Biology subjects

Correia, B.

Publications and source records attributed to Correia, B..

At least 19 recordsLinked to original sources

bioRxiv2026

A Unified 3D Generative Model for Synthesizable Structure-Based Drug Design

Traditional screening-based drug discovery is inherently limited by the astronomical scale of the chemical space. Generative modelling offers a compelling alternative to the classical search paradigm and enables rational, bottom-up design of novel and target-specific small molecules. However, its impact has been hampered by challenges in synthetic accessibility of the designed compounds and lack of large-scale experimental validation. Here, we introduce LDDM (Large Drug Discovery Model), a generative framework that supports a range of drug discovery tasks, including constrained and unconstrained docking, fragment linking and growing, and de novo design. We further introduce a programmable design algorithm that enables accurate design of synthetically accessible compounds satisfying various fine-grained objectives. We experimentally validated the designed or optimised ligands for five therapeutically relevant protein targets. In all cases, LDDM achieved high success rates, allowing us to identify molecules with confirmed binding affinity while synthesizing only a small number of generated compounds. The best designs were structurally characterised through NMR spectroscopy and X-ray crystallography, demonstrating high prediction accuracy. Overall, LDDM provides a scalable and flexible platform for the rapid and tailored design of small molecules and non-natural peptides for therapeutic applications.

bioinformatics↗

bioRxiv2026

Computational design of microbial and animal rhodopsin soluble analogues

The computational design of soluble analogues of membrane proteins has unlocked exciting opportunities for the integration of unique membrane protein functions into soluble proteins. Here, we use AF2seq to generate accurate soluble analogues of both animal and microbial rhodopsins, based on the membrane GPCR topology, and the microbial rhodopsin transmembrane fold. We characterize the analogues and demonstrate that they are well-folded and highly thermostable. Furthermore, they exhibit the expected red shift characteristic of retinal binding. Top-down mass spectrometry confirms the placement of retinal covalent attachment, while X-ray crystallography validates the structural fidelity of the microbial rhodopsin analogue. Notably, the microbial rhodopsin analogue retains the primary reaction of the retinal photocycle, closely matching that of the native membrane protein. Overall, this work advances the possibility to transfer unique membrane protein functions, such as retinal photoswitching, into the soluble proteome.

Synthetic Biology↗

bioRxiv2026

Induced Estrogen Receptor SUMOylation drives SERD activity

The ligand dependent transcription factor estrogen receptor (ER) is a key driver of and important drug target in breast cancer. Patients with advanced disease are typically treated with selective ER degraders (SERDs), whose therapeutic activity is commonly attributed to induced ER protein degradation. Yet the exact mechanism and relevance of degradation for clinical efficacy remain unclear. We show that SERDs directly induce ER SUMOylation, thereby triggering degradation via SUMO-targeted ubiquitin ligases (STUbLs). Inactivation of STUbLs prevents ER degradation and counterintuitively further sensitizes breast cancer cells to SERDs, rather than conferring resistance. SERD efficacy is independent of ER degradation, challenging the degradation-centric model of SERD action. Instead, SUMOylation recruits transcriptional co-repressors, turning SUMOylated ER into a dominant-negative repressor. Thus, SUMOylation rather than degradation is the direct consequence and driver of SERD activity. Our findings position SUMO-inducing drugs as a hitherto underappreciated yet clinically validated therapeutic modality with broad applicability.

Cancer Biology↗

bioRxiv2025

FragmentScope - exploring the fragment space with learned surface representations

Exploring fragment chemical space for ligand design remains a major challenge in early stage drug discovery. This task is particularly challenging due to the small size, low specificity, and weak binding affinities of low molecular weight (MW) fragments. We present FragmentScope, a computational pipeline that uses learned protein surface fingerprints to guide fragment placement and small molecule generation. By using a contrastive learning model trained on protein-ligand interactions, we built a database of surface-fragment pairs, which enables fast and accurate fragment placement in the target protein pocket. We demonstrate its effectiveness on benchmark datasets, achieving robust placement accuracy. FragmentScope also enables the design of small molecules based on the predicted fragments ensuring synthetic accessibility. We experimentally validated Fragmentscope across 5 different targets with binding assays and structural characterization. Our approach shows high success rates in fragment discovery and yielded promising leads for designed ligands. FragmentScope offers a scalable, structure-guided approach for narrowing chemical space and identifying prospective scaffolds, accelerating the early stages of small molecule design.

bioinformatics↗

bioRxiv2025

Leveraging protein representations to explore uncharted fold spaces with generative models

A major challenge in computational de novo protein design is the exploration of uncharted areas within the protein structural space. However, the large degrees of freedom of protein backbones complicate the sampling process during protein design. Machine learning-based models have made great strides in this problem, however due to their nature they tend to exploit rather than explore the data distribution used for training the neural networks. To address some of these challenges, we propose a new coarse grained protein structure representation generative method, DiffTopo, a diffusion model which increases the sampling efficiency and diversity. Combined with a backbone level protein generative model like RFdiffusion, novel protein folds can be generated rapidly, allowing for efficient exploration of the designable topology space. Interestingly, we have discovered that by mirroring the topological organization of native proteins using a pipeline named MirrorTopo, we can readily expand the known fold space. We generated and experimentally characterized 30 different novel topologies from DiffTopo and 6 different novel mirror topologies from MirrorTopo. The developed framework relying on low resolution sampling provides new means for fold exploration challenges, which could in principle enhance our knowledge of the first principles of protein structure and folding, as well as create new opportunities for functional design.

bioinformatics↗

bioRxiv2025

Rational design of chemically responsive cytokines for cancer immunotherapy

Cytokines are key signal mediators of the immune system, playing essential roles in regulating central immunological processes. Despite their unique ability to modulate the immune system, the translation of cytokine-based therapies to the clinic has been significantly hindered by severe toxicities resulting from the pleiotropy and off-target effects of many cytokines. Here, we present a general strategy for the design of switchable cytokines that enables precise control over cytokine activity using the clinically approved drug, Venetoclax. We rationally designed self-inhibited Switchable InterLeukins (SwILs) by embedding the chemically controlled interface (Bcl-2: BIM-BH3) into the cytokines structure, resulting in conditional cytokines activated by Venetoclax. We showed the broad applicability of this strategy across various cytokines, including IL-2, IL-15, and IL-10. In vivo studies with IL-15 demonstrated that the SwILs achieved Venetoclax-dependent tumor control comparable to that of the native cytokine. Additionally, we showed that this strategy can be expanded to respond to other tumor-intrinsic stimuli, such as tumor-specific proteases. Overall, SwILs offer control of cytokine activity, enhancing the safety and clinical applicability of cytokine-based therapeutics.

bioengineering↗

bioRxiv2025

Molecular basis for anti-jumbo phage immunity by AVAST Type 5

Jumbo phages protect their genomes from DNA-sensing bacterial defense systems by enclosing them within vesicles and nucleus-like compartments. Very little is known about defense systems specialized to counter these phages. Here, we show that AVAST Type 5 (Avs5) systems, part of the STAND superfamily and spanning three phylogenetic Avs5 clades, confer conserved immunity against jumbo phages. Using localization microscopy and biotin proximity labeling we demonstrate that Avs5 localizes to early infection vesicles, where it senses an essential, early expressed phage protein named JADA--Jumbophage AVAST5 Defense Activator. Recognition of phage infection triggers the Sir2-like effector domain of Avs5 across all three Avs5 clades, resulting in rapid NAD+ hydrolysis, disruption of phage nucleus formation, and arrest of infection. These findings reveal a spatially coordinated bacterial immune strategy that targets an early vulnerability in jumbo phage infection.

microbiology↗

bioRxiv2025

Humanized Caffeine-Inducible Systems for Controlling Cellular Functions

Current cell therapies are limited by the lack of tools for controlling gene expression using humanized systems responsive to non-toxic stimuli. Starting from nanobodies that homodimerize in response to caffeine, we computationally designed inducible heterodimers and humanized the best-performing pairs. We used the resulting caffeine-inducible domains in engineered cytokine receptors for caffeine-inducible STAT3 signaling and in split transcription factors (caff-TFs) containing human-derived zinc-finger proteins. Heterodimerization of split transcription factors drastically enhanced their performance compared to homodimerization. We demonstrate that caff-TFs are compatible with lentiviral and retroviral delivery to Jurkat T-cells and enable inducible expression of therapeutic genes such as chimeric antigen receptors (CARs) in response to caffeine concentrations consistent with normal coffee consumption. By using the common, non-toxic molecule caffeine, and exclusively humanized protein components, these systems promise to be a safer alternative to existing systems and may be used in synthetic biology applications and for safer, more effective cell therapies.

synthetic biology↗

bioRxiv2025

BoltzDesign1: Inverting All-Atom Structure Prediction Model for Generalized Biomolecular Binder Design

Deep learning in structure prediction has revolutionized protein research, enabling large-scale screening, novel hypothesis generation, and accelerated experimental design across biological domains. Recent advances, including RoseTTAFold-AA and AlphaFold3, have extended structure prediction models to work with small molecules, nucleic acids, ions, and covalent modifications. We present BoltzDesign1, which inverts the Boltz-1 model, an open source reproduction of AlphaFold3, to enable the design of protein binders for diverse molecular targets without requiring model finetuning. By utilizing only the Pairformer and Confidence modules, our method significantly reduces computational costs while achieving outstanding in silico success rates and diversity in binder generation. Optimizing directly on the distogram allows us to shape the probability distribution of atomic distances, rather than adjusting a single structure, steering the design toward sequences that yield robust structures with well-defined energy minima. By leveraging a fully atomic model trained on a wide variety of macromolecules, we can generate diverse heterocomplexes with flexible ligand conformations--a capability not currently matched by existing methods. This approach enables the design of novel protein interactions with potential applications in biosensors, enzyme engineering, therapeutic development, and biotechnological innovations.

molecular biology↗

bioRxiv2025

Enhancing protein graph representations through static-dynamic fusion

Machine learning for protein modeling faces significant challenges due to proteins inherently dynamic nature, yet most graph-based machine learning methods rely solely on static structural information. Recently, the growing availability of molecular dynamics trajectories provides new opportunities for understanding the dynamic behavior of proteins; however, computational methods for utilizing this dynamic information remain limited. We propose a novel graph representation that integrates both static structural information and dynamic correlations from molecular dynamics trajectories, enabling more comprehensive modeling of proteins. By applying relational graph neural networks (RGNNs) to process this heterogeneous representation, we demonstrate significant improvements over structure-based approaches across three distinct tasks: atomic adaptability prediction, binding site detection, and binding affinity prediction. Our results validate that combining static and dynamic information provides complementary signals for understanding proteinligand interactions, offering new possibilities for drug design and structural biology applications.

bioinformatics↗

bioRxiv2025

Computationally designed stem-epitope mimetics elicit broadly reactive antibodies

Broad protection against diverse influenza viruses can be conferred by broadly neutralizing antibodies (bnAbs) targeting a conserved site on the hemagglutinin (HA) stem domain. However, the low immunogenicity of this antigenic region hinders the robust induction of such antibodies. Here, we showcase a structure-based immunogen design strategy focusing on the surface mimicry of antigenic sites. By leveraging the structural definition of a stem epitope, we apply computational protein design to develop epitope mimetics to focus the immune response against this site of viral vulnerability. The structurally complex antigenic site is displayed on heterologous protein scaffolds, retaining excellent binding towards known HA stem-specific bnAbs. Our epitope-mimetic induces stem-specific antibodies against highly divergent group 1 and 2 subtypes. The results provide a general framework for the design of novel immunogens eliciting focused immune responses which may be a valuable tool in the development of effective vaccine candidates against other variable pathogens.

biochemistry↗

bioRxiv2024

Structure-based Design of Chimeric Influenza Hemagglutinins to Elicit Cross-group Immunity

Antigenic variability among influenza virus strains poses a significant challenge to developing broadly protective, long-lasting vaccines. Current annual vaccines target specific strains, requiring accurate prediction for effective neutralization. Despite sequence diversity across phylogenetic groups, the hemagglutinin (HA) head domains structure remains highly conserved. Utilizing this conservation, we designed cross-group chimeric HAs that combine antigenic surfaces from distant strains. By structure-guided transplantation of receptor-binding site (RBS) residues, we displayed an H3 RBS on an H1 HA scaffold. These chimeric immunogens elicit cross-group polyclonal responses capable of neutralizing both base and distal strains. Additionally, the chimeras integrate heterotrimeric immunogens, enhancing modular vaccine design. This approach enables the inclusion of diverse strain segments to generate broad polyclonal responses. In the future, such modular immunogens may serve as tools for evaluating immunodominance and refining immunization strategies, offering potential to bridge and enhance immune responses in individuals with pre-existing immunity. This strategy holds promise for advancing universal influenza vaccine development. O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=105 SRC="FIGDIR/small/628867v2_ufig1.gif" ALT="Figure 1"> View larger version (25K): org.highwire.dtl.DTLVardef@18dec5borg.highwire.dtl.DTLVardef@98587forg.highwire.dtl.DTLVardef@1d9fb2dorg.highwire.dtl.DTLVardef@1f97884_HPS_FORMAT_FIGEXP M_FIG Graphical abstract: Overview of cross-group RBS transplantation approachPhylogenetically diverse HA strains can be incorporated into chimeric immunogens by RBS transplantation. The chimera are evaluated for cross-reactivity to subtype-specific antibodies and the ability to elicit neutralizing antibodies to multiple strains. C_FIG

biochemistry↗

bioRxiv2024

Mapping targetable sites on the human surfaceome for the design of novel binders

The human cell surfaceome, integral to cell communication and disease mechanisms, presents a prime target for therapeutic intervention. De novo protein binder design against these cell surface proteins offers a promising yet underexplored strategy for drug development. However, the vast search space and limited data on natural or competitive binders have historically limited experimental success. In this study, we systematically analyzed the entire human surfaceome, identifying approximately 4,500 targetable sites and introducing high-quality binder seeds tailored for protein design applications. To validate these seeds, we implemented two experimental approaches (protein scaffolding and peptide cyclization) on three representative targets (FGFR2, IFNAR2, and HER3). Our results revealed a high success rate, emphasizing the precision and therapeutic potential of these seeds, as well as the need for constant improvements of computational protein design pipelines utilizing machine learning and physics-based methods. Additionally, we present SURFACE-Bind, an interactive database offering open access to all generated data. The high-throughput computational design methods and target-specific binder seeds established here pave the way for a new generation of targeted therapeutics for the human surfaceome.

synthetic biology↗

bioRxiv2024

Drug-controlled CAR-T cells through the regulation of cell-cell interactions

CAR T-cell therapy is constrained by on-target, off-tumor toxicities as well as cellular exhaustion due to chronic antigen exposure. CARs comprising small-molecule controlled switches can enhance both safety and therapeutic efficacy but are limited by the scarcity of non-immunogenic protein elements responsive to non-immunosuppressive, clinically approved drugs with favorable pharmacodynamics. Here, we combine rational design and library-based optimization of a protein-protein interaction (PPI) of human origin to develop venetoclax-controlled Drug-Regulated Off-switch PPI (DROP)-CARs. DROP-CARs enable dose-dependent release of the tumor-targeting scFv and consequent T-cell dissociation from the target tumor cell. Additionally, we present proof-of-concept for a dual DROP-CAR controlled by different small molecules, as well as for logic-gated synthetic receptors enabling STAT3 signaling. We demonstrate in vitro and in vivo function of DROP-CAR T cells and conclude that the approach holds promise for clinical application.

synthetic biology↗

bioRxiv2024

DiffTopo: Fold exploration using coarse grained protein topology representations

AO_SCPLOWBSTRACTC_SCPLOWA major challenge in the field of computational de novo protein design is the exploration of uncharted areas within protein structural space, i.e., generating "designable" protein structures that nature has not explored. However, the large degrees of freedom of protein structural backbones complicate the sampling process during protein design. In this work, we propose a new coarse grained protein structure representation method DiffTopo - an E(3) Equivariant 3D conditional diffusion model, which greatly increases the sampling efficiency. Combined with the RFdiffusion framework, novel protein folds can be generated rapidly, allowing for efficient exploration of the designable topology space. This opens up possibilities to solve the problem of generating new folds as well to functionalize de novo proteins through motif scaffolding, where functional or enzymatic sites can be introduced into novel protein frameworks.

biophysics↗

bioRxiv2023

A modular toolbox for the optogenetic deactivation of transcription

Light-controlled transcriptional activation is a commonly used optogenetic strategy that allows researchers to regulate gene expression with high spatiotemporal precision. The vast majority of existing tools are, however, limited to light-triggered induction of gene expression. Here, we inverted this mode of action and created two complementary optogenetic systems capable of efficiently terminating transcriptional activation in response to blue light. First, we designed highly compact regulators, by photo-controlling VP16 transactivation peptide exposure. Then, applying a two-hybrid strategy, we engineered LOOMINA (light off-operated modular inductor of transcriptional activation), a versatile transcriptional control platform for mammalian cells that is highly adaptable and compatible with various effector proteins. Leveraging the flexibility of CRISPR systems, we integrated LOOMINA with Cas9 as a DNA-binding domain to control transcription from various endogenous promoters with exceptionally high dynamic ranges in multiple cell lines, including neuron-like cells. Both functionally and mechanistically, LOOMINA represents a valuable addition to the optogenetic repertoire for transcriptional regulation.

synthetic biology↗

bioRxiv2023

Exploring "dark matter" protein folds using deep learning

De novo protein design aims to explore uncharted sequence-and structure areas to generate novel proteins that have not been sampled by evolution. One of the main challenges in de novo design involves crafting "designable" structural templates that can guide the sequence search towards adopting the target structures. Here, we present an approach to learn patterns of protein structure based on a convolutional variational autoencoder, dubbed Genesis. We coupled Genesis with trRosetta to design sequences for a set of protein folds and found that Genesis is capable of reconstructing native-like distance-and angle distributions for five native folds and three novel, so-called "dark-matter" folds as a demonstration of generalizability. We used a high-throughput assay to characterize protease resistance of the designs, obtaining encouraging success rates for folded proteins and further biochemically characterized folded designs. The Genesis framework enables the exploration of the protein sequence and fold space within minutes and is not bound to specific protein topologies. Our approach addresses the backbone designability problem, showing that structural patterns in proteins can be efficiently learned by small neural networks and could ultimately contribute to the de novo design of proteins with new functions.

bioinformatics↗

bioRxiv2023

Computational design of soluble analogues of integral membrane protein structures

De novo design of complex protein folds using solely computational means remains a significant challenge. Here, we use a robust deep learning pipeline to design complex folds and soluble analogues of integral membrane proteins. Unique membrane topologies, such as those from GPCRs, are not found in the soluble proteome and we demonstrate that their structural features can be recapitulated in solution. Biophysical analyses reveal high thermal stability of the designs and experimental structures show remarkable design accuracy. The soluble analogues were functionalized with native structural motifs, standing as a proof-of-concept for bringing membrane protein functions to the soluble proteome, potentially enabling new approaches in drug discovery. In summary, we designed complex protein topologies and enriched them with functionalities from membrane proteins, with high experimental success rates, leading to a de facto expansion of the functional soluble fold space.

bioinformatics↗