Monitoring the evolution of drug targets of pathogens to improve the design of therapies
- Autores: David Ferreiro Garcia
- Directores de la Tesis: Miguel Arenas Busto (dir. tes.)
- Lectura: En la Universidade de Vigo ( España ) en 2024
- Idioma: inglés
- Tribunal Calificador de la Tesis: Manuel Becerra Fernández (presid.), Diana Valverde Pérez (secret.), Filipe Adão Macedo Pereira (voc.)
- Programa de doctorado: Programa de Doctorado en Metodología y Aplicaciones en Ciencias de la Vida por la Universidad de Vigo
- Menciones:
- Internacional
- Calificación: Sobresaliente Cum Laude
- Materias:
- Enlaces
- Tesis en acceso abierto en: Investigo
- Resumen
Since the advent of molecular data, multiple studies have aimed to decode the evolutionary relationships between individuals, populations and species. Particularly, pathogens are a central target for researchers, seeking to discover and improve therapies against them. Therapies against pathogens are generally designed from sample-point molecular data and often ignore the molecular evolution of the pathogen drug targets and the population genetics of the pathogen populations. Indeed, some pathogens, such as the human immunodeficiency virus (HIV), can accumulate large genetic diversity in their populations. The evolutionary processes of the pathogens can reduce the efficiency of therapies through the emergence and establishment of resistant pathogen variants in the population. Thus, resistance to therapies remains nowadays a crucial threat to public health, triggering the development of therapies that aim to be active during long periods of time. In this concern, evolutionary biologists usually develop and apply models of evolution to predict molecular evolution. These predictions could be useful because anticipating pathogen variants could improve the design and selection of therapies against pathogens. An example was the development and application of centralized vaccines against HIV-1, like those based on ancestral sequences of the HIV-1 envelope. However, these vaccines were partially effective probably because of the use of unrealistic models of molecular evolution and the rapid and complex evolution of this virus. Another relevant approach was the consideration of the structure of the protein drug targets, in addition to their sequence, to improve the design of drugs and to evaluate the protein fitness. In general, proteins need to maintain their structure to remain active, so many evolutionary processes (i.e. coevolution among protein sites and variation of substitution patterns among sites) can be studied in detail by considering the protein structure. In the present Thesis, I investigated the evolution of the HIV protease (including variants resistant to common drugs) in terms of protein folding stability (Chapter 1), I developed a computational framework for selecting among site-dependent structurally constrained substitution models (SCS) of protein evolution through the approximate Bayesian computation approach (Chapter 2), I developed a substitution model of protein evolution for the HIV protease that considers information from both the folding stability and enzymatic activity (Chapter 3) and, finally, I explored the prediction of evolution of HIV-1 proteins based on a combination of a birth-death population genetics simulation method and SCS models of protein evolution (Chapter 4).
Particularly, in the Chapter 1, I investigated the evolution of the protein folding stability of the HIV-1 protease. This protein was selected because it is a common HIV-1 protein drug target. The evolution of proteins is driven by selection to maintain stable folding conformations. However, in the case of protein drug targets such as the HIV-1 protease, the protein needs to accommodate substitution events that allow resistance against therapies. I conducted a comprehensive study of the evolution of HIV-1 protease folding stability in intra-host and inter-host data from patients without and with treatments. I analyzed the folding stability of present and ancestral (reconstructed) protein variants and I examined the impact of a large number of drug resistance mutations with associated major resistance on the protein folding stability. I found that the stability of the HIV-1 protease fluctuated over time within a constant and wide range in all the studied evolutionary scenarios. I also found that a relevant number of resistant mutations of this protein do not significantly change the folding stability, but most of them produced more instable protein structures.
In order to obtain modeled molecules with biological meaning, it is crucial that the used model of evolution is as realistic as possible. For this reason, the selection of a substitution model of protein evolution is a mandatory step in evolutionary biology. Traditional methods for substitution model selection are based on phylogenetic likelihood. However, mathematical functions to calculate the phylogenetic likelihood are designed for every molecular site and assume that the evolutionary process is independent among sites. This assumption is clearly unrealistic but a number of substitution models of molecular evolution consider it to be included in likelihood-based methods. However, some substitution models that consider constraints on the protein structure involve site-dependent evolution, outperforming the traditional site-independent substitution models in terms of realism, but with the cost of not allowing its use in likelihood-based methods. In the Chapter 2, I developed a method for selecting the best-fitting substitution model among site-independent and site-dependent substitution models using the approximate Bayesian computation (ABC) approach, which is likelihood-free. We found that it accurately selects among site-dependent structurally constrained substitution models and empirical models. Furthermore, I implemented the ABC method into a computational framework called ProteinModelerABC that is freely available.
A number of structurally constrained substitution (SCS) models of protein evolution have been developed in the recent years because they provide a more realistic modeling compared to the traditional empirical substitution models. Most of these SCS models focus on the protein folding stability and ignore the protein activity, despite selection is commonly observed in nature to maintain the protein activity. In the Chapter 3, I developed a substitution model of protein evolution for the HIV-1 protease that considers both the protein folding stability and the enzymatic activity (based on the binding affinity of the enzyme substrate complex). We found that this new substitution model of protein evolution generally provides a better fitting with real data than the best-fitting empirical substitution model and an SCS model that ignores the enzymatic activity.
The development of more realistic substitution models of protein evolution is allowing us to obtain more accurate evolutionary inferences (i.e. phylogenetic reconstructions). Since we believe that accurate substitution models could allow the prediction of future (evolutionarily expected) pathogen variants, and some authors showed acceptable predictions of pathogen forecasting evolution, in the Chapter 4 I investigated the prediction of the evolution of HIV-1 proteins using SCS models. In particular, we combined the birth-death population genetics model (for simulating evolutionary histories) with SCS models (for simulating protein evolution) to forecast protein evolution, and I evaluated it with in-vitro and in vivo data of the HIV-1 matrix and protease proteins. The protocol was implemented by my thesis supervisor and allows the simultaneous simulation of the evolutionary history and molecular evolution based on the protein fitness derived from the folding stability. I found that the difference between the real (present) and the predicted sequences (prediction error) under the SCS model was generally low, especially for the amino acids located at the protein core.
In this PhD Thesis, I highlight the benefits of the development of advanced substitution models of protein evolution and their application to study the evolution of HIV-1 proteins, including a common drug target. In addition, I explored the prediction of evolution of HIV-1 proteins using advanced SCS models and I obtained an acceptable prediction accuracy. The predicted protein variants that are likely to occur in the future could be used to improve the design and selection of therapies in terms of longer duration of effectiveness.
