Abstract This thesis aims to study the interstellar medium (ISM) of nearby galaxies to characterize the phys­ical properties of the gas and dust. We especially focus our study on low-metallicity galaxies of the Local Universe, ideal candidates to study the in.uence of metal enrichment on the ISM properties of galaxies. Previous studies have shown that the Spectral Energy Distributions (SEDs) of low metallicity galaxies di.er signi.cantly from those of massive galaxies and that the dust-to-gas mass ratio (D/G) of the galaxy could be dependent of the metallicity. Observations of low-metallicity galaxies also often led to the detection of an excess at submillimeter (submm) wavelengths not always accounted for in usual SED models. Further studies and observations had to be performed to better cover the far-IR to submm range and probe the coldest phase of dust. We adopt a multi-wavelength approach to model and analyse the SEDs of 4 low-metallicity galaxies observed with LABOCA at 870 µm. We estimated the fraction of cool dust to be signi.cant compared to the total dust mass of the galaxies. Some D/Gs are incompatible compared to what is expected from the current chemical evolution model, revealing possible reservoirs of gas not detected by current HI or CO observations. I enlarged the .rst sample to a wider range of metallicities and showed that submm measure­ments signi.cantly a.ect the dust mass estimates of galaxies. For dustier galaxies for which the SED usually peaks at longer wavelengths, submm .uxes are crucial to constrain the position of the peak and the Rayleigh-Jeans slope of their SED. For low-metallicity galaxies, the submm wavelength domain harbours an excess that may imply a large amount of very cold dust. Our results con.rm that low-metallicity galaxies can exhibit a submm excess when observed at longer wavelengths. Obtaining a more precise inventory of the cold dust and resolve the main actors of dust evolution in massive star forming regions and molecular clouds was the logical following step. We obtain LABOCA observations of a resolved star forming region of the Large Magellanic Cloud (LMC), the N158/N159/N160 complex, situated in the south of the 30 Doradus complex. The proximity of the LMC enables us to resolve structures of a few parsecs at the LABOCA resolution to model the SEDs of individual and isolated regions across the complex and get a handle on the temperature distribution of the dust (hot, warm, cold). I am also comparing the dust distribution with the Hi, CO and Ha observations to spatially quantify the variations of the D/G. I .nally present the .rst Herschel observations of the galaxies NGC 6822 and NGC 1705, two low­metallicity galaxies observed as part of the Science Demonstration phase of the telescope launched in May 2009. In NGC 6822, we model the SEDs of some Hii regions and less active regions across the galaxy and .nd that the SEDs of Hii regions show warmer ranges of dust temperatures. We derive very high dust masses when graphite is used in our model to describe carbon dust. Using amorphous carbon, instead, requires less dust mass to account for the same submm emission due to its lower emissivity properties. This indicates that SED models including Herschel constraints may require di.erent dust properties than commonly used.