Text | China Science News reporter Li Chenyang
At the end of last year, Hu Junjie, a researcher at the Institute of Biophysics of the Chinese Academy of Sciences, and his team submitted a paper to the journal Nature Communications. Unexpectedly, the magazine editor was "stumped".
"I had a hard time finding the right reviewer for this paper." The editor admitted, "Because this research is too interdisciplinary and too cross-cutting. "Fortunately, although only 2 reviewers were found, they all gave positive reviews, which allowed the article to be published.
The question this paper seeks to answer is: How does the endoplasmic reticulum in cells form tubular shapes? In the process of exploring this difficult problem, researchers can be described as using "eighteen martial arts": chemical cross-linking, biological imaging, artificial intelligence (AI) structure prediction, 3D printing... Relevant expertise involves biochemistry, cell biology, structural biology, AI and other disciplines. This makes it difficult for editors to find academics with in-depth knowledge of all of these fields.
Writing such a paper is not for showmanship.
As early as 2008, Hu Junjie, who was still a postdoctoral fellow, published a paper in Science, pointing out that only a specific class of membrane proteins is required for the formation of the tubular structure of the endoplasmic reticulum. But because of the protein's own special properties, neither he nor many of the more well-known fellow scholars have been able to successfully resolve its structure.
"This represents a class of proteins that are difficult to resolve." Hu Junjie said, "In this recent paper, we combined a variety of methods, especially the innovative use of AI structure prediction and 3D printing technology, which has exploded in recent years, to cleverly solve this problem." It can be seen that this "eighteen martial arts" is really not less.
Later, they used a similar strategy to resolve another protein that plays an important role in the maintenance of lamellar endoplasmic reticulum morphology. Related papers were published in the Journal of Cell Science.
Hu Junjie's team fights protein like Lego. Photo courtesy of interviewee
Discover the secrets of "50 nanometers"
In the famous science fiction novel "The Hitchhiker's Guide to the Galaxy", there is a very classic meme: "The ultimate answer to the universe is '42'." ”
And in a world as complex as the universe, there is also such a mysterious number: 50 nanometers.
"The organelle of 'endoplasmic reticulum' is familiar to everyone." Hu Junjie told China Science News, "But few people notice that each cell has only one endoplasmic reticulum, and all endoplasmic reticulum structures, whether they look thick or glossy tubular, are all 'connected', in fact a net - just like longevity noodles, a bowl of noodles that looks full is actually only one." ”
The thickness of the lamellar endoplasmic reticulum in higher animal cells is 50 nanometers, and the diameter of the tubular endoplasmic reticulum is also 50 nanometers. Interestingly, it is the different proteins that determine the thickness of the former and the diameter of the latter, but they have reached a tacit understanding on the number "50 nanometers", which allows the tubular and lamellar endoplasmic reticulum in the cell to be smoothly connected. In single-celled eukaryotes such as yeast, this "tacit understanding" figure is 30 nanometers.
Hu Junjie's two recent successive papers explore the two protein structures that determine the thickness of the lamellar endoplasmic reticulum and the diameter of the tubular endoplasmic reticulum. The main tubulin he chose was called Yop1p, and the support protein in the flaky part was bridge nectin Climp63.
Tubular protein Yop1p is Hu Junjie's "old acquaintance". His postdoctoral research topic was the formation mechanism of tubular endoplasmic reticulum. In a 2008 paper published in Science, he pointed out that a family of proteins, including Yop1p, is a sufficient condition for the endoplasmic reticulum to form tubular structures, and purified Yop1p can form tubular membrane structures after recombination in vitro.
Since then, he has hoped to unravel the structure of these proteins and see how the morphology of the endoplasmic reticulum is shaped.
But he didn't expect that the realization of his dream would be 15 years later.
Integrate various advanced "weapons"
The structure of the Yop1p protein is difficult to resolve for several reasons.
Traditional protein structure elucidation methods mainly include two categories: cryo-electron microscopy and X-ray crystallography. YOP1P is too small a molecular weight to be suitable for cryo-electron microscopy. After it is purified from the membrane, it is always wrapped in a layer of detergent, making it difficult to form crystals. The greater difficulty is that the molecules of the Yop1p protein have some weak interaction forces that cause them to randomly form oligomers of varying numbers.
These "gluttonous snakes" of varying lengths further increase the difficulty of restoring the structure of individual proteins.
The road is blocked, how to go?
A "dark horse" in the field of structural biology in recent years - AI structure prediction technology has helped a lot. They first simulated the general structure of the Yop1p protein monomer through AI deep learning, and found that the transmembrane region and other elements in the protein molecule are locked to each other in a relatively stable position, in other words, each Yop1p molecule is a well-shaped building block.
Then, they used chemical crosslinking to find out the locations of weak interactions between proteins, so as to obtain a "drawing" for assembling Yop1p polymers.
Next, they printed the protein with 3D printing technology, and something interesting happened: the two protein models were aligned with each other according to the binding sites identified earlier, and with a "click", they were tightly stitched together.
"It's like building Lego." Hu Junjie said with a smile.
When the four proteins are spliced together in this way, a structure with a specific arc is seen, which Hu likes to call "scaffolding". When multiple "scaffolds" shape the membrane structure of the endoplasmic reticulum, they allow the membrane to bend naturally to form a tube. What is even more surprising is that the curvature of the Yop1p tetramer exactly matches the 30 nm membrane curvature required for yeast endoplasmic reticulum tubes.
After such a series of "combination punches", Hu Junjie's team finally successfully uncovered the pattern of Yop1p protein polymerization and the core mechanism of the curved membrane.
Later, they used a similar strategy to predict the structure of bridge nectin Climp63. They found that a single Climp63 protein is less than 30 nanometers long, and two Climp63 proteins form a "tail-tail opposite" dimer with a length of exactly 50 nanometers!
After 15 years, they finally obtained a protein structure that could not be resolved before. This process not only benefits from the advancement of technology, but also reflects the research ideas of flexibly using various methods and playing a good "combination box".
Reviewers say one of the biggest highlights of the paper is that "the authors used clever tactics."
In recent years, there has been a saying that structural biologists may be "out of work" as AI becomes more capable of predicting protein structure. Hu Junjie's work proves that AI is not a competitor for scientists, but the "best assistant".
Cross-border crossovers are more "playful" with science
Although both papers obtained protein structures that are almost impossible to resolve by traditional methods, Hu Junjie does not believe that these works are purely structural biology studies. "The focus of our research is still on the morphology and function of the endoplasmic reticulum." He said.
For example, REEP1, the homologous protein of Yop1p in mammals, is closely related to a human genetic disorder, spastic paraplegia. Their work has also revealed the intrinsic link between the tubular endoplasmic reticulum and these diseases.
As a scholar in the field of life sciences, Hu Junjie has a rich professional background. He studied biochemistry at Fudan University. After graduation, he went to the United States to study the structural basis of insulin signaling at New York University School of Medicine. He received his Ph.D. in pharmacology in 2005 and then did postdoctoral work in the Department of Cell Biology/Howard Hughes Medical Institute at Harvard Medical School, studying the formation mechanism of tubular endoplasmic reticulum. In 2012, at the age of 33, Hu Junjie won the inaugural Howard Hughes Medical Institute International Young Scientist Award, the youngest inductee at the time.
Biochemistry, cell biology, pharmacology, structural biology, biophysics... The accumulation of expertise in these different fields has become a valuable asset in Hu Junjie's scientific research career. Later, AI structure prediction, 3D printing... He is actively exposed to various emerging technologies and applies them to scientific exploration.
"I often encourage students to learn and try various technologies and methods, and often many scientific problems that are almost impossible to solve under traditional thinking are the key to breaking the game." Hu Junjie said.
Related paper information:
https://doi.org/10.1242/jcs.260976
https://doi.org/10.1038/s41467-023-38294-y