Cysteine-mapping efforts offer new tools for drug development and discovery

Abstract

The highly reactive amino acid cysteine has long been an attractive target for drug development. Much of that utility derives from cysteine’s critical biological roles and modifiability in different environments, from its sulfur-containing side chain’s ability to form strong disulfide bonds, and from its status as the most nucleophilic amino acid, which means that it readily donates electrons to binding partners. In addition, some cysteines have low pK_a values, which indicate stronger and more reactive acids. “That reactivity is also associated with function and sensitivities to different environments, and so I think there’s somewhat of a marriage between functional importance and then reactivity,” says Balyn Zaro, PhD, an associate professor of pharmaceutical chemistry at the University of California, San Francisco. That “perfect storm” combination makes cysteine an ideal target for drug developers, she says.

The dynamic nature of different cysteines within the same protein or in the same protein position but within different cells, however, has made determining which sites are better drug targets and which ones are more reactive a challenging task. New tools address those knowledge gaps in part by mapping the characteristics of cysteines in proteins within multiple cancer cell types and then making that information freely available to other researchers.

In 2024, researchers at the Mass General Cancer Center in Boston, Massachusetts, expanded on previous efforts by publishing an ambitious new resource named DrugMap.¹ This comprehensive atlas lists potential cysteine-based drug targets across 25 cancer types based on a systematic mapping of 416 cancer cell lines.

Coauthor Liron Bar-Peled, PhD, a principal investigator at the Krantz Family Center for Cancer Research at Mass General Cancer Center and associate professor of medicine at Harvard Medical School, says that the goal in drug discovery usually comes down to getting a small molecule to bind to a disease-associated protein. For kinases, a class of enzymes that can modify molecules by adding phosphates to them, researchers have figured out how to create anticancer drugs that effectively bind within the proteins’ pockets. For the majority of cancer-linked proteins, however, “we have no good starting places,” he says. “And because we don’t have a starting point, we often refer to 80% of oncogenes as ‘undruggable’ or very challenging.”

The DrugMap effort has suggested that the true “undruggable” percentage may be far less. “Because of its unique chemical properties, cysteine almost acts as Velcro or a chemical handle for us to attach small molecules,” Dr Bar-Peled says. That handhold can provide a key starting point in the molecule-developing stage of the drug discovery process. Other researchers previously developed chemical proteomics, or a suite of technologies that probe protein–molecule interactions and help to identify druggable cysteines. DrugMap dramatically expands the scale of mapped cell lines to ask whether a protein that is druggable in one cancer type, such as melanoma, may also be druggable in another, such as lung cancer.

 Dr Zaro’s research focuses instead on cysteine reactivity as a key consideration for drug design. “Because the cysteine is so reactive, your drug doesn’t have to be as reactive, which is appealing if you’re trying to make drugs that don’t accidentally engage too many things,” she says. “So it really makes it, in a lot of ways, the lowest hanging fruit from a drug discovery perspective.” Likewise, molecules such as covalent drugs that strongly bind cysteines could be used in lower doses or at reduced frequencies.

Her interest in the amino acid’s reactivity extends to more fundamental questions. “I was realizing that while people had developed these revolutionary tools that could be used for drug discovery, there was this whole basic science question of, when is a cysteine reactive?” she says. To address that question and the related issue of what drives that reactivity, her research group built its own data set called Cysteine.app.² The intention is to create a predictive model of what drives cysteine reactivity, she says. “It’s not an immediate drug discovery story, but we think it provides this long-term knowledge base that can lead to more discovery in the future.”

For both the DrugMap and Cysteine.app efforts, the researchers began by gently lysing the cells to break them open while preserving their protein complex structures as accurately as possible. For the DrugMap project, the collaborators then added drug-shaped fragments that include the binding scaffolds commonly found in drugs. “We ask, simply by proteomics, where do they bind?” Dr Bar-Peled says. A fragment bound with high affinity to a particular site on a protein often suggests that the site will be actionable—a designation that can be confirmed through a subsequent screen of small molecules. The “screening funnel,” he says, can essentially identify the parts of proteins that could be targeted by small molecule drugs. “It’s almost like our Google map for drug discovery.”

Beyond identifying new targets as well as flagging where off-target binding might occur, Dr Bar-Peled says that the research has yielded some welcome surprises. “I think we made this really remarkable discovery that chemical proteomics and drug discovery allows us to look at protein conformation within cells,” he says. Understanding how proteins appear within a tumor cell—still a blind spot in many cases—can provide a high-resolution glimpse of what is happening in that cell and offer more important context for drug design. “When you make a drug, you’re really drugging the conformation of that protein, and now we’re able to do that,” he says. “That was a completely unexpected finding, and I think what’s exciting is that this technology can allow us to do that.”

Instead of drug scaffolds, the Cysteine.app effort added low and high concentrations of a cysteine-reactive probe to lysed cells to determine the relative reactivity of cysteines across more than 12,000 proteins in nearly 60 cancer cell lines. Through the profiling experiments, the analysis identified more than 36,000 reactive cysteines and roughly 10,000 hyperreactive sites in which even low concentrations of the probe immediately engaged the amino acid; intriguingly, some cysteines were hyperreactive in certain tissues and cancer cell types, but not in others. If researchers can apply other techniques to detect a third category of relatively unreactive cysteines that do not engage the probe at all, Dr Zaro says, they could use machine learning methods to generate models that predict when a given cysteine will be reactive.

Many of the hyperreactive  cysteines found in highly conserved proteins may be functionally important, Dr Zaro says. For drug development, however, there could be a downside if hyperreactive cysteines bind too tightly with covalent drugs and become the off-targets of a drug directed at a separate site. “Are these  cysteines so ripe for reactivity with covalent drugs that we should be making sure that our drugs don’t engage them?” she says. The laboratory is currently investigating that question through a follow-up project. Alternatively, hyperreactive cysteines could be more prone to mutagenesis or treatment-acquired mutations. The hypothetical scenarios, Dr Zaro says, underscore the importance of context.

Beyond cysteine, her laboratory is starting to focus on arginine, which she finds just as compelling because of its association with nucleic acid binding sites, overabundance in protein interfaces, and interactions with transcription factors. Cysteine may be the “frontier,” she says, but some of the lessons learned could be applied to other amino acids and the search for additional drug targets.

The Cysteine.app site has seen abundant traffic so far, while Dr Bar-Peled says that drug industry groups have been “very heavy” users of DrugMap. Being part of the initial drug discovery process of “providing new starting points that people might not have thought of before” can be very rewarding, he says. He also welcomes synergistic efforts such as the Cysteine.app that are using his group’s underlying code to ask related questions. As part of Dr Bar-Peled’s expanding scope, he is asking new questions as well, such as, “Can we identify new pockets within [drug] targets that we have not seen?” In addition, the discovery tool could help him to ask other questions about tumor cell biology, such as how different environments can alter a protein’s druggability. “I think that these dual missions are readily available from this platform,” he says.