Funneling modulatory peptide design with generative models: Discovery and characterization of disruptors of calcineurin protein-protein interactions.

Design of peptide binders is an attractive strategy for targeting "undruggable" protein-protein interfaces. Current design protocols rely on the extraction of an initial sequence from one known protein interactor of the target protein, followed by in-silico or in-vitro mutagenesis-based optimization of its binding affinity. Wet lab protocols can explore only a minor portion of the vast sequence space and cannot efficiently screen for other desirable properties such as high specificity and low toxicity, while in-silico design requires intensive computational resources and often relies on simplified binding models. Yet, for a multivalent protein target, dozens to hundreds of natural protein partners already exist in the cellular environment. Here, we describe a peptide design protocol that harnesses this diversity via a machine learning generative model. After identifying putative natural binding fragments by literature and homology search, a compositional Restricted Boltzmann Machine is trained and sampled to yield hundreds of diverse candidate peptides. The latter are further filtered via flexible molecular docking and an in-vitro microchip-based binding assay. We validate and test our protocol on calcineurin, a calcium-dependent protein phosphatase involved in various cellular pathways in health and disease. In a single screening round, we identified multiple 16-length peptides with up to six mutations from their closest natural sequence that successfully interfere with the binding of calcineurin to its substrates. In summary, integrating protein interaction and sequence databases, generative modeling, molecular docking and interaction assays enables the discovery of novel protein-protein interaction modulators.

Conversion of methionine biosynthesis in Escherichia coli from trans- to direct-sulfurylation enhances extracellular methionine levels.

Methionine is an essential amino acid in mammals and a precursor for vital metabolites required for the survival of all organisms. Consequently, its inclusion is required in diverse applications, such as food, feed, and pharmaceuticals. Although amino acids and other metabolites are commonly produced through microbial fermentation, high-yield biosynthesis of L-methionine remains a significant challenge due to the strict cellular regulation of the biosynthesis pathway. As a result, methionine is produced primarily synthetically, resulting in a racemic mixture of D,L-methionine. This study explores methionine bio-production in E. coli by replacing its native trans-sulfurylation pathway with the more common direct-sulfurylation pathway used by other bacteria. To this end, we generated a methionine auxotroph E. coli strain (MG1655) by simultaneously deleting metA and metB genes and complementing them with metX and metY from different bacteria. Complementation of the genetically modified E. coli with metX/metY from Cyclobacterium marinum or Deinococcus geothermalis, together with the deletion of the global repressor metJ and overexpression of the transporter yjeH, resulted in a substantial increase of up to 126 and 160-fold methionine relative to the wild-type strain, respectively, and accumulation of up to 700 mg/L using minimal MOPS medium and 2 ml culture. Our findings provide a method to study methionine biosynthesis and a chassis for enhancing L-methionine production by fermentation.

Single-cell transcriptomic analysis of oral masticatory and lining mucosa-derived mesenchymal stromal cells.

AIM: To reveal the heterogeneity of ex vivo-cultured human mesenchymal stromal cells derived from either masticatory or lining oral mucosa. MATERIALS AND METHODS: Cells were retrieved from the lamina propria of the hard palate and alveolar mucosa of three individuals. The analysis of transcriptomic-level differences was accomplished using single-cell RNA sequencing. RESULTS: Cluster analysis clearly distinguished between cells from the masticatory and lining oral mucosa, and revealed 11 distinct cell sub-populations, annotated as fibroblasts, smooth muscle cells or mesenchymal stem cells. Interestingly, cells presenting a mesenchymal stem cell-like gene expression pattern were predominantly found in masticatory mucosa. Although cells of masticatory mucosa origin were highly enriched for biological processes associated with wound healing, those from the lining oral mucosa were highly enriched for biological processes associated with the regulation of epithelial cells. CONCLUSIONS: Our previous work had shown that cells from the lining and masticatory oral mucosae are phenotypically heterogeneous. Here, we extend these findings to show that these changes are not the result of differences in averages but rather represent two distinct cell populations, with mesenchymal stem cells more common in masticatory mucosa. These features may contribute to specific physiological functions and have relevance for potential therapeutic interventions.

A ligand-independent origin of abscisic acid perception.

Land plants are considered monophyletic, descending from a single successful colonization of land by an aquatic algal ancestor. The ability to survive dehydration to the point of desiccation is a key adaptive trait enabling terrestrialization. In extant land plants, desiccation tolerance depends on the action of the hormone abscisic acid (ABA) that acts through a receptor-signal transduction pathway comprising a PYRABACTIN RESISTANCE 1-like (PYL)-PROTEIN PHOSPHATASE 2C (PP2C)-SNF1-RELATED PROTEIN KINASE 2 (SnRK2) module. Early-diverging aeroterrestrial algae mount a dehydration response that is similar to that of land plants, but that does not depend on ABA: Although ABA synthesis is widespread among algal species, ABA-dependent responses are not detected, and algae lack an ABA-binding PYL homolog. This raises the key question of how ABA signaling arose in the earliest land plants. Here, we systematically characterized ABA receptor-like proteins from major land plant lineages, including a protein found in the algal sister lineage of land plants. We found that the algal PYL-homolog encoded by Zygnema circumcarinatum has basal, ligand-independent activity of PP2C repression, suggesting this to be an ancestral function. Similarly, a liverwort receptor possesses basal activity, but it is further activated by ABA. We propose that co-option of ABA to control a preexisting PP2C-SnRK2-dependent desiccation-tolerance pathway enabled transition from an all-or-nothing survival strategy to a hormone-modulated, competitive strategy by enabling continued growth of anatomically diversifying vascular plants in dehydrative conditions, enabling them to exploit their new environment more efficiently.

Discovery and characterization of small molecule inhibitors of cystathionine gamma-synthase with in planta activity.

The synthesis of essential amino acids in plants is pivotal for their viability and growth, and these cellular pathways are therefore targeted for the discovery of new molecules for weed control. Herein, we describe the discovery and design of small molecule inhibitors of cystathionine gamma-synthase, a key enzyme in the biosynthesis of methionine. Based on in silico screening and filtering of a large molecular database followed by the in vitro selection of molecules, we identified small molecules capable of binding the target enzyme. Molecular modelling of the interaction and direct biophysical binding enabled us to explore a focussed chemical expansion set of molecules characterized by an active phenyl-benzamide chemical group. These molecules are bio-active and efficiently inhibit the viability of BY-2 tobacco cells and seedlings growth of Arabidopsis thaliana on agar plates.

Screening Collagenase Activity in Bacterial Lysate for Directed Enzyme Applications.

Collagenases are essential enzymes capable of digesting triple-helical collagen under physiological conditions. These enzymes play a key role in diverse physiological and pathophysiological processes. Collagenases are used for diverse biotechnological applications, and it is thus of major interest to identify new enzyme variants with improved characteristics such as expression yield, stability, or activity. The engineering of new enzyme variants often relies on either rational protein design or directed enzyme evolution. The latter includes screening of a large randomized or semirational genetic library, both of which require an assay that enables the identification of improved variants. Moreover, the assay should be tailored for microplates to allow the screening of hundreds or thousands of clones. Herein, we repurposed the previously reported fluorogenic assay using 3,4-dihydroxyphenylacetic acid for the quantitation of collagen, and applied it in the detection of bacterial collagenase activity in bacterial lysates. This enabled the screening of hundreds of E. coli colonies expressing an error-prone library of collagenase G from C. histolyticum, in 96-well deep-well plates, by measuring activity directly in lysates with collagen. As a proof-of-concept, a single variant exhibiting higher activity than the starting-point enzyme was expressed, purified, and characterized biochemically and computationally. This showed the feasibility of this method to support medium-high throughput screening based on direct evaluation of collagenase activity.

Intracellular Protein-Drug Interactions Probed by Direct Mass Spectrometry of Cell Lysates.

Understanding protein-ligand interactions in a cellular context is an important goal in molecular biology and biochemistry, and particularly for drug development. Investigators must demonstrate that drugs penetrate cells and specifically bind their targets. Towards that end, we present a native mass spectrometry (MS)-based method for analyzing drug uptake and target engagement in eukaryotic cells. This method is based on our previously introduced direct-MS method for rapid analysis of proteins directly from crude samples. Here, direct-MS enables label-free studies of protein-drug binding in human cells and is used to determine binding affinities of lead compounds in crude samples. We anticipate that this method will enable the application of native MS to a range of problems where cellular context is important, including protein-protein interactions, drug uptake and binding, and characterization of therapeutic proteins.

Optimized small-molecule pull-downs define MLBP1 as an acyl-lipid-binding protein.

Abscisic acid (ABA) receptors belong to the START domain superfamily, which encompasses ligand-binding proteins present in all kingdoms of life. START domain proteins contain a central binding pocket that, depending on the protein, can couple ligand binding to catalytic, transport or signaling functions. In Arabidopsis, the best characterized START domain proteins are the 14 PYR/PYL/RCAR ABA receptors, while the other members of the superfamily do not have assigned ligands. To address this, we used affinity purification of biotinylated proteins expressed transiently in Nicotiana benthamiana coupled to untargeted LC-MS to identify candidate binding ligands. We optimized this method using ABA-PYL interactions and show that ABA co-purifies with wild-type PYL5 but not a binding site mutant. The Kd of PYL5 for ABA is 1.1 µm, which suggests that the method has sufficient sensitivity for many ligand-protein interactions. Using this method, we surveyed a set of 37 START domain-related proteins, which resulted in the identification of ligands that co-purified with MLBP1 (At4G01883) or MLP165 (At1G35260). Metabolite identification and the use of authentic standards revealed that MLBP1 binds to monolinolenin, which we confirmed using recombinant MLBP1. Monolinolenin also co-purified with MLBP1 purified from transgenic Arabidopsis, demonstrating that the interaction occurs in a native context. Thus, deployment of this relatively simple method allowed us to define a protein-metabolite interaction and better understand protein-ligand interactions in plants.

Inhibition of PD1:PD-L1 interaction by an E. coli-derived optimized PD1 variant.

Immune-checkpoint receptors are a set of signal transduction proteins that can stimulate or inhibit specific anti-tumor responses. It is well established that cancer cells interact with different immune checkpoints to shut down T-cell response, thereby enabling cancer proliferation. Given the importance of immune checkpoint receptors, a structure-function analysis of these systems is imperative. However, recombinant expression and purification of these membrane originated proteins is still a challenge. Therefore, many attempts are being made to improve their expression and solubility while preserving their biological relevance. For this purpose, we designed an E. coli-based optimization system that enables the acquisition of mutations that increases protein solubility and affinity towards its native ligand, while maintaining biological activity. Here we focused on the well-characterized extracellular domain of the 'programmed cell death protein 1' (PD1), an immune checkpoint receptor known to inhibit T-cell proliferation by interacting with its ligands PD-L1 and PD-L2. The simple ELISA-based screening system shown here enabled the identification of high-affinity, highly soluble, functional variants derived from the extracellular domain of human PD1. The system was based on the expression of a GST-tagged variants library in E. coli, which enabled the selection of improved PD1 variants after a single optimization round. Within only two screening rounds, the most active variant showed a 5-fold higher affinity and 2.4-fold enhanced cellular activity as compared to the wild type protein. This scheme can be translated toward other types of challenging receptors toward development of research tools or alternative therapeutics.

An ELISA for the study of calcineurin-NFAT unstructured region interaction.

Calcineurin is a phosphatase that targets the transcription factor, nuclear factor of activated T-cells (NFAT) dephosphorylates multiple sites along NFAT's regulatory domain. The calcineurin-NFAT complex interaction is mediated through two conserved binding motifs known as the PxIxIT and LxVP, which are located at the N- and C- terminus to the phosphorylation sites. The vast range of cellular processes regulated by the calcineurin-NFAT interaction has aroused great interest in the investigation of the structural aspects that govern their complex formation and in the discovery of protein-protein interaction inhibitors; the latter interfere with calcineurin-NFAT complex formation while keeping calcineurin's catalytic site free. To assist additional biophysical study of the calcineurin-NFAT structure-function relation and to screen for new inhibitors, we present a robust and cost-effective Enzyme Linked Immuno Sorbent Assay (ELISA) that is based on the interaction of calcineurin with the NFAT homology region. The latter includes the two calcineurin's binding sites, in addition to the phosphorylation sites. The ELISA experiment shown here can thus be applied towards the study of important structural aspects of the complex and for the discovery of new inhibitors. This will allow for a better understanding of T-cell activation switch.

Multidimensional NMR spectroscopy in a single scan.

Multidimensional NMR has become one of the most widespread spectroscopic tools available to study diverse structural and functional aspects of organic and biomolecules. A main feature of multidimensional NMR is the relatively long acquisition times that these experiments demand. For decades, scientists have been working on a variety of alternatives that would enable NMR to overcome this limitation, and deliver its data in shorter acquisition times. Counting among these methodologies is the so-called ultrafast (UF) NMR approach, which in principle allows one to collect arbitrary multidimensional correlations in a single sub-second transient. By contrast to conventional acquisitions, a main feature of UF NMR is a spatiotemporal manipulation of the spins that imprints the chemical shift and/or J-coupling evolutions being sought, into a spatial pattern. Subsequent gradient-based manipulations enable the reading out of this information and its multidimensional correlation into patterns that are identical to those afforded by conventional techniques. The current review focuses on the fundamental principles of this spatiotemporal UF NMR manipulation, and on a few of the methodological extensions that this form of spectroscopy has undergone during the years.

Engineering of methionine-auxotroph Escherichia coli via parallel evolution of two enzymes from Corynebacterium glutamicum's direct-sulfurylation pathway enables its recovery in minimal medium.

Methionine biosynthesis relies on the sequential catalysis of multiple enzymes. Escherichia coli, the main bacteria used in research and industry for protein production and engineering, utilizes the three-step trans-sulfurylation pathway catalyzed by L-homoserine O-succinyl transferase, cystathionine gamma synthase and cystathionine beta lyase to convert L-homoserine to L-homocysteine. However, most bacteria employ the two-step direct-sulfurylation pathway involving L-homoserine O-acetyltransferases and O-acetyl homoserine sulfhydrylase. We previously showed that a methionine-auxotroph Escherichiacoli strain (MG1655) with deletion of metA, encoding for L-homoserine O-succinyl transferase, and metB, encoding for cystathionine gamma synthase, could be complemented by introducing the genes metX, encoding for L-homoserine O-acetyltransferases and metY, encoding for O-acetyl homoserine sulfhydrylase, from various sources, thus altering the Escherichia coli methionine biosynthesis metabolic pathway to direct-sulfurylation. However, introducing metX and metY from Corynebacterium glutamicum failed to complement methionine auxotrophy. Herein, we generated a randomized genetic library based on the metX and metY of Corynebacterium glutamicum and transformed it into a methionine-auxotrophic Escherichia coli strain lacking the metA and metB genes. Through multiple enrichment cycles, we successfully isolated active clones capable of growing in M9 minimal media. The dominant metX mutations in the evolved methionine-autotrophs Escherichia coli were L315P and H46R. Interestingly, we found that a metY gene encoding only the N-terminus 106 out of 438 amino acids of the wild-type MetY enzyme is functional and supports the growth of the methionine auxotroph. Recloning the new genes into the original plasmid and transforming them to methionine auxotroph Escherichia coli validated their functionality. These results show that directed enzyme-evolution enables fast and simultaneous engineering of new active variants within the Escherichia coli methionine direct-sulfurylation pathway, leading to efficient complementation.

Nature-inspired peptide of MtDef4 C-terminus tail enables protein delivery in mammalian cells.

Cell-penetrating peptides show promise as versatile tools for intracellular delivery of therapeutic agents. Various peptides have originated from natural proteins with antimicrobial activity. We investigated the mammalian cell-penetrating properties of a 16-residue peptide with the sequence GRCRGFRRRCFCTTHC from the C-terminus tail of the Medicago truncatula defensin MtDef4. We evaluated the peptide's ability to penetrate multiple cell types. Our results demonstrate that the peptide efficiently penetrates mammalian cells within minutes and at a micromolar concentration. Moreover, upon N-terminal fusion to the fluorescent protein GFP, the peptide efficiently delivers GFP into the cells. Despite its remarkable cellular permeability, the peptide has only a minor effect on cellular viability, making it a promising candidate for developing a cell-penetrating peptide with potential therapeutic applications.

Real-time multidimensional NMR follows RNA folding with second resolution.

Conformational transitions and structural rearrangements are central to the function of many RNAs yet remain poorly understood. We have used ultrafast multidimensional NMR techniques to monitor the adenine-induced folding of an adenine-sensing riboswitch in real time, with nucleotide-resolved resolution. By following changes in 2D spectra at rates of approximately 0.5 Hz, we identify distinct steps associated with the ligand-induced folding of the riboswitch. Following recognition of the ligand, long range loop-loop interactions form and are then progressively stabilized before the formation of a fully stable complex over approximately 2-3 minutes. The application of these ultrafast multidimensional NMR methods provides the opportunity to determine the structure of RNA folding intermediates and conformational trajectories.

A novel computationally engineered collagenase reduces the force required for tooth extraction in an ex-situ porcine jaw model.

The currently employed tooth extraction methods in dentistry involve mechanical disruption of the periodontal ligament fibers, leading to inevitable trauma to the bundle bone comprising the socket walls. In our previous work, we have shown that a recombinantly expressed truncated version of clostridial collagenase G (ColG) purified from Escherichia coli efficiently reduced the force needed for tooth extraction in an ex-situ porcine jaw model, when injected into the periodontal ligament. Considering that enhanced thermostability often leads to higher enzymatic activity and to set the basis for additional rounds of optimization, we used a computational protein design approach to generate an enzyme to be more thermostable while conserving the key catalytic residues. This process generated a novel collagenase (ColG-variant) harboring sixteen mutations compared to ColG, with a nearly 4℃ increase in melting temperature. Herein, we explored the potential of ColG-variant to further decrease the physical effort required for tooth delivery using our established ex-situ porcine jaw model. An average reduction of 11% was recorded in the force applied to extract roots of mandibular split first and second premolar teeth treated with ColG-variant, relative to those treated with ColG. Our results show for the first time the potential of engineering enzyme properties for dental medicine and further contribute to minimally invasive tooth extraction.

Enhancement of Collagen-I Levels in Human Gingival Fibroblasts by Small Molecule Activation of HIF-1α.

Collagen is the most abundant protein in various mammalian tissues and has an essential role in various cellular processes. Collagen is necessary for food-related biotechnological applications such as cultivated meat, medical engineering, and cosmetics. High-yield expression of natural collagen from mammalian cells is challenging and not cost-effective. Thus, external collagen is obtained primarily from animal tissues. Under cellular hypoxia, overactivation of the transcription factor hypoxia-inducible factor (HIF) was shown to correlate with enhanced accumulation of collagen. Herein, we showed that the small molecule ML228, a known molecular activator of HIF, enhances the accumulation of collagen type-I in human fibroblast cells. We report an increase in collagen levels by 2.33 ± 0.33 when fibroblasts were incubated with 5 µM of ML228. Our experimental results demonstrated, for the first time, that external modulation of the hypoxia biological pathway can boost collagen levels in mammalian cells. Our findings pave the way for enhancing natural collagen production in mammals by altering cellular signaling pathways.

Application of In Silico Filtering and Isothermal Titration Calorimetry for the Discovery of Small Molecule Inhibitors of MDM2.

The initial discovery phase of protein modulators, which consists of filtering molecular libraries and in vitro direct binding validation, is central in drug discovery. Thus, virtual screening of large molecular libraries, together with the evaluation of binding affinity by isothermal calorimetry, generates an efficient experimental setup. Herein, we applied virtual screening for discovering small molecule inhibitors of MDM2, a major negative regulator of the tumor suppressor p53, and thus a promising therapeutic target. A library of 20 million small molecules was screened against an averaged model derived from multiple structural conformations of MDM2 based on published structures. Selected molecules originating from the computational filtering were tested in vitro for their direct binding to MDM2 via isothermal titration calorimetry. Three new molecules, representing distinct chemical scaffolds, showed binding to MDM2. These were further evaluated by exploring structure-similar chemical analogues. Two scaffolds were further evaluated by de novo synthesis of molecules derived from the initial molecules that bound MDM2, one with a central oxoazetidine acetamide and one with benzene sulfonamide. Several molecules derived from these scaffolds increased wild-type p53 activity in MCF7 cancer cells. These set a basis for further chemical optimization and the development of new chemical entities as anticancer drugs.

Collagenase Administration into Periodontal Ligament Reduces the Forces Required for Tooth Extraction in an Ex situ Porcine Jaw Model.

Minimally invasive exodontia is among the long-sought-for development aims of safe dental medicine. In this paper, we aim, for the first time, to examine whether the enzymatic disruption of the periodontal ligament fibers reduces the force required for tooth extraction. To this end, recombinantly expressed clostridial collagenase G variant purified from Escherichia coli was injected into the periodontal ligament of mesial and distal roots of the first and second split porcine mandibular premolars. The vehicle solution was injected into the corresponding roots on the contralateral side. Following sixteen hours, the treated mandibles were mounted on a loading machine to measure the extraction force. In addition, the effect of the enzyme on the viability of different cell types was evaluated. An average reduction of 20% in the applied force (albeit with a large variability of 50 to 370 newton) was observed for the enzymatically treated roots, reaching up to 50% reduction in some cases. Importantly, the enzyme showed only a minor and transient effect on cellular viability, without any signs of toxicity. Using an innovative model enabling the analytical measurement of extraction forces, we show, for the first time, that the enzymatic disruption of periodontal ligament fibers substantially reduces the force required for tooth extraction. This novel technique brings us closer to atraumatic exodontia, potentially reducing intra- and post-operative complications and facilitating subsequent implant placement. The development of novel enzymes with enhanced activity may further simplify the tooth extraction process and present additional clinical relevance for the broad range of implications in the oral cavity.

Native-unlike long-lived intermediates along the folding pathway of the amyloidogenic protein beta2-microglobulin revealed by real-time two-dimensional NMR.

Beta2-microglobulin (beta2m), the light chain of class I major histocompatibility complex, is responsible for the dialysis-related amyloidosis and, in patients undergoing long term dialysis, the full-length and chemically unmodified beta2m converts into amyloid fibrils. The protein, belonging to the immunoglobulin superfamily, in common to other members of this family, experiences during its folding a long-lived intermediate associated to the trans-to-cis isomerization of Pro-32 that has been addressed as the precursor of the amyloid fibril formation. In this respect, previous studies on the W60G beta2m mutant, showing that the lack of Trp-60 prevents fibril formation in mild aggregating condition, prompted us to reinvestigate the refolding kinetics of wild type and W60G beta2m at atomic resolution by real-time NMR. The analysis, conducted at ambient temperature by the band selective flip angle short transient real-time two-dimensional NMR techniques and probing the beta2m states every 15 s, revealed a more complex folding energy landscape than previously reported for wild type beta2m, involving more than a single intermediate species, and shedding new light into the fibrillogenic pathway. Moreover, a significant difference in the kinetic scheme previously characterized by optical spectroscopic methods was discovered for the W60G beta2m mutant.

HNCA-TOCSY-CANH experiments with alternate (13)C- (12)C labeling: a set of 3D experiment with unique supra-sequential information for mainchain resonance assignment.

Described here is a set of three-dimensional (3D) NMR experiments that rely on CACA-TOCSY magnetization transfer via the weak ³J(CαCα) coupling. These pulse sequences, which resemble recently described (13)C detected CACA-TOCSY (Takeuchi et al. 2010) experiments, are recorded in (1)H(2)O, and use (1)H excitation and detection. These experiments require alternate (13)C-(12)C labeling together with perdeuteration, which allows utilizing the small ³J(CαCα) scalar coupling that is otherwise masked by the stronger (1)J(CC) couplings in uniformly (13)C labeled samples. These new experiments provide a unique assignment ladder-mark that yields bidirectional supra-sequential information and can readily straddle proline residues. Unlike the conventional HNCA experiment, which contains only sequential information to the ¹³C(α) of the preceding residue, the 3D hnCA-TOCSY-caNH experiment can yield sequential correlations to alpha carbons in positions i-1, i + 1 and i-2. Furthermore, the 3D hNca-TOCSY-caNH and Hnca-TOCSY-caNH experiments, which share the same magnetization pathway but use a different chemical shift encoding, directly couple the (15)N-(1)H spin pair of residue i to adjacent amide protons and nitrogens at positions i-2, i-1, i + 1 and i + 2, respectively. These new experimental features make protein backbone assignments more robust by reducing the degeneracy problem associated with the conventional 3D NMR experiments.