“`html
Within every human cell, two meters of DNA is packed into a nucleus that measures merely one-hundredth of a millimeter in width.
To accommodate that minuscule area, the genome must collapse into a sophisticated arrangement known as chromatin, composed of DNA and proteins. The configuration of this chromatin subsequently influences which genes will be activated in a particular cell. Neurons, skin cells, and immune cells express distinct genes based on the accessibility of their genetic material for transcription.
Investigating these structures experimentally is a labor-intensive undertaking, complicating efforts to compare the three-dimensional genome configurations across various cell types. MIT Professor Bin Zhang is adopting a computational approach to tackle this issue, utilizing computer simulations and generative artificial intelligence to ascertain these configurations.
“Gene expression regulation hinges on the 3D genome architecture, so the aspiration is that if we can fully grasp these structures, we might understand the origins of cellular diversity,” explains Zhang, an associate professor in chemistry.
From the farm to the laboratory
Zhang first became captivated by chemistry when his older brother, who was four years his senior, acquired laboratory equipment and began conducting experiments at home.
“He would bring home test tubes and various reagents to perform experiments. Although I wasn’t quite aware of what he was doing back then, I was genuinely intrigued by all the vibrant colors and the smoke and scents that emerged from the reactions. That truly captured my interest,” says Zhang.
His brother eventually became the first individual from Zhang’s rural hometown to attend college. This event marked the first time Zhang contemplated the possibility of pursuing a path diverging from that of his parents, who worked as farmers in China’s Anhui province.
“As a child, I would have never envisioned a career in science or as a faculty member in the United States,” Zhang admits. “When my brother entered college, it significantly broadened my outlook, and I realized I didn’t have to replicate my parents’ journey and become a farmer. It sparked the thought that I could attend college and delve deeper into chemistry.”
Zhang enrolled at the University of Science and Technology in Hefei, China, where he majorly focused on chemical physics. He relished his academic pursuits and encountered computational chemistry and research, which grew to be his new passion.
“Computational chemistry merges chemistry with other disciplines I adore — mathematics and physics — and instills a sense of precision and logic into the otherwise more empirical principles,” he states. “I could utilize programming to resolve intriguing chemistry challenges and quickly test my ideas.”
After completing his undergraduate studies, he opted to further his education in the United States, which he regarded as “the apex of academia.” At Caltech, he collaborated with Thomas Miller, a chemistry professor employing computational methods to comprehend molecular processes such as protein folding.
For his doctoral research, Zhang investigated a transmembrane protein that functions as a conduit for other proteins to traverse the cell membrane. This protein, termed translocon, can also activate a side gate within the membrane, allowing proteins designated for membrane integration to exit directly into the membrane.
“It’s an astonishing protein, but its mechanism was not well understood,” Zhang remarks. “I constructed a computational model to decipher the molecular processes governing what features permit specific proteins to enter the membrane, while others are secreted.”
Shifting focus to the genome
Post-graduation, Zhang’s research attention transitioned from proteins to the genome. At Rice University, he undertook a postdoctoral fellowship with Peter Wolynes, a chemistry professor renowned for significant findings in the dynamics of protein folding. When Zhang joined the lab, Wolynes directed his focus toward the genome’s structure, and Zhang opted to follow suit.
In contrast to proteins, which typically exhibit highly organized regions amenable to analysis using X-ray crystallography or cryo-EM, DNA represents a globular molecule that does not lend itself to such evaluations.
A few years earlier, in 2009, researchers at the Broad Institute, the University of Massachusetts Medical School, MIT, and Harvard University developed a method to inspect the genome’s architecture by cross-linking DNA within a cell’s nucleus. By fragmenting the DNA into numerous small segments and sequencing it, researchers can then ascertain which segments are in proximity to one another.
Zhang and Wolynes utilized data generated by this technique, known as Hi-C, to investigate whether DNA gets tangled into knots when compressed in the nucleus, akin to how strands of Christmas lights may become entangled when packed into a storage box.
“If DNA behaved like a typical polymer, one would expect it to become twisted and form knots. However, that could be extremely detrimental for biology since the genome does not merely sit there idle. It must undergo cell division, and various molecular machinery must interact with the genome to transcribe it into RNA; knots would create many unnecessary obstacles,” Zhang explains.
They discovered, unlike Christmas lights, that DNA does not form knots even when densely packed within the cell nucleus, and they developed a computational model that enabled them to test hypotheses regarding how the genome manages to sidestep such entanglements.
Since joining the MIT faculty in 2016, Zhang has persisted in enhancing models of how the genome operates in three-dimensional space, employing molecular dynamic simulations. In one research domain, his lab investigates how variances in the genomic structures of neurons and other brain cells contribute to their distinct functions, while also exploring how improper folding of the genome might lead to ailments like Alzheimer’s.
Regarding establishing a connection between genome structure and function, Zhang is convinced that generative AI techniques will also play a crucial role. In a recent investigation, he and his students presented a novel computational model, ChromoGen, harnessing generative AI to anticipate the 3D configurations of genomic regions, based on their DNA sequences.
“I believe that in the future, we will have both elements: generative AI and theoretical chemistry-driven methods,” he states. “They beautifully complement one another and enable us to accurately construct 3D structures while comprehending how these configurations emerge from the fundamental physical forces.”
“`