Background
The emergence of next-generation sequencing (NGS) technology has provided a large amount of data, much of which is publicly available [
1,
2]. Specifically, RNA-Seq has been used for the estimation of RNA abundance [
3,
4], alternative splicing detection [
5‐
7], and the discovery of novel genes and transcripts. As such, RNA-Seq has become an important tool in cancer studies [
6], contributing to reduced costs and less time being spent in benchtop experiments, thus speeding up the resolution of biological problems. However, a challenge remains in achieving intelligible data analysis and efficient laboratory validation.
Triple-negative breast cancer (TNBC) is characterized by a lack of estrogen and progesterone receptor expression (
ESR and
PGR, respectively) and an absence of human epithelial growth factor receptor (
ERBB2) amplification. Approximately to 15–20% of breast malignancies are TNBC [
8]. Patients with TNBC often exhibit unfavorable histopathologic features at diagnosis, mainly consisting of a higher histologic grade, larger tumor size, and frequent metastasis to the lymph nodes [
9]. As a consequence, TNBC is associated with a shorter median time to relapse and death [
10]. TNBC represents an important clinical challenge since it does not respond to hormone therapy, which targets the abovementioned receptors [
11,
12]. Moreover, TNBC is highly heterogeneous [
13], indicating the necessity of identifying unifying molecular targets, which may help guide more efficient and less toxic therapeutic management [
14,
15].
Guanylate-Binding Protein-1 (
GBP1) is a member of the large GTPase family and is induced by interferons [
16] and inflammatory cytokines [
17].
GBP1 is also transcriptionally regulated by epidermal growth factor receptor (EGFR). In glioblastoma [
18,
19] and esophageal squamous cell carcinoma [
20],
GBP1 upregulation via the EGFR signaling pathway contributes to tumor proliferation and migration both in vitro and in vivo. Moreover, GBP1 is described as a component of the cytoskeletal gateway of drug resistance in ovarian cancer [
21,
22].
GBP1 expression is also linked to a lack of responsiveness to radiotherapy in some tumors [
23], and
GBP1 is overexpressed in pancreatic cancer that is refractory to oncolytic virus therapy [
24].
In this work, we utilized RNA-Seq data obtained from TNBC tissues as well as cell lines that were publicly available from The Cancer Genome Project (TCGA) and the Gene Expression Omnibus Portal (GEO), respectively, to search for new therapeutic targets for TNBC. To complement our findings, we also performed transcriptomics analyses of several TNBC cell lines. The obtained lists of overexpressed genes were inter-crossed and compared with data from normal tissues from the TCGA. Methylome and proteomic data were integrated to our analysis to give further support to our findings. Using this approach, we identified 243 genes, which were subsequently evaluated for their druggability potential. GBP1 was the second gene on the list, and knock-down of GBP1 in TNBC and non-TNBC cell lines showed that its expression is important for TNBC cell growth. In addition, we demonstrated that GBP1 expression is controlled by EGFR signaling in breast cancer cells. Thus, we present GBP1 as a new potential druggable target for TNBC with enhanced EGFR expression.
Discussion
Several works have used transcriptomic analysis to improve the classification of TNBC and to obtain new predictive markers and therapeutic targets for the disease [
25,
26,
37,
84‐
87]. In our approach, we integrated RNA-Seq data from normal and tumor tissues (obtained from TCGA) and from cell lines that were sequenced in-house or were available from the GEO databank. A unifying DE gene list was obtained from the comparisons of normal x TNBC tissues, TNBC x non-TNBC tissues and TNBC x non-TNBC cell lines. Methylome and proteomic data were integrated to our analysis to give further support to our findings. A total of 243 genes were shown to be exclusively overexpressed in TNBC tissues and established cell lines and, importantly, were more highly expressed in TNBC than in non-transformed breast epithelial tissues. Subsequently, we searched for novelty by removing genes that have already been strongly linked to TNBC by analyzing publications listed in PubMed. Finally, we subjected our list to druggability scoring using the multidisciplinary canSAR platform. With the canSAR platform, we were able to predict gene products that could be used as therapeutic targets based on protein structure availability, the presence of potential small molecule binding pockets and information regarding the pre-existence of bio-active compounds (drugs or chemical probes) that have already been tested on a target or its homologues. Thus, we combined transcriptomic and proteomic approaches to enhance our chances of identifying proteins with true potential to become new therapeutic targets.
Moreover, by comparing the GO signatures of the cell lines and tissue transcriptomic data, we showed that cell lines could serve as good surrogates for testing these potential new targets, and we used them to show that
GBP1 (the second highest ranked gene on the final list) knock-down selectively affected TNBC cell growth.
GBP1 expression is controlled by EGFR in glioblastoma [
18,
19] and esophageal squamous head and neck cancers [
20] and is important for proliferation and tumor invasion. In addition, GBP1 is linked to radiotherapy resistance in head and neck tumors [
23] and is a component of the cytoskeletal gateway of drug resistance in ovarian cancer [
21,
22], especially for paclitaxel, which is a common therapeutic choice for treating TNBC [
88]. Class III β-tubulin plays an important role in the development of drug resistance to paclitaxel by allowing the incorporation of GBP1 into microtubules. Upon entering the cytoskeleton, GBP1 binds to pro-survival kinases, such as Proto-oncogene Serine/threonine-protein kinase pim-1 (PIM1), and initiates a signaling pathway that induces resistance to paclitaxel [
89]. Indeed, a 4-aza podophyllotoxin derivative was demonstrated to act as a potent in vitro inhibitor of the GBP1:PIM1 interaction, which is a property that is maintained in vivo in ovarian cancer cells resistant to paclitaxel [
90]. Taken together, these findings confirm GBP1 as a druggable protein.
It is well known that the TNBC is a very heterogeneous breast cancer subtype [
91]. In saying so, it was not out of surprise that the tested TN cell lines responded heterogeneously to the
GBP1 knock down: Out of the 8 tested cell lines, while 4 presented increased cell death, 4 responded by decreasing cell proliferation in comparison to control. Moreover,
GBP1 expression level did not impact on patient’s 5 years survival as evaluated by the NCC-AUC model. Indeed, we observed that the cell lines that were more impacted by
GBP1 knock down are, following a molecular sub-classification of the disease [
37], Basal-like 1 (BL1) and 2 (BL2) cells (with the exception of MDA-MB-436). On the other hand, the cell lines that had only its proliferation affected after
GBP1 knock down are, all of them, mesenchymal (M) or mesenchymal stem-like subtypes (MSL) [
37]. Top gene ontologies for the BL1 and BL2 subtype are heavily enriched in cell cycle and cell division components and pathways, as well as growth factor signaling. Differently, the M and MSL subtype display gene ontologies that are heavily enriched in components and pathways involved in cell motility, ECM receptor interaction, and cell differentiation pathways. The MSL, in particular, presents enrichment of genes associated with stem cells and mesenchymal stem cell–specific markers, and low expression of claudins [
37]. We hypothesize that higher expression levels of
GBP1 may have a more severe impact on the survival of a subgroup of TNBC patients with specific molecular markers.
EGFR is overexpressed in a high proportion of the TNBC cases [
82,
92] and is a marker of a poor prognosis [
93‐
95]. Although EGFR has been successfully used as a therapeutic target for many tumor types [
96], unencouraging results have been obtained in clinical trials (in both mono- and adjuvant therapy protocols) conducted in TNBC patients [
97]. Failure to induce inhibition of Akt has been reported as a major cause of resistance to EGFR inhibitors [
97,
98]. Moreover, nuclear EGFR (nEGFR) can enhance resistance to anti-EGFR therapies and correlates with poor overall survival in breast cancer. Inhibition of nEGFR nuclear translocation leads to subsequent accumulation of EGFR on the plasma membrane, which greatly enhances the sensitivity of TNBC cells to cetuximab [
99]. We demonstrated that
GBP1 expression correlates with
EGFR expression (and protein levels) in both tissues and breast cancer cell lines. In most of the tested cell lines, we showed that the
GBP1 expression level responded to EGFR stimulation by epidermal growth factor.
Acknowledgements
The results published here are based on data generated by the TCGA Research Network:
http://cancergenome.nih.gov/. We would like to thank TCGA project organizers as well as all study participants. We thank the staff of the LaCTAD of the UNICAMP for performing the RNA-Seq runs. This work was facilitated by VVL (Viral Vector Lab Facility) and LBE (Bioassays Facility) at LNBio/CNPEM.