Nicholas J. Skelton
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Nicholas J. Skelton has a total of 51 co-authors. Here are the top co-authors by number of shared publications:
- Gabriele Scalia
RED Computational Sciences, Genentech Research and Early Development, San Francisco, USA
2 shared publications - Jan-Christian Hütter
RED Computational Sciences (gCS), Genentech, South San Francisco, CA, USA
2 shared publications - Tommaso Biancalani
RED Computational Sciences, Genentech Research and Early Development, San Francisco, USA
2 shared publications - Wayne J. Fairbrother
Early Discovery Biochemistry, Genentech, South San Francisco, CA 94080, USA
2 shared publications - Ziqing Lu
Genentech
2 shared publications - Alexander P. Wu
Genentech
1 shared publication
Latest Publications
- Open Access
Virtual phenotypic screening discovers novel scaffolds inhibiting the PI3K/mTOR pathway
Phenotypic drug discovery has yielded many first-in-class small-molecule drugs by discovering modulators of disease phenotypes in physiologically relevant cellular systems. However, high-content phenotypic assays lack the ultra-high-throughput scalability of target-based screens. Recent advances in virtual screening present an opportunity to address this bottleneck, but have been limited to simple phenotypes like viability, restricted to small repurposing libraries, or lack in-depth biological validation. Here, we present PhenoCompass, a multimodal co-embedding model that aligns compound structures and high-content phenotypic imaging to enable virtual phenotypic screening over billion-compound libraries. Following training on the Joint Undertaking in Morphology dataset with more than 100,000 Cell Painting compound profiles, retrospective validation with historical biochemical high-throughput screening data demonstrates that PhenoCompass ranks compounds according to their biochemical target engagement. Leveraging PhenoCompass, we performed a prospective screen of 3.8 billion Enamine REAL compounds for inhibitors of PI3K/mTOR pathway, a critical signaling cascade whose aberrant activation is a common tumor driver. This search identified 11 novel compounds with pathway-consistent Cell Painting readout and diverse scaffolds, a 54-fold enrichment over the training set. Orthogonal validation experiments using a FOXO3A reporter assay and direct kinase inhibition confirmed seven structurally novel inhibitors with distinct mechanisms of action. These results highlight the convergence of diverse molecular target profiles onto a shared morphological pathway signature and establish PhenoCompass as a robust framework for high-content phenotypic virtual screening.
Open AccessVirtual phenotypic screening discovers novel scaffolds inhibiting the PI3K/mTOR pathway
Phenotypic drug discovery has yielded many first-in-class small-molecule drugs by discovering modulators of disease phenotypes in physiologically relevant cellular systems. However, high-content phenotypic assays lack the ultra-high-throughput scalability of target-based screens. Recent advances in virtual screening present an opportunity to address this bottleneck, but have been limited to simple phenotypes like viability, restricted to small repurposing libraries, or lack in-depth biological validation. Here, we present PhenoCompass, a multimodal co-embedding model that aligns compound structures and high-content phenotypic imaging to enable virtual phenotypic screening over billion-compound libraries. Following training on the Joint Undertaking in Morphology dataset with more than 100,000 Cell Painting compound profiles, retrospective validation with historical biochemical high-throughput screening data demonstrates that PhenoCompass ranks compounds according to their biochemical target engagement. Leveraging PhenoCompass, we performed a prospective screen of 3.8 billion Enamine REAL compounds for inhibitors of PI3K/mTOR pathway, a critical signaling cascade whose aberrant activation is a common tumor driver. This search identified 11 novel compounds with pathway-consistent Cell Painting readout and diverse scaffolds, a 54-fold enrichment over the training set. Orthogonal validation experiments using a FOXO3A reporter assay and direct kinase inhibition confirmed seven structurally novel inhibitors with distinct mechanisms of action. These results highlight the convergence of diverse molecular target profiles onto a shared morphological pathway signature and establish PhenoCompass as a robust framework for high-content phenotypic virtual screening.
- Open Access
A high-throughput phenotypic screen combined with an ultra-large-scale deep learning-based virtual screening reveals novel scaffolds of antibacterial compounds
The proliferation of multi-drug-resistant bacteria underscores an urgent need for novel antibiotics. Traditional discovery methods face challenges due to limited chemical diversity, high costs, and difficulties in identifying structurally novel compounds. Here, we explore the integration of small molecule high-throughput screening with a deep learning-based virtual screening approach to uncover new antibacterial compounds. Leveraging a diverse library of nearly 2 million small molecules, we conducted comprehensive phenotypic screening against a sensitized Escherichia coli strain that, at a low hit rate, yielded thousands of hits. We trained a deep learning model, GNEprop, to predict antibacterial activity, ensuring robustness through out-of-distribution generalization techniques. Virtual screening of over 1.4 billion compounds identified potential candidates, of which 82 exhibited antibacterial activity, illustrating a 90X improved hit rate over the high-throughput screening experiment GNEprop was trained on. Importantly, a significant portion of these newly identified compounds exhibited high dissimilarity to known antibiotics, indicating promising avenues for further exploration in antibiotic discovery.
Open AccessA high-throughput phenotypic screen combined with an ultra-large-scale deep learning-based virtual screening reveals novel scaffolds of antibacterial compounds
The proliferation of multi-drug-resistant bacteria underscores an urgent need for novel antibiotics. Traditional discovery methods face challenges due to limited chemical diversity, high costs, and difficulties in identifying structurally novel compounds. Here, we explore the integration of small molecule high-throughput screening with a deep learning-based virtual screening approach to uncover new antibacterial compounds. Leveraging a diverse library of nearly 2 million small molecules, we conducted comprehensive phenotypic screening against a sensitized Escherichia coli strain that, at a low hit rate, yielded thousands of hits. We trained a deep learning model, GNEprop, to predict antibacterial activity, ensuring robustness through out-of-distribution generalization techniques. Virtual screening of over 1.4 billion compounds identified potential candidates, of which 82 exhibited antibacterial activity, illustrating a 90X improved hit rate over the high-throughput screening experiment GNEprop was trained on. Importantly, a significant portion of these newly identified compounds exhibited high dissimilarity to known antibiotics, indicating promising avenues for further exploration in antibiotic discovery.
- 31 January 2020
- Published by Taylor & Francis
- p. 55-86
Three-Dimensional Structures of the Chemokine Family
Members of the chemokine superfamily of proteins, which currently number nearly 30,1 are related by primary structural similarities, including a conserved four-cysteine motif.2–5 As indicated in the primary sequence alignments in Figure 1, the superfamily may be divided into two subfamilies in which the first two cysteines in the conserved motif are either separated by an intervening residue, the “C-X-C” branch, or are adjacent in the sequence, the “C-C” branch. Most of the C-X-C chemokines (Figure 1A) also contain a Glu-Leu-Arg (E-L-R) amino acid motif immediately preceding the C-X-C residues. In general, the E-L-R-C-X-C-containing proteins, including interleukin 8 (IL-8), melanoma growth stimulating activity (MGSA/GROα), GROβ, GROγ, neutrophil activating peptide 2 (NAP-2), and epithelial-derived neutrophil attractant 78 (ENA-78), are potent chemoattractants and 5657activators of neutrophils but not of monocytes. Exceptions to this generality include platelet basic protein (PBP), β-thromboglobulin (β-TG), and connective tissue activating peptide III (CTAP-III, also known as low-affinity platelet factor 4 or LA-PF4), that are inactive towards neutrophils; however, each can be converted into NAP-2 by N-terminal truncation.6,7 Platelet factor 4 (PF4), a C-X-C chemokine that does not possess the E-L-R motif, is unusual in that it induces migration of fibroblasts8 as well as both neutrophils and monocytes.9 The C-C chemokines do not chemotax neutrophils, but rather are chemoattractant for monocytes, basophils, eosinophils, and/or lymphocytes. Figure 1 Primary structure alignment of selected C-X-C (A) and C-C (B) chemokines. All sequences are for human proteins except bovine PFG (bPF4).
- 31 January 2020
- Published by Taylor & Francis
- p. 55-86
Three-Dimensional Structures of the Chemokine Family
Members of the chemokine superfamily of proteins, which currently number nearly 30,1 are related by primary structural similarities, including a conserved four-cysteine motif.2–5 As indicated in the primary sequence alignments in Figure 1, the superfamily may be divided into two subfamilies in which the first two cysteines in the conserved motif are either separated by an intervening residue, the “C-X-C” branch, or are adjacent in the sequence, the “C-C” branch. Most of the C-X-C chemokines (Figure 1A) also contain a Glu-Leu-Arg (E-L-R) amino acid motif immediately preceding the C-X-C residues. In general, the E-L-R-C-X-C-containing proteins, including interleukin 8 (IL-8), melanoma growth stimulating activity (MGSA/GROα), GROβ, GROγ, neutrophil activating peptide 2 (NAP-2), and epithelial-derived neutrophil attractant 78 (ENA-78), are potent chemoattractants and 5657activators of neutrophils but not of monocytes. Exceptions to this generality include platelet basic protein (PBP), β-thromboglobulin (β-TG), and connective tissue activating peptide III (CTAP-III, also known as low-affinity platelet factor 4 or LA-PF4), that are inactive towards neutrophils; however, each can be converted into NAP-2 by N-terminal truncation.6,7 Platelet factor 4 (PF4), a C-X-C chemokine that does not possess the E-L-R motif, is unusual in that it induces migration of fibroblasts8 as well as both neutrophils and monocytes.9 The C-C chemokines do not chemotax neutrophils, but rather are chemoattractant for monocytes, basophils, eosinophils, and/or lymphocytes. Figure 1 Primary structure alignment of selected C-X-C (A) and C-C (B) chemokines. All sequences are for human proteins except bovine PFG (bPF4).
- 28 September 2007
- Published by Wiley
- Vol. 161, 222-236
Ca2+ Binding in Proteins of the Calmodulin Superfamily: Cooperativity, Electrostatic Contributions and Molecular Mechanisms
In a large number of intracellular regulatory proteins of the calmodulin superfamily a pair of closely interacting helix–loop–helix Ca2+ binding sites (‘EF hands’) constitute the functional unit—an arrangement that enables cooperative binding. We have recently made detailed experimental studies of the binding of Ca2+ ions to calmodulin, its tryptic fragments TR1C and TR2C (which each constitute a globular domain of a pair of EF hands) and calbindin D9k. Macroscopic Ca2+ binding constants have been obtained over a range of ionic strengths (0 to 0.15 M KCl). For calmodulin the measurements indicate that the two separate globular domains TR1C and TR2C retain the Ca2+ binding properties they have in the intact molecule, with positive cooperativity within each domain. The absolute value of the free energy of interaction between the two sites in each domain, a measure of the cooperativity, increases with ionic strength and is ≥ 10 kJ mol−1 at 0.15 M KCl. Two-dimensional 1H NMR studies show that the addition of KCl does not alter the conformation of the protein. In the case of calbindin D9k several categories of mutants have been studied. One group encompasses the effect of protein surface charges 5 to 15 Å from the Ca2+ binding sites. Two-dimensional 1H NMR shows that neither the addition of KCl, nor mutations that neutralize the surface charges, change the protein conformation. Although the global structure of calbindin D9k is largely unchanged upon binding of calcium, the structure with only one cation bound is more similar to the (Ca2+)2 form. Interestingly, the dynamical properties of the Ca2+-free and the (Ca2+)2-forms of calbindin differ greatly. For example, the rate of NH/ND exchange of the Ca2+-free form is on average 200 times faster than that of the (Ca2+)2-form. The results obtained so far point to a non-negligible entropic contribution to the observed cooperativity of Ca2+ binding.
- 28 September 2007
- Published by Wiley
- Vol. 161, 222-236
Ca2+ Binding in Proteins of the Calmodulin Superfamily: Cooperativity, Electrostatic Contributions and Molecular Mechanisms
In a large number of intracellular regulatory proteins of the calmodulin superfamily a pair of closely interacting helix–loop–helix Ca2+ binding sites (‘EF hands’) constitute the functional unit—an arrangement that enables cooperative binding. We have recently made detailed experimental studies of the binding of Ca2+ ions to calmodulin, its tryptic fragments TR1C and TR2C (which each constitute a globular domain of a pair of EF hands) and calbindin D9k. Macroscopic Ca2+ binding constants have been obtained over a range of ionic strengths (0 to 0.15 M KCl). For calmodulin the measurements indicate that the two separate globular domains TR1C and TR2C retain the Ca2+ binding properties they have in the intact molecule, with positive cooperativity within each domain. The absolute value of the free energy of interaction between the two sites in each domain, a measure of the cooperativity, increases with ionic strength and is ≥ 10 kJ mol−1 at 0.15 M KCl. Two-dimensional 1H NMR studies show that the addition of KCl does not alter the conformation of the protein. In the case of calbindin D9k several categories of mutants have been studied. One group encompasses the effect of protein surface charges 5 to 15 Å from the Ca2+ binding sites. Two-dimensional 1H NMR shows that neither the addition of KCl, nor mutations that neutralize the surface charges, change the protein conformation. Although the global structure of calbindin D9k is largely unchanged upon binding of calcium, the structure with only one cation bound is more similar to the (Ca2+)2 form. Interestingly, the dynamical properties of the Ca2+-free and the (Ca2+)2-forms of calbindin differ greatly. For example, the rate of NH/ND exchange of the Ca2+-free form is on average 200 times faster than that of the (Ca2+)2-form. The results obtained so far point to a non-negligible entropic contribution to the observed cooperativity of Ca2+ binding.
- Open Access
- 1 February 1997
- Published by Elsevier in Journal of Biological Chemistry
- Vol. 272 (8) , 5105-5111
Model Peptide Studies Demonstrate That Amphipathic Secondary Structures Can Be Recognized by the Chaperonin GroEL (cpn60)
- 1 February 1997
- Published by Elsevier in Journal of Biological Chemistry
- Vol. 272 (8) , 5105-5111
Open AccessModel Peptide Studies Demonstrate That Amphipathic Secondary Structures Can Be Recognized by the Chaperonin GroEL (cpn60)
- 1 May 1996
- Published by Elsevier in Landscape and Urban Planning
- Vol. 34 (3-4) , 347-350
Management and vegetation of coastal sands in New South Wales, Australia, with special reference to Pleistocene beach sands
- 1 May 1996
- Published by Elsevier in Landscape and Urban Planning
- Vol. 34 (3-4) , 347-350
Management and vegetation of coastal sands in New South Wales, Australia, with special reference to Pleistocene beach sands