Yeast display finds tight-binding ACE2 macrocycles

Macrocyclic peptides have a persistent appeal: they can be small enough to feel “drug-like” but structured enough to bind surfaces that frustrate many small molecules. The catch is that finding good ones is often a tooling problem. You need a way to search a huge chemical space without synthesizing it all.

A new Journal of Medicinal Chemistry paper makes a straightforward claim with useful implications: yeast display can be a fast, quantitative engine for discovering potent macrocyclic peptide binders, not just a niche method for antibodies or protein fragments (PubMed).

Their demo target is human angiotensin-converting enzyme 2 (ACE2). ACE2 is famous in public-health conversations because it is the entry receptor for some respiratory viruses, but it is also a central enzyme in the renin-angiotensin system. In other words, it is a biologically important, structurally nontrivial target where “binding well” is not the whole story.

The bigger story is the search engine, not the target

Macrocyclic peptides often get described as a middle format between small molecules and antibodies. That framing is directionally right, but it can obscure the operational reality: macrocycles are often discovered through “library + selection” technologies.

Historically, many of the most productive approaches have been in vitro display systems (think mRNA display or related formats) that let researchers interrogate enormous libraries and then decode what stuck. Those systems are powerful, but they can be specialized.

Yeast display offers a different trade: it is a living system that can present peptide ligands and, crucially for this paper, support quantitative screening across millions of variants. If that works reliably, it broadens the number of labs that can run a serious macrocycle discovery campaign.

What the authors screened

The paper focuses on disulfide-cyclized macrocyclic peptides presented via yeast display. Conceptually, that matters because disulfide-cyclization is a practical way to impose ring constraints and shape, and yeast display can naturally connect phenotype (binding) to genotype (the DNA encoding each peptide).

They report screening millions of structurally diverse ligands, with the goal of finding molecules that inhibit human ACE2.

The concrete result: low‑nanomolar ACE2 inhibitors

From the abstract-level data, the best hits include two top macrocyclic peptides described as a “one-ring” and a “two-ring” macrocycle.

The headline numbers are hard to ignore:

the best one-ring macrocycle inhibited human ACE2 with a reported Ki of 1.9 nM
the best two-ring macrocycle inhibited human ACE2 with a reported Ki of 1.5 nM

Those are potencies that put the molecules in the same neighborhood as cyclic peptides found with more established in vitro selection workflows, which is exactly what you would want to see if the point is to validate yeast display as a discovery engine.

Structures that explain “why it binds”

Potency alone is a weak platform argument. The stronger version is: potency plus an explanation of how the molecule is achieving it.

The authors report crystal structures of both macrocycles bound to ACE2. The structures suggest two different “shape solutions”:

one macrocycle adopts a rigid β-hairpin
the other shows a cysteine-stabilized α-helix/α-helix motif

This is more than pretty structural biology. Macrocyclic peptides live and die by whether they can adopt stable, target-complementary conformations. Showing two distinct, well-defined folds helps make the case that the library and selection are not just finding sticky sequences, but discovering structured ligands.

The authors also note that both macrocycles use binding modes distinct from previously reported inhibitors. If that holds up in the full paper, it is a nice reminder that macrocycles can explore binding geometries that may be hard to reach with a classic small-molecule playbook.

What this does (and does not) say about ACE2 as a therapeutic target

It is tempting to read an “ACE2 inhibitor” headline and jump straight to disease implications. This is where it pays to separate the tool story from the clinical story.

ACE2 is not an arbitrary enzyme. It is implicated in blood pressure regulation and broader cardiometabolic physiology, and it has complicated biology. So “inhibiting ACE2” is not automatically a therapeutic win in the way that inhibiting a pathogen enzyme might be. Even in viral contexts, there is a difference between binding ACE2, blocking a viral interaction, and perturbing ACE2’s enzymatic role.

This paper’s cleanest contribution is upstream of those debates: it shows a practical route to find tight-binding, structurally characterized macrocycles against a meaningful human target.

The platform question: can yeast display generalize?

A skeptic could fairly say: a methods paper is only as good as its second example. ACE2 is one target. The harder question is whether this approach can produce comparable ligands across unrelated proteins that differ in topology, dynamics, and available binding surfaces.

That is also where yeast display has to earn its keep. It is not enough to pull a few strong binders in a single campaign. What you want, if you are building a repeatable engine, is something closer to a standard operating procedure:

“Here is how big a library we can screen in practice.”
“Here is the distribution of affinities we typically see.”
“Here is how often we get a stable, crystallizable complex.”
“Here is what early developability looks like (stability, protease sensitivity, selectivity).”

The current paper gets you partway there by pairing potency with structures. The next convincing step would be a series of targets, or at least early developability readouts that move the macrocycles from “binding molecules” toward “drug candidates.”

One quiet implication: macrocyclic peptide discovery is getting more industrial

When a technology lets you screen millions of variants quantitatively and then deliver low‑nanomolar ligands with distinct structural solutions, it changes the vibe of the field. Macrocyclic peptides stop being a boutique format that only certain groups can explore, and start looking like something that can be productized.

That does not mean macrocyclic peptides are easy. Constraints remain: manufacturing, stability, delivery, and the usual gulf between binding potency and real-world pharmacology.

But as discovery tooling improves, the bottleneck shifts. Instead of “can we find something that binds,” the interesting problems become “can we make it behave,” and “can we pick the right targets where a macrocycle is actually the best format.”