Modelación computacional de fármacos inhibidores de DPP-4 para el tratamiento de pacientes diabéticos con Covid-19
Cargando...
Fecha
Autores
Casamayor Méndez, Dainelis
Título de la revista
ISSN de la revista
Título del volumen
Editor
Universidad Central ¨Marta Abreu¨ de Las Villas. Facultad de Química y Farmacia. Departamento de Farmacia
Resumen
La enzima dipeptidil peptidasa 4 (DPP4) es una aminopeptidasa sérica que metaboliza las hormonas incretínicas GLP-1 y el GIP por lo que sus inhibidores han ganado un lugar clave en el tratamiento de la diabetes mellitus tipo II. Además, debido a la alta afinidad entre la DPP4 humana y el dominio de unión al receptor spike (S) del Síndrome Respiratorio Agudo Severo Coronavirus-2 (SARS-CoV-2), esta puede funcionar como receptor funcional para ingresar al huésped. Teniendo en cuenta que los pacientes diabéticos pueden tener suceptibilidad a las complicaciones de la Covid-19, la actividad dual de los inhibidores de DPP4 constituyen estrategias terapéuticas de elección para tratar la población diabética infectada con el virus. Por esta razón la motivación de este trabajo es obtener herramientas efectivas in silico que permitan el descubrimiento de potentes inhibidores de la DPP4. Primeramente, se establece un umbral para la inhibición de la DPP4 empleando un método de regresión piecewise (pKi=7,579) y a partir de este valor se conforman dos clases de compuestos (inhibidores y no inhibidores). Posteriormente, se llevan a cabo modelos de clasificación basados en técnicas de aprendizaje automatizado e inteligencia artificial, empleando los descriptores moleculares generados por el DRAGON. Fueron desarrollados modelos de máquinas de vectores soporte, redes bayesianas y redes neuronales MLP, donde este último muestra los mejores parámetros de precisión, exactitud, especificidad así como razón de falsos positivos. Finalmente se concluye que las herramientas computacionales propuestas constituyen una metodología eficiente para la identificación de nuevos fármacos con potencial actividad contra la enzima DPP4 para el tratamiento de pacientes diabéticos con Covid-19.
Palabras clave: Covid-19, Crivado virtual, Diabetes mellitus, Inhibidores de Dipeptidil peptidasa 4, QSAR.
The enzyme dipeptidyl peptidase 4 (DPP4) is a serum aminopeptidase that metabolizes the incretin hormones GLP-1 and GIP, which is why its inhibitors have gained a key place in the treatment of type II diabetes mellitus. Furthermore, due to the high affinity between human DPP4 and the spike (S) receptor binding domain of Severe Acute Respiratory Syndrome Coronavirus-2 (SARS-CoV-2), it can function as a functional receptor to enter the host. Taking into account that diabetic patients may be susceptible to complications from Covid-19, the dual activity of DPP4 inhibitors are therapeutic strategies of choice to treat the diabetic population infected with the virus. For this reason, the motivation for this work is to obtain effective in silico tools that allow the discovery of powerful DPP4 inhibitors. First, a threshold for the inhibition of DPP4 is established using a piecewise regression method (pKi = 7,579) and from this value two classes of compounds are formed (inhibitors and non-inhibitors). Subsequently, classification models based on automated learning techniques and artificial intelligence are carried out, using the molecular descriptors generated by DRAGON. Support vector machine models, Bayesian networks and MLP neural networks were developed, where the latter shows the best parameters of accuracy, accuracy, specificity as well as the false positive ratio. Finally, it is concluded that the proposed computational tools constitute an efficient methodology for the identification of new drugs with potential activity against the DPP4 enzyme for the treatment of diabetic patients with Covid-19. Key words: Covid-19, Virtual screening, Diabetes mellitus, Dipeptidyl peptidase 4 inhibitors, QSAR.
The enzyme dipeptidyl peptidase 4 (DPP4) is a serum aminopeptidase that metabolizes the incretin hormones GLP-1 and GIP, which is why its inhibitors have gained a key place in the treatment of type II diabetes mellitus. Furthermore, due to the high affinity between human DPP4 and the spike (S) receptor binding domain of Severe Acute Respiratory Syndrome Coronavirus-2 (SARS-CoV-2), it can function as a functional receptor to enter the host. Taking into account that diabetic patients may be susceptible to complications from Covid-19, the dual activity of DPP4 inhibitors are therapeutic strategies of choice to treat the diabetic population infected with the virus. For this reason, the motivation for this work is to obtain effective in silico tools that allow the discovery of powerful DPP4 inhibitors. First, a threshold for the inhibition of DPP4 is established using a piecewise regression method (pKi = 7,579) and from this value two classes of compounds are formed (inhibitors and non-inhibitors). Subsequently, classification models based on automated learning techniques and artificial intelligence are carried out, using the molecular descriptors generated by DRAGON. Support vector machine models, Bayesian networks and MLP neural networks were developed, where the latter shows the best parameters of accuracy, accuracy, specificity as well as the false positive ratio. Finally, it is concluded that the proposed computational tools constitute an efficient methodology for the identification of new drugs with potential activity against the DPP4 enzyme for the treatment of diabetic patients with Covid-19. Key words: Covid-19, Virtual screening, Diabetes mellitus, Dipeptidyl peptidase 4 inhibitors, QSAR.