Posted on by Lawrence Tabak, D.D.S., Ph.D.
Since joining NIH, I’ve held a number of different leadership positions. But there is one position that thankfully has remained constant for me: lab chief. I run my own research laboratory at NIH’s National Institute of Dental and Craniofacial Research (NIDCR).
My lab studies a biochemical process called O-glycosylation. It’s fundamental to life and fascinating to study. Our cells are often adorned with a variety of carbohydrate sugars. O-glycosylation refers to the biochemical process through which these sugar molecules, either found at the cell surface or secreted, get added to proteins. The presence or absence of these sugars on certain proteins plays fundamental roles in normal tissue development and first-line human immunity. It also is associated with various diseases, including cancer.
Our lab recently joined a team of NIH scientists led by my NIDCR colleague Kelly Ten Hagen to demonstrate how O-glycosylation can influence SARS-CoV-2, the coronavirus that causes COVID-19, and its ability to fuse to cells, which is a key step in infecting them. In fact, our data, published in the journal Proceedings of the National Academy of Sciences, indicate that some variants, seem to have mutated to exploit the process to their advantage .
The work builds on the virus’s reliance on the spike proteins that crown its outer surface to attach to human cells. Once there, the spike protein must be activated to fuse and launch an infection. That happens when enzymes produced by our own cells make a series of cuts, or cleavages, to the spike protein.
The first cut comes from an enzyme called furin. We and others had earlier evidence that O-glycosylation can affect the way furin makes those cuts. That got us thinking: Could O-glycosylation influence the interaction between furin and the spike protein? The furin cleavage area of the viral spike was indeed adorned with sugars, and their presence or absence might influence spike activation by furin.
We also noticed the Alpha and Delta variants carry a mutation that removes the amino acid proline in a specific spot. That was intriguing because we knew from earlier work that enzymes called GALNTs, which are responsible for adding bulky sugar molecules to proteins, prefer prolines near O-glycosylation sites.
It also suggested that loss of proline in the new variants could mean decreased O-glycosylation, which might then influence the degree of furin cleavage and SARS-CoV-2’s ability to enter cells. I should note that the recent Omicron variant was not examined in the current study.
After detailed studies in fruit fly and mammalian cells, we demonstrated in the original SARS-CoV-2 virus that O-glycosylation of the spike protein decreases furin cleavage. Further experiments then showed that the GALNT1 enzyme adds sugars to the spike protein and this addition limits the ability of furin to make the needed cuts and activate the spike protein.
Importantly, the spike protein change found in the Alpha and Delta variants lowers GALNT1 activity, making it easier for furin to start its activating cuts. It suggests that glycosylation of the viral spike by GALNT1 may limit infection with the original virus, and that the Alpha and Delta variant mutation at least partially overcomes this effect, to potentially make the virus more infectious.
Building on these studies, our teams looked for evidence of GALNT1 in the respiratory tracts of healthy human volunteers. We found that the enzyme is indeed abundantly expressed in those cells. Interestingly, those same cells also express the ACE2 receptor, which SARS-CoV-2 depends on to infect human cells.
It’s also worth noting here that the Omicron variant carries the very same spike mutation that we studied in Alpha and Delta. Omicron also has another nearby change that might further alter O-glycosylation and cleavage of the spike protein by furin. The Ten Hagen lab is looking into these leads to learn how this region in Omicron affects spike glycosylation and, ultimately, the ability of this devastating virus to infect human cells and spread.
 Furin cleavage of the SARS-CoV-2 spike is modulated by O-glycosylation. Zhang L, Mann M, Syed Z, Reynolds HM, Tian E, Samara NL, Zeldin DC, Tabak LA, Ten Hagen KG. PNAS. 2021 Nov 23;118(47).
COVID-19 Research (NIH)
Kelly Ten Hagen (National Institute of Dental and Craniofacial Research/NIH)
Lawrence Tabak (NIDCR)
NIH Support: National Institute of Dental and Craniofacial Research
Posted on by Dr. Francis Collins
To protect humans from COVID-19, the Pfizer and Moderna mRNA vaccines program human cells to translate the injected synthetic messenger RNA into the coronavirus spike protein, which then primes the immune system to arm itself against future appearances of that protein. It turns out that the immune system can also be trained to spot and attack distinctive proteins on cancer cells, killing them and leaving healthy cells potentially untouched.
While these precision cancer vaccines remain experimental, researchers continue to make basic discoveries that move the field forward. That includes a recent NIH-funded study in mice that helps to refine the selection of protein targets on tumors as a way to boost the immune response . To enable this boost, the researchers first had to discover a possible solution to a longstanding challenge in developing precision cancer vaccines: T cell exhaustion.
The term refers to the immune system’s complement of T cells and their capacity to learn to recognize foreign proteins, also known as neoantigens, and attack them on cancer cells to shrink tumors. But these responding T cells can exhaust themselves attacking tumors, limiting the immune response and making its benefits short-lived.
In this latest study, published in the journal Cell, Tyler Jacks and Megan Burger, Massachusetts Institute of Technology, Cambridge, help to explain this phenomenon of T cell exhaustion. The researchers found in mice with lung tumors that the immune system initially responds as it should. It produces lots of T cells that target many different cancer-specific proteins.
Yet there’s a problem: various subsets of T cells get in each other’s way. They compete until, eventually, one of those subsets becomes the dominant T cell type. Even when those dominant T cells grow exhausted, they still remain in the tumor and keep out other T cells, which might otherwise attack different neoantigens in the cancer.
Building on this basic discovery, the researchers came up with a strategy for developing cancer vaccines that can “awaken” T cells and reinvigorate the body’s natural cancer-fighting abilities. The strategy might seem counterintuitive. The researchers vaccinated mice with neoantigens that provide a weak but encouraging signal to the immune cells responsible for presenting the distinctive cancer protein target, or antigen, to T cells. It’s those T cells that tend to get suppressed in competition with other T cells.
When the researchers vaccinated the mice with one of those neoantigens, the otherwise suppressed T cells grew in numbers and better targeted the tumor. What’s more, the tumors shrank by more than 25 percent on average.
Research on this new strategy remains in its early stages. The researchers hope to learn if this approach to cancer vaccines might work even better when used in combination with immunotherapy drugs, which unleash the immune system against cancer in other ways.
It’s also possible that the recent and revolutionary success of mRNA vaccines for preventing COVID-19 actually could help. An important advantage of mRNA is that it’s easy for researchers to synthesize once they know the specific nucleic acid sequence of a protein target, and they can even combine different mRNA sequences to make a multiplex vaccine that primes the immune system to recognize multiple neoantigens. Now that we’ve seen how well mRNA vaccines work to prompt a desired immune response against COVID-19, this same technology can be used to speed the development and testing of future vaccines, including those designed precisely to fight cancer.
 Antigen dominance hierarchies shape TCF1+ progenitor CD8 T cell phenotypes in tumors. Burger ML, Cruz AM, Crossland GE, Gaglia G, Ritch CC, Blatt SE, Bhutkar A, Canner D, Kienka T, Tavana SZ, Barandiaran AL, Garmilla A, Schenkel JM, Hillman M, de Los Rios Kobara I, Li A, Jaeger AM, Hwang WL, Westcott PMK, Manos MP, Holovatska MM, Hodi FS, Regev A, Santagata S, Jacks T. Cell. 2021 Sep 16;184(19):4996-5014.e26.
Cancer Treatment Vaccines (National Cancer Institute/NIH)
The Jacks Lab (Massachusetts Institute of Technology, Cambridge)
NIH Support: National Cancer Institute; National Heart, Lung, and Blood Institute
Posted on by Dr. Francis Collins
Researchers recently showed that a computer could “learn” from many examples of protein folding to predict the 3D structure of proteins with great speed and precision. Now a recent study in the journal Science shows that a computer also can predict the 3D shapes of RNA molecules . This includes the mRNA that codes for proteins and the non-coding RNA that performs a range of cellular functions.
This work marks an important basic science advance. RNA therapeutics—from COVID-19 vaccines to cancer drugs—have already benefited millions of people and will help many more in the future. Now, the ability to predict RNA shapes quickly and accurately on a computer will help to accelerate understanding these critical molecules and expand their healthcare uses.
Like proteins, the shapes of single-stranded RNA molecules are important for their ability to function properly inside cells. Yet far less is known about these RNA structures and the rules that determine their precise shapes. The RNA elements (bases) can form internal hydrogen-bonded pairs, but the number of possible combinations of pairings is almost astronomical for any RNA molecule with more than a few dozen bases.
In hopes of moving the field forward, a team led by Stephan Eismann and Raphael Townshend in the lab of Ron Dror, Stanford University, Palo Alto, CA, looked to a machine learning approach known as deep learning. It is inspired by how our own brain’s neural networks process information, learning to focus on some details but not others.
In deep learning, computers look for patterns in data. As they begin to “see” complex relationships, some connections in the network are strengthened while others are weakened.
One of the things that makes deep learning so powerful is it doesn’t rely on any preconceived notions. It also can pick up on important features and patterns that humans can’t possibly detect. But, as successful as this approach has been in solving many different kinds of problems, it has primarily been applied to areas of biology, such as protein folding, in which lots of data were available for researchers to train the computers.
That’s not the case with RNA molecules. To work around this problem, Dror’s team designed a neural network they call ARES. (No, it’s not the Greek god of war. It’s short for Atomic Rotationally Equivariant Scorer.)
To start, the researchers trained ARES on just 18 small RNA molecules for which structures had been experimentally determined. They gave ARES these structural models specified only by their atomic structure and chemical elements.
The next test was to see if ARES could determine from this small training set the best structural model for RNA sequences it had never seen before. The researchers put it to the test with RNA molecules whose structures had been determined more recently.
ARES, however, doesn’t come up with the structures itself. Instead, the researchers give ARES a sequence and at least 1,500 possible 3D structures it might take, all generated using another computer program. Based on patterns in the training set, ARES scores each of the possible structures to find the one it predicts is closest to the actual structure. Remarkably, it does this without being provided any prior information about features important for determining RNA shapes, such as nucleotides, steric constraints, and hydrogen bonds.
It turns out that ARES consistently outperforms humans and all other previous methods to produce the best results. In fact, it outperformed at least nine other methods to come out on top in a community-wide RNA-puzzles contest. It also can make predictions about RNA molecules that are significantly larger and more complex than those upon which it was trained.
The success of ARES and this deep learning approach will help to elucidate RNA molecules with potentially important implications for health and disease. It’s another compelling example of how deep learning promises to solve many other problems in structural biology, chemistry, and the material sciences when—at the outset—very little is known.
 Geometric deep learning of RNA structure. Townshend RJL, Eismann S, Watkins AM, Rangan R, Karelina M, Das R, Dror RO. Science. 2021 Aug 27;373(6558):1047-1051.
Structural Biology (National Institute of General Medical Sciences/NIH)
The Structures of Life (National Institute of General Medical Sciences/NIH)
RNA Biology (NIH)
Dror Lab (Stanford University, Palo Alto, CA)
NIH Support: National Cancer Institute; National Institute of General Medical Sciences