Exploiting an Epigenetic Resistance Mechanism to PI3 Kinase Inhibition in Leukemic Stem Cells

This study identifies EZH1 upregulation as a resistance mechanism to PI3K inhibition in acute myeloid leukemia leukemic stem cells and demonstrates that combining a PI3K inhibitor with an EZH1/2 dual inhibitor overcomes this resistance to effectively target the disease in preclinical models.

The Gist

The Big Picture: The Unstoppable Leukemia Engine

Imagine Acute Myeloid Leukemia (AML) as a rogue factory that is churning out defective products (cancer cells) instead of healthy ones. The problem isn't just that the factory is running; it's that it has a "backup generator" and a "self-repair crew" that keeps it running even when doctors try to shut it down.

This paper is about a new strategy to not just turn off the main power switch, but to trick the factory into destroying its own backup systems, effectively shutting it down for good.

Part 1: The Main Power Switch (PI3K)

For a long time, scientists knew that AML cells rely heavily on a specific internal signal called the PI3K pathway. Think of this pathway as the main electrical grid powering the cancer factory. If you cut the power, the machines should stop.

Researchers tried using drugs to cut this power (PI3K inhibitors).

• The Good News: It worked at first! The cancer cells stopped multiplying, and some even started turning into normal, harmless cells (a process called differentiation).
• The Bad News: The cancer didn't stay dead. Like a smart factory, the cancer cells found a way to survive. They didn't mutate genetically (change their DNA); instead, they changed their "software" (epigenetics). They realized, "Hey, the main power is out, so let's switch to a different generator."

Part 2: The Sneaky Backup Generator (EZH2 and EZH1)

Here is where the story gets clever. The researchers discovered how the cancer survived the power outage.

1. The Main Generator (EZH2): Normally, the cancer relies on a protein called EZH2 to keep its "stem cells" (the master cells that can start new tumors) alive and powerful.
2. The Switch: When the PI3K power was cut, the cancer cells panicked. They realized they couldn't keep EZH2 running, so they shut it down.
3. The Backup (EZH1): But the cancer wasn't helpless. They had a twin brother protein called EZH1. Under normal circumstances, EZH1 is lazy and doesn't do much because EZH2 is doing all the work. However, once EZH2 was shut down, the cancer cells forced EZH1 to work overtime. EZH1 became the new "essential" protein keeping the cancer alive.

The Analogy: Imagine a car with two engines. Engine A (EZH2) is the big V8. Engine B (EZH1) is a tiny, inefficient motorcycle engine that usually just sits there. When the V8 is turned off, the car switches to the motorcycle engine. It's slow and inefficient, but it keeps the car moving just enough to survive.

Part 3: The Double-Strike Strategy

The researchers realized they couldn't just turn off the main power (PI3K) because the cancer would just switch to the backup engine (EZH1). They needed a two-pronged attack:

1. Step 1: Turn off the main power (PI3K inhibitor). This forces the cancer to rely entirely on the backup engine (EZH1).
2. Step 2: Smash the backup engine (EZH1/2 inhibitor).

The Result: The cancer is now in a "double bind." It has no main power, and its backup generator has been destroyed. The factory collapses completely.

What They Found in the Lab

• In Mice: When they gave mice this double-drug combo, the leukemia didn't just pause; it disappeared. The mice lived much longer, and in many cases, the cancer didn't come back.
• In Human Cells: They tested this on real blood samples from patients, including those with very dangerous types of leukemia (like those with TP53 mutations or those resistant to other common drugs like Venetoclax). The combo worked on almost all of them.
• Safety: Crucially, they checked healthy human cells. The combo hurt the cancer cells much more than the healthy cells, suggesting there is a "therapeutic window" where you can kill the bad guys without hurting the good guys.

Why This Matters

This paper is exciting for three main reasons:

1. It Targets the Root Cause: It focuses on Leukemic Stem Cells (LSCs). These are the "seeds" of the cancer. If you only kill the big trees (the bulk of the cancer) but leave the seeds, the forest grows back. This combo kills the seeds.
2. It Outsmarts Resistance: Instead of fighting the cancer's resistance, the researchers used the resistance against it. They forced the cancer to become dependent on a weak backup system, then destroyed that system.
3. It Uses Existing Drugs: They used a PI3K inhibitor (Copanlisib) and an EZH inhibitor (Valemetostat). Both drugs are already known to be relatively safe in humans. This means this new strategy could potentially move to human clinical trials faster than inventing a brand-new drug from scratch.

The Bottom Line

Think of this research as finding a way to cut the power to a villain's fortress, forcing them to rely on a weak, portable generator, and then blowing up that generator. It's a smart, two-step trap that stops the cancer from hiding and relapsing, offering new hope for patients who have run out of options.

Technical Summary

1. Problem Statement

Acute Myeloid Leukemia (AML) remains a devastating disease with high relapse rates, primarily due to the persistence of Leukemic Stem Cells (LSCs) that survive standard therapies. While the PI3K/AKT signaling pathway is hyperactivated in up to 80% of AML cases and is a known driver of LSC self-renewal, clinical trials of PI3K inhibitors have failed to yield approved therapies for AML. The failure is attributed to:

• Toxicity: The pathway is essential for normal hematopoiesis.
• Resistance: The emergence of non-genetic, adaptive resistance mechanisms that allow LSCs to survive PI3K inhibition.
• Isoform Redundancy: Uncertainty regarding which specific PI3K isoforms are critical for LSC maintenance versus normal cells.

The study aims to identify the specific PI3K isoform dependencies of LSCs, elucidate the mechanism of acquired resistance to PI3K inhibition, and develop a combinatorial therapeutic strategy to overcome this resistance and eradicate LSCs.

2. Methodology

The researchers employed a multi-faceted approach combining genetic models, pharmacological inhibition, and patient-derived samples:

• Genetic Models:
  • Generated conditional knockout mice for specific Class IA PI3K isoforms (Pik3ca, Pik3cb, Pik3cd) in hematopoietic cells.
  • Used the KMT2A-MLLT3 (MLL-AF9) retroviral model to induce aggressive AML and assess LSC self-renewal via serial replating assays.
  • Utilized NPM1c/NRAS and Patient-Derived Xenograft (PDX) models for in vivo efficacy testing.
• Pharmacological Interventions:
  • Tested isoform-selective inhibitors (Alpelisib for P110α, TGX221 for P110β, Idelalisib for P110δ) and dual inhibitors (Copanlisib for P110α/δ, Valemetostat for EZH1/2).
  • Synthesized Valemetostat in-house for in vivo use.
• Omics and Molecular Analysis:
  • Transcriptomics: Microarray analysis and Gene Set Enrichment Analysis (GSEA) on sorted GFP+ Granulocyte-Macrophage Progenitors (GMPs).
  • Proteomics: Mass spectrometry (LC-MS/MS) for histone modification profiling (H3K27me3, H3K27ac).
  • Western Blotting & qPCR: To monitor protein levels (EZH2, EZH1, p-AKT) and gene expression over time.
• Functional Assays:
  • Colony Forming Unit (CFU) Assays: To measure self-renewal and differentiation in mouse bone marrow and human patient samples (AML/MDS).
  • Flow Cytometry: To assess cell cycle, apoptosis (Annexin V/7-AAD), and surface markers (CD11b, CD15, CD115).
  • Limiting Dilution Transplantation: To quantify Leukemia-Initiating Cell (LIC) frequency.

3. Key Contributions & Findings

A. Isoform-Specific Dependency of LSCs

• P110α is Critical: Genetic deletion of Pik3ca (encoding P110α) significantly impaired LSC self-renewal and prolonged survival in KMT2A-MLLT3 AML models. In contrast, deletion of Pik3cb or Pik3cd had minimal impact on disease progression.
• Differentiation Induction: PI3K inhibition (specifically P110α) forced LSCs to undergo myeloid differentiation (monocytic/neutrophilic) and lose self-renewal capacity, both in vitro and in vivo.

B. Discovery of an Epigenetic Resistance Mechanism

• EZH2 Downregulation: While PI3K inhibition initially suppressed leukemia, resistant cells emerged. Mechanistic analysis revealed that PI3K inactivation leads to a time-dependent downregulation of EZH2 protein levels (a core component of the Polycomb Repressive Complex 2, PRC2).
• Compensatory Upregulation of EZH1: Despite the loss of EZH2 protein, H3K27me3 levels were maintained in resistant cells. The study identified that EZH1 (a homolog of EZH2) is compensatorily upregulated at the RNA and protein level in response to PI3K inhibition.
• Synthetic Lethality: This creates a new dependency: PI3K-inhibited cells become reliant on EZH1 for survival.

C. Therapeutic Strategy: Dual Inhibition

• Combination Therapy: The study demonstrated that combining a PI3K inhibitor (Copanlisib) with an EZH1/2 dual inhibitor (Valemetostat) overcomes the acquired resistance.
• Mechanism of Action: The combination prevents the compensatory upregulation of EZH1 and blocks the residual PRC2 activity, leading to:
  • Sustained suppression of proliferation.
  • Increased apoptosis (synthetic lethality).
  • Cell cycle arrest (G0).
  • Terminal myeloid differentiation.

4. Results

• In Vitro Cell Lines: The Copanlisib + Valemetostat combination effectively suppressed proliferation in diverse AML cell lines (including KMT2A-rearranged and non-rearranged) and venetoclax-resistant lines (MOLM13-VR). Single agents showed transient efficacy followed by regrowth.
• In Vivo Mouse Models:
  • In NPM1c/NRAS and KMT2A-MLLT3 models, the combination therapy significantly prolonged survival compared to single agents or vehicle.
  • LIC Depletion: Extreme Limiting Dilution Analysis (ELDA) showed a 10-fold reduction in Leukemia-Initiating Cell frequency in the combination group.
  • Toxicity: The combination showed a therapeutic window; while it reduced colony numbers in healthy mouse bone marrow, it did not eliminate all colony types, suggesting normal hematopoiesis was less affected than leukemic growth.
• Patient Samples:
  • The combination significantly reduced colony formation in primary AML and high-risk MDS patient samples, including those with complex cytogenetics, TP53 mutations, and venetoclax resistance.
  • Patient-derived cells were more sensitive to Valemetostat than established cell lines.
  • In a PDX model of venetoclax-relapsed AML, the combination provided a significant survival advantage with no observed weight loss (minimal toxicity).

5. Significance

This study provides a novel paradigm for treating AML by targeting the epigenetic resistance mechanisms that arise during signaling pathway inhibition.

• Overcoming Resistance: It identifies EZH2 downregulation/EZH1 upregulation as a specific non-genetic resistance mechanism to PI3K inhibition and proposes a rational combination therapy to exploit this vulnerability.
• Targeting LSCs: The regimen specifically targets the leukemic stem cell pool, which is the root cause of relapse, rather than just the bulk tumor.
• Broad Applicability: The efficacy across diverse molecular subtypes (including TP53 mutant and venetoclax-resistant AML) suggests this could be a universal therapeutic strategy.
• Clinical Translation: Since Copanlisib and Valemetostat are already FDA-approved (or approved in Japan) for other malignancies with known safety profiles, this combination offers a rapid pathway to clinical trials for AML patients who currently have limited treatment options.

In summary, the authors demonstrate that PI3Kα inhibition induces an adaptive shift to EZH1 dependency, which can be therapeutically exploited by combining PI3K and EZH1/2 inhibitors to eradicate AML stem cells and prevent relapse.