3D Regulatory Hubs as Structural Drivers of Oncogenic Programs in Glioblastoma

Glioblastoma (GBM) remains one of the most aggressive and treatment-resistant forms of brain cancer, with a five-year survival rate stubbornly below 10%. Despite decades of genomic research and an explosion of targeted therapy development, meaningful improvements in clinical outcomes have remained elusive. Part of the reason lies in GBM’s bewildering complexity—not just its mutational diversity, but the profound variability in how individual tumor cells behave, adapt, and survive under therapeutic pressure. Standard classification based on molecular subtypes, such as mesenchymal, classical, and proneural, has offered some clarity, but it has failed to translate into reliably effective, subtype-specific interventions. A growing body of evidence suggests that this is not merely a problem of which genes are mutated or expressed, but how the regulatory logic of those genes is organized in three-dimensional (3D) nuclear space. Traditional genomic studies view gene regulation as a linear process: enhancers activate nearby promoters, genes are transcribed, and cell identity is maintained or altered. However, this one-dimensional framework overlooks the physical reality of the genome folded inside the nucleus. Increasingly, it has become clear that enhancer-promoter communication is spatially orchestrated, not randomly or passively, but through complex networks of chromatin looping that bring regulatory elements into proximity, sometimes over hundreds of kilobases. This spatial architecture is not just a passive scaffold—it actively contributes to which genes are turned on, how strongly, and in what combinations.

To this account, new research paper published in Molecular Cell Journal and led by Professor Howard Fine and Associate Professor Effie Apostolou from the Weill Cornell Medicine and conducted by Sarah Breves, Dafne Campigli Di Giammartino, James Nicholson, Stefano Cirigliano, Syed Raza Mahmood, Uk Jin Lee, Alexander Martinez-Fundichely, Johannes Jungverdorben, Richa Singhania, Sandy Rajkumar, Raphael Kirou, Lorenz Studer, Ekta Khurana, and Alexander Polyzos investigated whether glioblastoma’s malignancy could be better understood through its 3D genomic organization. Specifically, they wanted to identify whether densely interconnected enhancer-promoter networks—termed “hyperconnected 3D regulatory hubs”—play a foundational role in driving the tumor’s transcriptional programs. They hypothesized that such hubs might serve as central organizing units that coordinate the expression of key oncogenes and possibly less-studied genes drawn into oncogenic networks through spatial proximity. Indeed, what makes this line of inquiry so compelling is its potential to explain phenomena that linear genomics cannot. For example, why do certain genes without recurrent mutations or obvious epigenetic marks still exhibit high expression in GBM? How can different tumors, each with distinct mutation profiles, converge on similar malignant phenotypes? And why do targeted therapies often yield only transient benefits? The answer may lie in the existence of 3D hubs that integrate and amplify oncogenic signals across gene networks, creating a system that is not only robust but remarkably adaptable.

To explore how spatial genome organization contributes to glioblastoma’s malignant behavior, the Weill Cornell Medicine researchers began by mapping enhancer-promoter interactions in four patient-derived glioblastoma stem cell (GSC) lines using H3K27ac HiChIP. This method allowed them to trace thousands of loops between active enhancers and promoters, revealing a dense web of regulatory interactions. These maps weren’t uniform; they varied across patients, aligning with known molecular subtypes like mesenchymal and classical, yet also exposing previously hidden layers of regulation. Importantly, the team identified specific genomic regions with exceptionally high connectivity—so-called “3D hubs”—that clustered numerous enhancers and promoters into regulatory centers. What made these hubs intriguing was that genes residing within them were not just active; they were highly co-regulated and disproportionately linked to oncogenic pathways. Using single-cell RNA sequencing, the researchers confirmed that genes within the same hub exhibited remarkably synchronized expression patterns across individual tumor cells—suggesting these hubs functioned as coordinated transcriptional units rather than coincidental groupings. To determine whether these structures had causal influence, not just correlation, the team employed CRISPR interference (CRISPRi) to silence specific hubs. In one experiment, they targeted a hub that included the proto-oncogene JUN, observing that not only did JUN expression fall, but so did several other genes connected to the same hub. These weren’t random transcriptional ripples; the entire local network dimmed, underscoring the hub’s role in maintaining coordinated gene activity.

The authors afterward turned their attention to a previously uncharacterized hub near GOLIM4—a region not traditionally associated with GBM. When they silenced this hub using a stable, inducible CRISPRi system, they found that all six connected genes were downregulated, and the GSCs shifted their transcriptional identity. This wasn’t just a subtle reprogramming; the cells adopted different states when grown in 3D brain organoid models, showing changes in clonogenic capacity and expression of genes tied to aggressive tumor behavior. What’s more, targeting single genes within the hub did not reproduce the full effect, suggesting that the hub’s function stemmed from its interconnected nature rather than any single node. Looking beyond glioblastoma, the researchers extended their analysis to 88 samples across 16 cancer types. They discovered that hyperconnected 3D hubs are not a GBM-specific anomaly but a recurring theme in tumor biology. Some were cancer-type-specific, while others spanned multiple malignancies, converging on universal oncogenic pathways like MYC and p53 signaling.

In conclusion, the study by Professor Howard Fine and Associate Professor Effie Apostolou reframes how we understand gene regulation in glioblastoma by demonstrating that the tumor’s aggressive behavior is orchestrated not just by linear genetic changes, but by complex three-dimensional structures in the nucleus. These hyperconnected 3D hubs act as regulatory epicenters, coordinating the expression of gene networks that define malignant identity. What makes this discovery especially significant is its ability to bridge a long-standing gap in cancer biology—explaining why tumors with vastly different mutations often exhibit remarkably similar behaviors.

We believe the implications are wide-reaching. First, it suggests that spatial genome architecture is a critical component of the oncogenesis regulatory logic. This perspective shifts the focus away from isolated genes and toward higher-order structures that govern entire transcriptional programs. Therapeutically, it opens the door to targeting regulatory hubs instead of individual genes—an approach that could simultaneously disrupt multiple oncogenic drivers and their supporting networks. Another key takeaway is the discovery that many of these hubs do not rely on structural mutations to form. This means that the epigenetic machinery, including transcription factors and chromatin organizers, plays a larger-than-expected role in shaping malignant phenotypes. Because such factors are more dynamic and potentially reversible, they offer more accessible intervention points than fixed genetic lesions. In fact, the study identified transcription factors enriched in hub-associated regions whose elevated expression correlates with poor prognosis across multiple cancers.  Moreover, the new finding that hub silencing can lead to coordinated transcriptional collapse—not just a downregulation of one gene, but a destabilization of an entire network. This has profound implications for designing therapies that aim to overcome the redundancy and adaptability that make glioblastoma so resistant to treatment. The idea that disrupting a single hub can alter cellular states and reduce clonogenic potential redefines how we might approach tumor targeting in the future.  Lastly, the recurrence of similar 3D hubs across diverse cancer types hints at a unifying principle behind oncogenic regulation—one that transcends tissue of origin. If validated further, this could become a cornerstone in developing universal, architecture-based cancer therapies. Rather than chasing every mutation, we might instead dismantle the regulatory frameworks that allow cancer cells to act in concert, survive stress, and evade control. This study lays the foundation for that possibility, with a clarity and precision that could reshape both diagnostics and therapeutic strategies in cancer biology.