Network insights on oxaliplatin anti-cancer mechanisms

Benign cellular pathways are complex, malignant ones more so [9]. One protein’s presence can have a drastic effect on a multitude of proteins and pathways and the interrelationships between seemingly unrelated proteins and pathways is only now coming to light [10]. Interfering with transcription of a single protein or even halting it altogether may have a positive result in patient care, yet over time the disease evolves to activate and upregulate other proteins and pathways rendering some cancers incredibly resistant to chemotherapeutic treatments [11]. What essentially occurs is the network component which has been altered or inhibited by treatment is compensated for by other members of a broader network. Thus, an emerging goal in cancer treatment is not just the targeting of single proteins, but engaging an entire network and its related networks to halt or reverse the disease [12]. Moreover, oxaliplatin has been implicated in Pt-protein adducts, and more and more significance is being placed upon this interaction as something more than just drug inactivation [13]. The effects of oxaliplatin protein adducts remain poorly understood but further research can yield results shedding light on key interactions relating to efficacy and toxicity.

Initial analyses of high throughput data might suggest at the outset that a drug works through activation of previously elucidated pathways [14]. To address this, we carried out microarray expression profiling and network modeling to understand the oxaliplatin response signatures using a genetically complex pancreatic cancer cell line model. The design and analysis of data has been provided in our previous publications [15, 16]. In this study, various mechanisms suggestive of oxaliplatin and its MDM2 inhibitor combination were investigated using systems science. However, single agent oxaliplatin protein network changes were not investigated. Using Capan-2 (wt-p53) model, we investigated the gene expression that was followed by Ingenuity network analysis (IPA) for pathway interactions. Table 1, shows data of known biochemical interactions of oxaliplatin. Among the 35 molecules included within the nucleotide excision repair pathway, nearly 80% were upregulated indicating DNA damage and its concurrent response. Similarly, in 56 molecules related to the RNA polymerase II complex assembly mechanism, nearly 80% were upregulated and nearly 70% of CHK related proteins involved in cell cycle checkpoint control were upregulated with more than 70% of death receptor signaling components activated as well. To our surprise we observed activation in competing carcinogenesis pathways as well. 379 molecules associated with broad cancer promoting mechanisms were found to be upregulated. In Table 2, disease prioritizing analyses showed up-regulation in 45 molecules associated to renal carcinoma, 47 molecules associated to small cell lung cancer, 39 molecules of non-small cell lung cancer signaling, 116 molecules implicated in colorectal cancer signaling, 53 molecules in prostate cancer signaling, 82 molecules of acute myeloid leukemia signaling, 46 molecules in thyroid cancer, 92 molecules of bladder cancer signaling, 65 molecules of pancreatic adenocarcinoma signaling, and 64 molecules of ovarian cancer signaling. Alternate pathways implicated in tumor growth also exhibited upregulation such as VEGF signaling with 49 molecules.

These results point to a remarkable complexity of drug action, but it also speaks to the multiple layers of competition to efficacy that exist. Network theory helps determine the interrelationships between these various pathways and in doing so, better targeting can be achieved simply by understanding the big molecular picture. Many of these pathways feature molecules which are related to other pathways mentioned, whether directly or indirectly. This crosstalk between components is only now being understood in a clinical context and has spurred the development of network pharmacology as a tool to better elucidate drug targets and better design their drugs. The existence of such complications explains the vast difficulties encountered in clinical trials with novel compounds. The nuanced differences between in vitro and in vivo models for the development of pharmaceuticals have been long known and have expressed themselves in unanticipated toxicities uncovered in initial trials. By further validating the network components of a disease, its relationship to other networks and a compound’s effect on these networks, toxicity can be better predicted based on a consistent system model which may be malleable to each patient’s own unique molecular disease network. Thus, precise response can be better predicted long before a drug enters the clinical trial stage.

Based on the vast body of science available and results from high throughput data, we are able to compile a preliminary explanation in a molecular network sense of why oxaliplatin as a single agent simply does not work (summarized in Table 3). In response to this, a more in depth analysis of combinations with oxaliplatin and other family member compounds such as cisplatin or their combination partners (e.g. FOLFOX) is required to better assess true efficacy in patients. In doing so, better combinations can be developed using already approved compounds present in the market.

FOS signaling was shown activated with treatment of oxaliplatin in relation to control. The FOS family of proteins consists of four members forming an AP-1 transcription factor complex, FOS, FOSB, FOSL1 and FOSL2. Early studies indicated marked expression of FOS in many malignancies including a possible link between increased c-fos mRNA expression and relapse of acute childhood lymphoblastic leukemia [17]. Similarly, in some non small cell lung carcinoma patients survivability decreased with overexpression of FOS products [18]. Furthermore, recent studies demonstrate the AP-1 complex being at a significant upstream point, possibly regulating the expression of critical pathway signaling components such as the little understood AKR1B10 marker in carcinogenesis [19]. In breast cancer, studies have implicated FOS as a critical member in the activation of MMP-9 by S1P, resulting in enhanced invasiveness and migration of breast cancer cells [20]. Concurrently, FOS was demonstrated as a possible prognostic marker for clinical decision making with BCL2 in endocrine breast cancer tumors [21].

Faulty NOTCH expression as well as related family members have been implicated in multiple human diseases since association with T-cell acute lymphoblastic leukemia in 1991 [22]. Many studies have demonstrated higher expression of NOTCH in human malignancies ranging from head and neck cancers, to cervical, lung and pancreatic cancers and also implicating it in epithelial-mesenchymal transition (EMT), a major signal of cancer-stem cells in resistant malignancies [23]. NOTCH signaling has been observed in crosstalk with NF-κB signaling, resulting in increased transcription of NOTCH targets and concurrent increase of carcinogenesis [24]. A multitude of other crosstalk targets have been elucidated, placing NOTCH at the center of a very complicated network. These findings as well as further elucidation of NOTCH signaling downstream effects render it an appealing target in cancer treatment [25]. Figure 3 demonstrates the array of related expression changes of NOTCH signaling with oxaliplatin treatments.

FAF1, a component of the Fas death-inducing signaling complex, was remarkaby under expressed relative to control. Contemporary observations have shown a repeated loss or under expression of FAF1 in some cancers, possibly indicating anti-tumor activity. Experiments have demonstrated a physical interaction between FAF1 and β-catenin whereas an increase in FAF1 resulted in decreased Wnt reporter activity and an increase in β-catenin cytosolic degradation [26]. Simultaneously, FAF1 down regulation has been associated with aberrant NF-κB function.

A member of the NCS family of proteins, VILIP-1 (a product of the VSNL-1 gene) is found expressed primarily in the tissues of the CNS as well as other peripheral tissues in humans and rats [27]. Various experiments carried out have demonstrated an inverse correlation between VILIP-1 expression and tumor aggression, whereby a down regulation is observed in various human squamous cell carcinomas (SCC) through activation of cAMP- or cGMP-signaling pathways [28]. In one elegant study, transfection of mouse SCC lines with VILIP-1 cDNA resulted in an observable reduction of invasiveness correlating with elevated cAMP levels and reduced MMP-9 and RhoA activity [29]. Loss of VILIP-1 expression paralleled clinicopathological features of SCC in terms of tumor invasiveness as well as local lymph node metastasis. Further evidence has linked VILIP-1 loss with tumor development in neuroblastoma and esophageal SCC and analyses have shown a marked VILIP-1 loss in prostate, lung, ovarian, renal, melanoma and leukemia cancer cells [30]. Taken together, these findings suggest an intricate involvement of VILIP-1 in tumor suppression, of which the precise mechanisms are still being explored.

As drugs are developed and approved, we still lack a coherent and consistent method of predicting efficacy and toxicity in a human model. Based on this preliminary data, we are able to construct roughly a theoretical predictive model of toxicity based on network response. Oxaliplatin serves only as a starting point for determining the veracity of a network approach to drug interactions. Fundamentally, the cancer network is dynamic and drug interactions contribute another complex degree of relationships. While the efficacy of oxaliplatin is not in dispute, the degree of efficacy or efficiency of treatment is something yet unexplored outside this preliminary review.

The degree to which treatment efficiency is successful would be highly dependent on advancements in patient and disease stratification. By characterizing the precise network regulators of a cancer type, further elucidating the nuanced differences between subtypes and further between patients, a targeting motif can be developed. Utilizing this same information, key predictors of patient response can be elucidated, further enhancing clinical efficacy of a drug or guiding combination treatments that may otherwise not already exist. In using oxaliplatin we determined the possibility of incidental activation of undesired networks, competing with the efficacy if the drug. By isolating some of these key nodes, a better and more effective targeting strategy can take place increasing the efficacy of a single agent even with already available treatments used off-label.