Transformaciones alquímicas de no-equilibrio aplicadas a la ingeniería de proteínas.
Loading...
Date
2025
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Universidad de Concepción
Abstract
La ingeniería de proteínas es una disciplina que busca diseñar, modificar y optimizar proteínas con funciones específicas. En este campo, la ingenie ría de enzimas es una de las áreas más importantes debido a la relevancia de estos biocatalizadores en la biotecnología, la industria y la medicina. Ex perimentalmente, estas nuevas enzimas suelen ser obtenidas a través de la evolución dirigida, realizando múltiples ensayos de prueba y error. Debido al alto costo monetario y capital humano que implica realizar tales búsquedas aleatorias, es que nace el diseño racional de enzimas. Estos enfoques utilizan métodos computacionales basados en principios fisicoquímicos para postular variantes con propiedades específicas. Estas propiedades pueden clasificarse en termodinámicas y cinéticas, donde estas últimas definen la velocidad y el mecanismo por el cual ocurre una reacción, requiriendo información acerca de la ruptura y formación de enlaces. Debido a esto, los métodos computacionales utilizados deben considerar explícitamente los grados de libertad electrónicos, lo que limita las opciones a enfoques como cálculos de mecánica cuántica o simulaciones multiestados QM/MM. El alto costo computacional de estas técnicas limita el número de variantes a estudiar a sólo unos pocos ejemplares.
En esta tesis se desarrolló una nueva metodología basada en transformaciones alquímicas de no-equilibrio, para obtener el cambio en la energía libre de activación ∆∆G‡, producto de mutaciones. Este nuevo método es al menos un orden de magnitud más rápido que las metodologías QM/MM actualmente establecidas, debido a que basa sus fundamentos en simulaciones de mecánica molecular.
En las transformaciones alquímicas es necesario describir las interacciones del estado de transición a través de la mecánica molecular. Para lograr esto, se propuso un esquema de derivación de parámetros enlazantes basado en la matriz Hessiana del estado de transición y una modificación del método de Seminario. Los términos no enlazantes por su parte fueron derivados utilizando la densidad electrónica y el esquema de partición MBIS, basado en la teoría de átomos-en-moléculas (AIM).
Las transformaciones alquímicas en conjunto con la derivación de campos de fuerza fueron validadas en dos sistemas enzimáticos: Crotonil-CoA Carboxilasa/Reductasa (CCR) y Dihidrofolato Reductasa (DHFR). Los resultados mostraron que las transformaciones alquímicas logran reproducir tanto valores experimentales, como sus tendencias, con errores cercanos a la precisión química (1kcal/mol). Además, se evaluó la modificación de los parámetros no enlazantes de aminoácidos, logrando una mejora de hasta 3 kJ/mol en DHFR.
Una vez validada la metodología se aplicó a una enzima diseñada artificialmente para catalizar una reacción aldólica: la variante de Old Yellow Enzyme(OYE) Y187F. El mecanismo de esta nueva reacción aldólica en OYE se determinó mediante simulaciones multiestado QM/MM, utilizando el método de la cuerda adaptativa. Los resultados mostraron que la coenzima de mono nucleótido de flavina (FMN)transfiere un hidruroal sustrato 2-ciclohexenona, generando un enolato. Este intermediario a su vez ataca a una molécula de formaldehido, para formar la 2-metoxiciclohexenona posterior a su protonación. El camino de mínima energía libre obtenido por el método de la cuerda reveló que el paso limitante de la reacción era la transferencia de hidruro.
Identificado el paso limitante y caracterizado el estado de transición, se realizaron transformaciones alquímicas de no-equilibrio en OYE Y187F, para encontrar variantes catalíticamente más activas. Los resultados predigieron al menos 2 variantes que reducen la energía libre de activación, de las cuales, la doble mutante Y187F/T130W viene a ser la variante más prometedora con un ∆∆G‡ de-8.8 kJ/mol.
En resumen, esta tesis logró demostrar que las transformaciones alquímicas de no-equilibrio, permiten predecir variaciones de la constante de velocidad de reacción catalítica de enzimas con una precisión similar a la de métodos basados en primeros principios, pero al menos, a un décimo de su costo computacional. Esto abre la posibilidad de realizar un análisis a gran escala, lo que facilitaría la búsqueda de nuevas variantes enzimáticas en el campo de la ingeniería de proteínas.
Protein engineering is a discipline that designs, modifies, and optimizes proteins for specific functions. Within this field, enzyme engineering is one of the most important areas due to the relevance of these biocatalysts in biotech nology, industry, and medicine. Experimentally, new enzymes are typically obtained through directed evolution, performing multiple trial-and-error es says. Given the human and financial resources required for such random pro cesses, rational enzyme design has emerged. This approach uses computatio nal methods and physical principles to propose variants with specific proper ties. Since kinetic properties require knowledge of the reaction mechanism, computational methods are limited to first-principles-based approaches, such as quantum mechanics (QM) calculations or multi-state QM/MM simula tions. However, their high computational cost restricts the number of variants to be studied to only a few. This thesis proposes a new methodology based on non-equilibrium alche mical transformations to obtain the change in activation free energy, ∆∆G‡, induced by mutations. This new method is at least one order of magnitude faster than the currently well-established QM/MM methodologies, as it relies on purely classical molecular mechanics. The alchemical transformations involve describing molecular interactions of transition states with molecular mechanics. To achieve this, a bonding pa rameter derivation scheme is proposed based on the Hessian matrix of the transition state and the modified Seminario method. Non-bonded interaction parameters are derived using the electron density of substrates and the MBIS partitioning scheme, based on the Atoms-in-Molecules (AIM) theory. The alchemical transformations and force field parameterization were validated in two enzymatic systems: Crotonyl-CoA Carboxylase/Reductase (CCR) and Dihydrofolate Reductase (DHFR). The results show that the al chemical transformations successfully reproduce both experimental values and trends, with errors close to chemical accuracy (1 kcal/mol). Additionally, modifications to the non-bonded parameters of mutating amino acids in the alchemical transformations are evaluated, improving predictions by up to 3 kJ/mol in the DHFR system. Once the methodology was validated, it was applied to an artificially de signed enzyme: the Y187F variant of Old Yellow Enzyme (OYE). Since this artificial variant catalyzes a new-to-nature aldol reaction in OYE, multi-state QM/MM simulations were performed using the adaptive string method to disclose the reaction mechanism. The results show that a flavin mononucleo tide (FMN) coenzyme transfers a hydride to 2-cyclohexenone, generating an enolate. This intermediate subsequently attacks formaldehyde molecule leading to 2-methoxycyclohexenone after protonation. The minimum free energy path obtained using the string method indicates that the rate-limiting step of the reaction is the hydride transfer. After identifying the rate-limiting step and characterizing the transition state, non-equilibrium alchemical transformations were performed on OYE Y187F to find catalytically more active variants. The results predict at least twovariants that reduce the activation free energy. The Y187F/T130W double mutant is the most promising candidate, achieving a ∆∆G‡ of-8.8 kJ/mol. In summary,this thesis demonstrates that non-equilibrium alchemical trans formations predict enzyme kinetic properties with a precision comparable to first-principles-based methods but at one-tenth of the computational cost. This breakthrough enables high-throughput analyses, fascilitating the disco very of new enzymatic variants in the field of protein engineering.
Protein engineering is a discipline that designs, modifies, and optimizes proteins for specific functions. Within this field, enzyme engineering is one of the most important areas due to the relevance of these biocatalysts in biotech nology, industry, and medicine. Experimentally, new enzymes are typically obtained through directed evolution, performing multiple trial-and-error es says. Given the human and financial resources required for such random pro cesses, rational enzyme design has emerged. This approach uses computatio nal methods and physical principles to propose variants with specific proper ties. Since kinetic properties require knowledge of the reaction mechanism, computational methods are limited to first-principles-based approaches, such as quantum mechanics (QM) calculations or multi-state QM/MM simula tions. However, their high computational cost restricts the number of variants to be studied to only a few. This thesis proposes a new methodology based on non-equilibrium alche mical transformations to obtain the change in activation free energy, ∆∆G‡, induced by mutations. This new method is at least one order of magnitude faster than the currently well-established QM/MM methodologies, as it relies on purely classical molecular mechanics. The alchemical transformations involve describing molecular interactions of transition states with molecular mechanics. To achieve this, a bonding pa rameter derivation scheme is proposed based on the Hessian matrix of the transition state and the modified Seminario method. Non-bonded interaction parameters are derived using the electron density of substrates and the MBIS partitioning scheme, based on the Atoms-in-Molecules (AIM) theory. The alchemical transformations and force field parameterization were validated in two enzymatic systems: Crotonyl-CoA Carboxylase/Reductase (CCR) and Dihydrofolate Reductase (DHFR). The results show that the al chemical transformations successfully reproduce both experimental values and trends, with errors close to chemical accuracy (1 kcal/mol). Additionally, modifications to the non-bonded parameters of mutating amino acids in the alchemical transformations are evaluated, improving predictions by up to 3 kJ/mol in the DHFR system. Once the methodology was validated, it was applied to an artificially de signed enzyme: the Y187F variant of Old Yellow Enzyme (OYE). Since this artificial variant catalyzes a new-to-nature aldol reaction in OYE, multi-state QM/MM simulations were performed using the adaptive string method to disclose the reaction mechanism. The results show that a flavin mononucleo tide (FMN) coenzyme transfers a hydride to 2-cyclohexenone, generating an enolate. This intermediate subsequently attacks formaldehyde molecule leading to 2-methoxycyclohexenone after protonation. The minimum free energy path obtained using the string method indicates that the rate-limiting step of the reaction is the hydride transfer. After identifying the rate-limiting step and characterizing the transition state, non-equilibrium alchemical transformations were performed on OYE Y187F to find catalytically more active variants. The results predict at least twovariants that reduce the activation free energy. The Y187F/T130W double mutant is the most promising candidate, achieving a ∆∆G‡ of-8.8 kJ/mol. In summary,this thesis demonstrates that non-equilibrium alchemical trans formations predict enzyme kinetic properties with a precision comparable to first-principles-based methods but at one-tenth of the computational cost. This breakthrough enables high-throughput analyses, fascilitating the disco very of new enzymatic variants in the field of protein engineering.
Description
Tesis presentada para optar al grado de Doctor en Ciencias con mención en Química
Keywords
Reacciones químicas, Ingeniería de proteínas, diseño de proteínas