Building bridges between Denmark-based and Broad researchers

When Maddie Murphy, a computational associate at the Broad Institute of MIT and Harvard, first learned she’d have the opportunity to pursue her research in Copenhagen, she envisioned a two- or three-week visit. Then her supervisor, Ray Jones, told her she could stay for two months—and truly immerse herself. “At first I was scared to go,” said Murphy. “Just on a human level, being away from everyone for that long.” 

Murphy spent September and October of 2024 in the lab of Robin Andersson, an associate professor at the University of Copenhagen. Her research grew out of a project helmed by Andersson, Jones, Jesse Engreitz, and their collaborators at the Novo Nordisk Foundation Center for Genomic Mechanisms of Disease (NNFC) at Broad: building computational models to predict which specific stretches of DNA called enhancers regulate which genes—and in which cell types. Untangling the links between enhancers and the genes they regulate in specific cell types is critical to learning which genes are at the root of metabolic disease.

“Many of the genomic regions strongly linked to cardiometabolic disease are enhancer regions,” said Jones, a scientific advisor in Melina Claussnitzer’s lab, which studies the genetic basis of metabolic disease. Claussnitzer is the director of the Broad Diabetes Initiative and the NNFC’s associate director of scientific strategy.

In Copenhagen, Murphy worked closely with Andersson and Wei-Lin Qiu, co-lead author on a paper in which Andersson’s lab introduced their computational models. Qiu helped Murphy gain insights about modeling gene regulation. Andersson suggested several datasets that Murphy could use to train her models. Murphy appreciated the chance to focus on a single project—and to be in charge of it. “I could spend almost all my time learning,” she said. “I loved it. Having more independence in my work made me think more about getting a Ph.D.” 

Experiences like Murphy’s have been integral to the mission of NNFC since it launched in 2021. The NNFC facilitates collaborations between the Broad and Danish researchers investigating the genetics and gene regulation of common complex disease, including type 2 diabetes and obesity. 

Thus far, more than 20 scientists from Denmark have visited the Broad, collaborating and learning new methods. Likewise, Broad scientists have gone to Copenhagen to attend symposia with their Danish NNFC colleagues and advance their metabolic research. More than 60 Broad scientists and 70 Denmark-based researchers convened last fall at the third annual NNFC symposium in Copenhagen. 

“We’ve created a center architecture and culture that enables synergies between Denmark-based and Broad-based strengths, so that together we can work at the forefront of cardiometabolic disease,” said Kasper Lage, the NNFC’s managing director. “This architecture enables researchers to exchange ideas and transfer technology across institutions.” 

Denmark’s the spot

One of the first Denmark-based researchers to visit the Broad was Andersson himself, something he calls “an incredibly positive and inspiring experience” that helped him strengthen existing collaborations and build new ones. “The Broad’s drive to tackle important research questions left a lasting impact on me.”  

One of those research questions was how to advance the computational modeling of gene regulation, a key NNFC research area. To train their initial models, Andersson, Jones, Engreitz, Qiu and their NNFC collaborators used data from single-cell analysis technologies and CRISPR genetic screenings, which enable scientists to target and edit DNA sequences and study the effects of those edits. 

Murphy’s project was to train models with a different type of data: expression quantitative trait loci (eQTL), which captures associations between gene variants and levels of gene expression. “The idea was to use eQTL data as a training set to identify regulatory relationships between enhancers and genes,” Murphy said. “We knew eQTLs are widely studied, including one of our own prior studies, and thus a more accessible option for training enhancer-gene models.” 

“Every person’s DNA has slight differences, and some of these variations affect how active a gene is,” added Andersson. “These connections, called eQTLs, have been mapped at scale using data from thousands of individuals, with major contributions from the Broad, through projects like GTEx.” (GTEx, or the Genotype-Tissue Expression Portal, is a public resource for studying tissue and cell-specific gene expression and regulation.)

The most salient aspect of the eQTL datasets was the range of cell types and tissues they covered. Studying multiple cell types is crucial for capturing biological complexity and cell-type-specific differences. A model trained on limited data might perform well in one context but struggle to make accurate predictions in another. “By training our models on a wide range of cell types, we aim to make them more generalizable and reliable when applied to new, unseen data,” said Andersson.

Enhancers are DNA regions that modulate transcription — the production of the messnger RNA molecules from genes— that serve as blueprints (“transcripts”) to make proteins. Genetic variants associated with disease can cause subtle perturbations in transcription that ultimately contribute to disease, but the gene regulatory code is enormously complex, necessitating computational models to interpret and predict single-cell datasets in metabolically relevant cell types at an unprecedented scale and depth. 

“By advancing computational models that map enhancers and their target genes in specific cell types, the NNFC aims to propel our understanding of mechanisms driving metabolic disease,” said Lage. “The ultimate goal is to translate genetic findings into mechanisms that can be targeted by new medicines.”

Across the sea

While Murphy was in Copenhagen building models for mapping enhancers, Svenja Hansson, a Ph.D. student at the Novo Nordisk Foundation Center for Basic Metabolic Research (CBMR) at the University of Copenhagen, was in Cambridge, MA, beginning a six-month visit to the Broad. Hansson’s project also concerned enhancers. She wanted to learn more about the enhancers that play a role in muscle-brain crosstalk after exercise—and subsequently link those enhancers to their target genes in skeletal muscle. This would contribute to a better understanding of how exercise benefits brain health.

In prior experiments, Hansson identified 656 exercise-responsive enhancers that contain a disease variant associated with brain function. Her initial goal was to experimentally perturb 100 of those enhancers. Her supervisor, Judhajeet Ray—a research scientist on the NNFC’s Human Gene Regulation Map (HGRM) team—urged her to think bigger. “He asked me, why not target all of them?” Hansson said.   

Hansson had never done such large-scale experiments before. “I’m a wet lab person,” she said. “I’d done low-throughput and bulk cellular experiments, but nothing at scale.” Though she’d brought with her from Denmark her established muscle cell models and planned to annotate putative enhancers in those cells using single-cell analysis technologies, she quickly saw the complexity of these experiments. She realized that she had to focus more on developing standard operating procedures in order to generate high-quality single-cell data.

She started over. With help from her new Broad colleagues, she harvested new cell samples and optimized nuclei isolation protocols to ensure they were intact enough to withstand the rigors of a single-cell, “multiome” workflow. Thiago Batista, a research scientist in the Claussnitzer lab, helped Hansson understand what was wrong with her prior set of data. Amy Guillaumet and Amrita Sule, Broad scientists in the Gene Regulation Observatory who work on dGTEx, advised her on optimizing the nuclei isolation from the new samples. 

During her time at Broad, Hansson successfully generated high-quality multiomic data that enabled her to annotate regulatory elements overlapping with the variant containing exercise-responsive enhancers. She returned to Denmark in late February. 

Her next step will be to finish designing the guide RNAs to perturb the 656 enhancers and link these to their target genes. The results will form the basis of her dissertation, which she’s hoping to submit by the end of 2025. “At the Broad, I learned you can gain so much by interacting with other scientists and sharing protocols,” she said. “The collaborative atmosphere was really a strength—it allowed me to take my project to the next level.” 

In addition to all the collaborative science Hansson experienced at the Broad, she enjoyed her free time in the U.S. She kayaked the Charles River, hiked Mount Monadnock and the White Mountains, and traveled to New York to walk in Central Park and see The Great Gatsby musical on Broadway. 

Likewise, Murphy savored her free time in Denmark. “I enjoyed biking everywhere, eating pastries all the time, and doing a lot of flea market shopping,” she recalled. “I cried on the plane ride home.” She hopes to return to Copenhagen to pursue her Ph.D.