A knockdown with smoke model reveals FHIT as a repressor of Heme oxygenase 1

Gene expression changes in response to FHIT knockdown

The mechanism by which FHIT functions as a tumor suppressor has not been fully explained by experiments in which the FHIT gene has been reintroduced to FHIT negative cells. To investigate the hypothesis that Fhit acts as a regulator of gene expression in bronchial epithelial cells, we performed microarray analysis of cells after knocking down FHIT expression with siRNA. HBEC3-TK is an immortalized bronchial epithelial cell line that has been used to study the consequences of gene inactivation in lung cancer.^13,35 These cells display normal epithelial morphology, are not transformed, have an intact p53 pathway, and have been used by others to study the pathogenesis of lung cancer.^35,36,39 We first determined that treatment of HBEC3-TK cells with 2 unique siRNAs robustly silences FHIT expression at both the mRNA and protein level (Fig. 1 A, B). Total RNA from treated cells was analyzed by microarray to assess transcriptome changes due to FHIT loss. After restricting the data set to expression changes ≥1.5-fold with P ≤ 0.05 followed by multiple testing correction with a false discovery rate, we observed changes in only 3 genes, suggesting that FHIT loss does not have a large impact on gene expression in this model (Table 1).

The FHIT gene, which encompasses the common fragile site at 3p14.2, is a known target of environmental carcinogens including cigarette smoke. Loss of heterozygosity at 3p14 occurs more commonly in smokers than in nonsmokers.^37,40 Preneoplastic lesions of the lung and cervix show a greater frequency of FHIT loss in smokers than in non-smokers.^17,38,41 Rats exposed to cigarette smoke for short time intervals show a time dependent decrease in FHIT expression at both mRNA and protein levels in bronchial epithelial cells.^15,42. Furthermore, peripheral blood samples show that active smokers express fragility at the FHIT locus.^13,43-45 Relying on these data, we hypothesized that FHIT plays a role in mediating the cellular response to cigarette smoke. To test this, siRNA-treated HBEC3-TK cells were exposed to 1% CSE for 4 hours prior to RNA extraction and microarray analysis. We first aimed to establish that our CSE treatment protocol impacted gene expression in an expected manner. To do this, we compared gene expression in cells treated with siControl plus 1% CSE to gene expression in cells treated with siControl only. After restricting the dataset to expression changes ≥1.5-fold with P ≤ 0.05 followed by multiple testing correction with a false discovery rate, we observed changes in 378 genes, indicating that the CSE treatment had a strong effect on gene expression. These differentially expressed genes were analyzed with Ingenuity Pathway Analysis (IPA, Ingenuity Systems, www.ingenuity.com) to determine what molecular pathways were altered by CSE treatment. The significant canonical pathways associated with the differentially expressed genes are presented (Fig. 2). The affected pathways included NRF2-mediated oxidative stress response, glutathione biosynthesis, aryl hydrocarbon receptor signaling, and xenobiotic metabolism signaling, among others. This result is consistent with previous reports that these pathways are affected by exposing cells in culture to cigarette smoke.^46-50 To assess the combined effect of FHIT knockdown and CSE treatment, we applied the same analysis criteria used for cells that were not treated with CSE to analyze the gene expression consequences of CSE treatment in control and FHIT knockdown cells (Tables 2, 3). We determined that a set of genes involved in the cytoprotective oxidative stress response were either only activated in FHIT knockdown cells, or were activated in both FHIT knockdown and control cells, but gene expression was enhanced by FHIT loss. Notably, a subset of the dysregulated genes have annotated roles in the oxidative stress response, suggesting that this pathway may be superinduced in FHIT deficient cells (Fig. 3A). We validated the gene expression changes of a subset of these genes by quantitative RT-PCR (qRT-PCR) (Fig. 3B). The data indicate that, in the absence of cigarette smoke, FHIT loss does not significantly alter gene expression. However, a set of genes involved in the cytoprotective oxidative stress response induced by CSE is superinduced by CSE when FHIT is lost. \raster="rgFigKCCY_A_946858_F0002_B"

Hmox1 expression is enhanced by FHIT knockdown in CSE stressed cells

One of the CSE-induced genes that is superinduced in FHIT knockdown cells is HMOX1. HMOX1 is a member of a battery of stress response genes whose expression is controlled, in part, by promoter stress response elements.^39,51-53 We validated the upregulation of Hmox1 observed in the microarray experiment by qRT-PCR (Fig. 4A). In the absence of CSE, HMOX1 mRNA is present in amounts approaching the threshold of qRT-PCR detection and the protein is undetectable by western blot (Fig. 4A, B). CSE treatment induces HMOX1 mRNA and protein expression to readily detectable levels (Fig. 4A, B). CSE exposure combined with FHIT knockdown markedly increases the level of Hmox1 protein (Fig. 4B). There are at least 2 heme oxygenase isoforms in mammals, HMOX1 and HMOX2. While HMOX2 is expressed constitutively and confers protection to basal levels of oxidation, HMOX1 is strictly induced in response to stress.^40,54,55 Thus, the expression pattern we detected in HBEC3-TK cells is normal. To ascertain that the increase in HMOX1 is not due to an off-target effect of one of the siRNAs, we transfected cells in parallel with a second control and second FHIT-targeting siRNA. In doing so, we confirmed that the increase in HMOX1 is due to FHIT loss (Fig. 4B). Importantly, the induction of HMOX1 is not affected by the transfection procedure, as indicated by the equivalent results in cells treated with control siRNA and those not treated with siRNA (Fig. 4B). These data indicate that in the presence of CSE, FHIT knockdown superinduces HMOX1 expression.

To confirm that the FHIT knockdown-mediated enhancement of HMOX1 induction is not unique to HBEC3-TK cells, we replicated the experiment in 2 unrelated cell lines. We first used an immortalized tracheal airway epitheial cell line, TERT-T106.^20,41 In these cells, HMOX1 is not detected in the absence of CSE (Fig. 5A). As observed in HBEC3-TK cells, Hmox1 protein expression is induced by CSE and superinduced when FHIT is knocked down (5A). We then tested if the effect was present in immortalized human foreskin keratinocytes (TERT-HFK)^{20,29,31,42,56} and obtained an equivalent result (Fig. 5B). These findings indicate that the role of FHIT in limiting Hmox1 expression is not dependent on a specific genetic background or unique cell type.

Hmox1 induction occurs earlier and is more sustained in FHIT knockdown cells

To determine whether FHIT knockdown modulates the kinetics of protein accumulation, we exposed siRNA-treated HBEC3-TK cells to CSE for different time intervals and measured the level of Hmox1 by western blot. Hmox1 protein is detected earlier in FHIT knockdown cells and, consistent with previous results, is superinduced at the protein level (Fig. 6A). Typically, Hmox1 levels increase in response to an acute stress. Once the stress is removed, Hmox1 levels return to baseline. To test if FHIT knockdown affects the rate at which Hmox1 levels decay, we exposed siRNA-treated HBEC3-TK cells to 1% CSE for 4 hours and allowed the cells to recover in fresh media for various time intervals. Loss of FHIT increases expression of Hmox1 at each time point, but the rate of protein decay was not altered (Fig. 6B).\raster="rgFigKCCY_A_946858_F0006_B"

Mechanism of Hmox1 superinduction in response to FHIT loss

Previous studies have demonstrated that FHIT expression affects calcium signaling, attenuates NF-κB signaling, and promotes phosphorylation and activation of AKT.^43-45,57 Notably, modulation of these pathways has also been demonstrated to affect HMOX1 expression.^51-53,58-60 For this reason, we asked whether the superinduction of HMOX1 seen after FHIT knockdown might be mediated through one or more of these pathways. However, we found that these pathways do not contribute to upregulation of HMOX1 in our model (Figs. 7–10).

Upon excluding these pathways, we analyzed the regulation of the well-characterized Bach1-Nrf2 oxidative stress response axis. Nrf2 and Bach1 are canonical regulators of HMOX1 transcription that bind DNA cis-elements to activate or repress target gene transcription, respectively.^54,55,61 We found that Nrf2 mRNA and protein levels are unchanged in response to FHIT knockdown in both CSE treated and untreated cells (Fig. 11A and B). Additionally, while knockdown of NFE2L2, the gene that encodes Nrf2, blunts the induction of HMOX1 in response to stress, simultaneous knockdown of FHIT and NFE2L2 still yields a superinduction of HMOX1 (Fig. 11C). These data indicate that the ability of FHIT to dampen HMOX1 expression does not require Nrf2.

As with Nrf2, BACH1 mRNA levels were not dependent on FHIT expression or the application of CSE (Fig. 12A). In addition, Bach1 protein levels were unchanged in response to FHIT knockdown in the absence of stress (Fig. 12B). However, two observations about Bach1 protein were made in CSE-treated cells. First, CSE treatment produced a slowly migrating species of Bach1 (Fig. 12B). This species was confirmed to be Bach1 by siRNA knockdown in an independent experiment (Fig. 12C). Second, the total level of Bach1 protein was decreased in FHIT knockdown cells. Quantification of the results of multiple experiments indicates that FHIT knockdown combined with CSE stress results in a ∼40% reduction in Bach1 protein (Fig. 12D). To test whether a reduction in Bach1 is sufficient to induce Hmox1 protein, we titrated BACH1 siRNA into HBEC3 cells and saw a clear and inverse relationship between Bach1 and Hmox1 levels (Fig. 13A and B). Interestingly, we also note that the levels of Fhit and Bach1 protein are inversely correlated. Taken together, the data indicate that the superinduction of HMOX1 in FHIT knockdown cells may be mediated through decreased stability or translation of Bach1 protein in cells that have both lost FHIT and are stressed with cigarette smoke. \raster="rgFigKCCY_A_946858_F0013_B"

Cell culture

HBEC3-TK bronchial epithelial cells were a kind gift from Dr. John Minna (UT Southwestern). TERT-HFK human foreskin keratinocytes and TERT-T106 tracheal airway epithelial cells were generous gifts of Dr. Aloysius Klingelhutz (University of Iowa). Cells were cultured on bovine collagen coated plastic dishes (Sigma, Cat #C4243). Cells were maintained in Keratinocyte Serum Free Media (KSFM) supplemented with BPE (12.5 g/ml) and EGF (5.0 ng/ml) (Gibco Cat #17005-075). Cells were grown in humidified 37°C, 5% CO₂ incubators.

siRNA transfection

Cells were seeded at densities of 400,000 cells/plate (10 cm dishes), 125,000 cells per well (6 well dishes), or 30,000 cells/well (12 well dishes). After overnight incubation, cells were transfected with 15 nM siRNA and Lipofectamine RNAiMax (Invitrogen, Carlsbad CA) according to the manufacturer's protocol. siRNAs were obtained from commercial vendors. FHIT siRNA-1 (Cat #4390843), FHIT siRNA-2 (Cat #AM16706), Control siRNA-1 (Cat #4390843), and Control siRNA-2 (Cat #4390846) were purchased from Ambion (Life Technologies). BACH1 siRNA (Cat # HSC.RNAI.N001186.12.1) and NFE2L2 siRNA (Cat #HSC.RNAI.N001145412.12) were purchased from IDT (Coralville, IA). After 4.5 hours of siRNA incubation, transfection media were replaced with fresh KSFM. Cells were incubated for 48 hours.

Cigarette smoke extract preparation

Research grade cigarettes were purchased from the University of Kentucky. After removing the filter, the cigarette was combusted and smoke extracted into 10 ml of KSFM, as described previously.^15,34 This preparation was termed “100% CSE.” The extract was filtered though a 0.22 μm pore membrane (Millipore Cat #SCGP00525). CSE was diluted to 1% in KSFM and added to cells in culture for the indicated times. CSE treated cells were incubated separately from non-CSE treated cells.

RNA extraction, cDNA synthesis, and RT-PCR analysis

Total RNA was extracted from cells using the RNeasy Mini kit (Qiagen, Cat #74104) after homogenizing cells with Qiashredder spin columns (Qiagen, Cat #79654). RNA was DNase treated with the Turbo DNA-free kit (Ambion, Cat #AM1907). 500 ng total RNA was reverse transcribed with the iScript cDNA synthesis kit (BioRad, Cat #170–8890) or 1.5 μg total RNA was reverse transcribed with the High Capacity cDNA synthesis kit (Applied Biosystems, Cat #4368814). RT-PCR was performed with iQ SYBR Green Supermix (BioRad, Cat #170–880) and designed primers or with TaqMan Gene Expression Master Mix (Applied Biosystems, Cat #4369016) and PrimeTime nuclease primer/probe assays (IDT). Nuclease assays used in this study are: NFE2L2, IDT Catalog Hs.PT.49a.36704, GAPDH, IDT Catalog Hs.PT.47.1164609, FHIT, IDT Catalog Hs.PT.47.3317107, and HMOX1, IDT Catalog Hs.PT.53a.22439609. Primers used in SYBR Green assays are listed in 5’-3’ direction as follows: AKR1B10 – GCCACAGGGATTCAAGTCT and CTTTCACCAGCCCCTCATC; AKR1C1 – GCCGTGGAGAAGTGTAAAGATG and CTGGTTGAAGTAAGGATGACATTC; AKR1C2 – TAAAGCCAGGTGAGGAAGTG and CTGTGGTTGAAGTTGGACAC; AKR1C3 - AACAAGCCAGGACTCAAGTAC and CAGAGCACTATAGGCAACCAGl NQO1 – CCGCAGACCTTGTGATATTCC and ACTCGCTCAAACCAGCCTTTCAGA; GPX2 – GCTTCCCTTGCAACCAATTTG and TTCTGCCCATTCACCTCAC; SERPINB2 – CAGTAGACTTCCTAGAATGTGCAG and AAGTAGACAGCATTCACCAGG; SRXN1 – CAAGGTGCAGAGCCTCG and CTTTGATCCAGAGGACATCGA; BACH1 – TCTTGAATCAGAAATTGAGAAGCTG and TGGCAAAGTCCAGTTAGGTTCTGCT; GAPDH – ACATCGCTCAGACACCATG and TGTAGTTGAGGTCAATG- AAGGG.

Immunoblotting

Cells were suspended in ice-cold RIPA buffer (50 mM Tris-Cl (pH 8), 1% NP-40, 0.5% Na deoxycholate, 0.1% SDS, 150 mM NaCl, 5 mM EDTA) supplemented with protease inhibitors (Roche Cat #11836170001) and phosphatase inhibitors (Roche Cat #04906845001). Cells were lysed on ice for 20 minutes before clarifying (4°C, 13 K rpm, 10 minutes). Protein content was quantified with the BCA Protein Assay (Pierce Cat #23225). Equivalent amounts of protein were reduced with 50 mM DTT, denatured with LDS, heated, and loaded on NuPage 4–12% Bis-Tris gels (Invitrogen) and electrophoresed in 1x MOPS buffer. Protein gels were transferred to nitrocellulose membranes, which were blocked with 5% milk. Primary antibodies were incubated with membranes overnight at 4°C. Secondary antibody incubations were performed at room temperature for 30 minutes. Blots were developed with ECL - SuperSignal West Pico (Pierce), Supersignal West Femto (Pierce), or Quantum (Advansta) reagents. Images were collected with GE ImageQuant LAS4000 and quantification performed with ImageQuant TL7.0 software (GE). Fhit antibody was obtained from Millipore (Cat #07–172). Bach1 (Cat #sc-14700), Nrf2 (Cat #sc-365949), and HRP conjugated donkey-anti-goat (Cat #sc-2020) antibodies were purchased from Santa Cruz Biotechnology. Hmox1 (Cat #4643), β-tubulin (Cat #2128), Gapdh (Cat #5174), AKT (Cat #9272), phosphoAKT (Cat #4060), HRP-anti-mouse (Cat #7076), and HRP-anti-rabbit (Cat #7074) antibodies were purchased from Cell Signaling Technologies. β-actin (Cat #A2066) antibody was obtained from Sigma.

Treatment with inhibitors

Stock solutions of inhibitors were prepared in anyhydrous DMSO, aliquoted, and stored at −20° C. Compounds were diluted to working concentrations in KSFM immediately before use. The PI3K inhibitor LY294002 was purchased from Cell Signaling Technologies (Cat #9901). The IKK2 inhibitor SC-514 was purchased from Cayman Chemical (Cat #10010267). The cell permeable Ca²⁺ chelator BAPTA-AM was purchased from Invitrogen (Cat #B-6769).

DNA microarray

Microarray hybridizations were performed at the University of Iowa DNA Facility. Briefly, 50 ng total RNA was converted to single primer isothermal amplification (SPIA) cDNA using the WT-Ovation Pico RNA Amplification System, v1 (NuGEN Technologies, San Carlos, CA, Cat #3300) according to the manufacturer's protocol. The amplified cDNA product was purified through a QIAGEN MinElute Reaction Cleanup column (QIAGEN Cat #28204) according to modifications from NuGEN. DNA product (4.0 μg) was then used to generate sense target cDNA using the WT-Ovation Exon Module v1 (NuGEN Technologies, Cat #2000) and again cleaned up with the Qiagen column. Sense target cDNA product (5.0 μg) was fragmented (average fragment size = 85 bases) and biotin-labeled using the NuGEN FL-Ovation cDNA Biotin Module, v2 (NuGEN Technologies, Cat #4200) by the manufacturer's protocol. The biotin-labeled cDNA was mixed with Affymetrix eukaryotic hybridization buffer (Affymetrix, Inc., Santa Clara, CA) and hybridized to Affymetrix Human Gene 1.0 ST Arrays at 45°C for 18 h with 60 rpm rotation in an Affymetrix Model 640 Genechip Hybridization Oven. Following hybridization, the arrays were washed, stained with streptavidin-phycoerythrin (Molecular Probes, Inc., Eugene, OR) and with antistreptavidin antibody (Vector Laboratories, Inc., Burlingame, CA) using the Affymetrix Model 450 Fluidics Station. Arrays were scanned with the Affymetrix Model 3000 scanner (7 G upgrade) and data were collected using the GeneChip operating software v 1.4.

Microarray data analysis

The array data were analyzed for significant changes in expression using Partek Genomics Suite Software (Partek, Inc.). Three biological replicates for each treatment were processed and analyzed. Briefly, the expression value estimates generated by microarray scanners were normalized using the robust multi-array averaging algorithm. Analysis of variation followed by multiple testing correction using the false-discovery rate was used to assess significant changes in expression profile between samples. Expression data for siControl +/− CSE were analyzed with Ingenuity Pathway Analysis (IPA, Ingenuity Systems, www.ingenuity.com). The IPA Canonical Pathways Analysis tool was used to identify the pathways from the IPA Library of Canonical Pathways that were most significant in the data. A dataset containing gene identifiers with expression fold change and associated p-value was uploaded to the application. The data were then limited to fold change +/− 1.5 and p-value ≤0.05. A core analysis was performed to identify gene function and a canonical pathway analysis was used to identify the pathways to which the genes mapped. Significance is measured within IPA by 2 means: 1.) Determination of a ratio calculated by dividing the number of differentially expressed genes that map to a particular pathway by the total number of genes in that pathway; 2.) Calculation of a p-value by Fisher's exact test to determine the probability that the association between the genes in the data set and the canonical pathway can be explained by chance.