Ribosomes are the cell's protein factories, which read the genetic code and assemble the proteins that every organism needs to live. But as far as how ribosomes themselves were formed, tantalizingly little was known.
Now, scientists have captured a key part of this process, in motion. The findings, published in Nature, combine artificial intelligence, cryo-electron microscopy, and genetics to reveal, in unprecedented detail, how cells coordinate, regulate, and safeguard the creation of the small ribosomal subunit—a machine central to forming every protein.
"We finally have a molecular movie of ribosome formation—we've reached the stage where we can see how it all connects," says Sebastian Klinge, head of the Laboratory of Protein and Nucleic Acid Chemistry.
"Rather than relying on a few snapshots of different stages in ribosome formation, this movie allows us to see, at each step, how things are changing and what is affecting those changes," adds Olga Buzovetsky, a research associate in Klinge's lab. "This work moves us closer to revealing an overall mechanism of how these fundamental protein factories come together."
To build a ribosome
For more than a decade, Klinge has been trying to answer one of the most fundamental questions in molecular biology: how cells build ribosomes. Ribosomes are so central to life that every cell devotes much of its energy to making them, yet the process of assembling their thousands of molecular parts has long been too complex and fleeting to capture in action.
In recent work, the Klinge lab helped explain how this process is organized within the nucleolus—the cell's ribosome factory—showing that the nucleolus' distinctive layered architecture emerges from the way that RNA is processed and folded during ribosome formation. "We were basically trying to figure out how the nucleolus works," Klinge explains. "And we were able to show that it's not voodoo or magic—there are distinct protein-protein and protein-RNA interactions that govern the form and functions of organelles."
But for all this progress, one piece was still missing—a way to move beyond static snapshots of ribosome formation and capture the continuous sequence of molecular events that transform an immature ribosomal particle into a finished subunit ready for action. Nowhere was this gap more glaring than in the formation of the small ribosomal subunit (SSU). Klinge and colleagues knew that the SSU's precursor must undergo an elaborate series of remodeling and quality-control steps before maturing into the ribosomal subunit responsible for decoding genes and ultimately producing proteins. But the mechanisms governing these transitions were elusive. Snapshots had revealed that some factor must be ensuring that the pathway for SSU formation proceeds forward, and never backward; prior work had suggested that two helicase enzymes, Mtr4, a component of the RNA-degrading machinery known as the exosome and Dhr1 an assembly factor, were coordinating the process. But how it all fit together remained a mystery.
"For a long time in the field, all we knew is that some proteins would leave and some proteins would join," Buzovetsky says. "But we had no idea why or how it happened."
A molecular movie
To bridge this gap, Klinge's team combined artificial intelligence, structural biology, and genetics in an approach that began with computation rather than experiment. Using AlphaFold—a powerful AI program that predicts the 3D shapes of proteins and their interactions—they modeled more than 3,500 possible interactions among the molecules that build the ribosome. These predictions served as a roadmap, helping the researchers identify which interactions to test in the lab and design experiments that would capture the process in action. Guided by these AI insights, the team then engineered yeast cells so that key assembly proteins could be tagged and tracked. Using cryo-electron microscopy, they collected more than 200,000 snapshots and reconstructed sixteen distinct 3D structures showing each step in the small ribosomal subunit's assembly.
"This is one of our first papers with a real AI foundation," Klinge says. "Instead of doing some genetics and biochemistry to solve structures, we essentially started with AI and used its structural predictions to design genetic and biochemical experiments. Thus, AI has helped to speed up initial discovery and has allowed for establishing testable hypotheses."
The result was a near-continuous "molecular movie" of how this essential machine takes shape. Across sixteen stages, the researchers watched as the Mtr4 enzyme acted like a molecular motor, breaking down a stretch of RNA to push the process forward. This irreversible step triggered a chain reaction of rearrangements and protein releases that kept assembly moving in one direction. The movie also revealed the role of Utp14, a flexible protein that serves as a central coordinator by positioning and activating the helicase Dhr1. Once switched into its active state, Dhr1 unwinds and removes an RNA chaperone, completing the final step.
The study also revealed an elaborate system of built-in safeguards. The RNA exosome remains tethered to the growing ribosomal subunit throughout assembly, closely monitoring its progress. As construction nears completion, these connections gradually release, and the exosome shifts to quality-control mode, inspecting each finished particle to ensure that only fully functional ribosomes continue to the next stage.
"We've come so far," Klinge reflects. "In 2013, we had nothing but a list of names of important factors in ribosome formation. That led to a chronology that explained the order that these factors appeared, and that led to low resolution structures. Then came high resolution structures, then rough ensembles of states—now it's a continuous movie."
The findings mark both a scientific and technological milestone. "We are no longer looking at snapshots of the beginning, middle, and end of ribosome formation," Buzovetsky says. "We are instead able to understand how RNA and proteins are mediating interactions and talking to each other, throughout the process of ribosome biogenesis."
The study also showcases the power of AI-driven structural biology and redefines how complex molecular systems can be studied. By weaving together artificial intelligence, high-resolution imaging, and genetics, Klinge's team has created a new model for discovery—one capable of tracking molecular motion in real time. Next, the lab aims to use their AI-driven model to extend this work to even earlier stages of ribosome assembly and to the quality-control systems that prevent mistakes in ribosome formation. "With the tools we now have in hand, we are able to achieve the kind of resolution that we need to better understand these processes," Buzovetsky says.
Finally, the work offers a glimpse into a fundamental moment in biology—when molecular components come together to create something that can sustain life. Every living organism, from bacteria to humans, depends on ribosomes to make the proteins that drive growth, repair, and survival. By revealing this process in unprecedented detail, the study brings scientists closer to understanding how the chemistry of life becomes its machinery. In doing so, the researchers are not only uncovering how one essential machine is built, but also charting a path toward visualizing life's inner workings as they happen—one molecular frame at a time.
"The formation of ribosomes from non-living matter is probably the closest thing to the origins of life that we know of," Klinge says. "Ribosomes are not quite alive but, when we study their biogenesis, we get a glimpse at the point at which something that isn't alive begins to feel alive."
