Anticancer genes (NOXA, PAR-4, TRAIL) are de-regulated in breast cancer patients and can be targeted by using a ribosomal inactivating plant protein (riproximin)

Anticancer genes are an endogenous defense against transformed cells as they impose antineoplastic effects upon ectopic expression. Profiling the expression of these genes is fundamental for exploring their prognostic and therapeutic relevance in cancers. Natural compounds can upregulate anticancer genes in malignant cells and thus be useful for therapeutic purposes. In this study, we identified the expression levels of anticancer genes in breast cancer clinical isolates. In addition, the purified and sequenced plant protein (riproximin) was evaluated for its potential to induce anticancer genes in two breast cancer cell lines. Expression profiles of three anticancer genes (NOXA, PAR-4, TRAIL) were identified by immunohistochemistry in 45 breast cancer clinical isolates. Breast cancer cells were exposed to riproximin and expression of the anticancer genes was determined by microarray, real-time PCR and western blot methodologies. Lastly, a bioinformatic approach was adopted to highlight the molecular/functional significance of the anticancer genes. NOXA expression was evenly de-regulated among the clinical isolates, while PAR-4 was significantly down-regulated in majority of the breast cancer tissues. In contrast, TRAIL expression was increased in most of the clinical samples. Expression levels of the anticancer genes followed a distinct trend in accordance with the disease severity. Riproximin showed a substantial potential of inducing expression of the anticancer genes in breast cancer cells at transcriptomic and protein levels. The bioinformatic approach revealed involvement of anticancer genes in multiple cellular functions and signaling cascades. Anticancer genes were de-regulated and showed discrete expression patterns in breast cancer patient samples. Riproximin effectively induced the expression of selected anticancer genes in breast cancer cells.


Introduction
Anticancer genes (ACGs) comprise an endogenous defense against cancer cells, as they can induce antineoplastic effects when expressed ectopically. So far, a few genes (~10) have been assigned with the status of ACGs, while a systemic search for new members of this family is an on-going process [1,2]. Proteins translating from ACGs are complex in structures and interact specifically with their cellular counterparts to initiate cancer cell-specific death mechanisms. ACGs mediated antineoplastic effects include ERstress, mitotic catastrophe, apoptosis and autophagy [3][4][5][6][7]. Currently, ACGs are being investigated actively for their potential role in cancer pathogenesis and therapeutic relevance. As per available data, low expression of ACGs is associated with poor prognosis, low survival rates, relapse of the disease conditions and resistance towards treatment modalities. Increased expression of ACGs is reported to sensitize cancer cells towards radio/chemotherapy and often induces synergistic effects in combination with antineoplastic-agents [8][9][10][11][12][13][14][15][16][17]. As far as the therapeutic domain is concerned, most of the known ACGs are restricted to preclinical investigations till today. A predominant challenge is to find suitable delivery methods to have higher expression levels of these genes in cancer cells. In this context, ACG based immunotoxins, antibody conjugates and genetic engineering approaches are being tested [18,19]. Apart from these synthetic options, a highly neglected field is finding natural compounds, which can induce the expression of ACGs in cancer cells. Identification of natural compounds with a safe physiological/ biological profile is deemed crucial for exploiting ACGs as "the Achilles Heel of cancer cells".
Ximenia americana is a plant that grows in tropical and subtropical areas of African and American countries. Kernels of this plant have long been used in African traditional medicine by local healers as a treatment for cancer [20]. Almost fifteen years ago, our group isolated and purified different protein fractions from aqueous extracts of X. americana kernels and tested their corresponding anticancer properties. A protein fraction with typical size of ~ 60 kDa having substantial anticancer potential was identified and termed "Riproximin" [21,22]. DNA/ peptide sequences showed that riproximin is a member of a big family consisting of ribosome-inactivating proteins (RIPs). RIPs are known for their catalytic nature and inhibit the translation procedure irreversibly in target cells by altering the 28 S rRNA subunit [23,24]. As far as the molecular structure is concerned, riproximin is a heterodimer with two polypeptide chains (Aand B-chains) held together by an intermolecular disulphide bridge. The B-chain of riproximin has lectin-like properties and is responsible for binding to cell surface glycans, while the A-chain is accountable for subsequent catalytic activity [25][26][27]. Riproximin has demonstrated substantial antineoplastic effects against a variety of cancer cell lines including breast, colorectal, leukaemia, pancreatic and prostate, while sparing normal healthy cells [28][29][30][31][32]. In vivo studies are also documented where riproximin showed anticancer effects in colorectal and pancreatic cancer liver metastasis rat models [21,22,30,33]. Reported antineoplastic effects of riproximin include ER-stress, induction of unfolded protein response, cell cycle arrest, apoptosis and autophagy. In addition, riproximin inhibits various functional properties of cancer cells including proliferation, migration and colony formation [31,32].
In this study, we hypothesized that ACGs are differentially expressed in breast cancer and can be regulated via natural compounds for therapeutic purposes. To support this hypothesis, we identified the expressional levels of multiple ACGs (NOXA, PAR-4 and TRAIL) in 45 tumour samples of breast cancer via immunohistochemistry. Afterwards, the potential of riproximin to induce ACGs in two molecularly distinct breast cancer cell lines (MDA-MB-231 and MCF-7) was evaluated at transcriptome and proteome levels by using microarray, real-time PCR and western blot methodologies. Lastly, bioinformatic tools were applied to highlight the molecular/ functional importance of ACGs inside a cellular environment.

Clinical samples and histopathological analysis
A total of 45 female patients, 23-52 years of age, presenting with primary breast cancers at the oncology units of the Institute of Nuclear Medicine and Oncology (INMOL), Lahore, Services Hospital, Lahore and the Allied Hospital, Faisalabad, Pakistan were included. After a complete radiological examination and histological diagnosis of malignancy by true-cut biopsy, patients were recruited for the study after obtaining their written informed consent. Patients with recurrence or on follow-up for previous malignancy were excluded. Modified radical mastectomy specimens were fixed in 10% neutral buffered formalin and grossly examined according to the College of American Pathologist (CAP) protocol. Paraffin embedded tissue sections were prepared and stained with H&E stain for confirmation of the histological diagnosis and grading was carried out following the Elston-Ellis modification of the Scarff-Bloom-Richardson grading system [34].

Immunohistochemistry of NOXA, PAR-4 and TRAIL
About 4-6 μm sections were taken on positively charged albumin coated slides for immunohistochemistry. The slides were placed in a hot air oven at 70 °C for 30 min, followed by three washing steps in xylene, for 1 min each. Paraffin removal was continued by three washing steps in alcohol, 1 min each. Then, the sections were treated with H 2 O 2 for 5 min to inhibit endogenous peroxidase and rehydrated by washing in H 2 O. Epitope retrieval (unmasking of antigen) was done by incubation in ethylenediaminetetraacetic acid (EDTA) buffer for 50 min at pH8 and 90 °C. Slides were held at room temperature for 20 min and then washed in 1× Phosphate Buffered Saline (PBS, pH: 7.4) for 10 min. A blocking reagent was applied for 10 min and slides were again washed with 1× PBS for 1 min. Sections were incubated for 30 min with the primary antibodies for ER (cat#PA0151, Leica Biosystem), PR (cat#PA0312, Leica Biosystem), HER2neu (cat#PA0983, Leica Biosystem), NOXA (cat#PA5-19977, Invitrogen), PAR-4 (cat#PA5-77686, Invitrogen) and TRAIL (cat#PA5-102584, Invitrogen). After washing in PBS, sections were incubated for 20 min with the secondary antibodies (cat#DS9800, Leica Biosystem) and again washed with 1× PBS. 3,3-Diaminobenzidine (DAB) was used as chromogen for 5-8 min and again the slides were washed with 1× PBS for 5 min. For counter staining, slides were dipped in haematoxylin and cleaned in xylene for 5 min.

3
Finally, slides were mounted with Dibutylphthalate Polystyrene Xylene (DPX) and examined under a microscope.
For NOXA, positive staining was localized to the tumour cell cytoplasm and was scored based on intensity (0/1/2/3) and percent tumour cell positivity [grouped into quartiles (0-4)]. Intensity and immune-percent were multiplied to yield a final immune-score. For purposes of analysis, NOXA staining was dichotomized as negative/low versus elevated, where low was expanded to an immune-score of ≤ 4 while a score > 4 was considered as an elevated expression [35]. For PAR-4 staining, immunohistochemical staining was assessed semi-quantitatively by multiplying both the intensity score (0/1/2/3) and the proportion of positive tumour cells (0, 0%; 1, 0-10%; 2, 10-50%; 3, 50-100%), (maximum possible, 9). For the statistical analysis, the weighted scores were grouped into two categories where scores of 0-3 were considered negative and 4-9 positive. For TRAIL immunohistochemistry, a binominal category was adopted. The absence of TRAIL expression was defined as the presence of cytoplasmic staining in less than 10% of tumour cells, while the presence of staining in at least 10% of tumour cells was considered as evidence of TRAIL expression [36]. For ER and PR, the Tumour Allred scoring scheme was adopted according to previously reported criteria [37]. A cut-off to define receptor positivity for ER and PR was an Allred score ≥3, the internationally accepted cut-off. High scores were defined as Allred 6-8, and low scores as Allred 3-5. Regarding HER2neu membrane staining, a previously published scoring scheme was adopted [38].

Real-time PCR analysis
Cells of the breast cancer cell lines (MDA-MB-231 and MCF-7) were exposed to riproximin and expression modulations were identified at transcriptome level by qRT-PCR methodology for three ACGs (NOXA, PAR-4, TRAIL). To that purpose, the cell lines were cultured in 6-well culture plates (1.5 × 10 5 cells/well/2 ml medium) and exposed for 48 h to three inhibitory concentrations of riproximin (IC 25 , IC 50 , IC 75 ), which were selected based on our previous data [32]. Following the exposure period, cells were harvested, cell pellets collected, total RNA extracted by using a commercial kit (cat#K0731, Thermo-Fisher Scientific) and cDNA was synthesized (cat#K1622, Thermo-Fisher Scientific). Prepared samples were subjected to transcript detection of the ACGs by using a mixture of gene specific primers (NOXA: ATT ACC GCT GGC CTA CTG TG, CAA TGT GCT GAG TTG GCA CT, PAR-4: GCA TGC ACA CTA AAA ACC AAAA, TGT GTC CCA GTG TTA TTC TTCAA, TRAIL: ACG ACA AAC AAA TGG TCC AA, AGC TCA AAT ATT CCC CCT TGA) and SybrGreen/ROX master mix (cat#K0221, Thermo-Fisher Scientific) in a QuantStudio 3 real-time PCR system. All samples were processed in triplicate while expression levels of untreated cells were used as controls. Expression of a reference gene (HPRT1) was used to normalize the data, whereas fold changes were identified by the 2−△△Ct method.

Microarray analysis
Microarray analysis was performed to determine the modifications in gene expression with minor modifications to our previously published protocol [28]. Briefly, MDA-MB-231 cells were exposed to riproximin (IC 25 , IC 50 , IC 75 ) for 24, 48 and 72 h followed by RNA extraction with a RNeasy Mini kit (cat#74004, Qiagen). The quality of the extracted RNA was determined by the total RNA Nano chip assay on an Agilent 2100 Bioanalyzer (Agilent Technologies). RNA samples with sufficient RNA Integrity Number values (≥ 8.8) were selected for the expression profiling.

Western blot analysis
Effects of riproximin exposure on protein levels of ACGs (NOXA, PAR-4, TRAIL) in breast cancer cell lines were identified via western blot methodology. Briefly, the cells were cultured in 25cm 2 flasks (1.0 × 10 6 cells/flask/5 ml medium) and exposed to different concentrations of riproximin (IC 25 , IC 50 ) for 48 h. Following the exposure period, cell pellets were collected and lysed with RIPA lysis buffer. Extracted proteins were quantified via Bradford assay and a total of 30 µg protein/sample was subjected to electrophoresis on 4-12% gradient polyacrylamide SDS gels. Afterwards, proteins were transferred onto nitrocellulose membrane and incubated for 2 h at room temperature with specific primary antibodies for NOXA, PAR-4 or TRAIL (same as used for immunohistochemistry). Later on, membranes were rinsed with TBST buffer followed by incubation with alkalinephosphatase (AP) conjugated secondary antibody for 1 h at 1 3 room temperature. Immunoreactive proteins were visualized with BCIP/NBT tablets (cat# B5655, Sigma), bands were analysed by ImageJ software and levels of β-actin were used to normalize the protein expression data sets.

Bioinformatics analysis
The protein-protein interaction (PPI) network was created by importing the physical subnetwork of target genes from STRING via StringApp plugin into Cytoscape version 3.9.1. The maximum of 100 interactors with a minimum interaction score of 0.15 were imported to construct the network. Cytoscape plugin NetworkAnalyzer version 4.4.8 was used to assess the topological parameters. PPI network clusters was detected by Markov Cluster Algorithm (MCL) clustering using Clustermaker2 app version 2.2 in Cytoscape. Clustering was performed with inflation value of 3.0 and stringdb score as array source. The genes' list of the PPI network was imported into R and their corresponding ENTREZ IDs were fetched using "biomaRt" package version 2.52.0. Gene Ontology (GO) categories (Biological Process-BP, Molecular Function-MF and Cellular Component-CC) and KEGG pathways based functional enrichment was performed by "ClusterProfiler" package version 4.4.4 with the q-value cut off <0.01. The three GO categories were visualized by "enrichplot" package version 1.16.2, while KEGG based enriched biological pathways were visualized by "GOplot" package version 1.0.2.

Clinical diagnosis and molecular sub-types of breast cancer patients
In the current study, a total of 45 cases of adult female patients with breast lumps having complete radiological evaluation at the time of diagnosis were included. The age of these patients ranged from 23 to 52 years; the mean age was 44.1 ± 10.03 years. Taking age groups into consideration, n= 20 (45.3%) patients were <40 years of age and n = 25 (55.5%) patients were ≥ 40 years of age. The size of the tumour ranged from 1.8 to 21.5 cm with a mean value of 13.06 ± 4.2 cm. Regarding the histological type, invasive ductal carcinoma (IDC) was diagnosed in all 45 cases. The Nottingham scoring and grading system showed n = 8 (17.8%) cases in grade 1, n = 16 (35.6%) cases in grade 2 and n = 21 (46.6%) cases in grade 3. Clinical staging of these breast cancer cases was done according to the American Joint Committee on Cancer (AJCC) staging manual. There was n = 7 cases (15.5%) in stage I, n = 15 cases (33.3%) in stage II, n = 18 cases (40%) in stage III and 5 cases (11.1%) in stage IV of breast cancer. For the purpose of analysis, the patients were subdivided into early (stage I/II) and advanced (III/IV) stage groups as 22 (48.8%) and 23 (51.2%), respectively ( Table 5). The molecular subtyping of breast cancer in increasing order of frequency were Her-2 type (HR-/ HER2+), luminal B (HR+/HER2+), luminal A (HR+/ HER2-) and basal like (HR-/HER2-), respectively (Table 1).

Differential expression of anticancer genes in breast cancer tissues
Expression frequency of NOXA, PAR-4 and TRAIL (Table 2) and molecular subtypes of breast cancer samples (Table 3) was analysed in order to observe their statistical association. Histopathological expression levels of NOXA, PAR-4 and TRAIL genes were identified by immunohistochemistry. Examples of low or high expression are shown in Fig. 1. The results shown in Table 3 depict that higher expression of TRAIL and lower expression of PAR-4 correlated significantly (p < 0.05) with the luminal A and basal type of breast carcinomas, thus reflecting poor response to treatment and an unfavourable outcome. Expression of NOXA did not show any significant frequency distribution or correlation among the molecular subtypes of breast cancer. The cases of breast carcinoma with grade 1 showed higher frequency of TRAIL negative and NOXA elevated expression with no significant association. Grade 2 carcinomas showed a higher frequency of TRAIL positive, NOXA low/ negative and PAR-4 negative expression with no significant association. Grade 3 cases demonstrated significant association with higher TRAIL expression and negative PAR-4 expression ( Table 4). When the clinical stage groups were   statistical significance. In contrast, NOXA neither showed any significant frequency distribution nor statistical association with the clinical stage groups (Table 5).

Riproximin mediated induction of transcript/protein of anticancer genes
Riproximin is known for inducing anticancer effects and altering the expression of multiple genes in breast cancer cells [32]. In this study, we particularly evaluated the impact of purified riproximin exposure on expressional changes at transcriptome and protein levels of the ACGs in two breast cancer cell lines. To that purpose, cell lines were exposed to riproximin (IC 25 , IC 50 , IC 75 ) followed by qRT-PCR and western blot for expression analysis. At the transcriptome level, ACGs were upregulated by enlarge in the two cell lines with a distinct response towards riproximin treatment (Fig. 3A). Maximum induction in MDA-MB-231 cells was observed for NOXA (3.9fold), followed by PAR-4 (2.8fold) and TRAIL (1.5fold), respectively. In MCF-7 cells, TRAIL was the most up-regulated gene (9fold) followed by PAR-4 (8.9fold) and NOXA (3.4fold). Interestingly, maximum induction in the ACGs was observed in MDA-MB-231 cells when the highest concentration (IC 75 ) of riproximin was applied. Opposite to this phenomenon, maximum induction of ACG transcripts was observed in MCF-7 cells with lower concentrations (IC 25 , IC 50 ) of riproximin. As far as protein levels are concerned, a reasonable induction of NOXA was observed in MDA-MB-231 cells, while the other two ACGs (PAR-4 and TRAIL) were inhibited moderately. In MCF-7 cells, all the three ACGs were induced at protein levels at higher concentrations (IC 50 ) of riproximin. By enlarge, expressional modifications in NOXA and TRAIL genes followed the similar pattern at transcriptome and proteome levels in the two cell lines following riproximin exposure. In contrast, PAR-4 gene was substantially induced at transcriptome levels, while inhibited (MDA-MB-231) or moderately up-regulated (MCF-7) at protein levels (Fig. 3B). This, in turn indicates the presence of a potential negative feed-back mechanism creating hinderance in translation of PAR-4 transcripts or requirement of longer intervals to show proportional expression of this gene at protein levels. Over all, riproximin showed a considerable potential for upregulating the expression of multiple ACGs in breast cancer cell-specific manner.

Effects of anticancer genes on multiple molecular and functional arms
Physical interactions-based PPI network of each target gene was imported from STRING into Cytoscape separately and then merged to get the maximum number of interactors imported. The final merged network had 96 nodes with 761 In GO enrichment analysis significant enrichment of 440 BP, 28 MF, 14 CC, and 71 KEGG pathways was revealed (Fig. 4B-E). Interestingly, the majority of BP enriched were related to extrinsic and intrinsic apoptosis signalling pathways and their regulation (GO:0097191, GO:0097193 & GO:2,001,233). In the MF category major enriched terms were GTPase activity (GO:0003924) and G-protein beta-subunit binding (GO:0031681) with 21 and 13 genes, respectively. For the category of CC, PPI network was mainly enriched in heterotrimeric G-protein complex (GO:0005834), GTPase complex (GO:1,905,360) and the extrinsic component of cytoplasmic side of the plasma membrane (GO:0031234). KEGG pathway enrichment analysis revealed that beside being enriched in viral infections and pathways of cancer, the PPI network genes were enriched in a lot of key pathways such as PI3K-Akt signalling pathway, apoptosis, chemokine signalling pathway, Ras signalling pathway, TNF signalling pathway and Relaxin signalling pathway.

Discussion
Precise targeting of the molecular factors associated with cancer pathogenesis is prerequisite for an effective cure. For this purpose, identification of novel therapeutic targets and development of corresponding molecules is a continuous process. ACGs are an endogenous defense against transformed cells and are being investigated for their prognostic and therapeutic relevance. Expressional profiling of various ACGs in cancers, their association with disease progression as well as involvement in treatment responses is being studied by the scientific community. In parallel, novel delivery methods including antibody-based conjugates, immunotoxins development and genetic engineering strategies are being established to achieve ACGs mediated cancer cell-specific death. In addition to these synthetic strategies, naturally occurring compounds are also being considered for their potential to induce ACGs in cancer cells and to exploit resulting antitumor effects.
In this study, as a first step, we identified the protein expression levels of three ACGs (NOXA, PAR-4, TRAIL) in breast cancer tissues by immunohistochemistry techniques. To the best of our knowledge, this is the first study to investigate the expression profile of multiple ACGs at a time in breast cancer patients. Specifically, we focused on IDC samples, which is the most prevalent type of breast cancer. As far as the patient cohort is concerned, early onset of the disease (average ± 44 years age) with a considerable proportion of young patients (n = 20/45: <40 years age) was noticed, which is mainly attributed to the lack of awareness programs and late diagnosis locally. Molecular subtyping classified the majority of cases as triple negative (basal like), which itself predicts poor prognosis of these patients, followed by Luminal A type. As far as immunohistochemistry results are concerned, distinct ACG expression profiles were identified in breast cancer patient samples. An almost even distribution of NOXA expression (Neg/Low 21/45 = 47%, Elevated 24/45 = 53%) was found with no correlation to the molecular subtypes of breast cancer. However, a clear trend was found between NOXA expression and tumour grade, where Neg/Low expression ratios increased with increasing grade of the disease (Table 4). This, in turn, reflects continuous inhibition of NOXA expression overtime in patients with advanced stages of breast cancer. Similar findings have also been reported where low expression of NOXA was found associated with poor prognosis, reduced overall survival and chemo-resistance in breast cancer [8,9]. In our selected cohort, low expression of the PAR-4 gene was a dominant fact where 76% (34/45) of the samples were found negative for this ACG. Furthermore, a continuous decrease in PAR-4 expression was witnessed with the increasing severity of the disease. Precisely, 4/8 (50%), 12/16 (75%) and 18/21 (86%) cases of grade 1, 2 and 3 were found negative for PAR-4 expression, respectively ( Table 4). The data is in line with other available reports according to which low expression of PAR-4 is associated with poor prognosis in breast cancer [10,11]. Low expression of PAR-4 was also correlated significantly (p < 0.05) with the luminal A and basal type of breast carcinomas, thus depicting the poor response to treatment and an unfavourable outcome. Similar findings have been reported, where PAR-4 expression was found to be low in the highly aggressive, estrogen-receptor (ER) negative, basal-like, and high-grade (grade 3) breast cancers, which are all associated with poor clinical outcome [39]. Alvarez and colleagues found low PAR-4 expression associated with a poor response towards neoadjuvant chemotherapy and an increased risk of relapse in these breast cancer patients. Concurrently, higher expression of PAR-4 is also reported to sensitize triple-negative breast cancer cell lines to DNA damage-induced cell death by making cells more chemosensitive [40]. All in all, our data and available scientific reports suggest an overall low expression of PAR-4 in breast cancer patients with a substantial association to aggressive nature, chemo-resistance and relapse of the disease. TRAIL is one of the most studied ACG till today and is being targeted in the field of cancer therapeutics. A major hurdle is to overcome the development of TRAIL resistance achieved by tumour cells. Despite these existing challenges, the development of specific molecules (agonists) against the TRAIL receptors and recombinant TRAIL is an on-going process. As far as the expression profile is concerned, TRAIL was the only anticancer gene, for which a positive expression was found (at least 10% of tumour cells positively stained) in most clinical samples (32/45 = 71%, Table 2). The positive expression was related to the severity of the disease as shown by 3/8 (38%), 11/16 (69%) and 16/21 (76%) cases in grade 1, 2 and 3, respectively (Table 4). Higher levels of TRAIL correlated significantly (p < 0.05) with the luminal A and basal types of breast carcinoma. Grade 3 cases in the present study were significantly associated with positive TRAIL and negative PAR-4 expression, thus illustrating these two markers as related to tumour aggressiveness and poor prognosis in breast cancer patients (Table 4). Over and above to these facts, astonishing findings are being reported where combination of TRAIL along with platinum drugs (cisplatin) and liposomal formulations have shown to be effective against cancer stem cells and circulating breast cancer cells, which in turn, gives us hope that exploiting ACGs can be effective in controlling secondary tumour development and recurrence of the disease [14,41].
Several natural compounds have shown the potential to up-regulate ACGs in cancer cells followed by induction of multidirectional antineoplastic effects [42][43][44][45][46]. Naturally occurring agents comprise an attractive research domain for upregulating ACGs. Among the plant-based anticancer agents, RIPs belong to an important class of toxins, which have potential to be developed as therapeutic entities [47]. The effects of RIPs in cancer cells include the upregulation of endogenous defense genes, which is a neglected field of research, so far. In a previous study, we reported that riproximin induces significantly the expression of a wellknown ACG in breast cancer cells (MDA-7/IL24; ˃100fold) [32]. In this study, we extended these results by the impact of riproximin on the expressional profile of three further ACGs (NOXA, PAR-4, TRAIL). For this purpose, two breast cancer cell lines with distinct molecular subtypes (MDA-MB-231: triple negative, MCF-7: ER/PR positive) were selected and exposed to various concentrations of riproximin. Transcriptome and proteomic data revealed that riproximin can up-regulate the three ACGs in breast cancer cells (Fig. 3). Interestingly, a more prominent induction of two ACGs was found in MCF-7 (PAR-4: 8.9fold, TRAIL: 9fold) than in MDA-MB-231 cells (PAR-4: 2.8fold, TRAIL: 1.5fold). This is related to our previous report, in which riproximin induced significant cytotoxic effects in MCF-7 cells (48 h exposure) at much lower concentrations (IC 50 :0.38 ng/ml) than in MDA-MB-231 (IC 50 :3.6 ng/ml) cells [32]. From these findings, it can be deduced that breast cancer cells showing ER/PR positive status are more sensitive towards riproximin than triple negative breast cancer cells. One potential explanation is that riproximin interacts via its lectin binding domain (B-chain) preferably to surface glycans (NA2/NA3) and N-acetyl-D-galactosamine (GalNAc/Tn Antigen) of these cells [25]. Thus, the presence and saturation status of specific glyco-targets along with the potential involvement of other cell surface receptors can modify the subsequent binding capacity and internalization of riproximin. Other than this, the possibility of wide-ranged downstream signalling cascades being targeted by riproximin in a cell-specific manner cannot be ruled out. As far as riproximin mediated induction of ACGs is concerned, the phenomenon is not restricted to breast cancer cells only, as we recently witnessed that this plant protein can induce the ACGs in multiple colorectal cancer cell lines [48]. To summarize, riproximin mediated up-regulation of ACGs can be exploited in various malignancies, but this hypothesis needs to be corroborated in further pre-clinical investigations.
Bioinformatics is an emerging interdisciplinary field being exploited in modern day biology to understand complex cellular mechanisms including roles of potential genes/proteins within a cell and extrapolation of the wet-lab results. With this notion, we were interested to highlight the molecular significance of ACGs in cell environment by using various in silico approaches. In the PPI based clustering built with the help of STRING and Cytoscape software, the ACGs showed independent clusters, which in turn, reflects their minimal interdependency on each other (Fig. 4A). If so, this fact is of particular importance as it shows that upregulation of the ACGs via riproximin can initiate distinct and independent cascades to induce antineoplastic effects via multiple molecular directions. TP53 (a master regulator of apoptosis) and DIABLO (a proapoptotic mitochondrial protein) were the only two genes connected between NOXA and TRAIL clusters. The substantial involvement of NOXA and TRAIL in apoptotic routes is a known fact, thus linking bridges between these two ACGs via TP53 and DIABLO 1 3 is not a surprising observation. PPI networks of the ACGs also showed presence of key enzymes, transcription factors and kinases in the clusters, which means ACGs are involved in mechanistic routes responsible for enzymatic activities, gene expression regulations and phosphorylation steps, hence bear a wide-ranged responsibility inside cells. GO enrichment analysis revealed a huge number of BP (440), MF (28), CC (14) and KEGG pathways (71) being associated with the ACGs (Fig. 4B-E). Mainly these CC, MF and BP include GTPase complex formation, their corresponding activities and apoptotic signalling cascades, respectively. KEGG pathways data was also very enriched where ACGs were shown to be involved in vital mechanisms including survival (PI3K/Akt), apoptosis/necroptosis (caspases and BCL family), chemokine networking, TNF and RAS/RAF signalling cascades. All in all, the bioinformatic approach showed that the ACGs are extremely crucial factors involved in multidirectional aspects of a cell life and need due attention to explore their corresponding prognostic/therapeutic relevance in cancers.
Overall, the findings of this study reflected a distinct expression pattern of ACGs in breast cancer clinical isolates. Targeting these endogenous inhibitors of transformed cells can be instrumental in the field of cancer therapeutics. Along with synthetic approaches being tested to induce ACGs in cancer cells, naturally occurring compounds with a safe biological profile can be alternative options. In this context, riproximin with its already known substantial antineoplastic effects can be a lead compound for further development. However, further pre-clinical investigations are needed to understand events being operated by riproximin to induce ACGs.

Conclusion
Distinct expression profiles of the ACGs (NOXA, PAR-4, TRAIL) were identified in breast cancer clinical isolates. Low expression of these ACGs were predominantly evident in advanced stages of breast cancer. Targeting the ACGs via natural compounds is an important research area to be explored for therapeutic purposes. Riproximin, a plant-based ribosome inactivating protein with known antineoplastic effects, can be a lead compound in this context Author contributions AP conceived this study, supervised/analysed the experiments and wrote the manuscript draft. NN supervised clinical sampling/immunohistochemistry and drafted the relevant findings. TS, IS, KK and OS helped in transcriptomic expression profiling and preparation of figures. SMR performed the bioinformatic analysis of this study. SI, FS and BI provided the support in western blot analysis. MRB facilitated the microarray experiments and also supervised the drafting/editing of the manuscript.
Funding No specific funding was acquired for this research work.

Data availability
The datasets generated during the current study are available from the corresponding author upon reasonable request.

Conflict of interest
The authors declare that they have no conflict of interest.
Ethical approval For clinical investigations, informed consent was obtained from all patients and the study was conducted in accordance with the guidelines of the Ethics Review Committee of the University of Health Sciences, Lahore, Pakistan. Furthermore, the procedures performed were in accordance with the ethical standards of the institutional and with the 1964 Helsinki declaration and its later amendmentsor comparable ethical standards.