Penicillin binding proteins (PBPs) are well validated antimicrobial targets, but the prevalence of β-lactamase driven resistance and, more rarely, target-based mutations, necessitates new classes of PBP-targeting drugs. Here we describe the discovery and optimisation of bicyclic peptide (Bicycle®) inhibitors of E. coli PBP3 (EcPBP3) using a proprietary phage display platform, and their conjugation to linear antimicrobial peptides to confer outer membrane permeation. These molecules exhibited high-affinity binding to E. coli PBP3 and a viable spectrum of killing activity against clinically relevant species of the Enterobacterales. X-ray crystallography was used to explore the mode of binding to PBP3, enabling increased target affinity and improvement of in vitro stability. These compounds bind to the transpeptidase active site cleft of PBP3 and represent, to our knowledge, a novel non-β-lactam chemical class of high affinity, non-covalent penicillin binding protein inhibitors. This work demonstrates an approach to rapidly find binders to antimicrobial targets, combined with an entry mechanism to provide access to the Gram negative cell.

Glycan-mediated interactions play a crucial role in biology and medicine, influencing signalling, immune responses, and disease pathogenesis. However, the use of glycans in biosensing and diagnostics is limited by cross-reactivity, as certain glycan motifs can be recognised by multiple biologically distinct protein receptors. To address this specificity challenge, we report the enzymatic synthesis of a 150-member library of site-specifically fluorinated Lewisx analogues ('glycofluoroforms') using naturally occurring enzymes and fluorinated monosaccharides. Subsequent incorporation of a subset of these glycans into nanoparticles or a microarray revealed a striking spectrum of distinct binding intensities across different proteins that recognise Lewisx. Notably, we show that for two proteins with unique binding sites for Lewisx, glycofluoroforms exhibited enhanced binding to one protein, whilst reduced binding to the other, with selectivity governed by fluorination patterns. We finally showcase the potential diagnostic utility of this approach in glycofluoroform-mediated bacterial toxin detection by lateral flow.

Developing therapeutic antibodies is laborious and costly. Here we report a method for antibody discovery that leverages the Illumina HiSeq platform to, within 3 days, screen in the order of 10 8 antibody–antigen interactions. The method, which we named ‘deep screening’, involves the clustering and sequencing of antibody libraries, the conversion of the DNA clusters into complementary RNA clusters covalently linked to the instrument’s flow-cell surface on the same location, the in situ translation of the clusters into antibodies tethered via ribosome display, and their screening via fluorescently labelled antigens. By using deep screening, we discovered low-nanomolar nanobodies to a model antigen using 4 × 10 6 unique variants from yeast-display-enriched libraries, and high-picomolar single-chain antibody fragment leads for human interleukin-7 directly from unselected synthetic repertoires. We also leveraged deep screening of a library of 2.4 × 10 5 sequences of the third complementarity-determining region of the heavy chain of an anti-human epidermal growth factor receptor 2 (HER2) antibody as input for a large language model that generated new single-chain antibody fragment sequences with higher affinity for HER2 than those in the original library.

Biosensing or diagnostics using glycan sequences as targets is limited by glycan cross-reactivities. As binding sites of different proteins that all recognise a given glycan will not be identical, we introduce application of a library of synthetic analogues of a single glycan ligand as a powerful approach to obtain fingerprint binding profiles. We report the enzymatic synthesis of a 150-member library of fluorinated Lewisx analogues (‘glycofluoroforms’) using naturally occurring enzymes and fluorinated monosaccharide building blocks, and the incorporation of a subset into lipid-linked glycan probes or into glyconanoparticles for probing protein binding both in solid-phase high-throughput glycan microarray screening analyses and in solution-phase nanoparticle-based interaction studies. These fluorinated Lewisx analogues, which NMR studies showed to have very similar 3D structures compared to the nonfluorinated Lewisx, gave variously increased or decreased binding with a set of proteins, the different proteins having different preferences and tolerances for binding.

BackgroundSerum albumin binding is an established mechanism to extend the serum half-life of antibody fragments and peptides. The cysteine rich knob domains, isolated from bovine antibody ultralong CDRH3, are the smallest single chain antibody fragments described to date and versatile tools for protein engineering.MethodsHere, we used phage display of bovine immune material to derive knob domains against human and rodent serum albumins. These were used to engineer bispecific Fab fragments, by using the framework III loop as a site for knob domain insertion. ResultsBy this route, neutralisation of the canonical antigen (TNFα) was retained but extended pharmacokinetics in-vivo were achieved through albumin binding. Structural characterisation revealed correct folding of the knob domain and identified broadly common but non-cross-reactive epitopes. Additionally, we show that these albumin binding knob domains can be chemically synthesised to achieve dual IL-17A neutralisation and albumin binding in a single chemical entity. ConclusionsThis study enables antibody and chemical engineering from bovine immune material, via an accessible discovery platform.

A significant unmet need exists for the delivery of biologic drugs such as polypeptides or nucleic acids to the central nervous system for the treatment and understanding of neurodegenerative diseases. Naturally occurring bacterial toxins have been considered as tools to meet this need. However, due to the complexity of tethering macromolecular drugs to toxins and the inherent dangers of working with large quantities of recombinant toxins, no such route has been successfully exploited. Developing a method where a bacterial toxin's nontoxic targeting subunit can be assembled with a drug immediately prior to in vivo administration has the potential to circumvent some of these issues. Using a phage-display screen, we identified two antibody mimetics, anticholera toxin Affimer (ACTA)-A2 and ACTA-C6 that noncovalently associate with the nonbinding face of the cholera toxin B-subunit. In a first step toward the development of a nonviral motor neuron drug-delivery vehicle, we show that Affimers can be selectively delivered to motor neurons in vivo.

The treatment of infection by Gram-negative bacteria is increasingly challenging as resistance to existing antibiotics spreads. Constrained peptides, selected for high target specificity and affinity via library display technologies, are an emerging therapeutic modality in many disease areas and may be a fertile source of new antibiotics. Currently, the utility of constrained peptides and other large molecules as antibiotics is limited by the outer membrane (OM) barrier of Gram-negative bacteria. However, the addition of certain moieties to large molecules can confer the ability to cross the OM; these moieties function as intramolecular trans-OM "vectors". Here, we present a method to systematically assess the carrying capacity of candidate trans-OM vectors using a real-time luminescence assay ("SLALOM", Split Luciferase Assay for Live monitoring of Outer Membrane transit), reporting on periplasmic entry. We demonstrate the usefulness of our tools by constructing a 3800 Da chimeric compound composed of a constrained bicyclic peptide (Bicycle) with a periplasmic target, linked to an intramolecular peptide vector; the resulting chimera is a broad-spectrum inhibitor of pathogenic Gram-negative bacterial growth.

A significant unmet need exists for the delivery of biologics to the central nervous system for the treatment and understanding of neurodegenerative diseases. Naturally occurring toxoids such as the non-toxic B subunit of the cholera toxin have been considered as tools to meet this need. However, due to the complexity of tethering macromolecular drugs to toxins, and the inherent dangers of working with large quantities of recombinant toxin, no such route has been successfully exploited. Developing a method where toxoid and drug can be assembled immediately prior to administration could therefore be extremely useful. Using phage-display, two cholera toxin-binding antibody mimetics (Affimers) were identified that non-covalently associate with the non-GM1 binding face of the cholera toxin B subunit (CTB). The two unique interactions were characterised using a range of techniques to dissect the Affimer-CTB assembly process. Internalisation of the complex was demonstrated in tissue culture, and the system was used to deliver GFP to mammalian cells. Finally, the complex was shown to be successfully internalised into the motor neurones of the brainstem in a mouse model. A second route to modular assembly of a protein delivery system was also explored. By using a high affinity peptide staple, a cholergenoid-botulinum toxin chimera was produced that could be assembled in vitro and used to deliver the functional catalytic domain of botulinum neurotoxin to cultured neuronal cells.

High-throughput studies have been widely used to identify protein-protein interactions; however, few of these candidate interactions have been confirmed in vitro. We have used a combination of isothermal titration calorimetry and fluorescence anisotropy to screen candidate interactions within the pantothenate biosynthetic pathway. In particular, we observed no interaction between the next enzyme in the pathway, pantothenate synthetase (PS), and aspartate decarboxylase, but did observe an interaction between PS and the putative Nudix hydrolase, YfcD. Confirmation of the interaction by fluorescence anisotropy was dependent upon labelling an adventitiously formed glycine on the protein N-terminal affinity purification tag by using Sortase. Subsequent formation of the protein-protein complex led to apparent restriction of the dynamics of this tag, thus suggesting that this approach could be generally applied to a subset of other protein-protein interaction complexes.