It remains a quest for modern astronomy to answer what main mechanisms set the star formation rate (SFR) of galaxies. Massive galaxies present a good starting point for such a quest due to their relatively easy detection at every redshift. Since stars form out of cold and dense gas, a comprehensive model for galaxy evolution should explain any observed connection between SFR and the amount and properties of the molecular gas of the interstellar medium (ISM). In proposed models of that kind, an active galactic nucleus (AGN) phase is often invoked as the cause for the decrease or cease of star formation. This thesis consists of models and observations of gas and AGNs in massive galaxies at z _ 2, and how they may affect the overall SFR and the subsequent evolutionary trajectory of massive galaxies to z = 0. For an improved understanding of how observed gas emission lines link to the underlying ISM physics, a new code is presented here; SImulator of GAlaxy Millimeter/submillimeter Emission (SÍGAME). By post-processing the outputs of cosmological simulations of galaxy formation with sub-grid physics recipes, SÍGAME divides the ISM into different gas phases and derives the density and temperature structure of these, with locally resolved radiation fields. In the first study, SÍGAME is combined with the radiative transfer code LIME to model the spectral line energy distribution (SLED) of CO. A CO SLED close to that of the Milky Way is found for normal star-forming massive galaxies at z _ 2, but 50% smaller _CO factors, with the latter decreasing towards the center of each model galaxy. In a second study, SÍGAME is adapted to model the fine-structure line of singly ionized carbon, [CII] at 158 _m, the most powerful emission line of neutral ISM. Applying SÍGAME to the same type of galaxies, most [CII] emission can be traced back to the molecular part of their ISM. The observed L[CII]-SFR relation at z > 0:5 is reproduced and a similar relation is established on kpc scales for the first time theoretically. A third study uncovers the presence of AGNs among massive galaxies at z _ 2, and sheds light on the AGN-host co-evolution by connecting the fraction and luminosity of AGNs with galaxy properties. By analyzing a large survey in X-ray, AGNs of high and low X-ray luminosity are extracted among massive galaxies at z _ 2 via AGN classification methods, and stacking techniques of non-detections, in X-ray. Consequently, it is found that about every fifth massive galaxy, quenched or not, contain an X-ray luminous AGN. Interestingly, an even higher fraction of low-luminosity AGNs reside in the X-ray undetected galaxies, and preferentially in the quenched ones, lending support to the importance of AGNs in impeding star formation during galaxy evolution.