Acquired resistance to chemotherapeutic drugs remains a major hurdle to effective cancer treatment and eradication, and a better understanding of the genes and pathways that contribute to this is needed. Our data demonstrate that transposon mutagenesis provides a powerful, adaptable and cost-effective forward genetic approach for identifying resistance genes. The use of parallel screening in four separate cell lines with replicate samples to identify both common and cell-line restricted resistance gene candidates illustrates the potential of this system for gaining a deeper and more comprehensive view of resistance across the full spectrum of malignancy.
Transposon-based gene activation systems have a unique combination of properties that make them ideal for studies of tumor cell resistance. First, they are readily applied to gain-of-function genetic screens, unlike the vast majority of functional genetic interrogation approaches for mammalian genomes. This may be of particular relevance in tumors, where gene activation or amplification are common transforming events. The potential for long-range transcriptional activation and the need to hit only one allele of a gene in diploid cells provides higher effective genome coverage than gene inactivation strategies at comparable levels of mutagenesis. Furthermore, the presence of multiple effective mutations in single cells should allow for identification of ‘cooperating’ resistance genes, as suggested by our analysis of the ABCB1/MEIS1 interaction in resistant clones. A second advantage is the potential to identify resistance events occurring from changes in expression of uncharacterized or poorly understood genetic elements such as long intergenic non-coding (LINC) RNA or microRNAs. We have identified resistant cells bearing insertions in LINC-RNAs and unannotated transcripts and further investigation of these mutations could shed light on new genetic elements. Third, this system is readily transferable to new cell lines, including cells derived from patient samples, and libraries can be expanded and regenerated simply by re-transfection with the transposase. Hence transposon-based screens allow for the rapid generation of resistant clones to single drugs or combined therapies in specific tumor cells, providing insights into potential resistance mechanisms. These could then be used to guide design of new drug combinations and tailor treatments to particular tumor types.
Transposon-based screening has been used previously to identify potential mechanisms of resistance to the antibiotic puromycin and chemotherapeutic, vincristine . In those studies, transposon insertions were found primarily in Abcb1a/b (both drugs) and the closely related transporter Abcg2 (puromycin only), reinforcing our findings that overexpression of ABCB1 represents a major mechanism of drug resistance. However, additional candidate genes were not identified; this may be due to the absence of a splice donor in the transposon used, limiting the ability of the inserted promoter to activate gene expression, the presence of only one transposon per cell, or the relatively limited analysis of insertion events using capillary sequencing of isolated cell clones
Based on our finding that the transposon used here can exert a strong transcriptional activation effect at 64kb upstream from open reading frames, we estimate (using Poisson distribution) that libraries consisting of just 4.7 × 104 clones (2.8 × 105 insertions) or 7.3 × 104 clones (4.4 × 105 insertions) respectively could be potentially capable of activating 95% of genes by delivering at least one or multiple forward upstream insertions. Therefore in the case of HeLa cells we have approached meaningful close to genome-wide coverage for activation events. This is supported by our results with ABCB1, from which it is clear that all of our libraries had sufficient coverage to provide multiple insertions in a single strong resistance gene. We deduced that the low incidence of identification of other resistance genes could therefore reflect real differences in the ‘potency’ of individual genes to promote resistance, with only ABCB1 being sufficient while other events requiring additive effects of multiple genes to yield resistance in the high taxol concentration used here for selection. This is also supported by a prior transposon screen [41 described above] which identified only ABC-family transporters as potential resistance genes. Support for this also comes from our bioinformatics analysis, which revealed concordance of gene function or pathways between candidate resistance genes, with a strong enrichment in microtubule related biology previously linked to paclitaxel resistance.
However, even considering these potential limitations, our data identify new possible resistance genes and strengthen the evidence for previously identified candidates. As an example, clonal analysis of resistant cells strongly implicate MEIS1 as a modifier of ABCB1-mediated resistance, and this is further supported by our analysis of a large panel of tumor cells. MEIS1 is a class A homeodomain protein that acts as a cofactor for homeobox (HOX) proteins, and has been implicated as a critical downstream target of oncogenic fusion proteins in leukemia [42–45]. Although high expression promotes leukemia cell proliferation, silencing of MEIS1 increases resistance to the chemotherapeutic etoposide , in agreement with our findings in IMR32 cells. Our clonal analysis of mutations also reveal CXCR4, the receptor for the chemokine CXCL12 (stromal derived growth factor 1, SDF-1), as a potential resistance gene that functions independently of ABCB1 (Figure 5A). Up-regulation of CXCR4 is associated with increased metastasis and poor prognosis in various forms of cancer, in part due to effects on cellular phenotype, and is associated with chemotherapeutic resistance in numerous tumor models. For example, CXCR4 is upregulated in gefitinib-resistant non-small cell lung cancer cells and promotes epithelial-mesenchymal transition (EMT) and self-renewal activity . Likewise, CD133+ glioblastoma cancer stem cells with increased resistance to a range of chemotherapeutic agents, including paclitaxel, express high levels of CXCR4 , and high surface expression of this chemokine receptor is considered a marker of cancer stem cells . Of direct relevance to our results, increased CXCR4 expression and CXCL12/ CXCR4 signaling promote tumor cell resistance to the chemotherapeutic gemcitabine .
Other genes identified in our screen, such as ALK and PDE4D, increase tumor cell growth and protect from apoptosis, and may therefore promote resistance through these mechanisms.
Finally, further genes identified from independent hits in different cell lines, including the protocadherin PCDH15 and neuroblastoma breakpoint family member NBPF11 have not been implicated in tumor resistance, but based on the examples described above, may also have important roles.