Les processus d’oxydo-réduction impliquant des molécules organiques se retrouvent très fréquemment dans les protéines. Ces réactions comprennent généralement des transferts d’électrons et de protons qui se traduisent dans le bilan réactionnel par des transferts couplés proton-électron, des transferts simples d’hydrogène, d’hydrure... Une des principales méthodes pour élucider ces mécanismes est fournie par l'évaluation de grandeurs thermodynamiques et cinétiques. Expérimentalement, ces informations sont cependant obtenues avec une résolution temporelle souvent limitée à la milli/microseconde. Les simulations numériques présentées ici complètent, à des échelles de temps plus courtes (femto, pico, nanosecondes), ces données expérimentales. Il existe de nombreuses méthodes de simulations dédiées à l’étude de mécanismes redox dans les protéines combinant la description quantique des réactifs (QM) nécessaire à l’étude des changements d’états électroniques et la description classique de l’environnement (MM), l'échantillonnage de conformations se faisant grâce à des simulations de dynamique moléculaire (MD). Ces méthodes diffèrent par la qualité de la description du mécanisme réactionnel et le coût en temps de calcul. L’objectif de cette thèse est d’étudier les mécanismes de différents processus impliquant des transferts de protons et d’électrons en recherchant à chaque fois les outils adaptés. Elle comporte trois parties : i) l’évaluation de potentiels redox de cofacteurs quinones ; ii) la description du mécanisme d’oxydation du L-lactate dans l’enzyme flavocytochrome b2 ; iii) la décomposition d’un transfert formel d’hydrure entre deux flavines au sein de la protéine EmoB. Dans le cas du calcul des potentiels redox, nous utilisons une méthode notée QM+MM où la description électronique se fait en phase gaz au niveau DFT tandis que les simulations de MD s’effectuent classiquement. Nous appliquons l’approximation de réponse linéaire (ARL) pour décrire la réponse du système aux étapes de changement d’état de protonation ou d’oxydation de la fonction quinone ce qui aboutit au calcul du potentiel redox théorique. Nous avons ainsi pu établir une courbe de calibration des résultats théoriques en fonction des données expérimentales, confirmant la validité de l'ARL pour les cofacteurs quinones dans l’eau. La méthode a été étendue à la protéine MADH mais les limites de l’ARL ont été atteintes du fait des fluctuations importantes de l’environnement. L’étude de l’oxydation du L-lactate en pyruvate repose sur le calcul de surfaces d'énergie libre au niveau AM1/MM. Ces surfaces sont obtenues à l’aide de simulations de MD biaisées puis corrigées à l’aide de calculs d’énergies DFT. Différents chemins de réactions impliquant les transferts d’un proton et d’un hydrure du substrat vers une histidine et une flavine respectivement ont pu être identifiés. Ces transferts peuvent être séquentiels ou concertés suivant la conformation du site actif ou les mutations effectuées. Les surfaces concordent avec les effets observés expérimentalement. Les barrières obtenues restent cependant supérieures à celles attendues ouvrant la voie à d’autres simulations. La décomposition du mécanisme de transfert d’hydrure en transfert d’électron et d’atome d’hydrogène s’appuie sur de longues simulations classiques et des calculs d’énergies au niveau DFT contrainte (cDFT)/MM. La DFT contrainte permet de décrire les états diabatiques associés au transfert d’électron à différents stades du transfert d’hydrogène. En appliquant l’ARL, nous pouvons construire des paraboles correspondant aux états diabatiques et déterminer la séquence des évènements de transfert d'électron et d’hydrogène. La comparaison entre milieux protéique et aqueux nous a permis d’établir que le rôle de la protéine dans le transfert d'hydrure global est de bloquer le transfert d’électron en l’absence du transfert d’hydrogène empêchant ainsi la formation de flavines semi-réduites. / Redox processes involving organic molecules are ubiquitous in proteins. They generally imply global reactions such as Proton Coupled Electron Transfers, hydrogen atom or hydride transfers which can be decomposed into both electrons and proton transfers. Kinetic and thermodynamic information leads to a better understanding of these mechanisms. However, experiments are often limited to a milli- or microsecond timescales. We present here numerical simulations allowing modeling at shorter timescales (femto, pico or nanosecond) to complete experimental data. Many numerical methods combine quantum description (QM) of the active center and classical description (MM) of the environment to describe redox transformations into biological media. Molecular dynamics (MD) simulations allowed a conformational sampling of the global system. Nevertheless, depending on their level of description of the QM part, the methods can cost more or less CPU time to get a good conformational sampling. In this thesis, we have studied different redox mechanisms involving both proton and electron transfers with a particular care paid to the balance between quality of the electronic description and of conformational sampling. For each mechanism, the coupled proton and electron transfers are investigated differently. This manuscript thus falls into three parts: i) the evaluation of the redox potentials quinone derivatives ; ii) the mechanistic description of the L-lactate oxidation into pyruvate in the flavocytochrome b2 enzyme; iii) decomposition of the formal hydride transfer occurring between two flavins in EmoB protein. A QM+MM scheme is chosen to evaluate redox potential of quinone cofactors: the electronic behavior is described at DFT level in gas phase while classical MDs provide a large conformational sampling of the molecule and its environment. Deprotonation and oxidation free energies are estimated by applying the linear response approximation (LRA). We finally get a theoretical value of the redox potential for different quinocofactors in water and a calibration curve of these theoretical results in function of experimental data. This curve allowed predictions of quinone redox potentials in water with a good accuracy (less than 0.1 eV). We also try our method on the MADH protein containing a Tryptophan Tryptophilquinone cofactor. However, because of great fluctuations of the environment, the LRA is not suitable for this system. This underlines the limits of our methodology. The oxidation of L-lactate to pyruvate is described by free energy surfaces obtained at AM1/MM level. Biased MDs provide the AM1/MM profile which is then corrected at DFT level. Several reactions pathways have been noticed. They consist in sequential or concerted transfers of a proton from L-lactate to a histidine and a hydride from L-lactate to a flavin cofactor. The coupling between the two transfers depends on the conformation of the active site or on the mutations. The obtained surfaces fit qualitatively the experimental data but the theoretical activation barriers are too high. Other simulations must be explored: different methods, other mechanism... Finally, a combination of long classical MDs and constrained DFT (cDFT)/MM are employed to decompose a hydride transfer between two flavins into one hydrogen atom and one electron transfer. cDFT methodology allow us to describe diabatic states associated to the electron transfer during the hydrogen atom transfer. Applying the LRA, we can build parabola of the diabatic and determine the sequence of the two transfers. The comparison of our results in the EmoB protein or in aqueous medium shows that the protein allows the electron transfer only if the hydrogen atom transfer is happening. By this way, no semi-reduced flavin is created.

As transições eletrônicas que apresenta as energias mais baixas no retinal e em quatro retinais sintéticos foram analisadas em dois diferentes ambientes: no vácuo e ligados à proteína bacterioopsina por uma base de Schiff protonada, utilizando diversos métodos de química teórica. Os resultados aqui apresentados fornecem indicativos de que três estados eletrônicos estão envolvidos na formação da primeira banda de absorção dos aldeídos, enquanto que, no caso dos compostos ligados à proteína apenas dois estados estariam envolvidos. As análises discutidas neste trabalho também sugerem uma possível explicação para o envolvimento de dois estados eletrônicos excitados no processo de fotoisomerização do retinal ligado à bacterioopsina. / Low-lying electronic transitions of retinal and of four synthetic retinals were analyzed in two different environments: in vacuum and linked to bacterioopsin through a protonated Schiff base, employing several methods of theoretical chemistry. The results here reported suggest that the first absorption band of the aldehydes involves three electronic states, while for the case of protein linked compounds, two states would be embraced. Our discussions also provided a possible explanation about the involvement of two electronic excited states in the photoisomerization process of retinal linked to bacterioopsin.

In der vorliegenden Arbeit wurden Cluster- und QM/MM-Einbettungsmodelle für geträgerte Vanadiumoxidkatalysatoren auf Siliziumdioxid mit Hilfe von DFT-Rechnungen untersucht. Es wurden Strukturen, Schwingungen und die Stabilität gegenüber Wasser und Sauerstoff unter Reaktionsbedingungen betrachtet. Ferner wurde die Reaktivität bei der Oxidativen Dehydrierung (ODH) von n-Propan zu n-Propen und der Oxidation von Methanol zu Formaldehyd untersucht. Zur Durchführung der Einbettungsrechnung wurde ein angepasstes Shell-Model-Potential verwendet. Bezüglich der Schwingungen zeigte sich, dass sich der Einfluss der Einbettung hauptsächlich auf die Kopplung von Schwingungen beschränkt, aber die Frequenzen kaum verändert werden. Die lokale Struktur um das Vanadiumatom beeinflusst die Vanadylschwingung wenig. Die Stabilitätsberechnungen haben gezeigt, dass die untersuchten Modelle unter Reaktionsbedingungen vorliegen und hydroxylierte Spezies keine Bedeutung haben. Bei der Untersuchung der beiden Reaktionen wird für die Cluster- und die QM/MM-Einbettungsrechnungen jeweils derselbe Mechanismus gefunden. Der Vorteil der Einbettungsrechnungen besteht darin, dass die lokale Struktur um das aktive Zentrum variiert werden kann. Bei der ODH von Propan hat die lokale Struktur einen eher geringen Einfluss auf die Reaktionsenergien, denn die Reaktion verläuft hauptsächlich am Vanadylsauerstoff. Der Übergangszustand des geschwindigkeitsbestimmenden Schrittes liegt bei der Einbettungsrechnung dennoch energetisch höher. Grund ist eine sterische Hinderung durch die Hydroxylgruppen der Oberfläche. Bei der Oxidation von Methanol ist der Einfluss der lokalen Struktur größer, denn die Reaktion verläuft sowohl über den Vanadylsauerstoff als auch über die Brückensauerstoffatome zum Trägermaterial. Für beide Reaktionen wird ein Einfluss der Vanadiumbeladung auf die Reaktionsenergien gefunden. Bei höherer Beladung werden die Reaktionen exothermer. / In this work cluster models and models for QM/MM-embedding for supported vanadia catalysts on silica were studied with help of DFT-calculations. The structures, vibrations and stability towards water and oxygen under reaction conditions were examined. Furthermore the reactivities towards the oxidative dehydrogenation (ODH) of n-propane to n-propene and the oxidation of methanol to formaldehyde were tested. For the embedding an adapted shell-model-potential was employed. Regarding the vibrations it was shown that the influence of the embedding lies mostly in the coupling of vibrations and not in their frequencies. The local structure surrounding the vanadium atom has only minor influence. The stability calculations have shown that the tested model systems exist under reaction conditions while hydroxylated species do not exist. When studying the reactivity of the two reactions always the same mechanism is found for cluster and embedded calculations. The benefit of the embedded calculations is the possibility to vary the local structure surrounding the active center. For the ODH of propane the local structure has only small impact on reaction energies because the reaction takes mainly place at the vanadyl oxygen. The transition state of the rate determining step nevertheless is energetically higher in the embedded calculations due to steric hindrance caused by the hydroxyl groups of the surface. The impact of local structure is more important for the oxidation of methanol since in this case vanadyl oxygen and bridging oxygens to the support are involved in the reaction. For both reactions an influence on reaction energies of the vanadia loading is found. With more loading the reactions becomes more exothermic.

Absorção e Emissão de radiação eletromagnética por moléculas na região do UV-Vis fornece informações sobre os estados eletrônicos excitados, sendo propriedades de grande interesse devido a sua relação com processos biológicos, bem como suas possíveis aplicações em diagnósticos e tecnologia. Essas propriedades são sensíveis ao meio em que as moléculas se encontram, tornando-se assim natural a busca por métodos teóricos que possibilitam descrever essas interações. Neste trabalho, nós usamos a metodologia Sequential-Quantum Mechanics/Molecular Mechanics para estudar o espectro de absorção e de emissão de moléculas de relevância biológica, quando em solução. Simulações clássicas Monte Carlo foram usadas pra construir uma configuração eletrostática média do líquido para posterior cálculo das propriedades, feito através dos métodos multiconfiguracionais CASSCF e CASPT2. Cuidados especiais foram tomados para incluir a polarização eletrônica que o soluto sofre devido à presença do solvente. Nossa contribuição é a adaptação do método do Gradiente de Energia Livre para permitir a obtenção de geometrias de estados excitados do soluto em solução. A técnica foi implementada em um programa e aplicada com sucesso nos sistemas estudados. As perspectivas agora se abrem para a obtenção de intersecções cônicas em meio solvente, permitindo assim o estudo de decaimentos não-radiativos em sistemas solvatados. / Absorption and emission of electromagnetic radiation by molecules in the UV-Vis region yields information about the electronic excited states, being properties of great interest due its relation with biological process, as well as its possible applications in diagnostics and technology. These properties are sensible to the environment in which the molecules are, making natural the search for theoretical methods that describe these interactions. In this work, we have used the Sequential-Quantum Mechanics/Molecular Mechanics methodology to study the spectrum of absorption and emission of molecules of biological relevance, when in solution. Classical Monte Carlo simulations were carried out to construct an average solvent electrostatic configuration and used to represent the liquid in posterior quantum mechanics calculation of the properties, performed by using the multiconfigurational methods CASSCF and CASPT2. Special cares were taken to include the electronic polarization of the solute due the solvent. Our contribution is the adaptation of the Free Energy Gradient method to allow the determination of the solute excited states geometries in solution. The method was implemented in a computer program and successfully applied in the systems studied. The perspectives are now open to the calculation of conical intersection in solvent environment, enabling the study of nonradioactive decays in solvated systems.

L'ADN est en permanence exposé à un grand nombre d'événements dommageables déclenchées par des agents endogènes et exogènes. De nombreux travaux expérimentaux ont fourni des informations cruciales sur les propriétés structurelles et la réparation de certains des lésions de l'ADN. Cependant, il manque une vision mécanistique ou énergétique sur leur formation. La biochimie computationnelle a émergé comme un outil puissant pour comprendre les réactions biochimiques et les propriétés électroniques de systèmes complexes.Dans cette thèse, nous étudions la formation de lésions complexes intra-brin et inter-brin. Ces lésions tandem constituent une puissant menace à l'intégrité du génome, en raison de leur haute fréquence mutagenique. Tout d'abord, nous discutons l'attaque d'une liaison covalente entre un radical pyrimidinique. En comparant avec les bases isolees, nos simulations hybrides Car-Parrinello demontrent que la reactivité de la thymine et de la cytosine radicalaires sont inversees dans l'environnement B-helical. De plus, nos resultats montrent egalement une deformation plus importante pour la lesion G[8-5]C.Nous rationalisons également la plus grande réactivité des cytosines par rapport aux purines vers la formation multi-etapes de lésions complexes inter-brins par condensation avec un site C4' abasique. Ces résultats bases sur des simulations avec solvatation explicite et combines a la théorie de la fonctionnelle de la densité sont en accord avec les données expérimentales. / DNA is continuously exposed to a vast number of damaging events triggered by endogenous and exogenous agents. Numerous experimental studies have provided key information regarding structural properties of some of the DNA lesions and their repair. However, they lack in mechanistic or energetic information pertaining to their formation. Computational Biochemistry has emerged as a powerful tool to understand biochemical reactions and electronic properties of large systems.In this thesis we study the formation of inter- and intra-strand cross-links. These tandem lesions pose a potent threat to genome integrity, because of their high mutagenic frequency. First, we discuss the formation of complex defects which arise from the attack of a pyrimidine radical onto guanine. In comparison with the reactivity of isolated nucleobases, our hybrid Car-Parrinello Molecular Dynamics simulations reveal that the reactivity of hydrogen-abstracted thymine and cytosine is reversed within a B-helix environment. Further, our data also suggest a more severe distortion of the B-helix for G[8-5]C.Second, we rationalize the higher reactivity of cytosine vs. purines toward the multistep formation of inter-strand crosslinks with a C4' oxidized a basic site, which is in qualitative agreement with experiments on isolated nucleobases, using explicit solvent simulations combined to density functional theory.

A química computacional, uma importante e crescente área da química teórica, vem sendo empregada com grande sucesso no entendimento dos mais variados tipos de sistemas químicos. Focando na interação entre arsênio e carbono, o presente trabalho apresenta duas situações distintas: uma que engloba o menor sistema formado entre eles, o radical CAs, e outra que traz a interação da superfície C(100) do diamante com o átomo de arsênio e a sua forma hidreto, ou seja, os sistemas C(100)+As e C(100)+AsH3. No primeiro caso, um tratamento de alto nível de correlação eletrônica (MRCI/aV5Z) é feito, permitindo se obter as propriedades espectroscópicas associados aos estados eletrônicos de mais baixa energia do radical. No estudo envolvendo a superfície C(100), dois modelos distintos de aglomerados foram adotados na representação do sólido: o modelo QM (C15H16) e o QM/MM (C292H120). A aproximação híbrida é baseada no método SIMOMM que é indicado para o tratamento de superfícies sólidas. Os dados estruturais e energéticos dos pontos de mínimos identificados nas duas diferentes aproximações foram contrastados, sugerindo que o modelo de aglomerado QM/MM é capaz de representar melhor o problema real. / Computational chemistry, a growing and important area in theoretical chemistry, has been successfully employed to understand many types of chemical systems. Focusing on the interaction between carbon and arsenic, this work presents two distinct situations: one is the study of the smallest system formed between them, the CAs radical, and another one where the diamond C(100) surface interacts with an arsenic atom and its hydride form, i.e. , the C(100)+As and C(100)+AsH3 systems. In the first case, a high-level (MRCI/aV5Z ) calculation was performed, allowing us to obtain the spectroscopic properties of the low-lying electronic states. In the study of the diamond C(100) surface, two different cluster models were used to represent the solid: one QM (C15H16) and the other QM/MM (C292H120). The hybrid approach is based on the SIMOMM method, that is recommended for dealing with solid surfaces. Energetic and structural aspects associated with the minimum energy species were contrasted for both a pproaches, suggesting that the QM/MM model is able to better represent the real problem.

Neste trabalho estudamos o espectro de absorção da clorofila d, incluindo os efeitos do solvente metanol, utilizando a Teoria do Funcional da Densidade Dependente do Tempo em combinação com o método s-QM/MM. Diferentes abordagens para a descrição do meio solvente, que vão desde o modelo contínuo polarizável até a inclusão de moléculas explícitas do solvente, são utilizadas. Observamos que a inclusão do solvente desloca o espectro, em relação ao calculado em vácuo, para o vermelho. A inclusão de 20 moléculas explícitas de metanol e 880 representadas como as cargas pontuais do seus átomos para a descrição do meio solvente foi a que melhor concordou com os valores experimentais. Mediante cálculos de mecânica quântica obtivemos para o complexo composto por a molécula de clorofila d e apenas uma de metanol explícita um deslocamento do átomo de magnésio da clorofila d em relação ao anel de 0,31 Å. Mediante a analise da função de distribuição radial de pares obtida da simulação clássica comprovamos que esse átomo de magnésio é penta-coordenado. Além disso, é usada uma estrutura simplificada da clorofila d baseado na localização dos orbitais moleculares participantes das transições eletrônicas com o objetivo de diminuir o custo computacional dos cálculos de mecânica quântica. Os espectros calculados para a clorofila d não apresentaram diferenças significativas com os calculados para a simplificação proposta. Como complementação o espectro Raman da clorofila d isolada é calculado mediante DFT e alguns dos modos normais são caracterizados. / In this work we study the absorption spectrum of chlorophyll d, including the effects of the solvent methanol, using the Time Dependent Density Functional Theory in combination with the method s-QM/MM. Different approaches for the description of the solvent medium, ranging from the polarizable continuum model to the inclusion of explicit solvent molecules, are used. We note that the inclusion of solvent shifts the spectrum, relative to that calculated in vacuum, to the red side of the spectrum. The inclusion of 20 explicit molecules of methanol and 880 represented as point charges of their atoms for the description of the solvent medium was the one that best agreed with the experimental values. By quantum mechanics calculations we obtained for the complex composed of the molecule of chlorophyll d and only one explicit methanol molecule a displacement of the magnesium atom of chlorophyll d in relation to the ring of 0,31 Å. By means of the analysis of the radial distribution function obtained from the classic simulation we can see that this atom of magnesium is penta-coordinated. In addition, a simplified structure of chlorophyll d is used based on the location of the molecular orbitals involved in the electronic transitions in order to reduce the computational cost of quantum mechanics calculations. The spectra calculated for chlorophyll d did not show significant differences with those calculated for the proposed simplification. As a complement the Raman spectrum of isolated chlorophyll d is calculated by DFT and some of the normal modes are characterized.

A química computacional, uma importante e crescente área da química teórica, vem sendo empregada com grande sucesso no entendimento dos mais variados tipos de sistemas químicos. Focando na interação entre arsênio e carbono, o presente trabalho apresenta duas situações distintas: uma que engloba o menor sistema formado entre eles, o radical CAs, e outra que traz a interação da superfície C(100) do diamante com o átomo de arsênio e a sua forma hidreto, ou seja, os sistemas C(100)+As e C(100)+AsH3. No primeiro caso, um tratamento de alto nível de correlação eletrônica (MRCI/aV5Z) é feito, permitindo se obter as propriedades espectroscópicas associados aos estados eletrônicos de mais baixa energia do radical. No estudo envolvendo a superfície C(100), dois modelos distintos de aglomerados foram adotados na representação do sólido: o modelo QM (C15H16) e o QM/MM (C292H120). A aproximação híbrida é baseada no método SIMOMM que é indicado para o tratamento de superfícies sólidas. Os dados estruturais e energéticos dos pontos de mínimos identificados nas duas diferentes aproximações foram contrastados, sugerindo que o modelo de aglomerado QM/MM é capaz de representar melhor o problema real. / Computational chemistry, a growing and important area in theoretical chemistry, has been successfully employed to understand many types of chemical systems. Focusing on the interaction between carbon and arsenic, this work presents two distinct situations: one is the study of the smallest system formed between them, the CAs radical, and another one where the diamond C(100) surface interacts with an arsenic atom and its hydride form, i.e. , the C(100)+As and C(100)+AsH3 systems. In the first case, a high-level (MRCI/aV5Z ) calculation was performed, allowing us to obtain the spectroscopic properties of the low-lying electronic states. In the study of the diamond C(100) surface, two different cluster models were used to represent the solid: one QM (C15H16) and the other QM/MM (C292H120). The hybrid approach is based on the SIMOMM method, that is recommended for dealing with solid surfaces. Energetic and structural aspects associated with the minimum energy species were contrasted for both a pproaches, suggesting that the QM/MM model is able to better represent the real problem.

This thesis presents a starburst model for far-infrared/sub-millimeter/millimeter (FIR/sub-mm/mm) line emission of molecular and atomic gas in an evolving starburst region, which is treated as an ensemble of non-interacting hot bubbles which drive spherical shells of swept-up gas into a surrounding uniform gas medium. These bubbles and shells are driven by winds and supernovae within massive star clusters formed during an instantaneous starburst. The underlying stellar radiation from the evolving clusters affects the properties and structure of photodissociation regions (PDRs) in the shells, and hence the spectral energy distributions (SEDs) of the molecular and atomic line emission from these swept-up shells and the associated parent giant molecular clouds (GMCs) contains a signature of the stage evolution of the starburst. The physical and chemical properties of the shells and their structure are computed using a a simple well known similarity solution for the shell expansion, a stellar population synthesis code, and a time-dependent PDR chemistry model. The SEDs for several molecular and atomic lines ($^{12}$CO and its isotope $^{13}$CO, HCN, HCO$^+$, C, O, and C$^+$) are computed using a non-local thermodynamic equilibrium (non-LTE) line radiative transfer model. By comparing our models with the available observed data of nearby infrared bright galaxies, especially M 82, we constrain the models and in the case of M 82, provide estimates for the age of the recent starburst activity. We also derive the total H$_2$ gas mass in the measured regions of the central 1 kpc starburst disk of M 82. In addition, we apply the model to represent various stages of starburst evolution in a well known sample of nearby luminous infrared galaxies (LIRGs). In this way, we interpret the relationship between the degree of molecular excitation and ratio of FIR to CO luminosity to possibly reflect different stages of the evolution of star-forming activity within their nuclear regions. We conclude with an assessment of the strengths and weaknesses of this approach to dating starbursts, and suggest future work for improving the model.

This thesis presents a starburst model for far-infrared/sub-millimeter/millimeter (FIR/sub-mm/mm) line emission of molecular and atomic gas in an evolving starburst region, which is treated as an ensemble of non-interacting hot bubbles which drive spherical shells of swept-up gas into a surrounding uniform gas medium. These bubbles and shells are driven by winds and supernovae within massive star clusters formed during an instantaneous starburst. The underlying stellar radiation from the evolving clusters affects the properties and structure of photodissociation regions (PDRs) in the shells, and hence the spectral energy distributions (SEDs) of the molecular and atomic line emission from these swept-up shells and the associated parent giant molecular clouds (GMCs) contains a signature of the stage evolution of the starburst. The physical and chemical properties of the shells and their structure are computed using a a simple well known similarity solution for the shell expansion, a stellar population synthesis code, and a time-dependent PDR chemistry model. The SEDs for several molecular and atomic lines ($^{12}$CO and its isotope $^{13}$CO, HCN, HCO$^+$, C, O, and C$^+$) are computed using a non-local thermodynamic equilibrium (non-LTE) line radiative transfer model. By comparing our models with the available observed data of nearby infrared bright galaxies, especially M 82, we constrain the models and in the case of M 82, provide estimates for the age of the recent starburst activity. We also derive the total H$_2$ gas mass in the measured regions of the central 1 kpc starburst disk of M 82. In addition, we apply the model to represent various stages of starburst evolution in a well known sample of nearby luminous infrared galaxies (LIRGs). In this way, we interpret the relationship between the degree of molecular excitation and ratio of FIR to CO luminosity to possibly reflect different stages of the evolution of star-forming activity within their nuclear regions. We conclude with an assessment of the strengths and weaknesses of this approach to dating starbursts, and suggest future work for improving the model.