While life is found in a wide range of environments, many of the chemical functions required for homeostasis and reproduction are conserved. A large subset of these functions are fulfilled by proteins, which themselves have common requirements for both structural stability and chemical function. Organisms living in extreme environments, such as psychrophiles in the arctic ocean or thermophiles around hydrothermal vents, must thus develop adaptations to maintain protein structure and function under widely varying conditions. Here, we consider one ubiquitous enzyme class - the serine proteases - which are found in all known cells. Using genomic data from prokaryotes living under different environmental conditions, we model proteases from the S9, S11, S12, and S13 serine protease families. We predict three-dimensional structures of each enzyme using comparative modeling with in silico maturation to correct for known post-translational modifications, also correcting protonation states for each enzyme's native environment. Using atomistic molecular dynamics (MD) simulations, we obtain locally equilibrated explicit solvent structures for each enzyme, as well as longer trajectories for selected cases drawn from different thermal environments. We perform a comparative analysis of structural properties including surface exposures and secondary structure content across environmental conditions, showing both conserved and variable properties by environment. We also employ protein structure networks to examine structural cohesion, and active site networks to examine differences in active site structure and dynamics. Our analyses suggest patterns of adaptation within the serine proteases, in response to the challenges of life in distinct environmental conditions.