Post-translational modification of proteins by members of the small ubiquitin-related modifier (SUMO) protein family regulates multiple cellular processes, including transcription, replication, chromosome segregation and DNA repair, in various human diseases including cancer (Geiss-Friedlander and Melchior, 2007). The SUMO proteins include three ubiquitin-like peptidic modifiers; SUMO-1, -2 and -3. Their conjugation to substrates occurs through heterodimeric SUMO-activating E1 enzyme (AOS1/UBA2), a SUMO-conjugating E2 enzyme UBC9 (encoded by UBE2I) and various E3 protein ligases (e.g., PIAS1, PIAS2, PIAS3, PIAS4 and RANBP2) that facilitate reversible binding of SUMO proteins to the lysine residues of the target protein (Kerscher et al, 2006).

Although sumoylation controls many cellular functions, one well-recognised and established role is in the regulation of transcription through the modification of histones, transcription factors, chromatin-modifying enzymes, and basal transcription machinery. Dysregulation of the SUMO pathway has been observed in various cancers, and is often associated with adverse patient outcomes (Driscoll et al, 2010). Emerging evidence suggests that targeting sumoylation could be a potential therapeutic approach for cancer treatment. In this context, a specific panel of transcription factors implicated in the epithelial-mesenchymal transition (EMT) and in chemotherapeutic resistance, such as p53, MDM2, NF-κB, and ZEB2 (Du et al, 2016), as well as tyrosine kinase receptors such as IGF-1R (Sehat et al, 2010), has been shown to be directly targeted by SUMO-mediated conjugation.

In this issue of the British Journal of Cancer, Codony-Servat et al (2017) report a novel observation that phosphorylated nuclear IGF-1R (nIGF-1R) is expressed in ∼20% of metastatic colorectal cancers (mCRC) and 50% of patients harboring mutations within the BRAF gene. In these subsets of patients, the levels of phosphorylated nIGF-1R in pre-treated metastases were markedly increased compared with their matched untreated primary tumours. Moreover, the authors demonstrated that high expression of nIGF-1R significantly correlated with poor overall survival in CRC patients.

To make sense of these clinical findings, the authors performed functional studies and successfully garnered supporting evidence that chemoresistant CRC cell lines displayed significantly higher levels of nIGF-1R expression. The potential molecular mechanism underlying the translocation of IGF-1R into the nucleus was explored using CRC cells treated with various chemotherapeutic drugs, rendering them chemoresistant. Codony-Servat et al (2017) observed that the protein inhibitor of activated STAT3 (PIAS3) was the key mediator contributing to IGF-1R nuclear sequestration, pointing to an essential role of PIAS3, a SUMO E3 protein ligase, in this process.

Another intriguing feature of this study was the complexity of the ‘BRAF-like’ phenotype in CRC patients. Such a phenotype was defined by the presence of bona fide BRAF mutations in mCRC patients, as well as the presence of a gene-expression signature in a subset of patients that lacked BRAF mutations, which was very similar to the patients with BRAF mutations. In fact, both groups of patients with mCRC have previously demonstrated resistance to cetuximab treatment (Popovici et al, 2012). This BRAF-like phenotype often results in the upregulation of various genes implicated in sumoylation, including: RANBP2, an E3-SUMO ligase implicated in kinetochore function during mitosis (Vecchione et al, 2016); the nuclear internalisation of IGF-1R (Packham et al, 2015); and the activation of splicing genes such as the GTPase—an active form of RAC1 and RAC1b, which promotes Cyclin D1 and NF-KB activity (Matos et al, 2008). Conversely, the downregulation of sumoylation-associated genes such as AXIN-2, CDX2 and RNF43 is observed, owing to hypermethylation of their promoter regions in the presence of point mutations (Bond et al, 2016), as illustrated in Figure 1.

Figure 1
figure 1

Routes to targeted therapy resistance in BRAF-like phenotype patients. In addition to the presence of bona fide BRAF mutations, a substantial percentage of colorectal cancer patients with KRAS mutations, as well as those with double wild-type genotypes (2 WT) are enriched with a ‘BRAF-like’ phenotype. Such a phenotype potentially prevents sensitivity to: (1) EGFR inhibitors (panitumumab and cetuximab); (2) BRAF inhibitors (vemurafenib, dabrafenib and encorafenib); (3) MEK inhibitors (trametinib, cobimetinib, binimetinib and selumetinib); and (4) PI3K inhibitors (alpelisib and buparlisib). As illustrated in this figure, the BRAF-like phenotype overcomes AKT/MEK inhibition by directly targeting the NF-kB transcription factor or Cyclin D1 by overexpression of SUMO proteins or RAC1b. Upregulated genes are shown in red circles, while green circles depict downregulated genes in CRC patients with a BRAF-like phenotype.

Full size image

This study by Codony-Servat et al (2017) is provocative and raises several important questions. First, is the expression of RANBP2 and/or PIAS3 upregulated more in BRAF-mutant mCRC patients compared with KRAS-mutant or double wild-type genotypes? If so, what is the underlying rationale for such a BRAF-like phenotype, which certainly seems to be involved in chemoresistance in such patients? Second, the increase in expression of PIAS3 and nIGF-1R was higher in chemotherapy pre-treated cell lines than naïve cell lines when later treated with ganitumab or dasatinib—is this important? Finally, is the phosphorylated nIGF-1R just a (surrogate) biomarker for the BRAF-like phenotype or does it actually play an active role in chemotherapy and targeted therapy-mediated resistance? Although the last question would require a careful evaluation in future studies, it is already known that nIGF-1R functions as a transcription co-factor for LEF-1, which activates the expression of cyclin D1 and AXIN-2; as well as for histone H3 through recruitment of Brg1 and SNAI2 expression (Warsito et al, 2016). Of note, in this study using HeLa cells, phosphorylated nIGF-1R which was induced upon IGF-1 stimulation, was inhibited with IGF-1R kinase inhibition (Warsito et al, 2016). The finding that IGF-1R expression increases in the nucleus following ganitumab treatment in pre-treated colorectal cancer cells is quite striking and could, at least partially, help provide an explanation for its therapeutic failure in pre-treated colorectal cancer patients (Van Cutsem et al, 2014). These findings have important clinical implications as these reinforce the importance for understanding the complexity of second-line therapy for treating CRC patients (e.g., mechanism of action of ganitumab).

In conclusion, the study by Codony-Servat et al (2017) sets the stage for important treatment decision making. Recently, vinorelbine demonstrated pre-clinical activity in RANBP2 addicted BRAF-like CRC cell lines (Vecchione et al, 2016). In addition, SUMOylation inhibitors (Bogachek et al, 2016; Wagner et al, 2015) and curcumin have the potential to reverse EMT- and NF-kB-mediated chemotherapeutic resistance, and nuclear internalisation of IGF-1R, respectively. Therefore, a rational step would be to explore the combinatorial efficacy of these agents in pre-treated mCRC patients with phosphorylated nIGF-1R overexpression. Other strategies worth considering might include the combination of these drugs with BRAF and MEK inhibitors, in pre-treated BRAF-mutant patients. The ultimate golden nugget to glean from a study such as this would be that in the era of precision medicine, the identification of robust biomarkers that could help delineate specific phenotypes will be crucial for optimal drug development in mCRC. In other words, we should have realistic aspirations of solving one piece of the puzzle at a time, rather than hoping for the big prize anytime soon.