Montserrat Soler-López

A network medicine approach to human disease

High‐throughput interaction discovery initiatives are providing thousands of novel protein interactions which are unveiling many unexpected links between apparently unrelated biological processes. In particular, analyses of the first draft human interactomes highlight a strong association between protein network connectivity and disease. Indeed, recent exciting studies have exploited the information contained within protein networks to disclose some of the molecular mechanisms underlying complex pathological processes. These findings suggest that both protein–protein interactions and the networks themselves could emerge as a new class of targetable entities, boosting the quest for novel therapeutic strategies.

Interactome mapping suggests new mechanistic details underlying Alzheimer's disease

Recent advances toward the characterization of Alzheimers disease (AD) have permitted the identification of a dozen of genetic risk factors, although many more remain undiscovered. In parallel, works in the field of network biology have shown a strong link between protein connectivity and disease. In this manuscript, we demonstrate that AD-related genes are indeed highly interconnected and, based on this observation, we set up an interaction discovery strategy to unveil novel AD causative and susceptibility genes. In total, we report 200 high-confidence protein-protein interactions between eight confirmed AD-related genes and 66 candidates. Of these, 31 are located in chromosomal regions containing susceptibility loci related to the etiology of late-onset AD, and 17 show dysregulated expression patterns in AD patients, which makes them very good candidates for further functional studies. Interestingly, we also identified four novel direct interactions among well-characterized AD causative/susceptibility genes (i.e., APP, A2M, APOE, PSEN1, and PSEN2), which support the suggested link between plaque formation and inflammatory processes and provide insights into the intracellular regulation of APP cleavage. Finally, we contextualize the discovered relationships, integrating them with all the interaction data reported in the literature, building the most complete interactome associated to AD. This general view facilitates the analyses of global properties of the network, such as its functional modularity, and triggers many hypotheses on the molecular mechanisms implicated in AD. For instance, our analyses suggest a putative role for PDCD4 as a neuronal death regulator and ECSIT as a molecular link between oxidative stress, inflammation, and mitochondrial dysfunction in AD.

Impact of Methylation on the Physical Properties of DNA

There is increasing evidence for the presence of an alternative code imprinted in the genome that might contribute to gene expression regulation through an indirect reading mechanism. In mammals, components of this coarse-grained regulatory mechanism include chromatin structure and epigenetic signatures, where d(CpG) nucleotide steps are key players. We report a comprehensive experimental and theoretical study of d(CpG) steps that provides a detailed description of their physical characteristics and the impact of cytosine methylation on these properties. We observed that methylation changes the physical properties of d(CpG) steps, having a dramatic effect on enriched CpG segments, such as CpG islands. We demonstrate that methylation reduces the affinity of DNA to assemble into nucleosomes, and can affect nucleosome positioning around transcription start sites. Overall, our results suggest a mechanism by which the basic physical properties of the DNA fiber can explain parts of the cellular epigenetic regulatory mechanisms.

The Proton-Pump Inhibitor Lansoprazole Enhances Amyloid Beta Production

A key event in the pathogenesis of Alzheimer’s disease (AD) is the accumulation of amyloid-β (Aβ) species in the brain, derived from the sequential cleavage of the amyloid precursor protein (APP) by β- and γ-secretases. Based on a systems biology study to repurpose drugs for AD, we explore the effect of lansoprazole, and other proton-pump inhibitors (PPIs), on Aβ production in AD cellular and animal models. We found that lansoprazole enhances Aβ37, Aβ40 and Aβ42 production and lowers Aβ38 levels on amyloid cell models. Interestingly, acute lansoprazole treatment in wild type and AD transgenic mice promoted higher Aβ40 levels in brain, indicating that lansoprazole may also exacerbate Aβ production in vivo. Overall, our data presents for the first time that PPIs can affect amyloid metabolism, both in vitro and in vivo.

dSysMap: exploring the edgetic role of disease mutations

Subject terms: Protein databases• Protein function predictions• Proteins• Systems biology At a glance Figures View all figures Figure 1: The dSysMap resource. (a) Screenshot of the dSysMap representation of pontocerebellar hypoplasia. The interactive network view represents proteins affected by mutations related to the disease and their interactors. Mutations are represented as small colored circles either on the nodes or on the edges (interactions). The position of the small circles in the network reflects their structural location (buried, surface or interface mutations). By clicking on the nodes or edges inside the network, the user can instantly see the corresponding high-resolution structures in the interactive view on the right, in which the mutations are highlighted with a sphere representation. (b) Hypothesis on the edgetic role of two mutations related to pontocerebellar hypoplasia on EXOSC3 (G31A and D132A), which we tested experimentally. The two mutations selectively disrupt the interactions of EXOSC3 with EXOSC5 and EXOSC9 but do not affect its ability to interact with EXOSC1. The different components are highlighted in the structure of the human RNA exosome (PDB 2NN6). Full size image View in article Supplementary Fig. 1: Enrichment of mutated surface residues on known interaction interfaces. The table shows the contingency matrix of surface residues on the structure of human proteins that are mutated in human diseases and belong (or not) to interaction interfaces. Mutated residues on the surface of proteins are more likely to be located at known interaction interfaces. The enrichment is calculated as while the log odds ratio is calculated as The P-value is calculated with a Fishers exact test on the contingency matrix. Standard error is calculated as where A, B, C, D are the corresponding numbers of the contingency matrix (A and B in the first row, C and D in the second). Full size image View in article Supplementary Fig. 2: Effects of mutations on the same protein but on different interaction interfaces and mutations on different proteins but on the two sides of the same interaction interface. (a) The table shows the number of pairs of mutations, on interaction interfaces of the same protein, that are causing different phenotypes and can be found on interfaces mediated different interactions vs. those that mediated the same interaction. Pairs of mutations on the same proteins but on interfaces mediating different interactions are more likely to cause different phenotypes. (b) The table shows the number of pairs of mutations, on interacting proteins that are causing the same phenotype and can be found on the two sides of the same interfaces vs. those that are on interfaces mediating other interactions. Pairs of mutations on different but interacting proteins that are on opposite sides of the same interface are more likely to cause the same phenotype than pairs of mutations that are on different non interacting interfaces. The different quantities are calculated as detailed in the legend of Supplementary Figure 1. Full size image View in article Supplementary Fig. 3: Schematic case of an edgetic effect. Mutations S and T display an edgetic effect on the interaction neighborhood of protein P. (a) While normally P interacts with X, Y and Z, in the presence of mutation S (b), located on the interface of the P-X interaction (but not overlapping with the P-Y and P-Z interfaces), the interaction with X is lost while the other two are preserved. (c) Mutation T, instead, being at the P-Y interface selectively affects the corresponding interaction, while leaving unaltered the other two. Full size image View in article Supplementary Fig. 4: Experimentally tested cases. Every row represents one of the 5 cases that were tested. The first column shows the interactions that were tested and the mutations that can be smapped to the interaction interfaces (small magenta circles). The remaining three columns represent the experimental results. Solid lines represent interactions that could be detected in our Y2H screen. Dashed lines represent wild-type interactions that, although not identified in the Y2H screen, are supported either by crystallographic structures or by other experimental evidence (for example, the CASK-PLK2 interaction is confirmed by a Y2H experiment [Wang et al, Mol Sys Biol 7, 536 (2011)], as deposited in Intact [Kerrien et al, Nucleic Acids Res 40, D841 (2012)], IMEx interaction ID IM-15364-946). Please notice that in some of the cases interactions can be found on an interface that mediate more than one interaction, like for CASK, where the Y268H mutation is predicted to be at the interface with both CDK9 and PLK2 and, therefore, might affect both interactions. See the supplementary text for a description of each case. Nature Full size image View in article Supplementary Fig. 5: Validation of edgetic perturbations. (a) The Y268H mutation on CASK is located at the interface between CASK and CDK9 but is far from the interface with PRKD2. Both interactions are likely to happen through the kinase domain even if they are mediated by different interfaces. Consistently the CASKY268H retains the CASK-PRKD2 interaction but looses the interaction with CDK9. The CASK-CDK9 interaction is modeled after the kinase-kinase domain-domain structural template PDB ID 2wnt_A:B obtained from 3did [Mosca et al, Nucleic Acids Res 42, D374 (2914)], while CASK-PRKD2 interaction is modeled after the template PDB ID 3gok_K:H. (b) Both the ETFA-T266M and the ETFBR164Q mutants loose their ability to dimerize. The ETFB-R164Q mutant maintains its interaction with ACADM. As a matter of fact, the ETFB-R164Q mutation is mapped to a region in the structure of the ETFB-ACADM interaction (PDB ID 2a1t_D:S) that is far from their binding interface, while both T266M and R164Q mutations can be mapped on the binding interface of the ETFA-ETFB dimer (PDB ID 1efv_A:B). (c) The G31A and D132A mutations on EXOSC3 selectively disrupt its interactions with EXOSC5 and EXOSC9 but do not affect its ability to interact with EXOSC1. The different components are highlighted in the structure of the human RNA exosome (PDB ID 2nn6). Two separate views show (1) the interaction of EXOSC3 (pink, chain G) with EXOSC5 (light orange, chain D) and EXOSC9 (purple, chain A) and (2) the interaction of EXOSC3 with EXOSC1 (dark gray, chain I). The G31A mutation is located on the EXOSC3-EXOSC5 interface, while the D132A mutation is located on the EXOSC3-EXOSC9 interface. Both mutations are far from the EXOSC3-EXOSC1 interface. Full size image View in article Supplementary Fig. 6: Disease map associated with the Loeys-Dietz syndrome and connected disease phenotypes. The disease map associated to the Loeys-Dietz syndrome (LDS) is an example of annotated network reported by dSysMap (f). The network represents the proteins affected by mutations related to LDS and their first interactors. Different colors for the nodes and edges represent the different types of structural information available for proteins and interactions as detailed in the legend. Small colored circles on the edges and proteins represent mutations mapped on proteins and interactions. Three cases are enlarged from the network. (a) Interaction neighborhood of TGFBR1 (corresponding to the orange shaded circle in the network); structures are available for the kinase and the activin receptor domains. Mutated residues are shown in spheres-representation and colored based on their structural classification (black for buried residues, blue for surface residues and red for interface residues). The Activin receptor domain mediates the interactions with TGFB1, TGFB3 and TGFBR2. The kinase domain mediates the TGFBR1 homomeric interaction, the interaction with FKBP1A and is predicted to mediate the interaction with BMPR2. Red circles highlight the mutations that affect every interaction. (b) Interaction neighborhood of ACVR1 (corresponding to the gray shaded circle in the network). Mutations related to Fibrodysplasia ossificans progressive are located at the ACVR1 homodimeric interface, at the interface with FKBP1A and at the predicted interface with BMPR2 but not at the interface of the homology model for the ACVR1-TGFBR2 interaction. (c) Interaction neighborhood of SMAD3 (corresponding to the pink shaded oval in the network). Mutations for LDS on SMAD3 are found at the interface with SMAD4 (d). Also mutations for Juvenile polyposis syndrome on SMAD4 are found at the interface with SMAD3, even if on a different structural model (e). Full size image View in article

Towards Alzheimer's root cause: ECSIT as an integrating hub between oxidative stress, inflammation and mitochondrial dysfunction

Here we postulate that the adapter protein evolutionarily conserved signalling intermediate in Toll pathway (ECSIT) might act as a molecular sensor in the pathogenesis of Alzheimers disease (AD). Based on the analysis of our AD‐associated protein interaction network, ECSIT emerges as an integrating signalling hub that ascertains cell homeostasis by the specific activation of protective molecular mechanisms in response to signals of amyloid‐beta or oxidative damage. This converges into a complex cascade of patho‐physiological processes. A failure to repair would generate severe mitochondrial damage and ultimately activate pro‐apoptotic mechanisms, promoting synaptic dysfunction and neuronal death. Further support for our hypothesis is provided by increasing evidence of mitochondrial dysfunction in the disease etiology. Our model integrates seemingly controversial hypotheses for familial and sporadic forms of AD and envisions ECSIT as a biomarker to guide future therapies to halt or prevent AD.

Physical properties of naked DNA influence nucleosome positioning and correlate with transcription start and termination sites in yeast

BackgroundIn eukaryotic organisms, DNA is packaged into chromatin structure, where most of DNA is wrapped into nucleosomes. DNA compaction and nucleosome positioning have clear functional implications, since they modulate the accessibility of genomic regions to regulatory proteins. Despite the intensive research effort focused in this area, the rules defining nucleosome positioning and the location of DNA regulatory regions still remain elusive.ResultsNaked (histone-free) and nucleosomal DNA from yeast were digested by microccocal nuclease (MNase) and sequenced genome-wide. MNase cutting preferences were determined for both naked and nucleosomal DNAs. Integration of their sequencing profiles with DNA conformational descriptors derived from atomistic molecular dynamic simulations enabled us to extract the physical properties of DNA on a genomic scale and to correlate them with chromatin structure and gene regulation. The local structure of DNA around regulatory regions was found to be unusually flexible and to display a unique pattern of nucleosome positioning. Ab initio physical descriptors derived from molecular dynamics were used to develop a computational method that accurately predicts nucleosome enriched and depleted regions.ConclusionsOur experimental and computational analyses jointly demonstrate a clear correlation between sequence-dependent physical properties of naked DNA and regulatory signals in the chromatin structure. These results demonstrate that nucleosome positioning around TSS (Transcription Start Site) and TTS (Transcription Termination Site) (at least in yeast) is strongly dependent on DNA physical properties, which can define a basal regulatory mechanism of gene expression.

Fuzziness and noise in nucleosomal architecture

Nucleosome organization plays a key role in the regulation of gene expression. However, despite the striking advances in the accuracy of nucleosome maps, there are still severe discrepancies on individual nucleosome positioning and how this influences gene regulation. The variability among nucleosome maps, which precludes the fine analysis of nucleosome positioning, might emerge from diverse sources. We have carefully inspected the extrinsic factors that may induce diversity by the comparison of microccocal nuclease (MNase)-Seq derived nucleosome maps generated under distinct conditions. Furthermore, we have also explored the variation originated from intrinsic nucleosome dynamics by generating additional maps derived from cell cycle synchronized and asynchronous yeast cultures. Taken together, our study has enabled us to measure the effect of noise in nucleosome occupancy and positioning and provides insights into the underlying determinants. Furthermore, we present a systematic approach that may guide the standardization of MNase-Seq experiments in order to generate reproducible genome-wide nucleosome patterns.

Nucleosome architecture throughout the cell cycle

Nucleosomes provide additional regulatory mechanisms to transcription and DNA replication by mediating the access of proteins to DNA. During the cell cycle chromatin undergoes several conformational changes, however the functional significance of these changes to cellular processes are largely unexplored. Here, we present the first comprehensive genome-wide study of nucleosome plasticity at single base-pair resolution along the cell cycle in Saccharomyces cerevisiae. We determined nucleosome organization with a specific focus on two regulatory regions: transcription start sites (TSSs) and replication origins (ORIs). During the cell cycle, nucleosomes around TSSs display rearrangements in a cyclic manner. In contrast to gap (G1 and G2) phases, nucleosomes have a fuzzier organization during S and M phases, Moreover, the choreography of nucleosome rearrangements correlate with changes in gene expression during the cell cycle, indicating a strong association between nucleosomes and cell cycle-dependent gene functionality. On the other hand, nucleosomes are more dynamic around ORIs along the cell cycle, albeit with tighter regulation in early firing origins, implying the functional role of nucleosomes on replication origins. Our study provides a dynamic picture of nucleosome organization throughout the cell cycle and highlights the subsequent impact on transcription and replication activity.

Unravelling the hidden DNA structural/physical code provides novel insights on promoter location

Although protein recognition of DNA motifs in promoter regions has been traditionally considered as a critical regulatory element in transcription, the location of promoters, and in particular transcription start sites (TSSs), still remains a challenge. Here we perform a comprehensive analysis of putative core promoter sequences relative to non-annotated predicted TSSs along the human genome, which were defined by distinct DNA physical properties implemented in our ProStar computational algorithm. A representative sampling of predicted regions was subjected to extensive experimental validation and analyses. Interestingly, the vast majority proved to be transcriptionally active despite the lack of specific sequence motifs, indicating that physical signaling is indeed able to detect promoter activity beyond conventional TSS prediction methods. Furthermore, highly active regions displayed typical chromatin features associated to promoters of housekeeping genes. Our results enable to redefine the promoter signatures and analyze the diversity, evolutionary conservation and dynamic regulation of human core promoters at large-scale. Moreover, the present study strongly supports the hypothesis of an ancient regulatory mechanism encoded by the intrinsic physical properties of the DNA that may contribute to the complexity of transcription regulation in the human genome.