Engineered ribosomes could make new polymers

The molecular machines that cells use to build proteins are backed by a billion years of evolution. In that time, these machines—ribosomes—have become exceptionally good at forging amide bonds between standard α-amino acids to make peptides and proteins.

Ribosomes are so good at their job that researchers want to harness them to make other polymers. If they could coax these protein-translation machines to use building blocks other than the amino acids found in nature, they could discover polymers—whether proteins or novel materials—with entirely new properties and functions. For years, researchers were satisfied with engineering them to add nonstandard α-amino acids to proteins.

But in the past 15 or so years, researchers have begun trying to use ribosomes to add other types of monomers, such as amino acids with longer backbones. Using the ribosome to synthesize polymers other than proteins will make it easier for researchers to explore wider areas of chemical space and evolve materials with new properties, such as resistance to enzymes or new ways of folding.

“My goal is to generate new molecules and learn how to exploit the ribosome to program novel bond-forming reactions. It’s a fascinating chemical problem,” says Alanna Schepartz, a chemical biologist at the University of California, Berkeley, who leads C-GEM, the Center for Genetically Encoded Materials. C-GEM is a National Science Foundation–funded Center for Chemical Innovation that brings together a multi-institutional team of chemists, biochemists, structural biologists, synthetic biologists, and organic and materials chemists to repurpose the translation machinery so that it can be used for the programmed synthesis of new medicines and materials. “There is already evidence that extant ribosomes can synthesize novel molecules that are not oligomers of α-amino acids, including ones with interesting properties,” Schepartz says.

Before researchers can harness the ribosome to make such new materials, they need to determine how to co-opt nature’s finely tuned machine without breaking it. That is not an easy task.

Nature’s protein factories

Bacterial ribosomes are hulking complexes made of three RNA molecules and more than 50 proteins. Those components are divided between a small and a large subunit, which have separate roles in protein synthesis. In simple terms, the small subunit binds to and decodes the messenger RNA (mRNA), which contains the protein-making instructions. The large subunit catalyzes the formation of amide bonds in a growing peptide or protein chain.

A small collection of RNA molecules forms the core of the ribosome and is responsible for catalyzing peptide bond formation. Those ribosomal RNA (rRNA) molecules have size-based names that are derived from their rates of sedimentation in a centrifuge. Bacterial ribosomes, the subject of most engineering efforts, are smaller than ribosomes in eukaryotes. Their large subunit incorporates two rRNAs—23S and 5S—and the small subunit has just one, 16S.

In normal protein production, each amino acid has its own transfer RNA (tRNA) that brings it to the ribosome. But first, that amino acid must be attached to—or, as biologists refer to the process, “charged” onto—that tRNA, a union that requires a corresponding enzyme, called an aminoacyl-tRNA synthetase.

When scientists first tried to expand the kinds of building blocks that could be incorporated into proteins, they focused on making new pairs of tRNA and aminoacyl-tRNA synthetase. Peter Schultz and coworkers at Scripps Research in California reported in 2001 that they could borrow a pair from an archaebacterium and use it to incorporate a nonstandard amino acid in Escherichia coli. The key was reprogramming a few letters of the bacterium’s genetic code to recognize the archaeal tRNA.

Since then, researchers have used the method to introduce hundreds of nonstandard α-amino acids to proteins. Those monomers have been low-hanging fruit because ribosomes evolved to work with them. But scientists, driven by a desire to create polymers with different properties, are now identifying how to incorporate amino acids with longer—or completely different—backbones. C-GEM researchers led by Schepartz and Scott Miller of Yale University have succeeded in getting ribosomes to even accept polyketide-like 1,3-dicarbonyl monomers (ACS Cent. Sci. 2019, DOI: 10.1021/acscentsci.9b00460).

Hiroaki Suga of the University of Tokyo developed a way of charging a variety of nonstandard amino acids, including β- and cyclic amino acids, onto tRNA. Instead of using an aminoacyl-tRNA synthetase to charge amino acids onto the tRNA, the in vitro system, which Suga has dubbed “flexizyme,” uses an RNA catalyst that can work with many different amino acids and tRNAs to do the charging.

But just because a tRNA brings an amino acid to the ribosome doesn’t mean that the ribosome will accept it, especially in vivo. The ribosome will tolerate some nonstandard monomers but not others. And the more of a nonstandard monomer that one tries to incorporate in a single peptide or protein, the harder it becomes. Getting the ribosome to efficiently accept a wider variety of substrates requires tinkering with the ribosome itself in addition to designing novel enzymes to biosynthesize the required tRNAs.

Making mutants

In the late 1970s, Sidney Hecht wanted to determine the variety of chemistries that ribosomes are capable of—not just what they do normally but also the other reactions they might be able to catalyze. To do that, Hecht and his coworkers fed the ribosomes building blocks other than amino acids. The results were disappointing.

“The ribosome is the classic one-trick pony,” says Hecht, who is a biochemist at Arizona State University. “The ribosome knows how to make a peptide bond. It does it very well. It does a few other similar things, like making lactones instead of lactams. And it didn’t do anything else. No matter what you would do experimentally to try and urge it on, it didn’t do anything well enough to be interesting.”

It was becoming obvious to Hecht and others in the field that engineering other parts of the protein-synthesis apparatus wasn’t going to let them include the variety of nonstandard monomers they wanted. They were going to have to tackle the ribosome itself. But that’s easier said than done.

The most obvious place to start is the part of the ribosome that catalyzes the formation of peptide bonds. This site—found in the large subunit’s 23S rRNA and called the peptidyl transferase center (PTC)—is well conserved across species.

Schepartz notes that the challenge is not just generating a PTC that will promote bond formation between new monomer classes but doing so without destroying the myriad other things a ribosome has to do during translation, such as recognizing tRNA and moving the growing amino acid chain out of the catalytic site.

The RNA in the ribosome is so large—more than 3,000 nucleotides—that the best hope for finding functional ribosomes that work with other monomers is to create libraries of ribosomes with engineered active sites.

For example, to find ribosomes that can incorporate β-amino acids, Hecht screens libraries of ribosomes for ones whose activities are slowed down or stopped by a specific form of the natural product puromycin. The antibiotic looks like the 3′ end of an amino acid–charged tRNA, a feature that allows it to take the place of a charged tRNA and stop the translation of any proteins containing β-amino acids (Biochemistry 2011, DOI: 10.1021/bi2016124).

Active mutant ribosomes typically require multiple mutations in the rRNA. That’s because a single mutation will often kill a ribosome, whereas several mutations can compensate for one another and enable new functions.

Hecht generally changes multiple locations sequentially—but, he cautions, not too many—making all possible versions of the RNA sequence at each location. “The reason that we make large numbers of changes is we don’t want to get locked into an inactive ribosome by making a single change when 6 or 11 changes might have brought it back to a different activity,” he says.

“In order to generate libraries of ribosomes, one has to be pretty smart about where one introduces mutations, especially if the ribosome is to remain functional in a cell,” Schepartz says. Her team identified a ribosome called P7A7 that with 12 mutations was able to incorporate a β-amino acid during protein synthesis in E. coli (J. Am. Chem. Soc. 2016, DOI: 10.1021/jacs.6b01023).

But the researchers were surprised when they tried to solve the structure of P7A7. C-GEM member Jamie H. D. Cate, a structural biologist at UC Berkeley, and coworkers found that they couldn’t get a cryo-electron microscopy image of an intact P7A7 ribosome (Biochemistry 2019, DOI: 10.1021/acs.biochem.9b00746). In the structures that they could get of the large subunit, the catalytic PTC was completely disordered.

Cate says the ribosome is nearly assembled—all the right parts are there. “But it just so happens that the peptidyl transferase center is the last thing to fold. You can get most of the way there and not realize you’re not all the way there,” Cate says. The quality of the electron microscope image was such that the researchers could see how some RNA bases paired and stacked with one another, “and then right next door is completely wiped out,” he says.

Schepartz and Cate think that inside cells, P7A7 benefits from crowded conditions and molecular chaperones that hold the ribosome together enough to function. Without those partners, it can’t assemble properly.

Like Hecht, Schepartz and coworkers have found that making one mutation often requires making another mutation for a ribosome to remain active. Schepartz’s graduate student Allison S. Walker used a method called statistical coupling analysis to identify patterns of residues extending from the PTC that coevolve and so need to be modified in tandem (Proc. Natl. Acad. Sci. U.S.A. 2020, DOI: 10.1073/pnas.1909634117).

“Her data provides a map for how to intelligently introduce mutations in the ribosome so that the PTC doesn’t explode,” Schepartz says.

The discovery of P7A7 was a good first step toward generating a fully functional engineered ribosome. Schepartz continues to tinker with the catalytic site and surrounding RNA to try to make it stabler. To that end, she has teamed up with Rhiju Das, a computational biologist at Stanford University and member of C-GEM. He and his coworkers developed a web-based game called Eterna to design RNA molecules and predict how they will fold. In the game, players solve puzzles related to RNA sequence and shape. Their collaborators can then synthesize the sequences the players make and see whether they match the predicted structures and, more importantly, whether they work.

“The ribosome, at its core, is an RNA machine. Its enzymatic active site is basically all RNA,” Das says. “A lot of the problems in trying to redesign the ribosome to do other things come down to RNA design problems.”

But redesigning this huge machine is not trivial. “Tools for redesigning RNA, much less RNA enzymes, are way behind tools for proteins, particularly in the computational sense,” Das says.

Das is also collaborating with Michael C. Jewett of Northwestern University, who is not part of C-GEM, to design “superfolder” ribosomes with improved stability. “The E. coli ribosome is fragile,” Das says. “You make changes to it, and then it falls apart, loses function. We’re trying to make a superfolder ribosome that is more stable structurally and more stable to changes in temperature or magnesium or sequence,” all conditions the group is exposing the ribosome to in order to make it do other things.

And C-GEM researchers are pursuing perhaps the most ambitious ribosome-related puzzle: designing a mini ribosome. “What if you could create a tiny version of a ribosome that’s just an RNA enzyme that polymerizes proteins?” Das asks. “If you had such mini constructs, it would be much easier to create selections for ribosomes that make β-proteins or polyaramids or other polyesters or other compounds. We’ve already run a pilot round of this with some really exciting results.”

Parallel tracks

Any cell that uses engineered ribosomes still needs to make its normal proteins. To relieve the engineered ribosomes of the burden of making new molecules and the normal cellular apparatus, researchers are engineering ribosomes that can work in parallel with normal ribosomes. In such systems, the engineered ribosomes make new polymers, and the normal ribosomes make everything else.

“If you want to engineer the ribosome to incorporate exotic monomers more efficiently that the ribosome hasn’t evolved to incorporate over the last billion-plus years, you have to deal with the fundamental constraint that the ribosome is necessary for cell viability,” says Jewett, who is working on such engineered ribosomes, also called orthogonal ribosomes.

In fact, most ribosomal mutations end up killing cells. So Jason W. Chin, a synthetic biologist at the MRC Laboratory of Molecular Biology and the University of Cambridge, thought it might be a good idea to have both wild-type ribosomes and engineered ribosomes in the same cell.

Accomplishing that involved tailoring both the mRNA and the ribosome. Chin’s team looked for a way to ensure the engineered ribosome recognized only specific mRNAs. They ended up changing the part of the mRNA called the Shine-Dalgarno sequence, which binds to the ribosome, as well as tweaking the part of the ribosome involved in that binding (Nat. Chem. Biol. 2005, DOI: 10.1038/nchembio719). “We set up directed evolution experiments that allowed us to find a ribosome that was directed to a different message,” Chin says. That allows the researchers to use engineered ribosomes for specific mRNA molecules while letting the normal ribosomes deal with the rest of the proteins.

The researchers focused on engineering the 16S rRNA, the part of the small subunit that reads the genetic message and pairs up the mRNA and tRNA. For the large subunit, they just left both rRNAs as is.

That work allowed them to evolve better versions of the small subunit, but they wanted to be able to evolve the small and large subunits as a pair. That task required finding a way to link the two subunits.

In 2015, independent teams, one led by Jewett and Alexander Mankin of the University of Illinois at Chicago and the other led by Chin, reported that they could connect the two subunits by synthesizing the 23S and 16S rRNAs—part of the large and small subunits, respectively—as one long RNA molecule with a linker between them. Jewett and Mankin published their version in Nature (2015, DOI: 10.1038/nature14862); a month later, Chin’s version appeared in Angewandte Chemie International Edition (2015, DOI: 10.1002/anie.201506311). In both systems, another large subunit component, called the 5S rRNA, remains a separate molecule.

Both teams initially thought that covalently linking the two subunits would be enough to keep the engineered ribosome separate from the natural ribosomes. But Chin found that some versions of the connected subunits could cross-assemble with ones from wild-type ribosomes to form hybrids.

It turns out that the length and structure of the RNA linker matter, Chin says. “We showed by a bunch of biochemical measurements that both our designs and other people’s original design didn’t solve the problem,” Chin says. “If you optimized the orientation and structure of the linker, you could get something that appeared to solve the problem.” Chin’s team used those optimized linkers to evolve ribosomes that catalyze new reactions (Nature 2018, DOI: 10.1038/s41586-018-0773-z).

Ahmed H. Badran of the Broad Institute of MIT and Harvard is taking a different tack to engineering orthogonal ribosomes. His lab assembles ribosomes from rRNA, ribosomal proteins, or both from different microbial species; the researchers then integrate the ribosomes into standard E. coli (Nat. Commun. 2021, DOI: 10.1038/s41467-020-20759-z). He developed an assay to determine whether heterologous small subunits with engineered 16S RNAs pair up with matching large subunits or E. coli subunits when in a mixture. They found that even when the heterologous subunits aren’t linked they pair up as well as covalently linked subunits do. The finding suggests that the heterologous subunits may be well suited to working in parallel with normal E. coli ribosomes. “I would not be surprised if a ribosomal RNA from one organism plus ribosomal protein from a different organism yielded improved ribosomes or engineering starting points for new bioactivities,” Badran says.

Although working in cells is attractive as a final goal, there may be advantages to working outside cells and constructing ribosomes in test tubes instead. “You can bypass the evolutionary pressures that otherwise the ribosome would be under in a living organism,” Jewett says.

To make cell-free systems with engineered ribosomes, Jewett removes natural ribosomes from lysed cells and then adds back ribosomal proteins and DNA that encodes the engineered rRNA. Once the rRNA is transcribed, the normal ribosome-assembly processes take over.

“The benefit of a cell-free system is you don’t have to worry about transport across the cell wall,” Jewett says. “As we start to try to get to these really funky monomers, you can imagine some things not being transported efficiently through the cell wall.”

Jewett and coworkers have also shown that they can still evolve ribosomes, even in in vitro systems. They make the new ribosomes using cell-free synthesis and then select ones with desired properties for subsequent rounds of directed evolution (Nat. Commun. 2020, DOI: 10.1038/s41467-020-14705-2).

“We’re seeing we can coax the ribosome into using a variety of unusual monomers, such as backbone-extended monomers, and if we start using engineered variants of ribosomes, they could use those monomers even better,” Jewett says. “It really tees up this exciting opportunity to truly expand the palette of ribosomal monomers that can be used in genetically encoded polymerization.”

Jewett and coworkers have been able to get ribosomes to accept linear amino acids with as many as five extra CH₂ groups between the amine and carboxylic acid. The ability to incorporate amino acids with such extended backbones “could enable new functionalities,” Jewett says. “As you move to β or γ structures, you actually change in many respects the structural properties of the amino acid.”

And cyclic amino acids could offer even more opportunities. “Cyclic amino acids have a rigid structure, provide different types of helix geometries and peptide turn characteristics, and create different folds that haven’t existed in nature before,” Jewett says. “That could lead to all kinds of new handles for evolution to create enzymes, therapeutics, or polymers.”

Putting the ribosomes to use

Engineered ribosomes could have many applications for synthetic polymers and therapeutics. One promising area is in discovering and synthesizing sequence-defined polymers, ones where the monomers follow a precise order, says Christopher Alabi, a polymer scientist at Cornell University and another member of C-GEM. Alabi uses such polymers for biomedical and drug delivery applications but notes that they are tough to make using conventional synthetic methods.

“Based on the chemistry and the way we add the monomers in sequence, it’s really difficult to add chirality into the backbone,” Alabi says. “Making the monomers is not the issue. It’s incorporating the monomer in an asymmetric manner into the backbone every single time that is difficult.”

An even bigger advantage of using ribosomes to make synthetic polymers, Alabi says, would be the ability to harness biological evolutionary processes of mutation and selection for discovering a wider variety of polymers with desired properties. “We can’t do polymer evolution by hand. It’s just impossible,” he says. Even with as few as two or four types of monomers, the sequence space is more expansive than chemists could search manually. “You can’t do it one at a time. I mean, you could, but you’d need a lot of graduate students. It would be painful for anyone involved,” Alabi says. But using engineered ribosomes with directed evolution and selection methods should make it possible to evolve polymers for specific functions.

Polymer scientists might not even be looking in the right part of the vast chemical space for the properties they want, Alabi says. “Biology has come up with a smart way of answering that question by just doing the evolution. It’s not doing screening. It’s just evolving. And that’s a faster way of going through that space than trying to screen through that space,” he says.

Polymer chemists can imagine all sorts of bonds they’d like to evolve ribosomes to catalyze, but they’d settle for only amide formation, Alabi says. After all, polymers such as nylon and Kevlar are polyamides. “A difference between those materials and proteins is that they have different backbones and can fold into structures with vastly different properties than proteins,” Alabi says.

Samuel H. Gellman, a chemist at the University of Wisconsin–Madison, studies foldamers: molecules that contain β-amino acids. “We make all of our molecules synthetically. We have to,” he says. “But I would love to be able to produce longer versions biosynthetically. I would love to be able to create libraries.”

Methods such as the one Suga uses have made synthesizing the molecules easier, Gellman says, but it’s still difficult to make sufficient quantities to run experiments affordably.

There’s still much to do before these and other applications will be possible at scale. A big challenge will be making the process cheap enough to compete with existing technology

But Chin is optimistic. “I think you’ll see from our lab, probably in much less than 10 years, examples of polymer biosynthesis in cells,” he says.