Andrei Loas

Covalent peptide-encoded libraries are assembled through a Heck coupling method allowing modification of resin-bound peptides and enabling affinity-based selection of small molecules bearing structurally diversified acrylamide warheads.

Luciferases are widely used bioluminescent reporters, yet access to these enzymes has historically relied on recombinant expression. Using automated fast-flow peptide synthesis (AFPS), we synthesized Gaussia luciferase (GLuc, 18.1 kDa, 168 amino acids), picALuc (13 kDa, 120 amino acids), and LuxSit-i (13.8 kDa, 118 amino acids), representing a nature-derived disulfide-rich luciferase, an engineered luciferase, and a de novo-designed luciferase, respectively. Synthetic GLuc was purified, oxidatively folded, and recovered as a monomeric, enzymatically active enzyme with biophysical properties comparable to its recombinant counterpart. picALuc likewise required oxidative folding, whereas cysteine-free LuxSit-i was folded by denaturant removal using size-exclusion chromatography. Biochemical and biophysical analyses confirmed the structural integrity and activity of the three synthetic enzymes. Notably, LuxSit-i represents, to our knowledge, the original total chemical synthesis of a fully de novo-designed enzyme. Together, these proteins expand the known repertoire of fully synthetic enzymes produced by solid-phase peptide synthesis by approximately 14% and establish the initial examples of fully synthetic luciferases produced by solid-phase peptide synthesis that demonstrate enzymatic activity. These findings position AFPS as a platform for accessing functional enzymes across evolutionary and design spaces and may enable the development of luciferases incorporating noncanonical residues, site-specific modifications, or mirror-image sequences.

Carboranes are endowed with unique structural and electronic features and are applied across diverse research areas. General methods for their high-content incorporation, however, remain limited. We report the development of organic carborane reagents based on N-hydroxysuccinimide (NHS) esters as a versatile platform for selective conjugation to amine-containing residues of small molecules and polypeptides. Bifunctional carborane reagents obtained by this method enable peptide-driven multimerization into carborane–peptide polymers, an original class of hybrid materials. Therapeutic antibodies functionalized with these reagents yield boron-rich conjugates containing up to 13 carboranes per antibody molecule that retain native antibody activity. Linker length is critical for maximizing loading efficiency. Mass spectrometry mapping revealed up to 30 preferential lysine modification sites per antibody, located outside the complementarity-determining region(s). Prevalence-ranked lysine mapping of IgG antibodies under native-like conditions using NHS esters bearing hydrophobic (carborane) substituents provides a useful framework for the rational design and optimization of antibody–drug conjugates. These constructs, integrating therapeutic antibody targeting features with boron delivery capabilities, hold promise as multimodal agents for boron neutron capture therapy (BNCT).

Developing chemical toolkits for mRNA modification has remained an immense challenge, driven by the inherent difficulties in targeting mRNA molecules. Antisense oligonucleotides (ASOs) offer a promising framework for targeting specific mRNA sequences, yet they do not possess the capacity to alter the structure of mRNA except through enzyme-mediated hydrolysis. We developed a platinum­(IV)–ASO strategy that combines the sequence specificity of ASOs with the reactivity of platinum to functionalize nucleic acids, including short RNA and mRNA, in a selective enzyme-free manner. Access to Pt­(IV)–ASO constructs was made possible by an equatorial Pt­(IV) ammine derivatization strategy, allowing for the conjugation of carboxylic acids directly to the Pt core. Reactivity with 21-mer RNA and full-length mRNA by Pt–ASO constructs was demonstrated, and the conjugated products were characterized using a suite of orthogonal techniques, such as electrophoretic mobility shift assay, MALDI-TOF MS, temperature-dependent dissociation assay, and RT-qPCR. Constructs were optimized for their reactivity and selectivity, allowing for Pt­(IV)–PMO constructs with subnanomolar IC₅₀ values in an RNA competition assay. This Pt­(IV)–ASO platform facilitates new avenues for RNA modification by providing a strategy for functionalizing nucleic acids with potential applications in molecular biology research.

The use of encoding tags in combinatorial libraries accelerates hit generation by enabling high-throughput identification of small-molecule ligands. Peptide-encoded libraries (PELs) support the selection of structurally diverse small-molecule binders to proteins of interest. Here, we introduce a covalent PEL (coPEL) platform that incorporates cysteine-reactive scaffolds to identify irreversible protein binders. We leverage the chemical stability of PELs and the selective reactivity of palladium catalysts derived from dialkylbiaryl phosphine ligands to enable solid-phase Heck coupling reactions to rapidly diversify covalent acrylamide warheads. The optimized reaction conditions are high-yielding across a broad range of (hetero)aryl halides, ensuring robust performance and versatility within the coPEL platform. Screening a coPEL against the epidermal growth factor receptor (EGFR) tyrosine kinase, a key oncology target, yielded covalent small-molecule inhibitors with low-micromolar potency in vitro. This approach provides a complementary strategy for targeting diverse proteins and developing new classes of covalent inhibitors.

Hi1a is a venom peptide isolated from the Australian funnel-web spider Hadronyche infensa, containing an intricate "double knot" tertiary structure, and is currently under investigation for the treatment of ischemic stroke. The recombinant expression and chemical synthesis of Hi1a remain a significant challenge due to laborious protocols, delaying progress in translation to the clinic. We describe the first single-shot chemical synthesis of Hi1a (76 amino acids, AA) using automated fast-flow peptide synthesis (AFPS), enabling rapid access (10 mg). This work highlights the utility of AFPS as a technology that enables chemical production of biomolecules challenging to obtain by recombinant methods.

Despite their potential, the preparation of large synthetic macrocyclic libraries for ligand discovery and development has been limited. Here, we produce 100-million-membered macrocyclic libraries containing natural and nonnatural amino acids. Near-quantitative intramolecular disulfide formation is facilitated by rapid oxidation with iodine in solution. After use in affinity selection, treatment with dithiothreitol enables near-quantitative reduction, rendering linear peptide analogs for standard tandem mass spectrometry. We use these libraries to discover macrocyclic binders to cadherin-2 and anti-hemagglutinin antibody clone 12ca5. Structure-activity relationship studies of an initial cadherin-binding peptide [ CBP ; apparent dissociation constant ( K _d ) = 53 nanomolar] reveal residues responsible for driving affinity (hotspots) and mutation-tolerant residues (coldspots). Two original macrocyclic libraries are prepared in which these hotspots and coldspots are derivatized with nonnatural amino acids. Following discovery and validation, high-affinity ligands are discovered from the coldspot library, with NCBP-4 demonstrating improved affinity ( K _d = 29 nanomolar). Overall, we expect that this work will improve the use of macrocyclic libraries in therapeutic peptide development.

Programmed cell death protein 1 (PD-1) and programmed cell death ligand 1 (PD-L1) are rapidly synthesized with automated flow technology.

The chemical synthesis of proteins (CSP) has been an essential tool in studying and understanding the role of these biological polymers and in enabling the discovery of novel classes of inhibitors. However, CSP with commercially available synthesizers is typically limited to producing polypeptides of about 50 to 70 amino acids in length. Consequently, a wide range of protein targets have been inaccessible using these technologies, or they require cumbersome synthesis and purification of multiple peptide fragments. In this report, we employed a powerful combination of automated fast-flow peptide synthesis (AFPS), native chemical ligation (NCL), and high-throughput evaluation of refolding conditions to achieve the first chemical synthesis of both the wild-type and mirror-image forms of functional full-length cyclophilin A, which plays a vital role in proline cis-trans isomerization and other important processes. Functional assays confirmed that the chemically synthesized proteins retained their biological properties.

Affinity selection-mass spectrometry (AS-MS) is a ligand discovery platform that relies upon mass spectrometry to identify molecules bound to a biomolecular target. When utilized with large peptide libraries (10⁸ members), AS-MS sample complexity can surpass the sequencing capacity of modern mass spectrometers, resulting in incomplete data, identification of few target-specific ligands, and/or incomplete sequencing. To address this challenge, we introduce pyBinder to perform quantitation on AS-MS data to process primary MS¹ data and develop two scores to rank the peptides from the integration of their peak area: target selectivity and concentration-dependent enrichment. We benchmark pyBinder utilizing AS-MS data developed against antihemagglutinin antibody 12ca5, revealing that peptides that contain a motif known for target-specific high-affinity binding are well characterized by these two scores. AS-MS data from a second protein target, WD Repeat Domain 5 (WDR5), is analyzed to confirm the two pyBinder scores reliably capture the target-specific motif-containing peptides. From the results delivered by pyBinder, a list of target-selective features is developed and fed back into subsequent MS experiments to facilitate expanded data generation and the targeted discovery of selective ligands. pyBinder analysis resulted in a 4-fold increase in motif-containing sequence identification for WDR5 (from 3 to 14 ligands discovered), showing the utility of the two scores. This work establishes an improved approach for AS-MS to enable discovery outcomes (i.e., more ligands identified), but also a way to compare AS-MS data across samples, protocols, and conditions broadly.