Abstract

   Breast cancer is the second most common types of cancer worldwide.
   Molecular strategies have developed rapidly; however, novel treatments
   strategies with high efficacy and lower toxicity are still urgently
   demanded. Notably, biological networks estimated from microarray data
   and functional activity network analysis could be utilized to identify
   and validate potential targets. In this study, two microarray data
   ([45]GSE13477, [46]GSE31192) were firstly selected, and analyzed by
   multi-functional activity network analysis to generate the core
   protein-protein-interaction (PPI) network. Several potential targets
   were subsequently identified and c-Met and poly (ADP-ribose)
   polymerase-1 (PARP-1) were manually chosen as the key targets in breast
   cancer. Furthermore, virtual screening and molecular dynamics (MD)
   simulations were utilized to recognize novel c-Met/PARP-1 inhibitors in
   Specs products database. Three small molecules, namely, ZINC19909930,
   ZINC20032678 and ZINC13562414 were selected. Additionally, these
   compounds were synthesized, and two breast cancer cell lines,
   MDA-MB-231 and MCF-7 cells were used to validate our bioinformatic
   findings in vitro. MTT assay and Hoechst staining showed that
   ZINC20032678 significantly induced breast cancer cell death, which was
   mediated through apoptosis by flow cytometry. Furthermore, ZINC20032678
   was shown to target the active sites of the both targets and
   recruitment of downstream apoptotic signaling pathways, eventually
   inducing breast cancer cell apoptosis. Collectively, our findings not
   only offer systems biology approaches based drug target identification,
   but also provide the new clues for developing novel inhibitors for
   future breast cancer research.

   Keywords: breast cancer, c-Met, PARP-1, systems biology, apoptosis,
   drug discovery

Introduction

   Breast cancer is one of the most leading causes of cancer death among
   women. It is reported that the overall pathogenic incidence rate in
   women remains generally stable, whereas the breast cancer incidence has
   slight increased from 2004 to 2013 [47]^1.Currently, there are four
   main molecular subtypes of breast cancer, referring as luminal, HER2,
   normal-like and basal. Of note, targeted therapies have been raised
   great attention in science community. It can target specific targets
   expressed in/on the surface of cancer cells which are involved in
   carcinogenesis and/or tumor growth [48]^2. For example, tamoxifen for
   treatment of ER positive tumors and trastuzumab in the treatment of
   HER2 positive tumors in breast cancer have been extensively studied
   [49]^3. However, it is worth knowing that most basal-like breast
   cancers do not express ER, PR and HER2, the two subgroups are not
   mutually exclusive. And, 80% of basal-like breast cancers are
   triple-negative and 80% of triple-negative breast cancers exert a
   basal-type phenotype [50]^4. Therefore, fundamental breakthroughs,
   particularity new therapeutic targets and new drugs are most urgent
   needed in breast cancer research.

   It is known to all that the entire DNA microarrays has been used in
   cancer drug development and facilitate clinical applications for years
   [51]^5. Since measurement of the sequence and expression of potential
   targets is greatly facilitated by microarray technology, it is of great
   importance to generate new clues to gene function which can help to
   identify appropriate targets for therapeutic intervention. Hitherto, a
   great deal of drug targets discovery and validation have been well
   described by microarray analysis [52]^6^, [53]^7. Additionally, DAVID
   online database, an integrated biological knowledge base and analytic
   tool, can extract biological meaning from large gene/protein lists
   [54]^8. Moreover, STRING database includes known and predicted PPIs
   which stem from various databases [55]^9. Combining the microarray data
   and functional activity network analysis of differentially expressed
   genes by using DAVID (including KEGG pathway and GO analysis) and
   STRING were accelerate the drug target development [56]^10. To our
   knowledge, a number of hub proteins/targets were successfully
   identified by using abovementioned approaches [57]^11^, [58]^12.

   In this study, two microarray data ([59]GSE13477, [60]GSE31192) were
   firstly analyzed to obtain consensus results of differently expressed
   genes. Then, the up- regulated genes were selected to identify the
   candidate genes by using GO, KEGG pathway and STRING analysis.
   According to the network results and consulting numerous references,