One possible mechanism for AIPC action is the ligand independent activation of androgen receptor (AR) via cross talk with other signaling pathways. Experimental observations have, in fact, revealed that proteins in the mitogen activate protein kinase (MAPK) cascade act as kinases for this outlaw AR activation. Specifically, ligand independent activation of the transmembrane growth receptor Her2 has been observed to initiate the MAPK cascade, resulting in AR activation, increased PSA expression and an increased rate of proliferation in several prostate cancer model cell lines. The proportion of active Her2 is strongly dependent on the expression level of the inter cellular form of Prostatic acid phosphatase (PAcP). Not surprisingly, inter cellular PAcP expression negatively correlates with prostate cancer tumor grade.
Currently, mathematical modeling has not played a major role in the study of the molecular basis of disease progression and the development of therapeutics. In this study we explored the working hypothesis that ultrascale mathematical modeling of complex protein interaction networks may be used to identify critical AIPC mechanisms despite model uncertainties. We present the results of a kinetic model for the outlaw activation of AR via MAPK and the regulation of several key proteins. The first generation ordinary differential equation prostate model architecture was composed of a collection of independently verified subsystems, such as MAPK signaling and translation initiation, together with other molecular interactions supported by the experimental literature. In total, our prostate model described the dynamics of 212 species and 384 interactions, where mass action kinetics were assumed for each interaction rate. Due to the limited data, the model parameters were under defined and poorly constrained. To properly account for the effect of unconstrained parameters a Monte Carlo ensemble approach was employed for predictions and for the estimation of prediction errors. Although model parameters in the ensemble varied over several orders of magnitude, the predictions were constrained. This observation demonstrated the reliability of model predictions despite parametric uncertainty.
The preliminary model captured the androgen response of both the androgen-dependent and androgen-independent clones of the LNCaP prostate cancer cell-line. The model also qualitatively reproduced experimental data pertaining to key proteins, such as Her, PAcP, and the kinases of the primary MAPK cascade. Through a preliminary sensitivity analysis we observed that models of the more aggressive, androgen-independent, clones were notably more robust to perturbation. This result may provide insight into a cancer cell's ability to thrive in an environment where healthy cells would either growth arrest or die. However, our analysis, also, predicted that the relative sensitivity of certain mechanisms pertaining to translation in androgen-independent clones to be elevated compared to androgen-dependent clones. When taken together, these results suggest that the manipulation of the translation subsystem may have a larger impact on cancer cells relative to healthy cells, and therefore, suggests these mechanisms may be low side effect, novel therapeutic targets for the treatment of AIPC.