Transcriptional regulation is an inherently dynamic process that rely on the stochastic molecular events occurring on the regulatory region of a gene. From prokaryotes to eukaryotes, it can take various forms and result from the interaction of a variety of molecules and epigenetic factors. We propose a generic model for investigating the stochastic dynamics of any arbitrary regulatory structure and its impact on gene expression. It describes the stochastic molecular events and their mutual influence. In particular, it can represent the interplay between various epigenetic factors (chromatin opening/closing, nucleosome sliding, histone tails covalent modifications) and TFs molecules (including both cooperative/competitive aspects of their association/dissociation as well as their remodeling/enzymatic activity on epigenetic factors). Subsequent steps of gene expression (transcription, translation and degradation of transcripts and proteins) are modeled simply but explicitly to assess how promoter dynamics impacts on the expression level. Contrarily to classical thermodynamic approaches [Saiz et al. 2006, Mol Syst Biol], our kinetic one allows to represent systems with energy-dependant transitions (ie. ATP-dependent chromatin remodeling) and is therefore relevent for investigating eukaryotic regulation. This approach also makes it possible to predict dynamic aspects of the activity of the system, which turn out to be essential for the understanding of these structures. The study of this model shows that the spontaneous activity of a single gene can demonstrate much complexity due to the mere interplay between molecules. In particular, the regulatory structure can demonstrate a strongly periodic activity as it is observed in vivo on eukaryotic promoters. Indeed, it is known that regulation proceeds through cycles of recruitment of TFs and of modifications of epigenetic factors [Metivier et al. 2003, Cell]. How this property can arise and what it implies can be investigated theoretically with such a model. For instance, changing TFs concentration, as in classical regulation, can provide as a flexible way to tightly modulate several parameters of this dynamics (ie. frequency and coherence of periodicities). Moreover, an energetic cost can be precisely quantified for a given system. In addition to be necessary for any periodic activity, this cost appears to be linked to other aspects of the dynamics (such as flexibility or robustness). This have certainly important evolutionary outcomes and call for further investigations. This model can also be used to integrate various experimental approaches. Indeed, the activity of the system can be derived in terms of various indicators reproducing measures of experimental devices such as flux cytometry on clonal populations of cells, spectral analysis of single-cells gene expression timecourses, FRAP, FRET and time-resolved ChIP. This model appears as a unifying framework for confronting experimental evidences at various levels and time scales that provides different viewpoints on the same object. Thus it appears as central tool for systemic experimental investigations of regulatory structure dynamics and stochasticity in gene expression.