Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) is a pathogen of immense public health concern. Efforts to control the disease have only proven mildly successful, and the disease will likely continue to cause excessive fatalities until effective preventative measures (such as a vaccine) are developed. It is important to better understand SARS-CoV-2 pathogenesis and population susceptibility to infection in order to develop effective disease management strategies. To this end, physiologically relevant mathematical modeling can provide a robust in silico tool to understand COVID-19 pathophysiology and the in vivo dynamics of SARS-CoV-2. Guided by ACE2-tropism (ACE2 receptor dependency of the virus for infection) of the virus, and by incorporating cellular-scale viral dynamics and innate and adaptive immune response, we have developed a multiscale mechanistic model for simulating the time-dependent evolution of viral load distribution in susceptible organs of the body. Following calibration with in vivo and clinical data, we used the model to simulate viral load progression in a virtual patient with varying degrees of compromised immune status. Further, to understand the effects of physiological factors and underlying conditions on viral load dynamics, we conducted global sensitivity analysis of model parameters and ranked them for their significance in governing clearance of viral load from the body. Antiviral drug therapy, interferon therapy, and their combination was simulated to study the effects on viral load kinetics of SARS-CoV-2. The model highlights the importance of innate immunity (interferons and resident macrophages) in controlling viral load, and the significance of timing when initiating therapy following infection.