I am passionate about understanding the role of protein’s flexibility in life and disease: Protein flexibility includes structural adaptation, coordinated multi-domain motions, flexible linkers, disordered regions, and fully disordered proteins. On the other hand, the processes of life and disease include: cell signaling, catalysis, inhibition, activation, coordination, condensation, and many others. Therefore, structural dynamics is intrinsic and necessary to all biomolecules.

Below is a short description of my main research topics. Here, the most relevant articles are highlighted but all my papers are listed in the publications tab.

Protein fuzzy binding in aberrant protein condensates in disease

Structural dynamics is vital for cellular processes; hence, protein flexibility governs cell homeostasis in health and disease. In the last two decades, researchers built the basis for understanding intrinsically disordered proteins and we are now able to investigate protein condensation through the lens of structural biology. Specifically, my research focuses on understanding how fuzzy binding networks control and describe protein-protein interactions [1, 2], protein liquid-liquid phase separation [3], and how single or multiple mutations result in aberrant condensation in disease.

Modeling Intrinsically Disordered Proteins

With Dr Julie Forman-Kay and Dr Teresa Head-Gordon, we are developing a software framework to model proteins with high conformational heterogeneity using experimental data (NMR, SAXS, FRET, etc), Bayesian approaches, A.I., and statistical sampling incorporating new and the state-of-the-art knowledge in intrinsically disordered proteins (IDPs).

We have developed a main software suite, IDPConformerGenerator [1, 2], able to generate highly diverse conformers of peptide chains expected to be disordered. Also, we proposed and implemented a Bayesian framework for ensemble selection that accounts for uncertainties in experimental data [3]. Recently we have leveraged machine learning methods to generate data-accurate ensembles of IDPs and more accurate conformer models [4, 5, 6].

Fuzzy interactions regulating Src Family Kinases

At the BioNMR group in Barcelona, in a highly synergistic group, we described a new regulatory mechanism for Src Family Kinases (SFK). SFK’s activity is regulated by the contraction and expansion of their domains as controlled by localized phosphorylations that render available or hide the active site. Although SFKs have been studied for many decades, under the light of “fuzzy binding” and using advanced NMR techniques, we found the N-terminal disordered Unique Domain of SFK’s transfers signals from the cell membrane to the SFK receptor via a network of fuzzy interactions with the folder SH3 domain. Furthermore, the structural details of the Unique Domain fine-tune these fuzzy networks. For example, we found the 21-residue deletion in SFK Lyn isoform B is enough to change the regulatory behavior of this 512 residue protein. We proposed that such a membrane-UD-SH3 mechanism is common to the SFK family. Find here my publications on how flexible linkers and fuzzy binding networks regulate SFK’s activity.

Role of MMP linkers in collagenolysis

During my Ph.D., we described how the MMP’s flexible linker orchestrates the relative movements MMP’s catalytic and hemopexin folded domains by poising them favorable to cleave the triple helical collagen - a process necessary for the metastatic progression of tumor cells [1, 2]. My work was one of the first to combine several NMR techniques, crystallography, and software from the We-NMR consortium to solve a large biochemical unknown and provide models for MMP-1 free, and collagen-bound, in solution. This research was highlighted in the first issue of the paraNMR FP-7 letters, you can download my Ph.D. thesis from ResearchGate where the MMP-1:Collagen models are presented. I am the author of the MMP-1 dedicated chapter in the Handbook of Proteolytic Enzymes, 2024.

New mechanisms to combat antibiotic resistance

Antibiotic resistance by pathogenic bacteria is likely today’s top concern on health security. Therefore, identifying alternative routes to destroy infectious bacteria are crucial to develop future disease-control strategies. We demonstrated that TomB antitoxin activity is oxygen dependent, and it actively oxidizes the Hha toxin to reduce the latter’s activity through the oxidation of a conserved cysteine residue [1].

Software engineering for biophysical sciences

Because of my continuous interest in software architecture and in sharpening best practices in open-source software, I have led the development of (or significantly contributed to) several software suites applied to biophysical sciences.

I highlight my contribution to the HADDOCK3 project as a lead developer from July 2021 to August 2022, when I brought the project from a proof-of-concept to a fully-featured beta version. HADDOCK is the largest biomolecular docking software in academia with thousands of active users worldwide and is part of the - three-times European funded - BioExcel project. To engage and share with the community, I imparted talks on developing open-source software and program architecture.

In the software tab, you can find the list of software projects I created, participate, and maintain. Of general interest, my template repository is configured with the latest best practices in software development and distribution in Python. Find me also on GitHub.

Understanding COVID-19 infection through computational modeling

In the absence of experimental data, computational modeling allows us to have prior knowledge of protein systems and complexes that can guide experimental analysis and project design. Early on during the devastating COVID-19 pandemic, together with colleagues worldwide, we investigated the cross-species transmission capacity of SARS-CoV-2 [1] by modeling the ACE2 receptors of different domestic animal species with the virus’ RBD protein, thus shedding light on the infections reported in farms and zoos and helping predict possible cross-species transmission. Also, we generated a data set of 242 modeled ACE2 human variants bound to the RDB protein [2]. This data set can boost the design of therapies against Sars-CoV-2 and coronavirus in general by considering different world populations and is available openly online.