Attention-based generative models for de novo molecular design

Attention mechanisms have led to many breakthroughs in sequential data modeling but have yet to be incorporated into any generative algorithms for molecular design. Here we explore the impact of adding self-attention layers to generative β-VAE models and show that those with attention are able to learn a complex “molecular grammar” while improving performance on downstream tasks such as accurately sampling from the latent space (“model memory”) or exploring novel chemistries not present in the training data. There is a notable relationship between a model''s architecture, the structure of its latent memory and its performance during inference. We demonstrate that there is an unavoidable tradeoff between model exploration and validity that is a function of the complexity of the latent memory. However, novel sampling schemes may be used that optimize this tradeoff. We anticipate that attention will play an important role in future molecular design algorithms that can make efficient use of the detailed molecular substructures learned by the transformer.

An implementation of attention within the variational autoencoder framework for continuous representation of molecules. The addition of attention significantly increases model performance for complex tasks such as exploration of novel chemistries.     

Peptide-based inhibitors of protein–protein interactions: biophysical,structural and cellular consequences of introducing a constraint

Protein–protein interactions (PPIs) are implicated in the majority of cellular processes by enabling and regulating the function of individual proteins. Thus, PPIs represent high-value, but challenging targets for therapeutic intervention. The development of constrained peptides represents an emerging strategy to generate peptide-based PPI inhibitors, typically mediated by α-helices. The approach can confer significant benefits including enhanced affinity, stability and cellular penetration and is ingrained in the premise that pre-organization simultaneously pays the entropic cost of binding, prevents a peptide from adopting a protease compliant β-strand conformation and shields the hydrophilic amides from the hydrophobic membrane. This conceptual blueprint for the empirical design of peptide-based PPI inhibitors is an exciting and potentially lucrative way to effect successful PPI inhibitor drug-discovery. However, a plethora of more subtle effects may arise from the introduction of a constraint that include changes to binding dynamics, the mode of recognition and molecular properties. In this review, we summarise the influence of inserting constraints on biophysical, conformational, structural and cellular behaviour across a range of constraining chemistries and targets, to highlight the tremendous success that has been achieved with constrained peptides alongside emerging design opportunities and challenges.

This review summarizes the influence of inserting constraints on biophysical, conformational, structural and cellular behaviour for peptides targeting α-helix mediated protein–protein interactions.     

Tandem sequential catalytic enantioselective synthesis of highly-functionalised tetrahydroindolizine derivatives

An isothiourea-catalysed enantioselective synthesis of novel tetrahydroindolizine derivatives is reported through a one-pot tandem sequential process. The application of 2-(pyrrol-1-yl)acetic acid in combination with either a trifluoromethyl enone or an α-keto-β,γ-unsaturated ester in an enantioselective Michael addition–lactonisation process, followed by in situ ring-opening and cyclisation, led to a range of 24 tetrahydroindolizine derivatives containing three stereocentres in up to >95 : 5 dr and >99 : 1 er.

The isothiourea-catalysed enantioselective synthesis of tetrahydroindolizine derivatives containing three stereocentres is reported through a one-pot tandem sequential process.     

A step-economic and one-pot access to chiral Cα-tetrasubstituted α-amino acid derivatives via a bicyclic imidazole-catalyzed direct enantioselective C-acylation

C^α-Tetrasubstituted α-amino acids are ubiquitous and unique structural units in bioactive natural products and pharmaceutical compounds. The asymmetric synthesis of these molecules has attracted a lot of attention, but a more efficient method is still greatly desired. Here we describe the first sequential four-step acylation reaction for the efficient synthesis of chiral C^α-tetrasubstituted α-amino acid derivatives from simple N-acylated amino acids via an auto-tandem catalysis using a single nucleophilic catalyst. The synthetic efficiency is improved via a direct enantioselective C-acylation; the methodology affords the corresponding C^α-tetrasubstituted α-amino acid derivatives with excellent enantioselectivities (up to 99% ee). This step-economic, one-pot, and auto-tandem strategy provides facile access to important chiral building blocks, such as peptides, serines, and oxazolines, which are often used in medicinal and synthetic chemistry.

The first four-step sequential reaction for the synthesis of C^α-tetrasubstituted chiral α-amino acid derivatives via auto-tandem catalysis has been developed.     

Interrogating surface versus intracellular transmembrane receptor populations using cell-impermeable SNAP-tag substrates

Employing self-labelling protein tags for the attachment of fluorescent dyes has become a routine and powerful technique in optical microscopy to visualize and track fused proteins. However, membrane permeability of the dyes and the associated background signals can interfere with the analysis of extracellular labelling sites. Here we describe a novel approach to improve extracellular labelling by functionalizing the SNAP-tag substrate benzyl guanine (“BG”) with a charged sulfonate (“SBG”). This chemical manipulation can be applied to any SNAP-tag substrate, improves solubility, reduces non-specific staining and renders the bioconjugation handle impermeable while leaving its cargo untouched. We report SBG-conjugated fluorophores across the visible spectrum, which cleanly label SNAP-fused proteins in the plasma membrane of living cells. We demonstrate the utility of SBG-conjugated fluorophores to interrogate class A, B and C G protein-coupled receptors (GPCRs) using a range of imaging approaches including nanoscopic superresolution imaging, analysis of GPCR trafficking from intra- and extracellular pools, in vivo labelling in mouse brain and analysis of receptor stoichiometry using single molecule pull down.

Impermeable SNAP-tag substrates allow exclusive labelling of receptors on the cell membrane for nanoscopy, SiMPull and in vivo use.     

Fluorogenic probes for detecting deacylase and demethylase activity towards post-translationally-modified lysine residues

Reversible enzymatic post-translational modification of the ε-amino groups of lysine residues (e.g. N-acylation reactions) plays an important role in regulating the cellular activities of numerous proteins. This study describes how enzyme catalyzed N-deprotection of lysine residues of non-fluorescent peptide-coumarin probes can be used to generate N-deprotected peptides that undergo spontaneous O- to N-ester transfer reactions (uncatalyzed) to generate a highly fluorescent N-carbamoyl peptide. This enables detection of enzyme catalyzed N-deacetylation, N-demalonylation, N-desuccinylation and N-demethylation reactions activities towards the N-modified lysine residues of these probes using simple ‘turn on’ fluorescent assays.

We developed “turn-on” fluorescent probes that detect enzymatic lysine deacylation and demethylation critical for epigenetic and other cellular phenomena, using intramolecular O- to N-ester transfer reactions.     

Cleavable and tunable cysteine-specific arylation modification with aryl thioethers

Cysteine represents an attractive target for peptide/protein modification due to the intrinsic high nucleophilicity of the thiol group and low natural abundance. Herein, a cleavable and tunable covalent modification approach for cysteine containing peptides/proteins with our newly designed aryl thioethers via a S_NAr approach was developed. Highly efficient and selective bioconjugation reactions can be carried out under mild and biocompatible conditions. A series of aryl groups bearing different bioconjugation handles, affinity or fluorescent tags are well tolerated. By adjusting the skeleton and steric hindrance of aryl thioethers slightly, the modified products showed a tunable profile for the regeneration of the native peptides.

A cleavable and tunable covalent modification approach for cysteine by aryl thioethers via a S_NAr approach was developed. The highly efficient and selective bioconjugation reactions can proceed under the mild and biocompatible conditions.     

Synthesis and mechanistic investigations of pH-responsive cationic poly(aminoester)s

The synthesis and degradation mechanisms of a class of pH-sensitive, rapidly degrading cationic poly(α-aminoester)s are described. These reactive, cationic polymers are stable at low pH in water, but undergo a fast and selective degradation at higher pH to liberate neutral diketopiperazines. Related materials incorporating oligo(α-amino ester)s have been shown to be effective gene delivery agents, as the charge-altering degradative behavior facilitates the delivery and release of mRNA and other nucleic acids in vitro and in vivo. Herein, we report detailed studies of the structural and environmental factors that lead to these rapid and selective degradation processes in aqueous buffers. At neutral pH, poly(α-aminoester)s derived from N-hydroxyethylglycine degrade selectively by a mechanism involving sequential 1,5- and 1,6-O→N acyl shifts to generate bis(N-hydroxyethyl) diketopiperazine. A family of structurally related cationic poly(aminoester)s was generated to study the structural influences on the degradation mechanism, product distribution, and pH dependence of the rate of degradation. The kinetics and mechanism of the pH-induced degradations were investigated by ¹H NMR, model reactions, and kinetic simulations. These results indicate that polyesters bearing α-ammonium groups and appropriately positioned N-hydroxyethyl substituents are readily cleaved (by intramolecular attack) or hydrolyzed, representing dynamic “dual function” materials that are initially polycationic and transform with changing environment to neutral products.

The synthesis and degradation mechanisms of a class of pH-sensitive, rapidly degrading cationic poly(α-aminoester)s are described.     

Synthesis of 1,3-disubstituted bicyclo[1.1.0]butanes via directed bridgehead functionalization

Bicyclo[1.1.0]butanes (BCBs) are increasingly valued as intermediates in ‘strain release’ chemistry for the synthesis of substituted four membered rings and bicyclo[1.1.1]pentanes, with applications including bioconjugation processes. Variation of the BCB bridgehead substituents can be challenging due to the inherent strain of the bicyclic scaffold, often necessitating linear syntheses of specific BCB targets. Here we report the first palladium catalyzed cross-coupling on pre-formed BCBs which enables a ‘late stage’ diversification of the bridgehead position, and the conversion of the resultant products into a range of useful small ring building blocks.

Bicyclo[1.1.0]butanes (BCBs) are valuable precursors to four-membered rings and bicyclo[1.1.1]pentanes, and useful bioconjugation agents. We describe a versatile approach to access 1,3-disubstituted BCBs, which are otherwise challenging to prepare.     

Antimicrobial α-defensins as multi-target inhibitors against amyloid formation and microbial infection

Amyloid aggregation and microbial infection are considered as pathological risk factors for developing amyloid diseases, including Alzheimer''s disease (AD), type II diabetes (T2D), Parkinson''s disease (PD), and medullary thyroid carcinoma (MTC). Due to the multifactorial nature of amyloid diseases, single-target drugs and treatments have mostly failed to inhibit amyloid aggregation and microbial infection simultaneously, thus leading to marginal benefits for amyloid inhibition and medical treatments. Herein, we proposed and demonstrated a new “anti-amyloid and antimicrobial hypothesis” to discover two host-defense antimicrobial peptides of α-defensins containing β-rich structures (human neutrophil peptide of HNP-1 and rabbit neutrophil peptide of NP-3A), which have demonstrated multi-target, sequence-independent functions to (i) prevent the aggregation and misfolding of different amyloid proteins of amyloid-β (Aβ, associated with AD), human islet amyloid polypeptide (hIAPP, associated with T2D), and human calcitonin (hCT, associated with MTC) at sub-stoichiometric concentrations, (ii) reduce amyloid-induced cell toxicity, and (iii) retain their original antimicrobial activity upon the formation of complexes with amyloid peptides. Further structural analysis showed that the sequence-independent amyloid inhibition function of α-defensins mainly stems from their cross-interactions with amyloid proteins via β-structure interactions. The discovery of antimicrobial peptides containing β-structures to inhibit both microbial infection and amyloid aggregation greatly expands the new therapeutic potential of antimicrobial peptides as multi-target amyloid inhibitors for better understanding pathological causes and treatments of amyloid diseases.

We report a new “anti-amyloid and antimicrobial hypothesis” by discovering host-defense antimicrobial peptides of α-defensins containing β-sheet structures, which possess inhibition functions against amyloid aggregation and microbial infection.     

Rational design of an “all-in-one” phototheranostic

We report here porphodilactol derivatives and their corresponding metal complexes. These systems show promise as “all-in-one” phototheranostics and are predicated on a design strategy that involves controlling the relationship between intersystem crossing (ISC) and photothermal conversion efficiency following photoexcitation. The requisite balance was achieved by tuning the aromaticity of these porphyrinoid derivatives and forming complexes with one of two lanthanide cations, namely Gd³⁺ and Lu³⁺. The net result led to a metalloporphodilactol system, Gd-trans-2, with seemingly optimal ISC efficiency, photothermal conversion efficiency and fluorescence properties, as well as good chemical stability. Encapsulation of Gd-trans-2 within mesoporous silica nanoparticles (MSN) allowed its evaluation for tumour diagnosis and therapy. It was found to be effective as an “all-in-one” phototheranostic that allowed for NIR fluorescence/photoacoustic dual-modal imaging while providing an excellent combined PTT/PDT therapeutic efficacy in vitro and in vivo in 4T1-tumour-bearing mice.

We report here porphodilactol derivatives and their corresponding metal complexes as “all-in-one” phototheranostics by controlling the relationship between intersystem crossing (ISC) and photothermal conversion efficiency following photoexcitation.     

Mechanistic insights into copper-catalyzed aerobic oxidative coupling of N–N bonds

Catalytic N–N coupling is a valuable transformation for chemical synthesis and energy conversion. Here, mechanistic studies are presented for two related copper-catalyzed oxidative aerobic N–N coupling reactions, one involving the synthesis of a pharmaceutically relevant triazole and the other relevant to the oxidative conversion of ammonia to hydrazine. Analysis of catalytic and stoichiometric N–N coupling reactions support an “oxidase”-type catalytic mechanism with two redox half-reactions: (1) aerobic oxidation of a Cu^I catalyst and (2) Cu^II-promoted N–N coupling. Both reactions feature turnover-limiting oxidation of Cu^I by O₂, and this step is inhibited by the N–H substrate(s). The results highlight the unexpected facility of the N–N coupling step and establish a foundation for development of improved catalysts for these transformations.

Mechanistic studies provide valuable insights into Cu-catalyzed N–N coupling reactions relevant to energy conversion and pharmaceutical synthesis.     

Nickel-catalyzed asymmetric reductive arylation of α-chlorosulfones with aryl halides

We report an asymmetric Ni-catalyzed reductive cross-coupling of aryl/heteroaryl halides with racemic α-chlorosulfones to afford enantioenriched sulfones. The reaction tolerates a variety of functional groups under mild reaction conditions, which complements the current methods. The utility of this work was demonstrated by facile late-stage functionalization of commercial drugs.

In this work Ni-Catalyzed reductive cross-coupling between (hetero)aryl halides and racemic α-chlorosulfones to prepare enantioenriched α,α-disubstituted sulfones was demonstrated, allowing facile structural derivatization of drug precursors.     

CHIP-mediated hyperubiquitylation of tau promotes its self-assembly into the insoluble tau filaments

The tau protein is a highly soluble and natively unfolded protein. Under pathological conditions, tau undergoes multiple post-translational modifications (PTMs) and conformational changes to form insoluble filaments, which are the proteinaceous signatures of tauopathies. To dissect the crosstalk among tau PTMs during the aggregation process, we phosphorylated and ubiquitylated recombinant tau in vitro using GSK3β and CHIP, respectively. The resulting phospho–ub-tau contained conventional polyubiquitin chains with lysine 48 linkages, sufficient for proteasomal degradation, whereas unphosphorylated ub-tau species retained only one–three ubiquitin moieties. Mass-spectrometric analysis of in vitro reconstituted phospho–ub-tau revealed seven additional ubiquitylation sites, some of which are known to stabilize tau protofilament stacking in the human brain with tauopathy. When the ubiquitylation reaction was prolonged, phospho–ub-tau transformed into insoluble hyperubiquitylated tau species featuring fibrillar morphology and in vitro seeding activity. We developed a small-molecule inhibitor of CHIP through biophysical screening; this effectively suppressed tau ubiquitylation in vitro and delayed its aggregation in cultured cells including primary cultured neurons. Our biochemical findings point to a “multiple-hit model,” where sequential events of tau phosphorylation and hyperubiquitylation function as a key driver of the fibrillization process, thus indicating that targeting tau ubiquitylation may be an effective strategy to alleviate the course of tauopathies.

Multiple-hit model for tau aggregation, where sequential events of tau phosphorylation and hyperubiquitylation function as a key driver of the fibrillization process.     

Catalytic asymmetric synthesis of quaternary trifluoromethyl α- to ε-amino acid derivatives via umpolung allylation/2-aza-Cope rearrangement

In this study, we developed an efficient Ir-catalyzed cascade umpolung allylation/2-aza-Cope rearrangement of tertiary α-trifluoromethyl α-amino acid derivatives for the preparation of a variety of quaternary α-trifluoromethyl α-amino acids in high yields with excellent enantioselectivities. The umpolung reactivity empowered by the activation of the key isatin-ketoimine moiety obviates the intractable enantioselectivity control in Pd-catalyzed asymmetric linear α-allylation. In combination with quasi parallel kinetic resolution or kinetic resolution, the generality of this method is further demonstrated by the first preparation of enantioenriched quaternary trifluoromethyl β-, γ-, δ- and ε-amino acid derivatives.

In this study, we developed an efficient Ir-catalyzed cascade umpolung allylation/2-aza-Cope rearrangement for the preparation of a variety of quaternary trifluoromethyl α-ε-amino acids in high yields with excellent enantioselectivities.     

Biomimetic nanochannels for the discrimination of sialylated glycans via a tug-of-war between glycan binding and polymer shrinkage

Sialylated glycans that are attached to cell surface mediate diverse cellular processes such as immune responses, pathogen binding, and cancer progression. Precise determination of sialylated glycans, particularly their linkage isomers that can trigger distinct biological events and are indicative of different cancer types, remains a challenge, due to their complicated composition and limited structural differences. Here, we present a biomimetic nanochannels system integrated with the responsive polymer polyethyleneimine-g-glucopyranoside (Glc-PEI) to solve this problem. By using a dramatic “OFF–ON” change in ion flux, the nanochannels system achieves specific recognition for N-acetylneuraminic acid (Neu5Ac, the predominant form of sialic acid) from various monosaccharides and sialic acid species. Importantly, different “OFF–ON” ratios of the conical nanochannels system allows the precise and sensitive discrimination of sialylated glycan linkage isomers, α2–3 and α2–6 linkage (the corresponding ion conductance increase ratios are 96.2% and 264%, respectively). Analyses revealed an unusual tug-of-war mechanism between polymer-glycan binding and polymer shrinkage. The low binding affinity of Glc-PEI for the α2–6-linked glycan caused considerable shrinkage of Glc-PEI layer, but the high affinity for the α2–3-linked glycan resulted in only a slight shrinkage. This competition mechanism provides a simple and versatile materials design principle for recognition or sensing systems that involve negatively charged target biomolecules. Furthermore, this work broadens the application of nanochannel systems in bioanalysis and biosensing, and opens a new route to glycan analysis that could help to uncover the mysterious and wonderful glycoworld.

A glycan-responsive polymer-modified nanochannels system enables the precise discrimination of sialylated glycan linkage isomers via the different “OFF–ON” changes resulting from a “tug-of-war” between polymer-glycan binding and polymer shrinkage.     

A bifunctional iminophosphorane squaramide catalyzed enantioselective synthesis of hydroquinazolines via intramolecular aza-Michael reaction to α,β-unsaturated esters

An efficient synthesis of enantioenriched hydroquinazoline cores via a novel bifunctional iminophosphorane squaramide catalyzed intramolecular aza-Michael reaction of urea-linked α,β-unsaturated esters is described. The methodology exhibits a high degree of functional group tolerance around the forming hydroquinazoline aryl core and wide structural variance on the nucleophilic N atom of the urea moiety. Excellent yields (up to 99%) and high enantioselectivities (up to 97 : 3 er) using both aromatic and less acidic aliphatic ureas were realized. The potential industrial applicability of the transformation was demonstrated in a 20 mmol scale-up experiment using an adjusted catalyst loading of 2 mol%. The origin of enantioselectivity and reactivity enhancement provided by the squaramide motif has been uncovered computationally using density functional theory (DFT) calculations, combined with the activation strain model (ASM) and energy decomposition analysis (EDA).

The activation of both aromatic and aliphatic ureas as N-centered nucleophiles in intramolecular Michael addition reactions to α,β-unsaturated esters was achieved under bifunctional iminophosphorane squaramide superbase catalysis.     

“Eggs in egg cartons”: co-crystallization to embed molecular cages into crystalline lattices

Discrete (M₃L₂)_n cages assembled from a tripodal ligand (L) and metal ions (M: Cu(i) or Ag(i)) are embedded in networked coordination hosts formed by partial dissociation of the same discrete cages during the crystallization process. The resulting “eggs-in-an-egg-carton” structures provide unique examples of the co-crystallization of discrete and infinite coordination frameworks.

Discrete coordination cages were connected into the infinite lattices via shape-complementary co-crystallization with networked coordination hosts in the “eggs-in-an-egg-carton” styles.     

Selective α,δ-hydrocarboxylation of conjugated dienes utilizing CO2 and electrosynthesis

To date the majority of diene carboxylation processes afford the α,δ-dicarboxylated product, the selective mono-carboxylation of dienes is a significant challenge and the major product reported under transition metal catalysis arises from carboxylation at the α-carbon. Herein we report a new electrosynthetic approach, that does not rely on a sacrificial electrode, the reported method allows unprecedented direct access to carboxylic acids derived from dienes at the δ-position. In addition, the α,δ-dicarboxylic acid or the α,δ-reduced alkene can be easily accessed by simple modification of the reaction conditions.

Selective electrosynthetic α,δ-hydrocarboxylation of 1,3-dienes is reported, utilising non-sacrificial electrodes that provide access to the previously challenging δ-carboxylated regioisomer.