Cytosolic crowding is known to influence the thermodynamics and kinetics of in vivo chemical reactions. Crowders, including proteins, macromolecular assemblies and intracellular organelles, reduce the volume available to a diffusing substrate and thereby lower its effective diffusion constant relative to its rate in bulk solution. However, the nature of a substrate's interaction with crowders that occlude diffusion can further influence the effective diffusion rate. To probe the impact of crowding over sub-micron intracellular distances, we apply a multi-scale mathematical theory, homogenization, to estimate effective diffusion rates for ions and small bio-molecules diffusing in a densely-packed lattice of representative cytosolic proteins. Our simulations quantify how the crowded volume fraction, irregularity of protein shapes and distribution, as well as electrostatic interactions, influence the diffusion rates of small molecules. For non-interacting substrates that diffuse much faster than proteins crowders, the effective diffusion constant is dominated by volume fraction with minor contributions from crowder distribution and shapes, which permits a simplified description of the cytosol for most diffusion applications. Our results also shows that the diffusion coefficient of the diffuser can be accelerated or decelerated for attractive or repulsive forces; these effect of these interactions on diffusion strongly depend on the distance between crowders and the electric double layer thickness. For the first time, we derived an analytical formula to calculate the diffusion coefficient of a crowded system as a function of volume fraction and background salt concentration. The predicted values of diffusion coefficient evaluated from the analytical model are in good agreement with our homogenization and continuum models. We finally demonstrate the application of this methodology toward modeling calcium release units in highly-structured intracellular domains.