Despite the prevalence of cancer as the second leading cause of death worldwide, current cancer therapies carry debilitating side effects. These effects, ranging from anemia to "brain fog", necessitate novel treatments that prevent metastasis and inhibit existing proliferation. New hope has taken the form of inhibition of aquaporins (AQPs), a group of transmembrane water channels with tissue-specific distributions and a wide range of physiological functions, from cell migration required for angiogenesis to regulation of neural signal transduction. Past research suggests that the expression of AQP1, AQP5, AQP8, and AQP9 is significantly correlated to tumor type, severity, migration, and prognosis because AQPs are closely associated with cancer biological functions. The overall objective of this study is thus to review novel candidates for AQP-specific modulators that were identified in past studies. We hypothesized that mitigating in silico tumor growth, the chemical reagents HTS13286 and suberoylanilide hydroxamic acid (SAHA) effectively inhibit AQP9 and AQP5, respectively. This study aimed to do so by comparing the inhibition constant (Ki) of AQP9 and HTS13286 to the K _i of AQP9 and Phloretin (a known AQP9 inhibitor), and repeating the process with the K _i of AQP5 and SAHA and the K _i of AQP5 and acetazolamide (AZA), a known AQP5 inhibitor. A proof-of-concept in silico model identified effects of AQP modulation on tumor growth. The model and findings are of great importance in respect to the cancer research field, as they provide a fast, inexpensive, and precise method to identify therapeutic cancer targets.

A variety of materials were used during the process, including AutoDock Vina, PyMOL, PyRx, MGL Tools, MATLAB: SimBiology toolkit, and Protein Data Bank (PDB) files of AQP9, AQP5, SAHA, HTS13286, phloretin, and AZA. To simulate the docking, the required PDB files were downloaded from sites such as RCSB Protein Data Bank. PyRx AutoDock Vina was then run, with AQP9 as the macromolecule and HTS13286 as the ligand. From this docking, we recorded the binding affinities and took the one with the highest absolute value, which is traditionally accepted as the optimal configuration [1]. This procedure was repeated with AQP5 and SAHA, using PyMOL to visualize the binding modes and find the bound amino acids. The bound molecules were exported from PyMOL to LigPlot, and a map of the ligand bonds and residues were generated, in order to gain a better understanding of the complex. The second part of the study involved the MATLAB model, where the SimBiology model for PK/PD of anticancer drugs was first uploaded. A PK/PD model, based on in vivo animal studies and proposed by Simeoni et al., was utilized to quantify the effect of anticancer drugs on tumor growth kinetics. This model described the growth as a biphasic process with initial exponential growth followed by linear growth. They described the growth rate of the proliferating tumor cells using the equation (ƛ ₀ x ₁ ) / [1 + (ƛ ₀ / ƛ ₁ w)ɸ] ^{(1/ ɸ)} , where ƛ ₀ , ƛ ₁ , ɸ are tumor growth parameters, x ₁ is the weight of the proliferating tumor cells, and w is the total tumor weight. In the absence of a drug, the tumor is comprised only of proliferating cells — therefore, w = x ₁ . In the presence of an anticancer agent, Simeoni et al. assumed that a fraction of the proliferating cells transform into non-proliferating cells. The rate of this transformation was assumed to be a function of the drug concentration and an efficacy factor, k ₂ . The body’s immune functions eventually cleared the non-proliferating cells from the tumor system [4]. Based on this model, a two-compartment method was programmed in MATLAB to apply to AQPs and calculated two equations to model the effectiveness of the inhibitors on tumor growth, as discussed in Results. The specific small molecule inhibitor, AQP5, and the bonded complex were then added. We refined a reaction on MATLAB using the small molecule inhibitor and AQP as reactants and bonded complex as products to determine the concentrations. Lastly, the MATLAB model was run with the SAHA-AQP5 complex, as it would better resemble a general tumor model due to AQP5's implication in colorectal cancer [3].

The binding affinities (kcal/mol) were converted to the inhibitory constant (K_i) values, using the equation K_i = e^{(deltaG/(RT))}, where R is the universal gas constant (1.986 cal/mol K) and K_i is the concentration of an enzymatic inhibitor at which 50% inhibition of reaction rate is observed. Thus, the K_i reflects the potency of an inhibitor, with a smaller K_i indicating a stronger inhibitor and vice versa. SAHA is an effective inhibitor of AQP5 because the K_i of AQP5 & AZA (the known AQP5 inhibitor) was identical to the K_i of AQP5 & SAHA, at 1.138 nM. Furthermore, the K_i of AQP9 & Phloretin (the known AQP9 inhibitor) was greater than the K_i of AQP9 & HTS13286, demonstrating that HTS13286 is an effective inhibitor of AQP9, because a smaller amount of that reagent, as compared to Phloretin, is needed for 50% inhibition of AQP9.

Using MATLAB to calculate the first reaction, which was the amount of AQP5 not inhibited, the first reaction was found to be 0.0167 (Peripheral.AQP5 + Peripheral.inhibitor) 0.6. The value of 0.0167 was obtained from the K_i value for AQP5 & AZA and AQP5 & SAHA, and the 0.6 coefficient was the proportion of AQP that would be inhibited, based off of an in vivo study of AZA [2]. Reaction 2 reflected the weight of cells converted to non-proliferating cells, as calculated by k₂(bonded complex)x₁ where k₂ = 0.6923. k₂ was found from 1 - the proportion of cell viability with 100% inhibited AQP5, based on values from an AQP5 study [5]. When the MATLAB graph was run, the general trend showed that tumor weight decreased over the course of 40 days.

The data obtained from the molecular visualization software supported the hypothesis that HTS13286 effectively inhibits AQP9. Additionally, we discovered the optimal bonding configuration between the ligand and macromolecule for maximum inhibition (K_i = 1.138 nM). Because this is relatively similar to the K_i of those in previous studies (37.673 nM), we upheld those results. From the models, binding energies, and LigPlots of the simulation between SAHA and AQP5, we found that the K_i of SAHA and AQP5 was equal to the K_i of AZA and AQP5 (K_i = 16.99 nM), indicating that SAHA is an effective inhibitor of AQP5. Furthermore, we showed that a novel method, utilizing MATLAB, could reliably indicate the effectiveness of the inhibited complex on tumor growth based on PK/PD and existing modeling principles. This system has potential for great societal impact as it allows for more expedited and cost-effective identification of therapeutic cancer targets.