The major challenge of cell and developmental biology is to propose a multiscale model able to describe the mechanism through which macromolecules act on the cell, and in cascade on the tissues and organ functions. One of the key goals in this expanding field is to decipher how stress induced by mechanical forces or DNA damage induces cancer. Presently, the multiscale analysis is done by resorting to different tools adapted to imaging at the molecular (nanometer scale), cell (micron scale) or tissue (mm scale) levels. A microscope able to observe the ballet of proteins inside live biological tissues over large fields of view (the dream of all set-ups in one) would drastically increase the available information and open new doors in integrative biology. Widefield fluorescence microscopy is the most widespread tool for getting real time images of specific protein distribution in live specimen over large volumes of observation (hundreds of thousands of microns cube). Unfortunately, its resolution, about 300 nm transversally and 1000 nm axially at best, is not sufficient for an accurate study of the macromolecule organisation and dynamics within living tissues. Super-resolution fluorescence microscopes using saturation or pointillism yield images with a resolution below 50 nm but their toxicity, the time required for the data acquisition and processing restrict their use to small observation volumes and slow temporal dynamics. Structured Illumination Microscopy (SIM) is the best compromise between resolution (about 100 nm transverse and 300 nm axial) and practical implementation on live samples. Yet, as it requires the perfect knowledge of the illuminations, it cannot be used deep inside distorting samples such as biological tissues and its experimental implementation is cumbersome. Recently, we have proposed a technique that gathers the resolution of periodic SIM and the ease of use and field of view of standard fluorescence microscopy. Random Illumination Microscopy (RIM) reconstructs a Super-Resolved (SR) image of the sample from multiple low-resolution frames obtained under different speckle illuminations. It is based on a mathematical analysis, showing that a two-fold resolution gain, can be obtained from the second order statistics of the speckle images. Speckles being insensitive to scattering, distortions and aberrations, RIM is expected to succeed in cases where SIM fails. In the last two years, we have implemented a two-dimensional (2D) version of RIM in which the sample is viewed as a slice limited to the focal plane. This simplified approach yielded remarkable results with 120 nm transverse and 300 nm axial resolutions together with an SR-image rate about 1-5 Hz. This achievement positions 2D-RIM as one of the best super-resolved techniques for live imaging, in particular deep inside biological tissues where aberrations and scattering are redhibitory for SIM. However, it is obvious that 2D-RIM does not exploit the full capacity of RIM and a significant amelioration of the temporal and axial resolutions could be obtained by taking advantage of the structuration of the speckles and observation point spread function along the optical axis. In this project, we propose to extend RIM principle to the three dimensions with an appropriate mathematical analysis and data processing coupled to an up-graded instrumentation. Our objective is to provide images over large field of views with 100 nm transverse and 200 nm axial resolutions and reach frame rate about 10-30 Hz. 3D-RIM will be tested on two key biological issues for which all existing super-resolution microscopes are inoperative: the multiscale interaction of cells with the surrounding tissue during apoptosis, which requires high spatial resolution and large fields of view and the dynamic chromatin loop extrusion during DNA repair which requires high spatiotemporal resolution.