Background
The human PRDMs (PRDF1 and RIZ1 homology domain containing proteins), members of a kruppel-like zinc finger subfamily, share a conserved PR/SET domain, followed by zinc finger domains [
1‐
3]. Through regulation of the chromatin architecture, several PRDMs modulate gene expression, by either recognizing specific consensus sequences in promoters or acting as non-DNA binding cofactors [
1‐
3]. Particularly, some PRDMs are endowed with histone methyltransferase activity [
1‐
3]. PRDMs are involved in many developmental processes and play key roles in cell differentiation as well as in transducing several signaling pathways [
1‐
4]. Moreover, numerous studies suggest that the modification of their expression levels, sequence, or structure, has a relevant impact in many human diseases, including cancer [
1‐
3]. Commonly, most
PRDM genes express two main molecular variants, with (PR + product) and without (PR − product) the PR/SET domain, playing opposite roles, with the full-length product usually acting as a tumor suppressor, and the short one as an oncogene. The imbalance in their expression levels is observed in many cancers because of inactivating mutations or silencing of the PR + product and/or to increased expression of the PR − one [
1‐
3]. Likewise,
PRDM2/RIZ expresses two main molecular variants, the PR + isoform (
PRDM2a/RIZ1) and the PR − (
PRDM2c/RIZ2) [
1‐
3,
5]. The imbalance in their expression levels in favor of
RIZ2 is observed in many cancer types. Formerly, the full length
RIZ1 was extensively investigated in several cancers where it acts as a tumor suppressor, whereas few studies had explored the oncogenic properties of
RIZ2. Recently, we showed that
RIZ2 overexpression increased cell viability and growth, prompted the G2-to-M phase transition and organoids formation in HEK293 cells [
5,
6]. Consistently, our Exome- and RNA-Seq public datasets available at The Cancer Genome Atlas (TCGA) portal analysis revealed that a subset of
PRDMs, including
PRDM2, are frequently mutated and/or transcriptionally deregulated in certain tumor types, such as colorectal cancer (CRC) [
5,
7]. In addition,
PRDM2 is often target of frameshift mutations and aberrant DNA methylation in CRC [
5,
8‐
10]. Particularly, frameshift mutations in the C-terminal region of
PRDM2, affecting (A)8 or (A)9 repeats within exon 8, are found in one third of CRC with microsatellite instability [
11‐
13]. These frameshift deletion mutations, enriched in CRC exhibiting microsatellite instability (MSI), rise to a truncated protein lacking the C-terminal PR-binding motif that is essential for the methyltransferase activity of PR/SET domain [
14].
CRC is the third most deadly and fourth most diagnosed cancer worldwide. Despite the progress in early diagnosis and therapeutic options, CRC shows a poor prognosis with a 5 year survival rate of ~ 45%. The CRC first-line treatment involves a multimodal approach that usually comprises surgical resection and chemotherapy combined with monoclonal antibodies or proteins against vascular endothelial growth factor and epidermal growth factor receptor (EGFR) [
15]. Nevertheless, CRC often relapses. As such, new efforts are needed to improve the screening, the therapeutic options and outcomes in CRC patients. In this context, the discovery of new druggable biomarkers and targets is still a challenge in clinical management of CRC patients.
We recently provided novel insights on the RIZ2 tumor-promoting functions in the HEK-293 cell model, highlighting its putative mechanism in cancer initiation and progression [
6]. However, many relevant aspects of PRDM2 action in cancerogenesis remain to be elucidated, particularly those related to RIZ2 role in CRC pathogenesis. The present study aims to fulfill this gap and elucidate the tumor-promoting function of RIZ2 in CRC. To this aim, the biological outcomes of RIZ2 overexpression have been explored, using functional and transcriptome studies in CRC-derived DLD1 cells.
Methods
In silico analysis of RIZ1 and RIZ2 expression in colon cancer
Gene Expression Profiling and Interactive Analyses2 (GEPIA2) (
http://gepia2.cancer-pku.cn/#index) [
16] is an online resource for transcriptional profiles analysis, containing 275 colon adenocarcinoma (COAD) and 41 colon normal samples of “The Cancer Genome Atlas” (TCGA) [
17]. We performed a differential analysis of
RIZ1 and
RIZ2 expression through GEPIA2 and used a boxplot to illustrate the results with log
2 of transcript count per million [log
2(TPM + 1)] showing the expression level of both isoforms. We also provided the expression distribution of
RIZ1and
RIZ2 signatures in the 3 COAD subtypes.
Total RNAs were extracted from cells using Trizol solution (Thermofisher), according to the manufacturer’s instructions. The quality and quantity of RNAs were assessed by denaturing agarose gel electrophoresis and by spectrophotometry analysis (NanoDrop Technologies), after RNAse-free DNAse-I treatment (Boehringer Mannheim, Indianapolis, IN, USA). RNA was reverse transcribed with SuperScript III (Thermofisher) using 500 ng of total RNA. Quantitative RT-PCR analysis was performed as previously reported [
4]. Glyceraldehydes-3-phophate dehydrogenase (
GAPDH) or peptidylprolyl isomerase A (
PPIA) were used as housekeeping control genes [
18].
Cell culture and treatments
Human colorectal adenocarcinoma cell lines DLD1, HCT116, SW48, SW620 and HT29 were provided and grown as previously described [
19,
20].
When indicated 48 h before stimulation, 70% confluent growing cells were made quiescent using phenol red-free RPMI medium containing 0.1% charcoal-stripped serum (CSS), penicillin (100 U/ml) and streptomycin (100 U/ml).
The hEGF (Sigma Aldrich, St. Louis, Missouri, USA) was used at 100 ng/ml. The EGFR tyrosine kinase inhibitor, ZD1839, (Selleckem, Planegg, Germany) was used at 2 μM. The EGF neutralizing antibody (10605-R001; Sino Biological, Beijing, China) was used at 1.5 µg/ml.
Plasmids and DLD1 transfection
The pEGFP-C1 vector was purchased from Clontech (Palo Alto, CA, USA) and was used to clone in the BamH1 site, the sequence of RIZ2 open reading frame (NM_015866.4) as reported elsewhere [
4,
6]. The RIZ2 expressing plasmid, designated pEGFP_hRIZ2, was in frame with the EGFP coding sequence, which was positioned at the N-terminus, without in-frame stop codons. Based on previous observations, the produced fusion protein is likely to maintain the native RIZ2 functions [
4‐
6]. Plasmids were prepared with Plasmid Midi Kit (Qiagen Inc, Valencia, CA, USA), according to manufacturer’s instructions. Cells were transfected using Lipofectamine
™ 2000 Reagent in OptiMem I Reduced Serum Medium (Life Technologies, Carlsbad, CA, USA) for 6 h, following the manufacturer’s instructions. Stable clones were selected with 0.8 mg/ml Geneticin- G418 (Sigma-Aldrich). Transfection was verified by fluorescent microscopy and Western blot analyses [
4,
6].
Transcriptome profiling and bioinformatic analyses
Total RNA–seq procedure was performed as described previously [
21,
22]. Briefly, before use, total RNAs concentration, from three biological replicates of DLD1-pEGFP and DLD1-pEGFP_hRIZ2 cells, was measured using RNA HS kit on a Qubit fluorimeter (Life Technologies, Monza, Italy) while its quality and integrity assessed with the Agilent 4200 TapeStation System (Agilent Technologies, Milan, Italy). For RNA sequencing experiments, indexed libraries, from the biological replicates mentioned above, were prepared using 500 ng of total RNA as starting material using the Illumina Stranded Total RNA prep Ligation with Ribo-Zero Plus kit (Illumina Inc., San Diego, CA, USA). Final libraries were pooled and diluted to a final concentration of 1.4pMol and sequenced on NextSeq 500 (Illumina Inc) in a paired-end mode (2 × 75 base pairs). The raw sequence files generated (.fastq files) underwent quality control analysis using FASTQC (
http://www.bioinformatics.babraham.ac.uk/projects/fastqc/) and adapter sequences were removed using Trimmomatic version 0.38 [
23]. Filtered reads alignment on the human reference genome (GRCh38/hg38) using STAR v2.7.5a with standard parameters [
24]. Quantification of expressed genes was obtained with featureCounts [
25]. Differentially expressed RNAs were identified using DESeq2 [
26]. RNA expression was assessed when detected by at least ≥ 10 raw reads. Differential expression was reported as |fold change| (FC) ≥ 1.5 along with associated adjusted p-value or p-value ≤ 0.05 computed according to Benjamini–Hochberg as previously described [
27]. Functional annotation analyses of differentially expressed genes were performed according to IPA (Ingenuity Pathway Analysis, QIAGEN) [
28] and GEPIA2 [
16] that was used also for gene expression correlation analyses.
Immunofluorescence, DNA synthesis, cell number and WST-1 assay
DLD1-pEGFP and DLD1-pEGFP_hRIZ2 were analyzed for pEGFP expression by immunofluorescence microscopy as reported [
6].
Cells plated on gelatin-coated coverslips were made quiescent and after 72 h were rinsed with phosphate-buffered saline (PBS), fixed for 20 min with paraformaldehyde (4%, w/v, in PBS; Merck, Saint Louis, MO, United States), permeabilized for 10 min with Tween (0.1%, v/v, in PBS; Bio-Rad, Hercules, CA, United States), and incubated for 1 h with PBS containing FBS (1%, vol/vol). Cells were then incubated with the anti-EGF (1:100, ab131498 Abcam, Cambridge, United Kingdom) antibody for 2 h. After washings in PBS, the coverslips were incubated for 1 h at 37 °C with diluted (1:250 in PBS, containing 0.01% BSA) Texas red-conjugated AffiniPure anti-rabbit immunoglobulin G (IgG; Jackson ImmunoResearch Laboratories, West Grove, PA, United States). Nuclei were stained for 5 min with Hoechst 33258 (1 μg/ml; Merck). The number of EGF-positive cells was determined as percentage of EGF-positive cells on total cells.
The DNA synthesis was analyzed by BrdU incorporation. Cells were left unchallenged or challenged as indicated in the figures and in corresponding legends. After in vivo pulse with 100 μM BrdU (Sigma-Aldrich, St. Louis, MO, United States), BrdU incorporation into the newly synthesized DNA was analyzed and quantified [
29], using a DMLB (Leica, Wetzlar, Germany) fluorescent microscope equipped with HCX PL Apo × 63 oil and HCX PL Fluotar × 100 oil objectives. Images were captured using a DC480 camera (Leica) and acquired using the Leica Suite software.
DLD1-pEGFP and DLD1-pEGFP_hRIZ2 cells were counted in the Bürker chamber by optical microscopy. DLD1-pEGFP_hRIZ2 cells were compared with DLD1-pEGFP cells.
Cell proliferation was determined using the WST-1 reagent (Merck). Untreated or treated 1 × 103 cells in 96-well culture plates were incubated with the WST-1 reagent for 2.5 h. Thereafter, the formazan dye was quantified at 450 nm by a scanning multi-well spectrophotometer (Enspire; Perkinelmer, Waltham, Massachuttes, USA). The measured absorbance was correlated to the number of viable cells.
Clonogenic assay
DLD1-pEGFP and DLD1-pEGFP_hRIZ2 cells (1.5 and 3 × 10
2) were seeded into six-well plates and cultured at 37 °C for ~ 10 days until cells have formed sufficiently large clones (at least 50 cells). Fresh media were supplied every 3 days. Clones were counted after 30 min fixing with a mixture of 6% glutaraldehyde and 0.5% crystal violet [
6,
30]. The stained colonies were photographed and the number colonies with size ≥ 1 mm were counted using the ImageJ software (National Institutes of Health, USA) and expressed as mean ± S.E.M. Each assay was performed in at least three independent experiments in triplicate.
Wound scratch analysis, migration, invasiveness and phase-contrast microscopy
For wound scratch analysis, 1.5 × 10
5 cells were seeded in a 24-well plate and wounded with a 10 µl sterile pipette tip. Cells were washed with PBS and pre-incubated for 30 min with cytosine arabinoside (Sigma-Aldrich) at 50 µM (final concentration) to inhibit cell proliferation. Cells were then left untreated or treated, as indicated in the figures. Different fields were analyzed using DMIRB inverted microscope (Leica) equipped with N-Plan 10 × objective (Leica) [
31]. Phase-contrast images were captured using a DFC 450C camera (Leica) and acquired using Application Suite Software (Leica). Images are representative of at least three different experiments. The wound area was calculated using NIH Image J Software and expressed as % of residual area.
The migration and invasion assays were done using 2.5 × 10
4 cells in Boyden’s chambers with 8 μm polycarbonate membrane (Corning; Corning, NY, USA) pre-coated with Collagen or growth factor reduced and phenol red-free Matrigel (Corning; Corning, NY, USA), respectively. Cytosine arabinoside (at 50 µM) was included in the cell medium. After 7- or 18 h respectively, non-migrating or non-invading cells were removed from the upper surface membrane using a sterile cotton swab. Membranes were fixed for 20 min in 4% paraformaldehyde, stained with Hoechst, removed with a scalpel from the companion plate and mounted. Migrating or invading cells from at least 30 fields/each membrane were counted as described [
32]. Data are representative of at least three independent experiments.
3D cultures
Organoids were generated as reported [
31]. Cells (1.5 × 10
4) were mixed in each well with 250 μl of growth factor reduced and phenol red-free Matrigel (Corning) and 100 μl of organoid plating medium (DMEM/F12 medium, containing 5%FBS, penicillin (100U/ml), streptomycin (100 U/ml), diluted GlutaMAX (100X), 10 mM Hepes, B27 (50 × stock solution), 1 mM nicotinamide (Merck), 1.25 mM N-acetylcysteine (Sigma-Aldrich), and 10 μM Y-27632 (Merck-Millipore, Temecula, CA, USA). After 3 days, the organoid-plating medium was replaced with a similar medium without N-acetylcysteine and Y-27632. When indicated, the organoids were untreated or treated with the indicated compounds. Except for the experiments with the EGF neutralizing antibody [
29], the medium was changed every 3 days and different fields were analyzed using DMIRB Leica (Leica) microscope, equipped with C-Plan × 40 (Leica). Phase-contrast images were acquired using a DFC 450C camera (Leica). The relative organoid size was calculated using the same software and expressed as a fold increase over the basal organoid size.
Enzyme-linked immunosorbent assay
Cells (1 × 106) were seeded in 100 mm plates and made quiescent. Cell culture media were collected after 24-, 48- and 72 h and used to assay the EGF concentration, according to the manufacturer’s instructions. Human EGF (KHG0061; Invitrogen) ELISA kit was used. Data were analyzed using the curve-fitting statistical software Graph Prims Pad.
Lysates and western blot
Lysates were done as reported [
31]. The following reagents were used: anti-KMT8/Riz1/Riz2 antibody (ab3790; Abcam); mouse monoclonal anti-FAK (610,088; BD Transduction Laboratories), or anti P-Tyr 397 FAK (611722; BD Transduction Laboratories); anti-p42 extracellular signal-regulated kinase (ERK) (sc-1647; Santa Cruz), or anti p44 and p42 P-ERK (sc-7383; Santa Cruz); anti-tubulin (DM1A sc-32293; Santa Cruz), anti-GAPDH (#E-AB-20078; Elabscience) and rabbit polyclonal antibodies anti-EGFR (610016;Millipore), anti P-Tyr1068 EGFR (#2234S; Cell Signaling, Danvers, MA, United States) antibodies. The ECL system (GE Healthcare, Chicago, IL, United States) was used to reveal immunoreactive proteins.
Statistical analysis
Results from cellular and biochemical data are reported as mean ± SD. Three independent experiments in triplicates were performed. GraphPad Prism 9,5 software was utilized to perform Brown-Forsythe and Welch ANOVA tests. Significances are indicated in the corresponding legends.
Discussion
Human CRC is one of the most common types of malignancy and cancer-related death causes worldwide [
39,
40]. Despite the progress in early diagnosis and treatment, CRC still shows a poor prognosis and a low survival rate also due to high incidence of recurrence and drug-resistance. Besides, the incidence rates, even among young generations, are expected to continuously increase especially in developing countries, thus implying the need of further efforts in the development of innovative tools for improved prevention and treatment of CRC. In this context, the discovery of new biomarkers and targets is a crucial goal for CRC management.
CRC is caused by genetic alterations that target oncogenes, tumor suppressors and DNA repair genes [
41]. The pathogenic mechanisms include chromosomal instability (CIN), microsatellite instability (MSI) and CpG island methylator phenotype (CIMP), each of which is characterized by specific molecular profiles, stages of tumor development or progression, and, ultimately, prognosis. Moreover, also in the same genetic background, a peculiar feature of CRC is its heterogeneity, which explains the wide variability of response to systemic therapies characterized by some patients with satisfactory and sustained responses and other experiencing low sensitivity, relapse, rapidly progressive disease and poor prognosis. In the joint effort to define a classification able to stratify CRCs into distinct genetic subtypes and to program and optimize personalized first- and second-line chemotherapies for metastatic disease, the consensus molecular subtypes (CMS) have been described [
42]. The CMS classify CRCs into four subtypes according to their molecular profiles and taking into account, beyond mutations, the anatomical location of the primary neoplasia and specific cellular subsets within the microenvironment able to influence malignant cells in various ways. However, the possibility to effectively translate the CMS into clinical practice and to select patient-tailored treatments is a matter of debate and reveals that several key players in CRC occurrence and progression, or at least several molecular interactions, are still under-recognized [
43].
In CRC, common alterations affect crucial signaling pathways, as EGFR signaling pathway among the others [
44,
45]. Indeed, the current CRC first-line treatment involves an approach that may comprise monoclonal antibodies or proteins against EGFR, combined with surgical resection and chemotherapy [
15]. In this scenario, it is claimed to further deepen our understanding of CRC biology, spanning from gene expression control to signal transduction pathways and metabolic processes regulating CRC behavior.
Some
PRDMs are mutated or aberrantly expressed in CRC [
7]. Specifically,
PRDM2 gene is often target of aberrant DNA methylation and frameshift mutations in CRC; besides, most of the identified
PRDM2 mutations were enriched in cancers exhibiting MSI [
1,
5,
8‐
11,
46,
47]. Furthermore, our
in-silico analysis of transcript expression levels from TCGA-COAD dataset revealed an imbalance among
RIZ1 and
RIZ2 in favor of
RIZ2 in CRC (Additional file
2: Fig S1). Based on literature data and bioinformatics observations, as well as our previous findings in HEK293 model [
6], we hypothesized that
RIZ2/RIZ1 imbalance could also play a role in CRC. Consistently, forced
RIZ2 overexpression in DLD1 colon cancer cells was able to increase cell growth, colony and organoid formation thus confirming its oncogenic function [
6] (Figs.
1,
2). Additionally, DLD1
RIZ2 overexpressing cells showed a higher ability to reduce the wounded areas in wound scratch assays and invade collagen or Matrigel thick layers more efficiently than DLD1 control cells, suggesting a role for RIZ isoform imbalance in regulating colon cancer cell migration and invasion (Fig.
2). Interestingly, in the present study we have further investigated the possible action mechanism by analyzing the sequenced transcriptome of
RIZ2 overexpressing cells [
48]. As expected, these cells exhibited many differentially expressed genes (DEGs) involved in colon cancer progression and metastasis (Fig.
3). Noteworthy, among the enriched pathways, we noted the presence of EGFR signaling that might lead to malignant transformation and CRC progression [
15,
44,
45] (Fig.
4). Thus, we hypothesized that activation of EGFR signaling represents a putative mechanism explaining the
RIZ2 oncogenic properties.
EGF ligands, including the EGF-related peptides, activate EGFR and other family members. Enhancement of EGFR ligand expression, an autocrine loop mediated by an EGFR ligand itself, is the main mechanism implicated in cancer development and progression [
49‐
51]. CRC cells release EGF in tumor microenvironment [
52]. EGF, in turn, can bind the EGFR expressed by surrounding monocytes leading to the activation of Smad-PI3K-Akt-mTOR pathway and the polarization of monocytes into M2 macrophages [
52]. In such a way, colon cancer cells and tumor-associated macrophages are involved in a cross-talk that affects and enhances malignancy of colon cancer.
EGFR can be considered a biomarker of cancer aggressiveness, since metastatic cells express five times more EGFR than nonmetastatic cells [
53]. EGFR is overexpressed in 60% to 80% of colon cancers, leading to development of EGFR-targeted drugs [
54], an important tool for treating cancer. In addition, specific targeting of EGFR ligands (e.g. amphiregulin, heparin-binding EGF-like growth factor) by neutralizing antibodies or small molecules represents a promising therapeutic approach, as it inhibits the proliferation and metastatic events in breast [
55] and hepatocellular [
56] carcinoma cells.
Despite accumulating evidence suggests that the EGFR or transforming growth factor α (TGFα)/ EGFR signaling pathways play a critical role in colon cancer progression, the EGF autocrine loop in colon cancer is still poorly investigated. Targeting EGFR signaling through EGFR inhibitors profoundly impairs the tumor progression [
57]; however, monoclonal antibodies blocking the ligand binding to EGFR, have shown modest activity as a single agent in patients with metastatic colorectal cancer in clinical trials. Thus, combinatorial approaches against still unexplored intracellular targets (e.g. EGFR and RIZ2 signaling network) would improve the current clinical management of CRC patients.
Here, we show that the blockade of EGFR phosphorylation by ZD1839 interferes with proliferation, spheroid size and invasion of
RIZ2 overexpressing cells (Fig.
5). These cells also release higher EGF levels, as compared to control ones suggesting the possibility that the EGF-mediated autocrine loop modulation triggered by RIZ2 contributes to the maintenance of aggressive cell phenotype (Fig.
6). In turn, conditioned medium promotes the phosphorylation of EGFR and ERK in DLD1
RIZ2-overexpressing cells. A neutralizing anti-EGF antibody reverses these effects (Fig.
6). Of note, the molecular loop affects DLD1 cell migration and the selective tyrosine kinase inhibitor ZD1839 counteracts the EGF-induced phosphorylation of EGFR and ERK. At last, consistent with findings from wound scratch assay, EGF promotes FAK activation that is prevented by ZD1839.
Although not specifically evaluated in this work, the RIZ2-induced increase in EGF secretion could be a very intriguingly point and, at least partially, responsible of some features of CRC pathogenesis and progression that are not yet well dissected and not exclusively restricted to primary tumor cells. Through M2-polarization of tumor-associated macrophages, EGF secreted by colon cancer cells enhances the cancer-driving effect of the tumor microenvironment [
52], then promoting immuno-suppression, angiogenesis and neovascularization, as well as stromal activation and remodeling. Moreover, EGF suppresses the expression of LGR5 (a co-receptor for Wnt/β-catenin signaling and marker of both normal intestinal stem cells and cancer stem cells) finely regulated during the adenoma-carcinoma progression, the development and maintenance of CRC-derived metastasis [
58,
59]. The finding that
RIZ2 overexpression enhances the EGF/EGFR pathway strongly suggests a not-negligible role of
PRDM2 isoforms in the control of several oncogenic networks involved in CRC encompassing oncogenic pathway crosstalk, tumor microenvironment remodeling, tumor immune-escape, plasticity and stemness. However, the biological role of EGF in CRC is still not completely understood. Although it should act as a negative prognostic factor due to its activation of the MAPK signaling cascade, currently there are no literature studies demonstrating a correlation between serum EGF levels and prognosis in CRC. From a clinical point of view, it would also be interesting to search for a possible correlation between
RIZ2 expression and serum EGF levels in CRC patients as well as their relationship with patient outcomes. The prognostic value of this correlation in larger multi-center studies might enable clinicians to identify subgroups of patients who may benefit from targeted therapy.
Besides, in our opinion, the finding that
RIZ2 overexpression triggers the increase of EGF secretion in the setting of CMS1 background, such as in DLD1 cells, could represent a new element in the effort to better characterize the CRC molecular subtypes and to understand the players able to modify the tumor microenvironment, the immune-escape and, in turn, the optimized sensitivity to first and second line treatments [
33,
42].
These results add novel insights on the putative RIZ2 tumor-promoting functions in CRC. However, further investigations are required to explore the RIZ2 molecular mechanisms of action in CRC to exploit the full potential of PRDM2 knowledge for therapeutic applications. For instance, loss-of-function studies might be performed to corroborate our results. We have not used this approach here since it is still difficult to selectively silence
RIZ2 without targeting, at least in part, also
RIZ1 transcript [
6,
34]. In this context, one possibility could be to exploit the preferential presence of slightly different 3’ tails in
RIZ1 and
RIZ2 transcripts, as we recently described in lymphocytes [
4]. These differences could be useful to selectively target the different
PRDM2 transcripts. Besides, chromatin immunoprecipitation (ChIP) studies coupled with next-generation sequencing on overexpressing cells might provide us the profile of RIZ2 direct downstream targets to further elucidate the molecular bases of its function.
Additional genes and pathways, which have been found to be deregulated in these cells, may be also involved in the RIZ2/RIZ1 imbalance mechanism of action. For instance,
EGFL7 is very highly upregulated in
RIZ2-overexpressing cells (
FC = 34.322) (Additional file
1: Table S1). This gene encodes for the epidermal growth factor-like protein-7, which is known to facilitate tumorigenesis and angiogenesis through main cancer drivers from EGFR signaling to integrins, Notch receptor, or lysyl oxidase family members [reviewed in 60]. Interestingly, EGFL7 can also bind to the EGFR at cell membrane and this interaction enhances cell migration of hepatocellular carcinoma cells through FAK phosphorylation [
60,
61]. Again, EGFL7 competes with EGF for binding to EGFR [
60]. These findings, together with our present work, strongly suggest investigation of this mechanism also in CRC cells overexpressing
RIZ2.
To assess the clinical relevance of our findings, we performed the analysis of overall survival and disease-free survival of TCGA-COAD patients on GEPIA2. As shown in Additional file
2: Fig S3, the correlation of
RIZ2 expression with these clinical parameters was not statistically significant, probably because of the small number of currently available patients. However, a reduction trend of both overall survival and disease-free survival in TCGA-COAD patients with high
RIZ2 was observed; interestingly, this trend was more evident in patients with MSS status (Additional file
2: Fig S3). Thus, future studies on larger cohorts of patients are warranted to provide the applicability of these results in the management of CRC patients. Besides, an extended analysis of
RIZ2 transcript in further TCGA datasets revealed that it is overexpressed also in other gastroenterological tumors (Additional file
2: Fig S4) thus suggesting a broader role of RIZ2 in cancer biology. At last, analysis of RIZ2 role in CRC cell metabolism and identification of selected domains or functions might provide new insights for their possible clinical use in both diagnosis and therapy of CRC.