Functional relationship between CFTR and RAC3 expression for maintaining cancer cell stemness in human colorectal cancer

Abstract

Purpose

CFTR mutations not only cause cystic fibrosis, but also increase the risk of colorectal cancer. A putative role of CFTR in colorectal cancer patients without cystic fibrosis has so far, however, not been investigated. RAC3 is a nuclear receptor coactivator that has been found to be overexpressed in several human tumors, and to be required for maintaining cancer stemness. Here, we investigated the functional relationship between CFTR and RAC3 for maintaining cancer stemness in human colorectal cancer.

Methods

Cancer stemness was investigated by analysing the expression of stem cell markers, clonogenic growth and selective retention of fluorochrome, using stable transfection of shCFTR or shRAC3 in HCT116 colorectal cancer cells. In addition, we performed pathway enrichment and network analyses in both primary human colorectal cancer samples (TCGA, Xena platform) and Caco-2 colorectal cancer cells including (1) CD133+ or CD133- side populations and (2) CFTRwt or CFTRmut cells (ConsensusPathDB, STRING, Cytoscape, GeneMANIA).

Results

We found that the CD133+ side population expresses higher levels of RAC3 and CFTR than the CD133- side population. RAC3 overexpression increased CFTR expression, whereas CFTR downregulation inhibited the cancer stem phenotype. CFTR mRNA levels were found to be increased in colorectal cancer samples from patients without cystic fibrosis compared to those with CFTR mutations, and this correlated with an increased expression of RAC3. The expression pattern of a gene set involved in inflammatory response and nuclear receptor modulation in CD133+ Caco-2 cells was found to be shared with that in CFTRwt Caco-2 cells. These genes may contribute to colorectal cancer development.

Conclusions

CFTR may play a non-tumor suppressor role in colorectal cancer development and maintenance involving enhancement of the expression of a set of genes related to cancer stemness and development in patients without CFTR mutations.

Supplementary Information

The online version contains supplementary material available at 10.1007/s13402-021-00589-x.

Introduction

Colorectal cancer (CRC) is the second most common cause of cancer death in the world as reported by the World Health Organization (WHO). Moreover, the Pan American Health Organization informs (PAHO Official Report: Colorectal Cancer in the Americas, Situation and Challenges, 2016) that it is the fourth most common cancer, with more than 240,000 newly diagnosed people and approximately 112,000 people dying from this disease each year, in the Americas. The highest CRC incidence rates are observed in Canada, Uruguay and Barbados, while countries in Central America have the lowest rates. If no action is taken, the CRC incidence is expected to increase 60 % by 2030. Although genetic susceptibility to CRC is considered to be high, e.g. associated with familial adenomatous polyposis (APC mutation), the full spectrum of driver genes for CRC, especially their contribution to disease susceptibility and their underlying mechanisms of action, remain to be fully elucidated [1].

CFTR is an ion channel, mainly regulated by ATP and is related to cytoskeleton proteins [2]. More than 1,900 CFTR mutations have been identified that cause its malfunctioning, resulting in the genetic disorder of cystic fibrosis. The most common mutation is ΔF508, which prevents CFTR protein maturation and produces defects in the gating of the channel. Based on the observation that patients with cystic fibrosis have a high incidence of CRC, several studies have suggested a relationship between CFTR and CRC development and defined CFTR as a tumor suppressor [3]. Although several previous studies have proposed probable mechanisms by which the absence or mutation of CFTR may contribute to CRC development, its tumor suppressing mechanisms are yet not fully understood. Some evidence supports CFTR conductance as its most important biological function, involving osmoregulation, transport of major ions across epithelia and intestinal stem cell regulation [1, 4, 5]. Most of its functions are also linked to the well-defined hallmarks of cancer [4]. However, the role and probable modulation of CFTR expression in CRC development in patients that do not have cystic fibrosis are not well understood.

Cancer stem cells have the ability to initiate and perpetuate tumor development when they invade other tissues or when they are transplanted, thereby recapitulating the heterogeneity and histology of the parent tumor. These cells represent the most important side-population responsible for cancer propagation and resistance to anti-cancer treatment [4, 6, 7]. The nuclear receptor coactivator RAC3 (also called NCOA3, AIB1 or SRC3) has been found to be overexpressed in several types of cancer [8–11]. We have previously shown that it is required for CRC stemness maintenance [12, 13]. RAC3 is a NF-κB coactivator [14] whose overexpression inhibits apoptosis [15, 16] and senescence [17], and influences chemoresistance of CRC trough the inhibition of apoptosis and autophagy [18]. Its biological actions are not limited to the nucleus, but also include the cytoplasm [16, 19], and it can be upregulated by inflammatory cytokines [20].

The expression of CFTR is mainly increased in the crypts of the intestine, where normal stem cells are located [21, 22]. This tissue has the highest rate of epithelial turnover in the body, and it has been well documented that pathologic conditions involving mutation or loss of CFTR that enhance the rate of epithelial turnover and intestinal inflammation may contribute to increased CRC risk by enhancing the rate of stem cell divisions [22]. Although normal and cancer stem cells share gene expression profiles [3], the role and expression levels of CFTR in CRC cancer stem cells, required for cancer perpetuation and metastasis in patients without cystic fibrosis, have not been elucidated. Here, we investigated the putative relationship between CFTR and RAC3 expression in signaling, biological functioning and maintenance of CRC stem cells.

Materials and methods

Cell culture and reagents

Human colorectal cancer HCT116 cells (ATCC) were maintained in Dulbecco’s Modified Eagle Medium (DMEM)/F12 (Gibco Laboratories, Grand Island, NY, USA). The non-tumoral human embryonic kidney HEK293 cells (ATCC) were maintained in DMEM high glucose. All cultures were supplemented with 10 % fetal bovine serum (FBS) (Invitrogen), penicillin (100 U/ml) and streptomycin (100 µg/ml). The cells were cultured at 37 °C in a humidified atmosphere with 5 % CO₂. Unless stated otherwise, all reagents were obtained from Sigma Aldrich, Argentine or Santa Cruz Biotechnology, USA.

Immunofluorescence assay

Immunofluorescence assays were performed as previously described [16]. Briefly, cells were fixed with glutaraldehyde 1 % and formaldehyde 0.2 %, permeabilized with PBS-Triton 0.2 %, blocked with 10 % FBS and incubated 2 h at room temperature with 0.5 µg/ml of the following antibodies: anti-CFTR (R&D Systems), anti-Nanog (sc: 293,121), anti-Oct-3/4 (sc:5279), anti-c-Myc (sc-40) (Santa Cruz Biotechnology) and anti-CD133 (AC133) (Milteni). Finally, cells were incubated with fluorochrome-conjugated secondary antibodies and visualized using a fluorescence microscope (Olympus BX51) and photographed at 40x magnification.

Western blot analysis

Western blot assays were performed as previously described [20]. Briefly, cell monolayers were scraped off and treated with RIPA buffer containing pepstatin A, phenymethylsulfoyl fluoride and dithiothreitol, solubilized in SDS-PAGE sample buffer, separated by 8 % SDS-PAGE, and electro-transferred to nitrocellulose membranes. Next, the membranes were blocked for nonspecific binding with TBS 5 % milk and 0.05 % Tween-20 (T-TBS) and incubated overnight in T-TBS/0.5 % BSA with 0.05 µg/ml anti-CFTR or anti-RAC3 primary antibodies. Subsequently, the membranes were washed, incubated for 1 h with a HRP-conjugated secondary antibody and developed using chemiluminescence (Santa Cruz Biotechnology).

RT-PCR assay

In order to check the expression of CFTR in transfected cells, RT-PCR assays were performed as previously described [20]. Briefly, total RNA was isolated from HCT116 cells using the TRIzol protocol (Invitrogen). Next, reverse transcription was carried out using a SuperScript II kit (Invitrogen) following the manufacturer’s instructions. Gene expression analyses were performed using sequence-specific primers for: CFTR forward 5´-actattgccaggaagccatt-3’ and reverse 5’-caccggaactcaagcaagtg–3’ and β-actin forward 5´-cggaaccgctcattgcc-3´ and reverse 5´-aacacactgtgcccatcta-3´ as internal control.

Clonogenic assay

Clonal cell growth was assessed as previously described [12]. Briefly, single-cell suspensions were seeded into 6-well plates (200–800 cells/well) in DMEM/F12 supplemented with 10 % FCS. Medium was changed every 72 h. After 7 days of culture the number of colonies was determined using an inverted microscope.

Hoechst efflux assay

Hoechst efflux assays were performed as previously described [12, 23], with some modifications. Briefly, cells were stained with 5 µg/ml Hoechst 33,342 in the presence or absence of 10 µM nifedipine (Sigma) for 90 minutes at 37 °C. Then, the cells were washed and suspended in PBS. The Hoechst content in the presence or absence of nifedipine was determined by spectrophotometry at 350 nm and cell numbers were determined by counting using fluorescence microscopy.

Expression vector transfections

HEK293 cells were transfected with a PiggyBac transposon gene expression vector for RAC3 (pPB[Exp]-EGFP/Puro-CMV > hNCOA3[NM_181659.2]) and an expression vector (pRP[Exp]-mCherry-CAG > ORF_1785 bp). HCT116 cells were transfected with an expression vector (pRV-GFP-puromycin) for shRNA-RAC3 [18] or shRNA-CFTR (Table 1) and selected for stable expression using puromycin. Once transfected and selected for stable expression, the cultures were amplified and preserved by freezing in liquid nitrogen.

Table 1.

shCFTR sequences

Name	Sequence 5’ to 3’
shCFTR1 sense strand	gatcgggagacagaagaagaggtgcaagatacaatcaagagttgtatcttgcacctcttcttctgtctccttttttgaagct
shCFTR1 antisense strand	agcttcaaaaaaggagacagaagaagaggtgcaagatacaactcttgattgtatcttgcacctcttcttctgtctccc
shCFTR2 sense strand	gatcggaacacataccttcgatattcaagagatatcgaaggtatgtgttcttttttgaagct
shCFTR2 antisense strand	agcttcaaaaaagaacacataccttcgatatct cttgaatatcgaaggtatgtgttcc

Bioinformatics

Datasets and human colon samples

The bioinformatics studies were performed using public datasets, including platform GPL16791 (Illumina HiSeq2500), accession number GSE130226, mRNA expression in CFTR mutated/CFTR wild type Caco-2 cells. In this study 17 sub-clonal cell lines were generated using CRISPR/Cas9-mediated mutagenesis of CFTR. All lines maintained CFTR mRNA production and formation of tight junctions. CFTR+ lines displayed short circuit currents in response to forskolin, while the CFTR- lines did not [24]. The other platform was GPL96 (Affymetrix Santa Clara, CA, USA), accession number GSE24747, analysis of CD133+ or CD133- FACS sorted Caco-2 cells in Gene Expression Omnibus (GEO) public repository data bank at the National Centre for Biotechnology Information (NCBI). Additional studies were performed using samples from The Cancer Genome Atlas (TCGA) (https://www.cancer.gov/): GDC TCGA Colon Cancer (COAD) containing 571 colon tumor samples and normal colon tissues (GTEX, 308 samples).

Enrichment analysis

Different types of enrichment analyses were performed using GSE130226 and GSE24747 datasets: (1) in pathways by overrepresentation, where a selected set of upregulated or downregulated genes was analysed using the platform ConsensusPathDB from The Max Planck Institute for Molecular Genetics in Berlin, Germany http://consensuspathdb.org/ [25], (2) in pathways by expression, where all genes differentially modulated in their expression levels by a mutation of CFTR or as a ratio CD133-/CD133+ were analysed using the same platform as in item 1, and the list of genes was uploaded with its respective values of expression. In both cases the analysis included the sources Kegg and Wikipathways, and (3) biological activities were analysed using Gene Ontology (GO) (http://geneontology.org/) and STRING platforms (http://string-db.org/).

Network analysis

Network analysis for putative interactions between CFTR, RAC3 (NCOA3) and CD133 (PROM1) was first performed using the GeneMANIA tool in Cytoscape software, version 3.8.0 and then analysed using the STRING platform (http://string-db.org/). The networks for specific functional clusters in a pathway were performed and analysed using the STRING platform, where the list of gene sets and their respective expression values were uploaded.

Gene pattern and mutation analyses in human colon samples

The gene expression patterns and mutations of human CRC and control colon samples were analysed using the Xena platform (https://xenabrowser.net/) [26].

Results

CFTR expression is upregulated by RAC3 overexpression in CRC cells

Previously, RAC3 overexpression has been associated with CRC, increasing resistance to chemotherapeutic drugs [18], and with cancer stem cells [12]. Taking into account that CFTR is considered a tumor suppressor, but its expression is high in normal stem cells [22], we first decided to investigate whether its expression may be associated with RAC3 expression. To this end, we analysed CFTR protein expression in human HCT116 CRC cells. Normally, these cells overexpress RAC3, but they were modified by stable transfection with a vector constitutively expressing shRAC3, which inhibits RAC3 expression as well as cancer stem cell properties, as previously reported [12]. We found that RAC3 downregulation significantly inhibited CFTR expression compared with that of wild type HCT116 cells (Fig. 1a). Because we have previously demonstrated that RAC3 overexpression in non-tumoral cells may induce a tumoral phenotype, being capable of inducing tumors when inoculated in animals, we next investigated whether CFTR may also be modulated in such model. To this end, we measured CFTR protein levels in human HEK293 cells, wild type or constitutively overexpressing RAC3. We found that although CFTR was almost undetectable in wild type HEK293 cells, it was upregulated by RAC3 overexpression (Fig. 1b). In agreement with these results and our previous work [12], the cancer initiating ability appears to be accompanied by an increased CFTR expression, which is discordant with its well-described tumor suppressor role.

Fig. 1.

RAC3 modulates CFTR expression. a CFTR protein expression determined in HCT116 cells, wild type (wt) or carrying shRAC3. The bar diagram shows the mean ± SD relative densitometric units (RDU) compared to Tubulin from three independent experiments. At left RAC3, at right CFTR. b CFTR protein expression determined in HEK293 cells, wt or constitutively expressing RAC3. At left RAC3, at right CFTR. *p < 0.05 compared to control; Tukey’s test

CFTR downregulation inhibits HCT116 cancer stem cell properties

In order to determine whether this change in CFTR expression level is only a consequence of RAC3 expression, without specific effects of CFTR on mesenchymal or cancer stem cell properties, we inhibited CFTR expression in HCT116 cells through constitutive expression of a short hairpin RNA. We found that cells carrying two variants of shCFTR express significantly lower levels of CFTR protein compared to wild type cells, as determined by both Western blotting and immunofluorescence (Fig. 2a and b, respectively). Similar results were obtained using RT-PCR (data not shown). Using clonogenic assays, we found that the number of shCFTR colonies was significantly inhibited compared to those of wild type cells. Similar results were obtained using the CFTR inhibitor-172 in wild type HCT116 cells, confirming that the ability of HCT116 cells to initiate clonogenic growth may be dependent on CFTR activity (Fig. 2c). Based on the property of cancer stem cells to rapidly exclude Hoechst staining through their high ABC-MDR transporter expression [12, 23], we next decided to detect differences in the number of cancer stem cells between CRC cell populations expressing high or low levels of CFTR, in the absence or presence of nifedipine, a drug that blocks ABC transporter activity. We found that in the absence of the blocking drug the number of cells retaining Hoechst stain over 90 minutes was slightly increased in the shCFTR cell cultures (Fig. 2d). However, when ABC transporters were blocked, the stain retention showed similar values as those in cells without blocking. These results suggest that the amount of cancer stem cells in the shCFTR cell populations is lower than in wild type cells. Our results are in agreement with the expression levels of some cancer stem cell markers [12, 27] as shown in Fig. 2e. We found that CD133, c-Myc and Oct4 were significantly inhibited by CFTR downregulation, as detected by immunofluorescence.

Fig. 2.

A diminished CFTR expression inhibits HCT116 cancer stem cell properties. a Bar diagram showing mean ± SD relative densitometric units (RDU) of CFTR expression in HCT116 cells transfected with two variants of shCFTR compared to Tubulin from three independent experiments. An image of one representative experiment is shown at the bottom. b CFTR detection by immunofluorescence, 40x. c Image and bar diagram showing mean ± SD HCT116 cell colony growth, each from one clone, when seeded at densities of 800 cells per well, *p < 0.05; Tukey’s test. d Bar diagram showing mean values ± SD of intracellular Hoechst (H) determined by spectrophotometry at 350 nm in HCT116 cells in the presence or absence of nifedipine (N), wild type (wt), or carrying shCFTR. *p < 0.05 compared to Hoechst (H); t-test. e Expression of CD133 and pluripotency markers Nanog, c-Myc and Oct4 determined by immunofluorescence, 40x. The bar diagram corresponds to the mean of three independent experiments ± SD of CD133, Nanog, c-Myc and Oct4 positive HCT116 cells compared to Hoechst positive stained cells. *p < 0.05 compared to HCT116; Tukey’s test

CFTR absence in CRC cells induces simultaneous up- and downregulation of genes involved in cancer pathways

Since our results from HCT116 cells do not seem to be supported by previous work and the tumor suppressor role of CFTR [5], we used public repository datasets to explore gene expression patterns in another human CRC cell line, Caco-2, in the absence of functional CFTR (GSE130226) but expressing CFTR mRNA (it is not a knockout), where other authors have previously shown that CFTR knockout increased oncogenic properties [24, 28]. To additionally understand whether the genes that are up- or downregulated by the absence of functional CFTR could be associated with a cancer stem cell phenotype, we also analysed gene expression patterns in CD133+ and CD133- Caco-2 side populations (GSE24747), a known marker of this side population [27]. Interestingly, we found that the expression levels of CFTR and RAC3 were significantly lower in CD133- than in CD133+ cells (Fig. 3a and b, respectively). However, no significant differences in CD133 and RAC3 expression were observed between wild type and mutant CFTR cells.

Fig. 3.

High CFTR and RAC3 expression is associated with cancer stem cell marker CD133 expression in human colorectal cancer cells. Bar diagram showing mRNA expression of RAC3 (a) and CFTR (b) both from public repository microarray data of CD133+ or CD133- FACS sorted Caco-2 cells. *p < 0.05 compared to CD133-. Platform GPL96 Affymetrix, accession number GSE24747

Knowing that cancer stem cells are required for cancer development and perpetuation, we next decided to investigate how the change in CFTR expression in Caco-2 cells may contribute to tumor development, even without affecting the expression levels of at least a typical cancer stem cell marker like CD133. Firstly, we performed an enrichment of pathways by total gene expression analysis of CFTR mutant Caco-2 cells using the bioinformatics platform Consensus Path-DB from the Max Planck Institute for Molecular Genetics [25]. One of the most important pathways affected by the absence of CFTR turned out to be cancer, involving 242 genes that were changed in their expression levels from a total of 526 genes of the pathway (Table 2). By analysing this gene set using the STRING platform, we obtained a network of 240 nodes with 4177 edges (expected 1507), with an average node degree of 34.8, an average local clustering coefficient of 0.591 and a PPI enrichment p-value < 1.0e-16. Thereby, we confirmed that these proteins have more interactions among themselves than would be expected for a random set of proteins of similar size drawn from the genome. Such an enrichment indicates that the proteins are at least partially biologically connected, as groups, in this case pathways in cancer (Fig. 4a). Before performing a similar analysis for CD133-dependent gene expression, we calculated a Log Fold for the ratio CD133- to CD133+ induced gene expression, and next performed an enrichment analysis. By doing this, we found that some important pathways were cytokine-cytokine receptor interaction, cell cycle (concerning a high number of genes) and metabolic reprogramming in colon cancer, involving 40 from 42 genes of the pathway (Table 2). We analysed this last gene set using the STRING platform and obtained a network with 37 nodes, 358 edges (expected 20), an average node degree of 19.4, an average local clustering coefficient of 0.762 and a PPI enrichment p-value < 1.0e-16, confirming that these proteins are part of metabolic reprogramming in colon cancer (Fig. 4b).

Table 2.

Pathway analysis

Pathway	Genes contained number	Pathway size	% of genes contained	p-value	q-value
Enrichment regarding genes that are modulated by CFTR mut
VEGFA-VEGFR2 signaling pathway	131	236	55.51	3.09E-23	2.20E-21
MAPK signaling pathway - Homo sapiens	147	295	49.83	7.23E-26	8.11E-24
Focal Adhesion-PI3K-Akt-mTOR signaling pathway	146	302	48.34	1.06E-25	9.20E-24
Pathways in cancer - Homo sapiens	242	526	46.01	1.90E-41	1.49E-38
PI3K-Akt signaling pathway - Homo sapiens	157	354	44.35	1.64E-27	2.58E-25
Enrichment regarding the differential gene expression levels in CD133- respect to CD133+
Metabolic reprogramming in colon cancer	40	42	95.24	1.01E-04	4.56E-03
Cell cycle - Homo sapiens	114	124	91.94	8.80E-06	6.03E-04
Cellular senescence - Homo sapiens	144	160	90	4.88E-04	1.29E-02
Cytokines and inflammatory response	26	29	89.66	6.69E-03	9.59E-02
Cytokine-cytokine receptor interaction - Homo sapiens	246	294	83.67	9.05E-06	6.03E-04
Overepresentation analysis. Common genes that are overexpressed in CFTR mut and CD133+ tumoral colorectal cells
Pathways in cancer - Homo sapiens	95	526	18.10%	5.72E-03	1.27E-01
Transcriptional misregulation in cancer - Homo sapiens	42	186	22.70%	1.00E-03	7.01E-02
Proteoglycans in cancer - Homo sapiens	41	201	20.40%	8.61E-03	1.33E-01
Signaling pathways regulating pluripotency of stem cells - Homo sapiens	33	139	33.70%	1.53E-03	8.09E-02
TGF-beta signaling pathway	29	132	22%	9.05E-03	1.33E-01
Cytokine-cytokine receptor interaction - Homo sapiens	24	294	8.16%	2.20E-03	1.35E-04
Colorectal cancer - Homo sapiens	21	86	24.40%	7.28E-03	1.33E-01
MAPK signaling pathway - Homo sapiens	15	295	5.08%	5.79E-03	7.70E-02
Chemokine signaling pathway - Homo sapiens	12	189	6.35%	2.28E-03	4.08E-02
Overepresentation analysis. Common genes that are downregulated in CFTR mut and overexpressed in CD133- tumoral colorectal cells
Pathways in cancer - Homo sapiens	69	526	13.10%	1.71E-03	8.73E-02
VEGFA-VEGFR2 signaling pathway	34	236	14.40%	6.02E-03	1.11E-01
Focal adhesion - Homo sapiens	30	199	15.10%	4.98E-03	1.04E-01
NF-kB signaling pathway - Homo sapiens	18	95	19.10%	2.16E-03	8.73E-02
Target Of Rapamycin (TOR) signaling	9	36	25.00%	4.51E-03	9.99E-02
Overepresentation analysis. Common genes that are downregulated in CFTR mut and overexpressed in CD133+ tumoral colorectal cells - Source: KEGG
MAPK signaling pathway - Homo sapiens	64	295	21.70%	2.21E-03	5.90E-02
Cytokine-cytokine receptor interaction - Homo sapiens	63	294	21.40%	3.22E-03	6.45E-02
Nuclear Receptors Meta-Pathway	74	316	23.40%	9.47E-05	7.02E-03
Insulin signaling	48	160	30.00%	1.92E-06	7.12E-04
EGF-EGFR signaling pathway	45	162	27.80%	3.48E-05	4.97E-03
TNF signaling pathway - Homo sapiens	28	110	25.50%	4.01E-03	7.45E-02
Intestinal immune network for IgA production - Homo sapiens .	7	49	14.95%	3.97E-04	1.06E-02
Inflammatory bowel disease (IBD) - Homo sapiens .	7	65	10.85	2.80E-03	3.82E-02
Overepresentation analysis. Common genes that are overexpressed in CFTRmut and CD133- tumoral colorectal cells
RNA transport - Homo sapiens	42	171	24.60%	1.12E-07	0
Cell cycle - Homo sapiens	30	124	24.20%	9.77E-06	0.0004
Spliceosome - Homo sapiens	30	134	22.40%	4.81E-05	0.0014
Glycolysis/Gluconeogenesis - Homo sapiens	18	68	26.50%	1.70E-04	0.0038

Source: KEGG and Wikipathways

Fig. 4.

Gene networks involved in cancer development are affected by CFTR mutation or selective expression of CD133 in colorectal cancer cells. a Image corresponding to a network obtained using STRING analysis of 242 genes involved in cancer pathways affected by CFTR mutation (from 526 genes of pathways in cancer). The data are presented as the ratio of mRNA expression in CFTR mutated/CFTR wild type cells. Public repository data, platform GPL16791 Illumina HiSeq2500, accession number GSE130226. b Network obtained using STRING analysis of 40 genes differentially expressed in CD133+ or CD133- cells involved in metabolic reprogramming in colon cancer (from 42 genes of the pathway). The data are presented as the ratio of mRNA expression in CD133-/CD133+ cells. Public repository microarray data, platform GPL96 Affymetrix, accession number GSE24747. Dark red circles are the higher upregulated genes, turning to dark blue for the lower downregulated genes. c Gene patterns shared between wild type and mutant CFTR, CD133+ or CD133- side populations of colorectal cancer cells. The image represents a Venn diagram, where the green colour corresponds to genes that are upregulated in CFTR mutant (CFTR mut) cells, while the blue colour corresponds to genes that are downregulated in CFTR wild type (CFTR wt) cells. Light red and yellow correspond to genes upregulated in CD133- and CD133+ cells, respectively. Numbers correspond to the number of genes in each group

Although RAC3 and CD133 expression levels do not seem to be dependent of CFTR, still some genes up- or downregulated by changes in CFTR expression could be shared with CD133+ cells. Therefore, we next analysed all the upregulated genes that could be shared between CFTR wild type or mutant and CD133+ or CD133- side populations by generating Venn diagrams. Only genes with a 1.5 fold change between conditions were considered for the analysis. We found that both mutant and wild type CFTR cells shared with CD133+ cells 415 and 489 upregulated genes, respectively (Fig. 4c). Interestingly, a lower number of upregulated genes (139) for CFTR mutant cells and 174 for wild type cells were identified as being upregulated in CD133+ cells. In order to investigate in which pathways these gene sets are mainly involved, we performed an additional bioinformatics study of enrichment by over-representation. We found that the genes upregulated by the absence of CFTR and also expressed at high levels in CD133+ cells were most directly associated with cancer and colorectal development (Table 2). Unexpectedly, however, we found that there were genes upregulated in both wild type CFTR (those that are downregulated by the absence of CFTR) and CD133- cells that also may contribute to cancer development (Table 2).

In addition, we found that there are other upregulated genes in wild type CFTR and CD133+ cells that may also contribute to cancer development through inflammatory responses and nuclear receptor modulations (Table 2), while only the genes upregulated by CFTR mutation and expressed at high levels in CD133- cells were less directly related to cancer development (Table 2). These results suggest that mutation of CFTR may contribute to the development of CRC, at least through the induction or maintenance of expression of cancer stem cell-associated genes, usually expressed in CD133+ cells, even without changing the expression level of this molecule. In addition, another set of genes that are highly expressed in wild type CFTR cells was identified that may also contribute to cancer development, possibly when CFTR is upregulated.

CFTR is highly expressed in CRC patients without cystic fibrosis and correlates with the presence of cancer stem markers

Based on the notion that CFTR may play a role in CRC development, even without mutation, we wanted to investigate whether different CFTR expression levels may be found in CRC patients, and whether there is a correlation with CD133 expression or, in agreement with our results (Fig. 3), with RAC3 expression. To this end, we performed a bioinformatics analysis of 571 human CRC samples from The Cancer Genome Atlas using the Xena platform [26]. The samples were arranged according to CFTR mRNA expression level, after which RAC3 and CD133 expression levels were aligned with each CFTR arranged sample. Moreover, samples containing CFTR mutations were identified. Several types of CFTR mutations were found, but only in 4 % of the samples. Those containing deleterious or missense mutations were grouped (red group in the second column of Fig. 5b) and compared with those with silent or no mutations (blue). We found that the CFTR mRNA expression level was slightly increased in this last group, albeit not significant (Fig. 5a). Similar results were obtained when we analysed CD133 expression, whereas a significantly increased RAC3 expression was found to be associated with silent CFTR mutations or wild type CFTR (Fig. 5a). A higher RAC3 expression in samples with higher CFTR expression was observed, as indicated by red bands concentrated at the top of the third lane (in the silent or wild type CFTR group; Fig. 5b), although no evident correlation was found for CD133 (supplementary results).

Fig. 5.

CFTR is highly expressed in human colon adenocarcinomas from patients without cystic fibrosis and correlates with the expression of cancer stem cell markers. Bioinformatics analysis using the Xena platform. Human colon cancer samples: GDC TCGA Colon Cancer (COAD) were filtered excluding those undetermined for the selected variable (396 valid data). a Bar diagram showing mRNA expression levels of CFTR, RAC3 and CD133 in samples with CFTR mutations that affect its functionality (15 samples, missense or frameshift) (Welch’s t-test p = 0.08033 for CFTR, p = 0.002726 for RAC3 and p = 0.4957 for CD133). b Gene expression pattern image in which patient samples are ordered from high (red) to low (green) expression. c Gene expression pattern image of human colon adenocarcinomas (571 samples) and normal colon tissues (GTEX, 308 samples). d, e, f Bar diagrams showing expression of CFTR (Welch’s t-test p = 3.319e-57, t = -18.56), cancer stem marker CD133 (Welch’s t-test p = 2.297e-26, t = -11.16) and c-Myc (Welch’s t-test p = 1.132e-55, t = -18.10) in human colon adenocarcinomas compared to normal colon tissues

In order to better understand the impact of CFTR expression on CRC development, we performed an additional analysis, where human CRC samples were compared with normal tissues (GTEX, 308 samples). The expression levels of CFTR (Fig. 5c and d) and the stem cell markers CD133 (Fig. 5c and e) and c-Myc (Fig. 5c and f) were significantly increased in the CRC samples compared to the normal tissues (p = 3.319e-57 for CFTR, p = 2.297e-26 for CD133 and p = 1.132e-55 for c-Myc). Therefore, although CFTR could be considered a tumor suppressor, most of the human CRC samples of patients without cystic fibrosis or CFTR mutations that affect its biological function exhibit an increased expression of this molecule.

CFTR, RAC3 and CD133 are part of an inter‐regulatory network in CRC

In order to better understand the functional relationship between CFTR, RAC3 and CD133 in pathways contributing to CRC development, we performed an analysis of these genes looking for a possible cluster to which they may be associated. To this end, we used the GeneMANIA tool of the Cytoscape 3.8.0 software. We found that CFTR, RAC3 and CD133 require at least 22 nodes in order to be part of the same network. This network involves different regulatory interactions involving several biological functions, including epithelial differentiation and endothelial proliferation dependent on VEGF, such as that mediated by XBP1 through the AKT/GSK pathway [29], as determined by Gene Ontology analysis enrichment (Fig. 6). Next, we analysed this network using the STRING application and found that these genes are part of a network containing 33 nodes with 250 edges, with an expected number of edges of 75. The average node degree was 15.2, although the three input genes were not considered in the size as a measure of degree and are represented as black similar size nodes in Fig. 6. The average local clustering coefficient was 0.799, and the PPI enrichment p-value < 1.0e-16. The three genes are part of a high score network, with several nodes and edges of interaction between them, including a very short way involving at least three genes, XBP1, SLC9A3R1 and STIP1, all of which are differentially regulated in both CFTR mutant cells and CD133+ or CD133- side populations.

Fig. 6.

CFTR, RAC3 and CD133 are part of an inter-regulatory network. Network obtained by entering three genes (black and same size) using the GeneMANIA tool within Cytoscape 3.8.0 software. The size of the nodes corresponds to the degree of each one of them. PROM1 is the gene name of CD133, NCOA3 is the gene name of RAC3

Discussion

CFTR functions as a chloride (Cl⁻) and bicarbonate (HCO3⁻) anion channel. It is located in the apical surfaces of luminal epithelia and its deficiency is associated with cystic fibrosis, with the most severe manifestations occurring in the lung, but also in the intestine, affecting nutrient absorption, inflammatory responses and inducing obstruction [4]. The most common mutation associated with cystic fibrosis is ΔF508, and there is ample evidence showing that this mutation increases the risk of cancer in the gastrointestinal tract [4, 30, 31]. In fact, CFTR has been defined as a tumor suppressor [5]. About 5 % of CRCs is inherited in an autosomal dominant fashion involving, among others, APC and several DNA mismatch repair genes. Since cystic fibrosis is not the most frequent disorder contributing to CRC, the role of CFTR in patients with CRC has so far not been clarified. Here, we show that although mutation of CFTR induces the expression of genes involved in signaling networks related to cancer, there are other genes that also contribute to CRC development, probably when they are overexpressed in CFTR overexpressing cases (Fig. 7). If this were not the case, normal CFTR expression could be considered as a CRC risk factor, which it is not. Therefore, we conclude that CFTR may contribute to CRC development not only through mutation, but also through increased expression in patients without cystic fibrosis, probably enhancing the expression of a set of genes involved in cancer pathways distinct from those involved in cystic fibrosis. In agreement with this notion we observed, when we inhibited CFTR expression in human CFTR wild-type CRC cells, a decreased expression of cancer stem cell markers and an enhancement of features such as clonogenic growth and a lower activity of ABC-MDR transporters.

Fig. 7.

Role of CFTR in cancer. a List of gene sets involved in cancer pathways that are selectively upregulated in wild type or mutant CFTR cells. b Image showing two pathways by which up- or downregulation of CFTR may induce colorectal cancer

CFTR is expressed along the entire length of the intestinal tract, with a gradient of decreasing expression from proximal (duodenum) to distal (ileum) in the small intestine. CFTR expression is strongest at the base of the crypt, the location of the intestinal stem cell compartment [32, 33]. Although the origin of cancer stem cells has not been clearly defined yet, it is thought that they may be derived from normal stem cells or from trans-differentiation of other mature differentiated cells [3, 6, 7, 34]. The increased expression of CFTR in normal stem cells could be in agreement with our current results with HCT116 cells, where inhibition of CFTR induced loss of cancer stem cell characteristics. In addition, through bioinformatics analysis, we found that CD133+ cells express higher levels of CFTR than CD133- cells.

Interestingly, it has been reported that CFTR is expressed in human Philadelphia chromosome-positive acute leukaemia cells, and that application of CFTRinh-172, a CFTR inhibitor, downregulates the expression of CFTR, p-BCR-ABL and classical Wnt/β-catenin signaling [35]. Additionally, in placental trophoblastic cells and human airway cells, CFTR has been found to be functionally expressed and to promote cell migration [36].

There are several mechanisms by which the absence of CFTR may contribute to cancer development through altering homeostasis and the balance of acid-base and pH equilibrium, the expression of genes involved in proliferation, migration and epithelial to mesenchymal transition [30]. Interestingly, the authors of this latter work explored the aetiology of high cancer risk in ΔF508-cystic fibrosis patients. They collected mouse ΔF508-CFTR embryonic stem cells and used them as an in vitro model to study the role of CFTR. By doing this, they found that the self-renewal properties are intact in ΔF508-CFTR embryonic stem cells, but that the expression levels of intestine progenitor and major intestine differentiated cell markers were significantly upregulated in ΔF508-CFTR differentiated cells. Moreover, they found that these cells may induce teratomas. Although these results may be interpreted as contradictory to ours, it is important to consider that the authors used a model where CFTR was knocked out in normal pluripotent stem cells, while we inhibited its expression in CRC cells developed without CFTR mutation. Moreover, although some stem properties are shared between normal and cancer cells, it is well known that the ability of pluripotent cells to induce teratomas does not imply overt cancer development [3, 7].

Cancer perpetuation requires cancer stem cells in order to migrate, invade and initiate metastasis [3]. We have previously found that RAC3 is required for maintaining stemness in CRC cells [12]. Here, we show that its downregulation induces decreased CFTR expression in CRC cells, while its overexpression is associated with an increase of this channel in non-tumoral cells. Moreover, when we analysed the expression of these molecules in primary human tumor samples, we noted a correlation between the expression levels of both molecules. Although the molecular mechanism by which this co-regulation happens was not investigated, a role of RAC3 at the transcriptional level as a nuclear receptor coactivator, with the ability to increase the transactivation of several transcription factors, cannot be excluded [14, 19].

In cells with a CFTR mutation, we identified 95 upregulated genes associated with cancer pathways, that are similarly expressed in CD133+ cells. In addition to previously described signals involved in increased proliferation and inflammatory responses [4, 24, 37, 38], these may be important pathways by which CFTR mutation contributes to CRC development and perpetuation, directly enhancing the expression of cancer stem cell-associated genes [39]. Although alteration of CFTR expression did not modify the expression of RAC3, nor another known marker of cancer stem cells, CD133 [27, 39], we have previously shown that high expression of RAC3 correlates with high CD133 expression and cancer stem cells in CRC [12]. Therefore, our results suggest a hierarchical order driving the expression of genes associated with CRC development in patients without cystic fibrosis, where a high RAC3 expression may induce an increase in CFTR expression and then an increase in the expression of genes that participate in cancer pathways (Fig. 7). Notably, we found that a set of 69 cancer pathway genes that are upregulated in wild type compared to mutant CFTR cells, are not shared with CD133+ cells. However, taking into account that the expression levels of CFTR may vary in human colorectal samples, we may conclude that although wild type CFTR is not involved in CRC development, an increased expression of the channel may induce biological responses involving EGF, mTOR, AKT and Wnt signaling, which may contribute indirectly to perpetuation and differentiation of cancer stem cell side populations, as revealed by the analysis of CD133 and c-Myc expression in primary human tumor samples, which correlated with high CFTR expression. In addition, another 489 genes were found to be upregulated in wild type CFTR cells compared to mutant CFTR cells, and shared with CD133+ cells, which are strongly involved in inflammatory responses and whose role favouring a tumoral microenvironment and tumor initiation have been previously reported [13, 40].

Taken together, we show that CFTR may be a target of RAC3, whose expression may be upregulated by inflammatory cytokines [20], driving an increase in the expression of genes that may contribute to cancer stemness and CRC development in patients without CFTR mutation or cystic fibrosis.

Abbreviations

ABC transporters

ATP-Binding Cassette transporters

ATP

Adenosine Triphosphate

BSA

Bovine Serum Albumin

CFTR

Cystic Fibrosis Transmembrane Conductance Regulator

EGF

Epidermal Growth Factor

MDR

Multiple Drug Resistance

NF-ҡB

Nuclear Factor Kappa B

OCT4

Octamer-Binding Transcription Factor 4

PBS

Phosphate Buffered Saline

PPI

Protein-Protein Interaction

shCFTR

Short Hairpin for CFTR

shRAC3

Short Hairpin for RAC3

TNF

Tumor Necrosis Factor

Author contributions

Alejandra Graciela Palma: methodology, investigation, visualization, formal analysis; Mileni Soares Machado: methodology, investigation, visualization; María Cecilia Lira: investigation, formal analysis; Francisco Rosa: investigation, formal analysis; María Fernanda Rubio: Review and editing, visualization; Gabriela Marino: formal analysis, visualization; Basilio Aristidis Kotsias: conceptualization, writing, review and editing; Mónica Alejandra Costas: supervision, conceptualization, funding acquisition, project administration, original draft writing, formal analysis, visualization.

Funding

This work was supported by Consejo Nacional de Investigaciones Científicas y Técnicas [PIP 11220150100118CO] and Agencia Nacional de Promoción Científica y Tecnológica [PICT 1424, PBID 2014].

Data availability

Not applicable.

Code availability

Not applicable.

Declarations

Conflicts of interest/competing interests

We wish to confirm that there are no known conflicts of interest associated with this publication and that there has been no significant financial support for this work that could have influenced its outcome. 

Ethics approval

Not applicable. 

Consent to participate

Not applicable.

Consent for publication

Not applicable.