Phenotypic screening for small molecules that lower PrP in cultured cells

Although a phenotypic screen in mouse cells identified two small molecules that selectively lower PrP via a proteasome-dependent post-translational mechanism, their failure to reduce PrP in human cell lines and in vivo highlights the significant translational challenges of mechanism-agnostic screening for prion disease therapeutics.

The Gist

The Big Picture: A "Wanted" Poster for a Dangerous Protein

Imagine the brain as a bustling city. In a disease called prion disease, a specific protein called PrP (think of it as a "bad citizen") starts folding into the wrong shape. Once it folds wrong, it acts like a zombie, turning all the healthy PrP proteins into zombies too. This causes the city (the brain) to collapse rapidly.

Scientists know that if they can just lower the number of PrP proteins in the city, the disease stops. They have already tried big, heavy tools like "genetic scissors" (CRISPR) and "molecular mail" (antisense oligonucleotides) to delete the instructions for making PrP. These work, but they are hard to deliver.

The researchers in this paper wanted to find a tiny, simple pill (a small molecule) that could slip into the city, find the PrP protein, and quietly remove it without causing a riot.

The Search: Looking for a Needle in a Haystack

To find this magic pill, the scientists set up a massive screening lab.

• The Haystack: They had a library of 3,492 different chemical compounds. These weren't random junk; they were a curated list of drugs with known "jobs" (mechanisms of action).
• The Test: They grew mouse brain cells in a petri dish. To make the test fair, they added a "security camera" system:
  • PrP was tagged with Red Light.
  • A harmless protein (GFP) was tagged with Green Light.
  • If a pill turned off the Red Light but kept the Green Light on, it was a potential winner. If it turned off both, it was just a toxic poison that killed the whole cell.

The Discovery: Two "Magic Bullets" (That Turned Out to be Duds)

Out of the 3,492 compounds, the screen found three that looked promising. Two of them, named EYH and LCZ, were the stars of the show.

• The Good News: In the mouse cells, these two compounds were incredibly specific. They acted like a precision sniper, shooting down the Red Light (PrP) while leaving the Green Light (harmless protein) and the cell itself completely untouched.
• The Mechanism: They figured out how they worked. The compounds didn't stop the cell from making the protein (like turning off the factory). Instead, they acted like a garbage disposal, telling the cell's internal recycling system (the proteasome) to eat the existing PrP proteins faster than usual.

The Plot Twist: The "Mouse Trap" vs. The "Human City"

This is where the story takes a sad turn. The researchers thought they had found a cure, but they hit a massive wall when they tried to move from the lab to the real world.

1. The Species Barrier (Mouse vs. Human)
The compounds worked beautifully in mouse cells. But when they tested them on human cells, the magic disappeared.

• Analogy: Imagine you found a key that opens a specific lock on a toy car (the mouse cell). You are thrilled! But when you try that same key on a real car (the human cell), it doesn't fit at all. The human cells were either immune to the drug, or the drug became so toxic at the doses needed to work that it killed the cells.

2. The Body Barrier (The Blood-Brain Wall)
Even though the drugs worked in a petri dish, they failed when tested in live mice.

• The scientists fed the drugs to mice for two weeks.
• The Result: The drugs successfully entered the mouse's body and even reached the brain in high concentrations. But, the PrP levels in the brain didn't drop.
• Analogy: It's like sending a delivery truck full of "garbage collectors" to a city. The trucks arrived at the city gates (the brain), but for some reason, they couldn't get out of the truck to do the job. Or, the "garbage" (PrP) in the live mouse brain was behaving differently than in the petri dish.

The Conclusion: Why This Matters

The paper ends with a bittersweet lesson for the future of medicine.

1. The Challenge of "Black Box" Hunting: The researchers tried to find a drug by just looking for the result (lower PrP) without knowing exactly how the drug worked. This is called "phenotypic screening." While they found two drugs that worked in the lab, they couldn't figure out exactly how they worked, and they couldn't make them work in humans.
2. The Shift in Strategy: Because of these failures, the authors suggest that for prion disease, we should stop guessing and start engineering. Instead of hoping a random pill will work, we should use platform technologies (like the genetic tools mentioned earlier) where we know exactly how the drug works and how to deliver it.

In a nutshell: The scientists built a sophisticated trap to catch a "bad protein" using mouse cells. They caught two beautiful butterflies (EYH and LCZ), but when they tried to release them into the wild (human bodies), the butterflies died or refused to fly. It was a beautiful experiment that taught them that finding a simple pill for this complex disease is much harder than it looks, and we might need to rely on more precise, engineered tools instead.

Technical Summary

1. Problem Statement

Prion diseases are fatal neurodegenerative disorders driven by the accumulation of misfolded prion protein (PrP). A validated therapeutic hypothesis is that reducing the levels of endogenous PrP (PrP lowering) can halt disease progression. While platform technologies like antisense oligonucleotides (ASOs) and siRNA have shown efficacy, they face delivery challenges (e.g., intrathecal administration). Small molecules offer a potential alternative due to their ability to diffuse through tissues without compartment-restricted delivery. However, previous high-throughput screens for PrP-lowering small molecules have failed to yield compounds with in vivo efficacy, often plagued by non-specific cytotoxicity, pan-assay interference compounds (PAINS), and poor reproducibility.

2. Methodology

The authors designed a rigorous, mechanism-agnostic phenotypic screening campaign to identify small molecules that selectively lower PrP while sparing cell viability and general protein expression.

• Cell Model: They generated stable mouse Neuro-2a (N2a) cell lines expressing:
  • Cytosolic GFP: To monitor general protein synthesis and cell health.
  • GFP-GPI: A GPI-anchored protein sharing the biosynthetic pathway with PrP, serving as a counter-screen for non-specific suppression of membrane proteins.
• Screening Strategy:
  • Library: A curated "MoA box" library of 3,492 compounds with annotated mechanisms of action (to minimize PAINS and aid interpretation).
  • Assay: High-content imaging in 1536-well plates. Cells were treated with compounds at 1 µM and 10 µM.
  • Readouts: Quantification of PrP signal (membrane), GFP signal (cytosol), and cell count (viability).
  • Triage: Hits were selected based on a 10-fold therapeutic index (lowering PrP without affecting GFP or viability). Hits underwent progressive dose-response confirmation (4-point, then 8-point) and Western blot validation.
• Characterization:
  • Proteomics: Tandem Mass Tag (TMT) mass spectrometry on whole-cell lysates to assess global protein changes.
  • Mechanism Studies: RT-qPCR for mRNA levels, transfection of exogenous PrP ORF in PrP-negative cells (RK13), and co-treatment with degradation pathway inhibitors (MG132 for proteasome, Bafilomycin A1 for lysosome).
  • Human Validation: Testing in various human cell lines (U251-MG, HEK293, A375, HT29, U87MG) to assess translatability.
  • In Vivo Testing: Oral gavage administration (100 mg/kg daily for 14 days) in wild-type mice, followed by PrP quantification in brain and colon via ELISA and pharmacokinetic (PK) analysis via UPLC-MS/MS.

3. Key Contributions

• Development of a Robust Screening Platform: The integration of a dual-reporter system (PrP vs. GFP-GPI) in a high-throughput format allowed for the specific selection of PrP-lowering compounds while filtering out general cytotoxins.
• Identification of Selective Hits: The screen successfully identified two compounds, EYH and LCZ, that selectively lower PrP in mouse N2a cells with high specificity compared to the broader proteome.
• Mechanistic Insight: The study elucidated that these compounds act via a post-translational mechanism, specifically promoting the proteasomal degradation of PrP, rather than inhibiting transcription or translation.
• Critical Assessment of Phenotypic Screening: The paper provides a rigorous case study on the limitations of mechanism-agnostic screening, highlighting the disconnect between in vitro potency in murine cells and in vivo efficacy or human cell activity.

4. Key Results

• Hit Identification: From 3,492 compounds, three initial hits (EYH, LCZ, Y-320) were identified. Y-320 was discarded due to broad proteomic toxicity (affecting ~24% of the proteome).
• Selectivity of EYH and LCZ:
  • Proteomics: EYH and LCZ showed high specificity. EYH downregulated only 21 proteins (0.2%), and LCZ downregulated 51 (0.6%). PrP was the #1 (EYH) and #2 (LCZ) most downregulated protein.
  • Specificity: Unlike Y-320, these compounds did not broadly suppress membrane or secreted proteins, suggesting a specific target or pathway.
• Mechanism of Action:
  • Post-Translational: PrP mRNA levels were minimally affected. Compounds lowered PrP in RK13 cells (lacking endogenous PrP regulation) and in non-dividing primary neurons within 24 hours, indicating clearance of existing protein.
  • Proteasomal Degradation: Co-treatment with the proteasome inhibitor MG132 caused an accumulation of unglycosylated PrP, confirming proteasomal clearance. Lysosomal inhibition did not rescue PrP levels.
  • Trim66: Both compounds upregulated Trim66, but genetic manipulation of Trim66 (overexpression/knockdown) did not alter PrP levels or compound potency, ruling it out as the primary effector.
• Failure in Human Models:
  • EYH and LCZ showed diminished potency in human cell lines (U251-MG, HEK293, etc.), requiring concentrations >20 µM (vs. 0.2–2 µM in N2a) to see effects, often accompanied by cytotoxicity.
  • No human cell line met the criteria for a CRISPR loss-of-function screen to identify the effector gene.
• Failure in Vivo:
  • Despite achieving brain tissue concentrations near or above the in vitro EC50 (EYH: 1.7 µM; LCZ: 1.75 µM) after 14 days of oral dosing (100 mg/kg), neither compound reduced PrP levels in the mouse brain or colon.
  • In the colon, EYH treatment actually resulted in a nominal increase in PrP.

5. Significance and Conclusion

• Challenges of Phenotypic Screening: The study underscores the difficulty of translating in vitro phenotypic hits to in vivo therapeutics, particularly when the mechanism of action is unknown. The compounds likely rely on murine-specific biology or high concentrations that induce off-target effects not present at therapeutic doses.
• Therapeutic Implications: The failure of EYH and LCZ to engage the target in vivo despite adequate exposure suggests that mechanism-agnostic small molecule screens for PrP lowering face profound hurdles.
• Strategic Shift: The authors conclude that while small molecules are attractive, the current success rate is low. They advocate for a shift in investment toward rational platform technologies with defined mechanisms of action (e.g., ASOs, siRNA, viral vectors, and engineered transcriptional repressors) that have already demonstrated in vivo efficacy in prion disease models. This paper serves as a cautionary tale and a validation of the need for mechanism-based drug discovery in this field.