Inferred inactivation of the Cftr gene in the duodena of mice exposed to hexavalent chromium (Cr(VI)) in drinking water supports its tumor-suppressor status and implies its potential role in Cr(VI)-induced carcinogenesis of the small intestines

Abstract

Carcinogenicity of hexavalent chromium [Cr (VI)] has been supported by a number of epidemiological and animal studies; however, its carcinogenic mode of action is still incompletely understood. To identify mechanisms involved in cancer development, we analyzed gene expression data from duodena of mice exposed to Cr(VI) in drinking water. This analysis included (i) identification of upstream regulatory molecules that are likely responsible for the observed gene expression changes, (ii) identification of annotated gene expression data from public repositories that correlate with gene expression changes in duodena of Cr(VI)-exposed mice, and (iii) identification of hallmark and oncogenic signature gene sets relevant to these data. We identified the inactivated CFTR gene among the top scoring upstream regulators, and found positive correlations between the expression data from duodena of Cr(VI)-exposed mice and other datasets in public repositories associated with the inactivation of the CFTR gene. In addition, we found enrichment of signatures for oncogenic signaling, sustained cell proliferation, impaired apoptosis and tissue remodeling. Results of our computational study support the tumor-suppressor role of the CFTR gene. Furthermore, our results support human relevance of the Cr(VI)-mediated carcinogenesis observed in the small intestines of exposed mice and suggest possible groups that may be more vulnerable to the adverse outcomes associated with the inactivation of CFTR by hexavalent chromium or other agents. Lastly, our findings predict, for the first time, the role of CFTR inactivation in chemical carcinogenesis and expand the range of plausible mechanisms that may be operative in Cr(VI)-mediated carcinogenesis of intestinal and possibly other tissues.

1. Introduction

Hexavalent chromium [Cr (VI)] compounds are widely used in metallurgical, chemical and refractory industries [1]. Their carcinogenic properties have been supported by a number of epidemiological studies (reviewed in [2]). Occupational studies of workers exposed to a variety of airborne Cr (VI) compounds provided most of the evidence for an increased risk of lung cancers, and suggestive evidence for an increased risk of sinonasal cancers. More recently, a meta-analysis of 47 occupational cohort studies found in exposed male workers an increased risk of developing cancers of the oral cavity, respiratory system, pharynx, stomach and prostate, as well as increased risk of death due to cancers of the lung, larynx, bladder, kidney, testes, bone and thyroid gland [3]. These occupational studies implicated a number of Cr(VI) compounds of diverse nature and variable solubility, suggesting that the carcinogenic effects are attributable to hexavalent chromium generally, rather than to only some specific Cr(VI) compounds .

The carcinogenicity of Cr(VI) compounds have also been shown by experimental studies, using several animal models and routes of exposure (reviewed in [2]). Most notable among them were 2-year rodent studies, which were designed by the National Toxicology Program (NTP) in an effort to identify potential hazards of human exposure to Cr (VI) in drinking water [4, 5]. These NTP studies found a significantly increased incidence of squamous cell carcinomas of the oral cavity in F344/N rats, as well as an increased combined incidence of adenomas and carcinomas of the small intestines in B6C3F1 mice, in animals exposed to sodium dichromate dihydrate (SDD) in drinking water.

Cellular and molecular changes that could explain the induction of cancers by the Cr(VI) compounds are not completely understood. For this reason, studies that can generate more mechanistic evidence of Cr(VI) compounds are still needed to support causality of associations determined by epidemiological studies [6], and to determine human relevance of the results produced by animal studies. Numerous mechanistic studies performed in vitro and/or in vivo so far have suggested the role of mutagenesis, inflammation, or cytotoxicity followed by regenerative proliferation as the carcinogenic mode of action (MoA) of Cr(VI) (reviewed in [2, 7]). These modes of action were attributed, at least in part, to underlying oxidative stress [8], DNA reactivity of Cr-species generated by the reduction of Cr(VI), and/or epigenetic changes, such as, e.g., epigenetic silencing of the hMLH1 gene, resulting in impaired DNA mismatch repair [9].

In this paper we present systems biology evidence that oral exposure to hexavalent chromium in drinking water is also associated with inactivation of the cystic fibrosis (CF) gene Cftr (cystic fibrosis transmembrane conductance regulator) in murine small intestines. Considering the recently reported tumor-suppressor role of the CFTR gene in murine and human intestinal cancers [10], our finding expands the range of plausible mechanisms that may be operative in Cr(VI)-mediated carcinogenesis of intestinal and possibly other tissues. In addition, our finding supports the human relevance of intestinal carcinogenesis detected in mice by the NTP mouse study and implies possible vulnerable groups that may be at increased risk of Cr(VI)-induced carcinogenicity.

2. Materials and Methods

2.1. Gene expression data

“Transcriptomic data to assess hexavalent chromium mode of action in mice and rats”, deposited in the Gene Expression Omnibus (GEO) (URL: https://www.ncbi.nlm.nih.gov/geo/) as a SuperSeries GSE87259, were used in this study. We analyzed a subset of the data for the duodena of B6C3F1 mice continuously exposed to drinking water containing sodium dichromate dihydrate (SDD) at target concentrations of 0 mg/L (control), or 0.3, 4, 14, 60, 170 and 520 mg/L SDD until study termination at days 8 or 91, when the animals were euthanized and duodenal mucosa specimens were collected for gene expression analysis. This subset consists of 70 microarrays from the Agilent-014868 Whole Mouse Genome Microarray 4x44K platform. The generation and the use of this dataset has been previously described in peer-reviewed articles [11–14]. Normalized gene expression data for exposed and corresponding control mice for triplicated experiments were retrieved using the GEO2R interactive web tool [15]. Exposure levels and durations were selected to reflect previously reported findings from the 2-year and 91-day NTP rodent studies of hexavalent chromium in drinking water (reports TOX72 and TR546; URL: https://ntp.niehs.nih.gov/whatwestudy/topics/hexchrom/index.html) as well as contemporary trend of short-term toxicogenomic studies.

2.2. Upstream Regulator Analysis

We applied the Upstream Regulator Analysis (URA) feature [16] of the Ingenuity Pathway Analysis suite (IPA) v. 2.4 (Qiagen Bioinformatics, Redwood City, CA, USA) to infer upstream regulatory molecules that are likely responsible for the observed gene expression changes. This inference is based on their direct or indirect relationships with differentially expressed genes, using prior knowledge from an expertly-curated knowledge base to reconstruct gene expression networks. Cut-off values of p=0.05 and log₂FC= 0.585 were used for differentially expressed genes in this analysis. The analysis was performed using the following settings: Core analysis; expression analysis by IPA; use information from Ingenuity Knowledgebase; consider direct and indirect relationships; causal network analysis. Consistency between the effects of potential upstream regulators on the transcriptional status of downstream genes was determined for each reconstructed network using Z-scores calculated as previously reported [16]. Z-scores for each potential upstream regulator reflect the confidence with which the upstream regulator is activated or inhibited. P-values that reflect likelihood of a chance association between the differentially expressed genes and genes regulated by given up-stream regulators were determined by Fisher’s exact test.

2.3. Identification of correlated gene expression studies

The Base Space Correlation Engine (BSCE) [17] from Illumina, Inc. (San Diego, CA, USA) was used to identify annotated gene expression studies in public repositories that correlate with gene expression changes in duodena of Cr(VI)-exposed mice. This suite employs rank-based directional enrichment (a running Fisher statistical test) to determine correlations between query datasets and collections of genes that are associated with specific biological functions (biogroups) or collections of ranked genes that correspond to specific treatments or conditions in gene expression experiments (biosets). Probe sets with absolute fold change ≥1.2 and p≤0.05 were uploaded to the BSCE and analyzed using the Meta-Analysis application.

2.4. Gene Set Enrichment Analysis

The expression data for duodena of mice exposed to SDD for 8 days or 91 days were analyzed for enrichment of “hallmark” (H) and oncogenic signature (C6) collections of gene sets from the Molecular Signatures Database (MSigDB) [18] using the Gene Set Enrichment Analysis (GSEA) method [19]. The gene sets enriched at p<0.01 and FDR q<0.1 were visualized using the EnrichmentMap Cytoscape Application 3.3 [20]. Parameters for the Enrichment Map were the following: Gene sets enriched in Cr(VI) exposed vs. control groups; Data Set Edges: Automatic; Metric: Overlap; Cutoff: 0.5.

3. Results

3.1. The IPA Upstream Regulator Analysis (URA) suggests decreased activity of Cftr in the duodena of exposed mice

Upstream Regulator Analysis identified several upstream regulators that can explain observed differences in gene expression between exposed and control mice (Figures 1 and 2, and Supplemental files 1 and 2). The Cftr gene was identified among the top scoring upstream regulators, ranking 3^rd and 15^th for 8-day and 91-day exposures, respectively. The Cftr gene was selected as a target of analysis because of its previously suggested role in small intestine and colon carcinogenesis [10] without a known association with hexavalent chromium toxicity.

Figure 1. Upstream regulators inferred from differentially expressed genes in the duodena of mice exposed to hexavalent chromium in drinking water for 8 days.

Upstream regulators are ordered based on absolute values of Z-scores (A) and p-values (B). CFTR ranks 3^rd (A) and 11^th (B) among the inferred upstream regulators. A – color coding indicates Z scores; B- color coding indicates enrichment p-values. Dots: (A) |Z|<2; (B): p>0.05

Figure 2. Upstream regulators inferred from differentially expressed genes in the duodena of mice exposed to hexavalent chromium in drinking water for 91 days.

Upstream regulators are ordered based on absolute values of Z-scores (A) and p-values (B).CFTR ranks 15^th (A) and 12^th (B) among the inferred upstream regulators. A – color coding indicates Z scores; B- color coding indicates enrichment p-values. Dots: (A) |Z|<2; (B): p>0.05

The inactivated status of Cftr (Z-score < -2.0 and p-value p<0.05) was consistently predicted for 8-day exposures at 0.3 ppm, 60 ppm, 170 ppm and 520 ppm (Figure 1). For 91-day exposures, inactivation of Cftr was predicted for exposures ≥ 170 ppm at the same significance thresholds (Figure 2).

Histopathologic findings previously reported as significantly occurring in duodena of mice exposed to SDD for 8 days included vilous cytoplasmic vacuolization (≥170 mg/L), villous atrophy and crypt cell hyperplasia (at 520 mg/L). In mice exposed for 91 days, the findings included vilous cytoplasmic vacuolization (≥ 60 mg/L), histiocytic infiltration of villous lamina propria and crypt cell hyperplasia (≥170 mg/L) [21].

The inactivated status of Cftr was predicted based on the over-expressed or under-expressed status of genes that are known by the IPA knowledgebase to be regulated by Cftr (Figure 3). Intriguingly, the inactivated status of Cftr was not accompanied by significant down-regulation of Cftr expression on the mRNA level. In contrast, the Cftr gene was overexpressed on the mRNA level in the duodena of mice exposed to the two highest concentrations of SDD in drinking water for both 8-day and 91-day durations of exposure (Supplemental files 3 and 4). This suggests that the CFTR is inactivated through posttranscriptional mechanisms.

Figure 3.

Heatmap depicting differentially expressed genes that support inactivated status of Cftr in the duodena of mice exposed to hexavalent chromium for 8 days (A) and 91 days (B). Color coding indicates log expression values (green-red) and CFTR upstream activator z-score (blue-white-red).

This inactivated status of CFTR is unlikely to be a promiscuous non-specific event that could be attributable to a range of duodenal tissue-damaging toxicities. This is due to the fact that no histopathologic changes were reported in the duodena of mice exposed to SDD for 8 days at levels ≤ 60 ppm of SDD in drinking water. On the other hand, the NTP 2-year study found a significant occurrence of histopathologic changes at exposure levels ≥ 14.3 mg/L, including diffuse hyperplasia (starting at 14.3 mg/L), histiocytic infiltration (starting at 172 mg/L), adenomas (starting at 172 mg/L) and carcinomas (at 516 mg/L) [5].

Since our results suggest inactivation of Cftr in mice exposed to the levels of SDD as low as 0.3 mg/L and 60 mg/L, the inactivation of Cftr is not attributable to tissue damage and represents an early effect of Cr(VI) exposure [21].

3.2. Gene expression data for the duodena and ilea of Cr(VI)-exposed mice correlate with studies relevant to the inactivation of Cftr

The BSCE identified positive correlations between 8-day and 91-day expression data for duodena of mice exposed to Cr(VI) and other datasets associated with the inactivated status of Cftr. Top scoring among them were datasets generated for ilea and duodena of Cftr-knockout mice (Table 1) [22, 23].

Table 1:

Top 3 biosets relevant to the CFTR and positively correlated with gene expression data for the duodena of mice exposed to Cr(VI) in drinking water. Retrieved by the BSCE from public repository

Expression dataset		Ilea of high weight mice – CFTR knockout _vs_ wildtype (GSE5715) Na=3358		Ilea of mice – CFTR knockout _vs_ wildtype (GSE5715) Na=3648		Small intestine from cystic fibrosis mouse CFTR knockout _vs_ wildtype (GSE765) Na=587
Exposure level SDD	Exposure duration (Number of genes)
		p-valueb	Common genes	p-valueb	Common genes	p-valueb	Common genes
0.3 mg/L	8 days (Na=2135)	5.5x10−15	395	1.0x10−14	446	6.4x10−17	122
	91-days (Na=1057)	0.2444	151	0.194	159	0.2792	29
4 mg/L	8 days (Na=1813)	3.7x10−12	307	4.9x10−11	344	5.5x10−5	82
	91 days (N=969)	0.1198	139	0.1756	150	0.3574	29
14 mg/L	8 days (Na=3366)	5.8x10−7	585	7.2x10−8	648	0.3524	159
	91 days (Na=1195)	0.0148	181	0.0062	181	0.277	24
60 mg/L	8 days (Na=5871)	4.9x10−43	1198	3.7x10−39	1300	1.3x10−17	304
	91 days (Na=3324)	5.4x10−25	670	3.8x10−22	721	0.0018	122
170 mg/L	8 days (Na=6506)	5.1x10−53	1338	1.4x10−49	1442	4.4x10−42	377
	91 days (Na=5063)	9.1x10−52	1078	7.8x10−49	1145	2.0x10−28	297
520 mg/L	8 days (Na=7212)	1.1x10−73	1569	1.9x10−70	1678	1.4x10−36	385
	91 days (Na=4613)	7.9x10−49	1002	1.2x10−46	1059	1.3x10−27	269

^aNumber of genes in a bioset

^bOverlap p-value

For instance, in mice exposed to 60 mg/L SDD in drinking water for 8 days, 447 genes were found up-regulated (p=1.7x10⁻¹⁹) and 513 were down-regulated (p=7.7x10⁻⁶⁰) concordantly between the duodena of Cr(VI)-exposed mice and the duodena of Cftr-knockout mice (Figure 5). Since this exposure level and duration has not been found to be associated with histopathologic changes in the duodena of Cr(VI)-exposed mice [21], this result supports the inactivation of Cftr by Cr(VI) rather than an indirect effect of Cr(VI)-mediated tissue damage.

Figure 5. Correlation of biosets for the duodena of mice exposed to Cr(VI) and the ilea of mice with Cftr-knockout.

Exposure: 60 mg/L SDD for 8 days. Depicted are numbers of unique and overlapping genes between the two datasets and statistical significance of their overlaps. P-values are determined using the running Fischer test that is implement in BSCE.

3.3. Gene Set Enrichment Analysis (GSEA) implies involvement of molecular changes relevant to carcinogenesis in the duodena of mice orally exposed to Cr(VI)

Gene expression analysis using the GSEA method and the Hallmark (H) or Oncogenic Signature (C6) collections of annotated gene sets identified numerous gene sets relevant to carcinogenesis (Figure 6 and 7 and Supplemental figures S2 and S3). The duodena of mice exposed for 8 days and 91 days display significant enrichment for gene sets “Myc_Targets_v1”, “Myc_Targets_v2”, “E2F_Targets”, “G2M_Checkpoint”, “DNA_Repair”, and “CSR_Late_UP.v1_UP” consistent with sustained proliferative signaling, which represents a hallmark of cancer [24].

Figure 6. Hallmark Gene Sets (H) enriched in the duodena of mice exposed to SDD for 8 days at 0.3–520 mg/L.

(p-value<0.01 and FDR q-value<0.1. Color coding: concentration of SDD. Description of gene sets is available at: http://www.gsea-msigdb.org/gsea/msigdb/index.jsp

Figure 7. Hallmark Gene Sets (H) enriched in the duodena of mice exposed to SDD for 91 days at 0.3–520 mg/L.

(p-value<0.01 and FDR q-value<0.1. Color coding: concentration of SDD. Description of gene sets is available at: http://www.gsea-msigdb.org/gsea/msigdb/index.jsp

In addition, gene sets enriched in the duodena of mice exposed for 8 days support angiogenesis, impaired apoptosis and epithelial-mesenchymal transition (Figure 6) that also represent hallmarks of cancer.

Enrichment of the “Cholesterol_Homeostasis” gene set found for 8-day and 91-day exposures and several exposure levels (Figure 6 and 7) implies activation of cholesterol biosynthesis that is associated with intestinal crypt hyperproliferation and tumorigenesis [25]. Activation of cholesterol biosynthesis is further supported by up-regulation of a master regulator of sterol synthesis Srebf2 [26] in duodena of mice exposed to SDD at ≥ 170 mg/L for 8 days or ≥ 60 mg/L for 91 days (Supplemental files 3 and 4).

Enriched gene sets from the Oncogenic Signature collection imply oncogenic activation of Kras, Src, Shh and PI3K/AKT/mTOR signaling and inactivation of signaling mediated by tumor-suppressors Pten and Rb (Supplemental figures S2 and S3). Taken together, the results support duodenal carcinogenicity in mice of Cr(VI) ingested in drinking water through activation of oncogenic signaling, inactivation of signaling mediated by tumor-suppressors, sustained cell proliferation, impaired apoptosis and tissue remodeling.

4. Discussion

4.1. General

Our results computationally infer inactivation of Cftr, activation of oncogenic signaling, sustained cell proliferation, tissue remodeling and impaired apoptosis in the duodena of mice exposed to a soluble salt of hexavalent chromium in drinking water. This finding is strongly supported by consistent changes in the expression of several genes in the duodena of mice exposed to Cr(VI) and the small intestines of the Cftr null mice [23]. Of 71 genes reported by Norkina et al. as differentially expressed between Cftr null and Cftr wild-type mice [23], 40 were found to be differentially expressed in the duodena of mice exposed to SDD for 8 days at 170 mg/L. These 40 genes, which represent cytochrome P450 enzymes, and genes related to the cytoskeleton, innate immune response, inflammation, metabolism, and transport, displayed consistent changes in expression for Cftr null mice and mice exposed to Cr(VI) in drinking water (Supplemental Table S1), providing additional support for our finding. Our results support the previously suggested tumor-suppressor role of Cftr and open up new avenues in understanding the complexity of toxicological effects of hexavalent chromium. These new avenues include the human relevance of animal toxicological studies of this toxicant, and identification of potentially vulnerable populations for public health interventions. Their implications necessitate more corroborated discussion, which is presented in the next sections.

4.2. CFTR in health and disease

The CFTR gene product is a member of the ATP-binding cassette (ABC) family of membrane transport proteins [27] that serves as a phosphorylation-regulated ATP-gated Cl⁻ and HCO₃⁻-conducting anion channel [27]. It plays a fundamental role in epithelial physiology in many tissues, including the lungs, pancreas, sweat glands and intestinal tract, as well as in male and female reproductive tracts through the control of composition, hydration and pH of epithelial fluid [28].

Consistent with its critical role in epithelial tissue homeostasis, impairment of CFTR activity is associated with disease. Loss-of-function mutations in the CFTR gene cause cystic fibrosis (CF), which is the most common lethal autosomal-recessive disease in Caucasians. It represents a clinically heterogeneous disease, that typically presents with progressive deterioration of lung function due to airway obstruction, inflammation and infections; however, it can also include pancreatic insufficiency, hepatobiliary disease and other abnormalities. In addition, primary intestinal complications that can develop in CF patients include meconium ileus, distal intestinal obstruction syndrome, and intussusception [29].

This clinical variability is, at least in part, attributable to variable CFTR genotypes [30]. From over 2000 identified CFTR gene mutations [31], 352 were causally linked to CF [32], but the majority of the CF cases worldwide are caused by the F508del mutation [33]. Intriguingly, some CFTR mutations, which were not found to cause classic CF, have been associated with mono- or oligosymptomatic CFTR-related diseases, such as obstructive azoospermia, idiopathic pancreatitis or disseminated bronchiectasis, which expand the landscape of diseases associated with inactivation of CFTR [34].

The CF-causing mutations can have different deleterious effects on the CFTR protein, such as (i) impaired synthesis of the full length protein, aberrant protein maturation due to misfolding, misprocessing and mis-localization, (iii) impaired ion channel regulation/gating, (iv) impaired anion conductance, and (v) reduced CFTR stability (reviewed in [33]). The broad spectrum of the known functional impairments of the CFTR protein implies the necessity of diverse therapeutic approaches that would target these specifically impaired functions. However, it also implies diverse toxicologically relevant mechanisms, which can impair CFTR functions and induce adverse outcomes associated with inactivation of CFTR.

Although the main cause of mortality and morbidity in CF is the lung pathology [35], morbidity and mortality in CF patients is not limited only to chronic airway infections and their complications. In fact, the growing body of evidence suggests that CF patients have an increased risk of gastrointestinal cancers [36–38], cancers of the esophago-gastric junction, biliary tract, and small intestines; testicular cancers, lymphoid leukemia [37], and colon cancers [37]. The highest risk of developing bowel cancer (cancer of the small intestines and the colon combined) was found for patients with a severe CF phenotype due to the F508del mutation, which indicates that the degree of CFTR impairment correlates with the risk of cancer development, at least in the colon and small intestines.

Available data suggest that the risk of cancers in CF patients depends on tissue type/site, which implies different sensitivities of tissues to CFTR-mediated tumorigenesis. Similar to cystic fibrosis, other CFTR-associated pathologies also display different occurrences and/or severities in different tissues. The localization and severity of the impairment also varies across specific CFTR mutations, which reflects their ability to confer a mild or strong disease phenotype. These observations support the expectation that the impairment of CFTR poses different risks of cancer development in different tissues.

Furthermore, epidemiological data indicate that, at least for some cancers, CFTR impairment can contribute to the carcinogenic risk of some environmental or lifestyle exposures. This was shown by significantly younger age at diagnosis of pancreatic adenocarcinomas in ever smokers carrying one pathogenic allele of CFTR compared to ever smokers not carrying CFTR mutations [39]. Considering the fact that tobacco smoke is an inhibitor of CFTR [40–42], the observation of the increased risk in smokers with CFTR mutations is intriguing and implies that impaired CFTR can further increase the carcinogenic risk of environmental exposures that inhibit CFTR activity.

4.3. CFTR as a tumor suppressor

The tumor suppressor status of the CFTR gene (reviewed in [43]) has been suggested based on the results of epidemiological [36–39, 44, 45], clinical [10, 46, 47], and experimental studies [10, 47–50].

For instance, using a murine model with an intestinal-specific Cftr gene knock-out, Than et al. [10] demonstrated that Cftr-deficient mice have a significantly increased risk of developing tumors in the colon and small intestines. Furthermore, the loss of Cftr activity was shown to enhance intestinal tumorigenesis in ApcMin mice that carry a mutated tumor-suppressor gene Apc. These findings demonstrate that impairment of CFTR leads to the tumorigenesis in murine small intestines.

Interestingly also, our search of the Catalogue of Somatic Mutations in Cancer (COSMIC v92, released 27-AUG-20) [51] identified 241 confirmed somatic missense and nonsense mutations of the CFTR gene predicted to be pathogenic (by FATHMM-MKL model) in 290 unique tumor samples (Supplemental file 5). Of them, 13 are annotated in the CFTR2 database [52] as either causal for CF or having variable clinical consequences. These somatic mutations were found in a range of cancers with the highest prevalence in cancers of the skin (8.17%), thyroid (3.6%), urinary tract (2.27%), endometrium (1.94%) and large intestines (1.72%). Their predicted damaging effect and relatively frequent occurrence in at least some cancers further supports the tumor-suppressor status of the CFTR gene.

The tumorigenicity of impaired Cftr in animal models supports the relevance of the Cr(VI)-mediated inactivation of Cftr for the development of small bowel tumors in mice exposed to Cr(VI) in drinking water. Furthermore, enhancement of the tumorigenicity of the Apc mutations by Cftr inactivation implies that carriers of APC mutations may be more susceptible to the tumorigenicity induced by events that inactivate CFTR, including Cr(VI) exposure. These findings open an avenue for identification of vulnerable groups, such as APC mutations carriers, that can be more sensitive to the Cr(VI)-mediated carcinogenicity.

This reasoning likely extends to humans, because (i) CFTR reportedly acts as a tumor-suppressor in the human colon [10], and (ii) germline mutations in the APC gene or its regulatory sequences in humans are known to cause APC-Associated Polyposis Conditions, including familial adenomatous polyposis (FAP), which are associated with a high risk of colon cancer and an increased risk of cancers at other sites, including the duodenum, thyroid gland and stomach [53].

Furthermore, the human relevance of Cr(VI)-mediated inactivation of Cftr inferred from murine duodena is supported by the similarity of expression profiles between murine duodena exposed to Cr(VI) and duodenal polyps of FAP patients (Supplemental Section 1, Supplemental figure S1 and Supplemental Table S2). This result indicates that duodena of mice exposed to Cr(VI) and human duodenal pre-cancerous adenomatous polyps [54] share common canonical pathways and upstream regulators, including the inactivated status of CFTR inferred by the IPA analysis.

The human relevance is also supported by the similarity of expression profiles of human HepG2 cells exposed to Cr(VI) in vitro and expression profiles of lungs and colons from the Cftr-knockout mice (Supplemental Section 2 and Supplemental figures S4–S6). The results indicate that Cr(VI) exposure of human cells in vitro induces gene expression changes consistent with the Cftr inactivation in animal models.

4.4. Acquired impairment of CFTR in the development of cancer and other diseases

Inhibition of CFTR activity by chemicals was previously shown to induce adverse outcomes, including cancers, with a known or suspected association with CFTR impairment. For instance, administration of CFTRinh-172, a low-molecular-weight inhibitor of CFTR-mediated Cl^—conductance, increased the proliferation of prostate epithelial cells in rats [55]. Similarly, CFTRinh-172 stimulated DNA replication and the migratory potential of endometrial adenocarcinoma cells in vitro [56]. Moreover, CFTRinh-172 was shown to inhibit cisplatin-induced apoptosis in CFTR-expressing renal proximal cells in vitro and cisplatin-induced kidney damage in rats, but it did not interfere with apoptosis in fibroblasts that do not express CFTR [57]. Lastly, treatment with CFTRinh-172 induced epithelial-to-mesenchymal transition (EMT) in human lung cancer cells in vitro [47]. Taken together, chemical inhibition of CFTR can contribute to tumorigenesis through increased cell proliferation, inhibited apoptosis and an EMT-driven invasive phenotype, which are known hallmarks of cancer [24]. This finding is consistent with the findings of our analysis and supports the role of Cr(VI)-mediated inactivation of CFTR in gene expression changes relevant for hallmarks of cancer.

Tobacco smoke has been shown to inhibit CFTR in human airway cells [40–42] and in human intestinal cells in vitro [40], but also in the nasal airway in human subjects [40]. Acquired dysfunction of CFTR among long-term smokers is considered to play a role in the development of chronic obstructive pulmonary disease (COPD) [58]; however, the role of CFTR dysfunction in the pathogenesis of tobacco-related cancers has not yet been suggested. In light of the clinical, epidemiological and experimental studies supporting a tumor-suppressor role of CFTR and tumor-promoting role of CFTR inactivation, it is plausible that this mechanism is also involved in tobacco-related carcinogenesis, in addition to other mechanisms, including genotoxicity induced by polycyclic aromatic hydrocarbons (PAHs), 4-(methylnitrosamino)-1-(3-pyridyl)-1-butanone (NNK) and possibly other carcinogens [59].

Acquired inhibition of Cftr was also suggested to play a role in the chloroprene-induced lung carcinogenesis in mice. Specifically, Cftr was found to be a hub gene for the largest of the 12 gene modules, which were identified by the weighted gene co-expression network analysis of differentially expressed genes in the lungs of mice following inhalational exposures to chloroprene [60].

Glibenclamide (glyburide) is a hypoglycemic agent of the sulfonylurea class, which was also shown to inhibit CFTR [61]. Results of several prospective and retrospective epidemiological studies suggest that the use of this agent in the management of diabetes mellitus type 2 is associated with a higher risk of cancer than is the use of other sulfonylurea hypoglycemic agents [62], which also supports the potential role of CFTR inhibition in cancer development.

Likewise, several metals or metalloids represent toxicants that can inactivate CFTR. These include cadmium and arsenic, which inhibit CFTR in rodents and humans, as well as mercury (Hg^II) and zinc, which inhibit the activity of shark CFTR but not of its human homologue [63]. The fact that these metals/metalloids also inhibit CFTR provides additional support for the Cr(VI)-mediated inactivation of CFTR reported in this article.

Exposure to cadmium compounds at levels not affecting cell viability decreases the abundance of the CFTR protein and impairs chloride transport in primary and continuous human epithelial airway cells in vitro, and decreases CFTR protein levels in the lungs of nasally-exposed mice [64].

Intriguingly, exposure to cadmium compounds reportedly also increased the risk of lung cancer, and with lower confidence, also the risk of prostate cancer in several occupational studies [65]. Moreover, various cadmium compounds induced tumors in several animal species, including lung cancer in rats exposed by inhalation, leukemia and benign testicular tumors in rats exposed by oral administration, and various tumors in rodents that received cadmium compounds by subcutaneous injections (reviewed in [65, 66]). Considering the evidence supporting the carcinogenicity of Cd²⁺ and its ability to decrease the level of CFTR protein, we infer that the inactivation of CFTR is likely involved in the carcinogenic mechanism of cadmium and its salts. Since tobacco smoke is known to be a source of cadmium exposure [67], it is tempting to speculate that cadmium is, at least in part, involved in tobacco smoke carcinogenicity. Indeed, smoking is associated with increased concentrations of cadmium in urine, blood and lungs (reviewed in [68]), and the risk of developing some lung diseases in long-term smokers, such as COPD, appears to be associated with local accumulation of cadmium in the lungs. The proportion of lung cancer deaths attributable to cadmium from tobacco smoke has been assessed to be 0.2%-1.6% with upper-bounds: CI95 1.6%-8.8% [69]. Nevertheless, the specific contribution of cadmium to the development of lung diseases, including lung cancer, in the presence of other components of tobacco smoke, remains to be elucidated [70]. It should be also noted that other toxicants in tobacco smoke, such as acrolein, can also modulate CFTR activity [71].

Arsenic is another metal/metalloid found to inhibit CFTR activity. Arsenic compounds were found to decrease total and plasma membrane-bound CFTR protein levels in human airway epithelial cells in vitro [72], as well as CFTR-mediated Cl- secretion in killfish [73]. Importantly, these effects were observed at low, environmentally relevant exposures. For instance, the levels of arsenic that decreased CFTR protein levels in human airway epithelial cells were well within the range of values measured in the blood of humans exposed to arsenic in drinking water and below the levels needed to induce cytotoxicity [72]. Arsenic-induced downregulation of CFTR, which resulted in impaired mucociliary clearance, was implicated in a reduced response to respiratory infections that was reported by epidemiological [74, 75] and animal studies [76, 77] of environmentally relevant exposures to arsenic. Since inorganic arsenic compounds demonstrated carcinogenicity in epidemiological and animal studies via inhalation and oral exposure (reviewed in [2]), we posit that inhibition of the tumor-suppressor CFTR may be also involved in its carcinogenic mechanism of action.

Taken together, modulation of CFTR by some chemicals is associated with tumorigenesis, which supports the relevance of CFTR inhibition for development of some cancers. Inhibition of CFTR by other carcinogenic metals and metalloids further supports our findings for Cr(VI)-mediated inactivation of CFTR and supports its mechanistic relevance for the carcinogenicity of hexavalent chromium.

4.5. Possible mechanism of CFTR inactivation-mediated tumorigenesis

Several molecular events downstream of CFTR have been implicated in tumorigenesis (reviewed in [78]); however, the mechanisms of its tumor-suppressive effect are not yet fully understood.

For instance, intestinal crypts in CFTR-deficient mice displayed increased Wnt/β-catenin signaling and increased proliferation of Lrg5+ intestinal stem cells (ISC) relative to the wild type controls. As a result, suppression of ISC proliferation by CFTR appears to be protective and its loss can contribute to the increased risk for intestinal tumors [79]. Consistent with this explanation of increased risk, our analysis revealed a higher abundance of stem cells relative to the differentiated villous cells in the duodenal tissues of Cr(VI)-exposed mice compared to control animals (Supplemental table S3). This finding is supported by a strong inverse correlation between expression profiles of the duodena of Cr(VI)-exposed mice and expression profiles of differentiated enterocytes vs. Lgr5-expressing ISC (Supplemental table S3). Intriguingly, this correlation supporting the abundance of ISC has been also found for the early (8-day) timepoint in a group exposed to the lowest concentration of SDD in drinking water (Supplemental figure S7). Additional support for a higher abundance of stem cells in the duodena of Cr(VI)-exposed mice comes from a histopathologic study, which reported crypt cell hyperplasia in mice exposed to SDD for 8 days (520 mg/L) or 91 days (≥170 mg/L) [21]. The role of Wnt/β-catenin signaling is further supported by results of the Upstream Regulator Analysis (URA) for expression data from the duodena of mice exposed to SDD. The URA analysis for the 8-day exposure identified the following activated upstream regulators (Z>2): Wnt1 for 170 mg/L and 520 mg/L (Supplemental figure S8), CTNNB1 for 520 mg/L, and TCF4 for 0.3 mg/L, 170 mg/L and 520 mg/L (Supplemental file 1).

Ion channels and transporters, including CFTR, have been proposed to regulate miRNAs in response to changes in the extracellular environment [80]. For this reason, it is tempting to consider the tumor-suppressor role of CFTR as a possible regulator of miRNAs that promote or inhibit tumorigenesis.

Indeed, numerous miRNAs are reportedly regulated by CFTR (reviewed by [80]), some of which have been previously implicated in cancers. Specifically, in endometrial adenocarcinoma, inhibition of CFTR by CFTRinh-172 resulted in decreased expression of miR-125b and enhanced proliferation and migration of the cells. This effect could be reversed by transfection of miR-125b, which in turn results in a decreased expression of VEGF-A and MMP11 [56]. However, our data show significant down-regulation of Vegfa in the duodena of mice exposed to Cr(VI) at several exposure levels and no evidence for differential expression of Mmp11 (Supplemental Table S4). These findings argue against the role of miR-125b in gene expression changes observed in our data.

Similarly, in prostate cancers, the tumor-suppressor role of CFTR was linked to the up-regulation of hsa-miR-193b, which inhibits urokinase plasminogen activator (uPA; PLAU) involved in the progression of many cancers [81]. Our results show the up-regulation of PLAU gene in the duodena of mice exposed to the highest exposure levels for 8 days (520 mg/L) and 91 days (≥170 mg/L) (Supplemental table S4). Furthermore, the CFTR knock-down in breast cancer cells reportedly induced epithelial-to-mesenchymal transition (EMT) through up-regulated PLAU gene [82], which is consistent with the results of our GSEA suggesting EMT in the duodena of Cr(VI)-exposed mice (Figure 6). Taken together, the expression profiles of the duodena of mice exposed to Cr(VI) in drinking water suggest that the inferred inactivation of CFTR, EMT and observed up-regulation of uPA may be linked together through up-regulation of miR-193b, but this model needs further verification.

Unrelated to its ion transportation activity, CFTR has been shown to physically interact with various proteins, and some of these interactions may be important to its tumor-suppressive activity. These proteins include ZO-1, E-cadherin, MRP2 and afadin (reviewed in [83]). Specifically, the interaction between CFTR and afadin was found to be critical for cell-cell junctions and suppression of the CFTR gene expression was shown to increase migration and invasion of cancer cells via activation of the ERK signaling pathway [84].

Inactivation of CFTR has also been mechanistically linked to the inhibition of apoptosis induced by oxidative stress. Specifically, human airway epithelial cells and HeLa cells expressing wild-type CFTR displayed higher sensitivity to treatment with hydrogen peroxide than cells expressing mutated CFTR. This difference in sensitivities was attributed to the fast depletion of glutathione in CFTR-proficient cells and subsequent activation of pro-apoptotic BAX protein [85]. As a result, accumulation of glutathione in cells with inactivated CFTR due to its impaired transport may prevent apoptotic death in cells exposed to oxidative stress, and in this way increase the chance of malignant transformation of surviving cells.

4.6. Possible mechanisms of Cr(VI)-triggered inactivation of CFTR

The mechanism by which Cr(VI) triggers inactivation of CFTR remains to be elucidated. Although several plausible mechanisms for this process can be suggested based on previous studies, their limitations and remaining uncertainties warrant further investigation.

The unfolded protein response (UPR), activated by chemically-induced endoplasmic reticulum (ER) stress, is one of the plausible processes that could explain Cr(VI)-triggered inactivation of CFTR. The UPR was previously shown to decrease CFTR on the mRNA and protein level through the ER-associated protein degradation pathway (ERAD) [86]. However, while our results indicate enrichment of the UPR gene set in the duodena of Cr(VI)-exposed mice (Figures 6 and 7), we have not demonstrated down-regulation of CFTR on the mRNA level, which makes the mechanistic role of the UPR less supported in our results.

Likewise, the role of HIF-1a in the transcriptional repression of CFTR [87] and CFTR promoter hypermethylation [45] have been previously suggested; however, these mechanisms also imply down-regulation of CFTR mRNA, which is not consistent with the results of our differential expression analysis for the duodena of Cr(VI)-exposed versus control mice (Supplemental files 3 and 4).

Post-transcriptional down-regulation of CFTR by miRNAs has been previously demonstrated in the context of tobacco smoke and cadmium exposures, which suggests the potential relevance of miRNAs for the down-regulation of CFTR in the duodena of Cr(VI)-exposed mice. MiR-101 and miR-144, which are up-regulated by tobacco smoke and cadmium exposures in bronchial epithelial cells, were shown to target the CFTR 3’UTR and repress its expression on the protein level. Moreover, up-regulation of miR-101 and down-regulation of the CFTR protein were detected in mice exposed to tobacco smoke. And likewise, miRNA-101 was found upregulated in humans with severe chronic obstructive pulmonary disease (COPD), a condition also linked to CFTR deficiency [88]. Taken together, miR-101 and miR-144 can be important direct regulators of CFTR at least in bronchial cells. Interestingly, our analysis of genes down-regulated in the duodena of mice exposed for 8 days to SDD at 520 mg/L identified an over-representation of an mRNA sequence motif that is complementary to the seed sequence of mmu-miR-101a, mmu-miR-101b and mmu-miR-144 (Supplemental section 2 and Supplemental figure S9). This finding supports the possible role of miR-101 and miR-144 in the Cr(VI)-mediated inactivation of CFTR in murine duodena.

Activation of the TGF-β signaling has been attributed to the downregulation of CFTR expression and a significant reduction in cAMP-dependent current in nasal polyp cells from non-CF patients [89]. Intriguingly, this condition with hyperplastic inflammatory lesions, which is prevalent in CF-patients, has been associated with increased risk of nasopharyngeal cancers and cancers in the nasal cavity and paranasal sinuses [90]. Expression data from the duodena of mice exposed to Cr(VI) for 8 days indicate an activation of upstream regulators TGFB1, TGFB2, TGFB3, TGFBR1 and SMAD3 known to be involved in TGF-β signaling (Supplemental file 1), which is indicative of a possible role of this mechanism in Cr(VI)-mediated inactivation of CFTR in small intestines. Importantly, TGF-β signaling plays a major role in intestinal tumorigenesis, where it acts initially as a tumor suppressor and later as a tumor promoter [91]; therefore, its activation inferred in our study can be relevant to both CFTR inactivation and intestinal cancer development.

Oxidative stress has been also shown to suppress CFTR activity and expression on the mRNA and protein levels. This was implicated as a mechanism serving to preserve the cellular antioxidant glutathione through decreased CFTR-mediated transport [92]. We noted that a collection of 26 genes known to be responsive to oxidative stress was significantly enriched in mice exposed to SDD for 91 days. In contrast, however, the data for the 8-day exposure indicate that this collection was enriched in control mice (Supplemental section 3). Likewise, transcription factor Nfe2l2 (NRF2) that regulates the cellular defense against oxidative stress [93] was found activated in duodena of mice exposed to SDD at ≥ 60 mg/L for 91 days, but inactivated in mice exposed at < 14 mg/L for 8 days (Supplemental files 1 and 2).

This result suggests that there was a lower amount of ROS in the duodena of mice exposed for 8 days, but a higher amount of ROS in the duodena of animals exposed for 90 days (Supplemental table S5). This can be possibly explained by the accumulation of the antioxidant glutathione in cells exposed to Cr(VI) for a shorter period of time due to the inactivation of CFTR-mediated transport of glutathione. On the other end, glutathione may have been depleted in cells after prolonged exposure to Cr(VI) even if CFTR was inactivated, which is consistent with the observed up-regulation of the Gclc gene that encodes an enzyme catalyzing a rate-limiting step in glutathione synthesis, as well as other genes involved in glutathione metabolism (Supplemental section 3 and Supplemental figure S10). Consequently, ROS are not likely triggering the inhibition of CFTR in the duodena of Cr(VI)-exposed mice. Instead, inhibition of CFTR is triggered by another mechanism that appears to have decreased the amount of ROS in cells exposed for shorter time periods through the decreased transport of glutathione.

5. Conclusions

In conclusion, our systems biology analysis suggests there is an inactivation of CFTR in the duodena of mice exposed to Cr(VI) in drinking water. This was likely due to post-transcriptional regulation of the CFTR protein, since we did not detect down-regulation of CFTR mRNA expression, but rather its up-regulation in the two highest exposure groups. Discordant expression of CFTR on mRNA and protein levels have been previously reported by other investigators [94] and the transcriptional up-regulation of CFTR coupled with its inactivation has been previously suggested as a compensatory mechanism for the loss of the functional CFTR gene product [95]. In our study, inferred inactivation of CFTR was accompanied by molecular signatures consistent with the hallmarks of cancer, which supports the previously suggested tumor-suppressive role of CFTR and implies the potential role of its inactivation in Cr(VI)-mediated carcinogenesis in the murine small intestines. Our results support the human relevance of Cr(VI)-mediated inactivation of CFTR and indicate that individuals with impaired CFTR and/or APC activity due to genetic conditions or other environmental exposures may be more vulnerable to the adverse effects of Cr(VI) or other exposures known to inactivate CFTR.

This study is a subject to potential limitations that are inherent to transcriptomics studies and their biological interpretations. Above all, our inference of inactivated status of CFTR in duodena of mice exposed to Cr(VI) is contingent on the IPA knowledgebase built from published reports and methodology for prediction of upstream regulators from differential expression data [16]. The strength of our predictions is supported by internal consistency across exposure levels and external consistency with other annotated expression datasets or findings reported by other investigators (discussed in sections 4.3–4.6 of this report). Nevertheless, functional studies are needed to definitively conclude the role of CFTR in Cr(VI)-induced intestinal carcinogenesis.

Supplementary Material

Supplemental Material

Supplemental material: Supplemental sections S1-S4; Supplemental figures S1-S10; Supplemental tables S1-S5.

Sup 2

Supplemental file 2: Upstream regulators identified for 91-day exposure data as significant (|Z|≥2 or p<0.05) for at least one Cr(VI) exposure level

Sup 1

Supplemental file 1: Upstream regulators identified for 8-day exposure data as significant (|Z|≥2 or p<0.05) for at least one Cr(VI) exposure level

Sup 5

Supplemental file 5: Somatic mutations of CFTR gene (deleterious by FATHMM-MK2) identified in tumors and deposited in COSMIC database

Sup 4

Supplemental file 4: List of differentially expressed genes for 91-day exposure data (|log₂(Fold Change)| ≥0.585 and p<0.05)

Sup 3

Supplemental file 3: List of differentially expressed genes for 8-day exposure data (|log₂(Fold Change)| ≥0.585 and p<0.05)

Figure 4. Differentially expressed genes supporting inactivated status of Cftr in the duodena of mice exposed to hexavalent chromium for: (A) 8 days at 60 ppm; (B) 91 days 170 ppm.

Color coding: Green shapes- down-regulated; red shapes- up-regulated; blue shape- inferred inactivated; blue edges- activating relationship consistent with prediction; brown edges- inactivating relationship consistent with prediction; yellow edge- relationship inconsistent with prediction. More details on figure legend are available at: https://qiagen.secure.force.com/KnowledgeBase/KnowledgeIPAPage?id=kA41i000000L5rTCAS