Prospecting for pH sensors in host and pathogen systems

Human cells contain several types of membrane-enclosed compartments. Such compartmentalisation gives metabolic advantages, isolating regions (organelles) and processes within the cytoplasm of a cell interior, for example genome organisation, energy generation, and protein degradation. Intracellular organelles also allow unequal distributions of small molecules and ions to be set up by pumps made of proteins residing in the membranes. Notably, gradients of pH (differences in the concentration of hydrogen ions) exist between several organelle interiors and the cytoplasm. Functioning of a cell depends on control of pH, as does its relationship, across the outer cell membrane, with the external environment. Two areas in which human health intersects with an understanding of pH are cancer growth and viral infection. Metabolic changes in tumour cells lead to alteration in the balance of pH between cell interior and exterior. Separately, many viruses (including SARS-CoV-2) make use of a low pH organelle to release their genome into a host cell. In both of these diseases modern techniques are being used for large-scale data acquisition. As a tumour grows, mutations accrue (called somatic mutations) that enable adaptation to the altered pH environment, and these are recorded (10,000s) and deposited in databases. Biochemical characterisation of SARS-CoV-2 is occurring at pace, including structural analysis with newly-developed techniques that are particularly suited to large assemblies such as viruses.

Our work addresses the question of how to make a model that predicts the key elements of pH response in biology, i.e. molecular pH sensors. For viruses this would allow prediction of which viruses use low the pH infection route, where their pH sensors are located, and potentially lead to new antiviral strategies. In cancer adaptation to pH, somatic mutations are known, but which directly interact with pH-relevant pathways is mostly unknown. We aim to map mutation to atomic structure, and use the model to predict sites of pH sensing. Early work indicates that such a pipeline will reveal insights into metabolic changes in tumour growth, again with the potential to consider new therapeutic avenues. Predictions will be tested experimentally with a small number of exemplar systems.

The method is a synthesis of mostly available tools for making structural models and predicting pH-dependence from structure. Modellers have access to a hierarchy of techniques, based on their level of detail and computational resource required. Our model will be based in the so-called coarse-grained part of this hierarchy, where the level of detail is scaled back to allow for calculations that are fast enough to provide a quick turnaround, essential for a web application. In this context, pH sensors are suited since they are focussed in a subset of amino acids in proteins. Our aim, through model development, benchmarking with known data, and testing in our own laboratory, is to enable any user, world-wide, to predict pH sensors in systems of interest to them. This approach will enable generation of testable hypotheses from the wealth of genomic and biochemical data being collected.

As measurement technologies advance rapidly in biology, it remains a challenge to find the most productive ways in which molecular simulation and modelling methods can add value. Bioinformatics is crucial for processing and classifying 'omics data. Modelling complements structural analysis, facilitating discovery of the optimal fit to experimental data, latterly in the burgeoning set of cryo-EM structures. It is beyond these points that clear rationale for use of predictive computational methods by experimental groups is often lacking. This project addresses that gap, for pH-dependence and molecular pH sensors. A user-friendly web interface will be developed for predicting pH sensing regions in proteins, from structure or sequence-based structural models. Focus will be on two application areas, with experimental validation. First, many viruses exploit the endosome entry pathway, where acidic pH sensing drives viral genome release. Cryo-EM is fuelling a renewed structural interest of viruses, so that a wealth of data exists for user interrogation of pH sensing. Second, somatic mutations in tumour growth will be mined for those predicted to underpin adaptation in altered intra and extracellular pH environments. Here, large-scale data acquisition (deep sequencing of cancer genomes) couples with metabolic adaptation (and potential therapeutic intervention), via molecular analysis of pH sensing. Calculations on this scale require a coarse-grained approach, central to which are pKa prediction schemes, allied to static and dynamic (elastic network model) analysis of interfaces. Prior to experimental validation of pH sensors, benchmarking will be made against literature biophysical data. Although our focus is on viral infection and tumour adaptation to altered pH, equally important is delivery of a simple to use web tool and downloadable open source code to encourage widespread usage in hypothesis creation.