ligand.

4. Key Results

A. Binding and Structural Characterization

• All four miniproteins bound CCL25 with high affinity ($pK_d$ ranging from 6.30 to 6.82).
• VUP25101, VUP25107, VUP25112: These binders target the C-terminal helix of CCL25, directly occluding the CRS1 interface. Structural modeling confirmed steric clashes with the receptor N-terminus.
• VUP25111: This binder targets the $\beta$1-strand and 40s-loop of CCL25. It does not block CRS1 or CRS2 directly but creates a complex that clashes with the Extracellular Loop 2 (ECL2) of CCR9.

B. Inhibition of Receptor Activation

• General Inhibitors (VUP25101, 107, 112): These effectively blocked CCL25 binding to both CCR9 and ACKR4. They completely inhibited:
  • Arrestin recruitment.
  • G protein (mGi) recruitment.
  • GRK3 recruitment.
  • Receptor internalization.
  • MOLT-4 cell migration.
• The Biased Binder (VUP25111):
  • Arrestin: Strongly inhibited CCR9 arrestin recruitment ($pIC_{50} \approx 5.77$).
  • G Protein: Did not inhibit mGi recruitment or cAMP production, even at high concentrations.
  • GRK: Partially inhibited GRK3 recruitment, suggesting phosphorylation still occurs.
  • Internalization: Inhibited CCR9 internalization (likely arrestin-independent mechanisms), but with lower potency.
  • ACKR4: Showed no inhibition of ACKR4 activation or arrestin recruitment.

C. Cell Migration

• VUP25101 and VUP25107 completely abolished CCL25-mediated MOLT-4 cell migration.
• VUP25111 had no effect on migration, consistent with the finding that CCR9-mediated migration in these cells is strictly dependent on G protein signaling, which remains active in the presence of VUP25111.

5. Significance

• Therapeutic Potential: The study validates targeting chemokine ligands as a viable strategy for modulating GPCR signaling, offering a solution to the historical difficulty of targeting chemokine receptors directly.
• Bias Engineering: The discovery of VUP25111 demonstrates that de novo proteins can be engineered to convert a balanced natural agonist (CCL25) into a biased agonist. This allows for the selective suppression of specific pathways (e.g., arrestin-mediated desensitization/internalization) while preserving others (e.g., G protein-mediated migration), or vice versa.
• Receptor Specificity: The ability to inhibit CCR9 without affecting the scavenging receptor ACKR4 is crucial. Blocking ACKR4 could lead to elevated systemic chemokine levels, potentially worsening the disease state. This ligand-targeting approach avoids that pitfall.
• Tool Development: These miniproteins serve as powerful research tools to dissect the structural basis of chemokine receptor activation, specifically revealing that CRS1 blocking is sufficient for inhibition and that ECL2 interactions are critical for arrestin coupling.

In conclusion, this paper establishes a new paradigm for GPCR drug discovery: using AI-designed miniproteins to target the ligand rather than the receptor, achieving high specificity, potent inhibition, and the ability to fine-tune signaling bias.