Structure-guided targeting of the GATAD2A-CHD4 interaction within the MBD2-NuRD complex results in high levels of HbF in adult erythroid cells

Scientists have found a way to "turn on" fetal hemoglobin in adult blood cells by disrupting a specific protein connection in the NuRD complex. This is achieved using a small peptide that prevents two proteins, GATAD2A and CHD4, from sticking together. This discovery could lead to new small-molecule therapeutics for sickle cell disease and beta-thalassemia. Such drugs might provide a scalable alternative to expensive gene therapies.

The search for a scalable hemoglobin switch

The regulation of hemoglobin—the protein in red blood cells that carries oxygen—is a fundamental process of human development. Humans undergo a sequential switch. We begin with embryonic hemoglobin, move to fetal hemoglobin (HbF), and eventually transition to adult beta-like hemoglobin (HbA) after birth. In patients with sickle cell disease (SCD) or beta-thalassemia, this adult hemoglobin is either defective or insufficient. However, if clinicians can artificially maintain high levels of HbF into adulthood, that fetal hemoglobin can substitute for the missing or broken adult versions. This helps mitigate the severe symptoms of these disorders.

Current therapeutic frontiers have focused heavily on BCL11A. This is a transcription factor (a protein that controls the rate of genetic transcription) that acts as a master silencer of the HbF gene. While CRISPR-based gene therapies targeting BCL11A have shown clinical success, they face massive economic and logistical hurdles. These treatments are highly specialized and expensive. They are difficult to deploy in resource-limited regions where the burden of SCD is highest. Consequently, there is a pressing need for small-molecule therapeutics. These are drugs that can be taken orally or via simple injection to prevent the silencing of the fetal gamma-globin (HBG) genes.

Breaking the NuRD silencing machinery

The researchers identified the MBD2a-NuRD chromatin remodeling complex as a prime candidate for such a drug. The NuRD complex functions as an epigenetic silencer. It does not change the DNA sequence itself. Instead, it alters the "packaging" of the DNA (chromatin) to prevent genes from being read. Specifically, the complex is composed of two functional sub-units. One is a histone deacetylase core complex (HDCC), which modifies the proteins around which DNA is wrapped. The other is a chromatin remodeling complex, which physically moves those proteins to block access to the gene.

The study focuses on the structural "bridge" that holds these two sub-units together. Using AlphaFold 3 and existing crystal structures, the authors mapped the interface between two specific proteins. These are GATAD2A (part of the HDCC) and CHD4 (the ATPase responsible for the remodeling activity). They identified a precise mechanical vulnerability. This is a short alpha-helix within the CR2 domain of GATAD2A that nests into the C-terminal domains of CHD4 [Figure 1A].

The mechanism for disruption follows a clear logic: 1. Identification: Mapping the physical contact points where the GATAD2A helix meets the CHD4 domains. 2. Destabilization: Introducing mutations (specifically replacing leucine with proline) to break the shape of the GATAD2A helix. This makes it unable to dock with CHD4. 3. Dissociation: Once the docking site is compromised, CHD4 fails to associate with the rest of the NuRD complex. This leaves the HBG promoter (the regulatory region of the gene) unprotected. 4. Activation: Without the NuRD complex to hold the "gate" shut, the chromatin shifts to an active state. This allows HbF to be produced.

Evidence of successful reactivation

The authors validated this mechanism through a hierarchy of experiments. They moved from purified proteins to engineered cell lines and finally to primary human cells.

First, they used biophysical assays to prove the interaction was real and targetable. Isothermal Titration Calorimetry (ITC) showed that the isolated CR2 helix binds to CHD4 with high affinity ($K_D = 11 \pm 2$ nM). This value describes how tightly the molecules stick together. They further demonstrated that adding an excess of the wild-type peptide could destabilize the entire ternary complex. This reduced its thermal melting temperature by $3.6^\circ\text{C}$ [Figure 2A].

The functional impact was even more striking. In HUDEP-2 cells (an adult erythroid cell model), the authors used prime editing—a high-precision genome editing method—to introduce mutations into the endogenous GATAD2A gene. The paper reports that these mutations resulted in a 140-fold increase in HBG mRNA expression. This pushed HbF levels to approximately 40% of total hemoglobin. This is a massive jump compared to the less than 1% found in control cells [Figure 3C, 3D, 3F].

Crucially, the authors demonstrated that this wasn't just a side effect of cell death. In primary human CD34+ hematopoietic stem cells, the expression of a 31-amino acid peptide mimicking the CR2 helix induced HbF levels as high as 75% [Figure 5D, 5F]. This induction occurred without perturbing erythroid differentiation. This means the cells still developed into healthy red blood cells [Figure 5G].

Limitations and the road to the clinic

While the results are robust, the study is a proof-of-concept. Several engineering challenges remain unsolved.

First, the primary tool used to induce HbF was a peptide. Delivering peptides into cells is notoriously difficult. They are often degraded by proteasomes (cellular recycling centers). They also struggle to cross the cell membrane. The authors suggest that macrocyclic peptides (circularized structures that are more stable and permeable) could solve this. However, the transition from a linear peptide to a viable drug is a significant leap.

Second, the NuRD complex is not exclusive to the hemoglobin gene. It is a general regulator of the genome. The authors note that there is a potential for unintended off-target toxicities. This could happen if a drug disrupts NuRD function globally across other genes. Although the study showed that the peptide only significantly affected 89 genes in their RNA-seq analysis [Figure 6A], the long-term safety of systemic NuRD inhibition remains unknown.

Finally, the study relies on the assumption that the GATAD2A-CHD4 interface is the most efficient target. While they successfully dissociated CHD4 from the core complex without affecting the binding of other components like MBD2, the complexity of the NuRD interactome remains. Subtle off-target effects could emerge in more complex physiological environments.

The verdict

The findings represent a high-confidence "yes" for the feasibility of targeting the NuRD complex to treat hemoglobinopathies. By identifying a specific, high-affinity protein-protein interaction interface, the authors have moved toward surgical, structure-guided intervention. The jump from $<1\%$ to $\sim75\%$ HbF in primary cells is a massive functional gain. This justifies the pursuit of macrocyclic peptide or small-molecule inhibitors. The next milestone will be determining whether these "molecular wedges" can be engineered to be both cell-permeable and sufficiently selective.

Figures from the paper