PMP is a poorly studied disease, owing both to its rarity, which hinders the collection of samples, and to the intrinsic complexity of these samples, mostly composed of mucus with very little cellularity, which complicates their study at the molecular level. Hence, most available studies have focused on the clinical management and treatment of PMP patients, especially on using of CRS + HIPEC [2]. Consequently, molecular understanding of PMPs is shallow compared to other tumors, mostly limited to broad genetic and transcriptomic characterization. Genomic information points to known oncogenes and tumor suppressor genes KRAS, GNAS, TAT4, TGFBR1, TP53 and SMAD3/4 as the most frequently mutated in PMP, commonly linked to bad prognosis [5].5 At the transcriptional level, mucin related genes MUC-5AC, MUC-2, and TFF1/2 have been found overexpressed, while protein levels of TOPO1, TS and ERCC1 have been proposed as biomarkers in IHC studies. However, the limited number of samples hampered the identification of more proteins in a reliable way [19]. In this context, this study represents to best of our knowledge, the first approach to precisely explore a central regulatory system for RNA biology in PMP, and reveals that splicing machinery is severely disrupted in tumoral tissue, which may entail pathophysiological implications. Moreover, our result uncovered four components particularly altered, whose expression levels allowed to distinguish PMP tumoral tissue from non-tumoral tissue, and were associated with pivotal prognostic features, such as overall survival.
Our results are nicely in line with recent evidence from our group and others that dysregulation of the splicing machinery is a widespread feature among multiple diseases, from diverse cancers [10–15, 20], to metabolic disorders [21, 22], and chronic inflammatory diseases [23]. This study adds PMP to this growing line of evidence while, at the same time, provides precise information on the distinct, tumor-specific alterations that occur in these rare tumors, which may become useful for the understanding and future management of the disease.
HNRNPK is not the most commonly altered splicing factor in cancer, but various studies have previously found dysfunctions of in other tumors [24], including pancreatic adenocarcinoma [25], where its alteration is linked to activation of RAS family members, such as KRAS, a gene that, interestingly, is also found altered in PMP [5]. Likewise, several studies have addressed the role of PTBP1 in cancer, whereas its interacting co-factor RAVER1 is less known. Of note, PTBP1 has been linked to colorectal cancer [26], a disease closely related to PMP. Similarly, MBNL1 is a well-known player in oncogenesis [27], such as in digestive system tumors [28].
Splicing machinery alterations in PMP were not restricted to a given portion of it but included key components of the major and minor spliceosome and splicing factors, most of which were downregulated in tumoral tissue compared to control tissue, being PTBP1 and RAVER1 the upregulated exception. Some of the alterations were sufficiently profound to clearly distinguish between normal and tumoral tissues at RNA level. These changes open the possibility to study their potential clinical value as biomarkers for tumor detection and even fluorescence-guided surgery [29], a technique that would be useful in these mucinous tumors that can fill the entire peritoneum and permeate all organs, which would allow inadvertent tumor implants to be detected.
Prognostic biomarkers are lacking for PMP management. In this sense, tumors presenting high levels of HNRNPK and MBNL1 displayed a better overall survival stratified by cases with optimal cytoreduction (CC0/CC1). While for HNRNPK, MBNL1 these results appear reasonably predictable, unfortunately, we are limited by a reduced dataset and further research in this line will be required to precisely ascertain the role of splicing factors in PMP. This limitation may also be related to the lack of patent differences in the expression profile of the splicing machinery between high- and low-grade tumors, the latter showing more robust differences with control tissues, due to the higher heterogeneity of high-grade tumors. Intriguingly, however, the lack of molecular differences between high- and low-grade tumors also occurs at the mutational level in KRAS and GNAS mutations (the most commonly mutated genes in PMP), whereas TP53 and PI3K are more frequently mutated in high-grade tumors [30, 31]. Taken together these differences, and lack thereof, it seems conceivable that the molecular underpinnings of low- and high-grade tumors are not globally distinct but may pertain to a limited set of genes, whose elucidation would provide valuable prognostic information.
In search of further molecular information, we pursued enrichment analyses with REACTOME, Gene Ontology and STRING tools. A study of HNRNPK, MBNL1, PTBP1 and RAVER1 with GO and STRING analyses provided predictable results linking them to RNA modification and splicing processes, which reinforced the suitability of the approach. Most importantly, analysis of the four genes using the REACTOME database associated their altered expression with key cancer-related molecular pathways. Namely, pathways related to PTK6 (Protein Tyrosine Kinase 6 or Brk), whose activity and regulation are strongly associated with multiple cancers, like colon [26] and breast cancer [32, 33], oral squamous cell carcinoma [34] or hepatocellular carcinoma [35] by different mechanisms. Of note, one of these mechanisms relates to the appearance of a splicing isoform (Alt Brk) that may act by preventing tumor progression [32, 33]. Reelin signaling is clearly related to proliferation, migration and invasion, inducing epithelial-mesenchymal transition in certain cancers [36, 37]. Disabled-1 (Dab1) encodes critical reelin signaling adaptor and is a possible target of the splicing factor PTBP1, which promotes the generation of a shorter isoform and influences cell migration [38]. Other altered pathways include CD28 receptor signaling, which plays a key role in T cells activation and function, and whose stimulation has been proposed as a new way of immunotherapy for some cancers [39]. Actually, the enrichment analysis also reveals links to other interleukins as 4, 13 and 20 families, further suggesting a relationship with immunoregulatory processes and inflammation that would mark the tumor microenvironment. Finally, FTL3 tyrosine kinase receptor mechanisms have been linked to a variety of tumors, including hepatocellular carcinoma [40] and acute myeloid leukemia [41]. In the latter, it is worth noting that the numerous splicing-related alterations have enabled to establish different tumors groups, based on the pathways to which differentially spliced genes belong. Obviously, the biocomputational findings obtained herein will require further examination in vitro and in vivo in appropriate models to assess their actual relevance and potential value in PMP. However, these results already provide original information to explore novel opportunities to better understand these rare tumors.