NRF2 negatively regulates primary ciliogenesis and hedgehog signaling

Primary cilia are lost during cancer development, but the mechanism regulating cilia degeneration is not determined. While transcription factor nuclear factor-erythroid 2-like 2 (NRF2) protects cells from oxidative, proteotoxic, and metabolic stress in normal cells, hyperactivation of NRF2 is oncogenic, although the detailed molecular mechanisms by which uncontrolled NRF2 activation promotes cancer progression remain unclear. Here, we report that NRF2 suppresses hedgehog (Hh) signaling through Patched 1 (PTCH1) and primary ciliogenesis via p62/sequestosome 1 (SQSTM1). PTCH1, a negative regulator of Hh signaling, is an NRF2 target gene, and as such, hyperactivation of NRF2 impairs Hh signaling. NRF2 also suppresses primary cilia formation through p62-dependent inclusion body formation and blockage of Bardet–Biedl syndrome 4 (BBS4) entrance into cilia. Simultaneous ablation of PTCH1 and p62 completely abolishes NRF2-mediated inhibition of both primary ciliogenesis and Hh signaling. Our findings reveal a previously unidentified role of NRF2 in controlling a cellular organelle, the primary cilium, and its associated Hh signaling pathway and also uncover a mechanism by which NRF2 hyperactivation promotes tumor progression via primary cilia degeneration and aberrant Hh signaling. A better understanding of the crosstalk between NRF2 and primary cilia/Hh signaling could not only open new avenues for cancer therapeutic discovery but could also have significant implications regarding pathologies other than cancer, including developmental disorders, in which improper primary ciliogenesis and Hh signaling play a major role.


Introduction
Nuclear factor-erythroid 2-like 2 (NRF2) is a transcription factor that mediates cellular redox, metabolic and protein homeostasis [1,2]. Under physiological conditions, NRF2 is negatively regulated by Kelch-like ECH-associated protein 1 (KEAP1), a substrate adaptor protein of the Cullin3 (Cul3)-Ring-Box 1 (Rbx1) E3-ligase complex that targets NRF2 for ubiquitylation and degradation by the 26S proteasome [3]. KEAP1 functions as a molecular sensor through its cysteines, especially C151, controlling activation of the NRF2 pathway [4]. Upon  promotion of cancer. Knockdown (KD) of NRF2 enhanced, whereas pharmacological activation or overexpression of NRF2 suppressed, primary cilia formation. PTCH1, a negative regulator of Hh signaling, was demonstrated to have a functional ARE, and increasing NRF2 prevented SMO translocation and suppressed Hh signaling in a PTCH1-dependent manner. Furthermore, NRF2 suppressed primary ciliogenesis by enhancing the expression of p62/ sequestosome 1 (SQSTM1), resulting in the sequestration and mislocalization of Bardet-Biedl syndrome 4 (BBS4), a positive regulator of cilia formation. Our data not only reveal a previously unidentified role of NRF2 in controlling key cellular processes (primary ciliogenesis and Hh signaling) but also uncover a mechanism by which NRF2 hyperactivation promotes tumor progression via primary cilia degeneration and aberrant Hh signaling.

NRF2 deletion enhances primary ciliogenesis and Hh signaling
To evaluate the potential role of NRF2 in primary cilia formation, primary cilia were examined in mouse embryonic fibroblasts (MEFs) isolated from Nrf2 wild-type (WT) (Nrf2 +/+ ) and Nrf2 knockout (Nrf2 −/− ) mice. Immunofluorescence (IF) staining for acetylated tubulin (Ac-Tub) and ADP-ribosylation factor-like protein 13B (ARL13B) (two markers for primary cilia) clearly showed more ciliated cells in the Nrf2 −/− MEF cells compared to the Nrf2 +/+ MEF cells (Fig 1A), as well as higher protein levels of Ac-Tub and ARL13B (Fig 1B, S1A Fig). Consistent with the MEF data, the percentage of ciliated cells and the level of Ac-Tub or ARL13B were significantly higher in NRF2 −/− BEAS-2B and NRF2 −/− H838 cell lines compared to their respective NRF2 +/+ controls (Fig 1A and 1B, S1A Fig). The expression level of several genes that play a crucial role in intraflagellar transport (intraflagellar transport-20 [IFT20], intraflagellar transport-88 [IFT88], and Kinesin Family Member 3A [KIF3a]) were also enhanced by NRF2 knockout (Fig 1C and 1D, S1B Fig), indicating that primary cilia are negatively regulated by NRF2. To determine if the lack of cilia was a result of decreased ciliogenesis or improper break down of primary cilia, ciliary disassembly was evaluated by measuring colocalization of the primary components (NudE Neurodevelopment Protein 1 [NDE1], oral-facial-digital syndrome 1 [OFD1], and Aurora A) of the cilium disassembly complex (CDC) with the primary cilia itself, as reported previously [36,37]. As shown in S2 Fig, the percentage of cilia that exhibited colocalization of each CDC component was similar between NRF2 +/+ cells and NRF2 −/− cells. Furthermore, activation of the complex was also evaluated by measuring colocalization of Ac-Tub with active Aurora A (phospho T288); however, similar to the nonphosphorylated form, there was no significant difference in colocalization between NRF2 +/+ cells and NRF2 −/− cells. While this suggests that the negative effect of NRF2 on primary cilia formation is most likely through inhibition of ciliogenesis, future studies to further clarify NRF2 regulation of primary cilium assembly/disassembly are still needed. Because primary cilia are essential for transduction of the Hh signal, the effect of NRF2 on Hh signaling was also examined. Hh signaling is primarily mediated by the transcription factors GLI2 and GLI3, which coexist as N-terminal repressor (R) and full-length activator (FL) forms. Thus, the ratio of FL/R can be used to evaluated the activation of the Hh signal pathway [35,38]. Interestingly, the GLIFL/R ratio of both GLI2 and GLI3 was significantly increased in NRF2 −/− compared to NRF2 +/+ cell lines, while the expression of SMO was unchanged (Fig 1E,  S1C Fig). To further evaluate the effect of NRF2 on Hh signaling, GLI transcriptional activity was also assessed by GLI luciferase assay in the presence or absence of Sonic hedgehog (Shh), a known Hh pathway activator. As expected, the GLI luciferase activity was increased by Shh in a dose-dependent manner in both NRF2 +/+ and NRF2 −/− cell lines; however, the basal level in  NRF2 −/− cells was higher than NRF2 +/+ cells (Fig 1F), and the response to Shh was slightly diminished in NRF2 −/− BEAS-2B and H838 cells compared with the WT cells (S1D Fig). NRF2-mediated suppression of primary ciliogenesis and Hh signaling is accompanied by enhanced PTCH1 expression and impaired ciliary entrance of SMO NRF2 up-regulation, either by bixin treatment (C151-dependent activator) or ectopic expression of NRF2, increased NAD(P)H Quinone Dehydrogenase 1 (NQO1) (a well-defined NRF2-target gene) and PTCH1 levels but had no effect on KEAP1; in contrast, the level of Ac-Tub and ARL13B gradually decreased over time (Fig 2A, S3A Fig). The ratio between GLIFL and GLIR (GLI2FL/GLI2R and GLI3FL/GLI3R), as well as GLI luciferase activity, was also reduced with bixin treatment and NRF2 overexpression, whereas there was no change in SMO protein levels (Fig 2B and 2C, S3B Fig). In addition, the percentage of ciliated cells was also significantly reduced upon NRF2 up-regulation, as seen in cells with KEAP1 knockout, bixin treatment, or NRF2 overexpression ( Fig 2D). Furthermore, NRF2 up-regulation, either pharmacologically or genetically, inhibited the ciliary entrance of SMO ( Fig 2E). In addition, the effect of bixin on inhibition of primary ciliogenesis and Hh signaling is NRF2-dependent, as the inhibitory effects of bixin were lost in NRF2 −/− cells as compared with the WT control (S4 Fig). To exclude the possibility that the negative effect of NRF2 on primary cilia is due to an indirect effect of NRF2 on cell proliferation, cell cycle status was further analyzed via propidium iodide (PI) staining and fluorescence-activated cell sorting (FACS). The percentage of cells in the different phases of the cell cycle were similar regardless of NRF2 status, with over 70% of the cells being in G0/G1 phase (S5 Fig).
https://doi.org/10.1371/journal.pbio.3000620.g001 NRF2 negatively regulates primary cilia and Hh signaling PLOS Biology | https://doi.org/10.1371/journal.pbio.3000620 February 13, 2020 A t test was used to compare the various groups, and p < 0.05 was considered statistically significant. � p < 0.05 compared between the two groups. Ac-Tub, acetylated tubulin; ARL13B, ADP-ribosylation factor-like protein 13B; GAPDH, glyceraldehyde-3-phosphate PTCH1 deletion rescues the ciliary entrance of SMO but only has a partial effect on NRF2-mediated suppression of primary cilia and Hh signaling To explore whether PTCH1 is critical in NRF2-mediated repression of Hh signaling and primary cilia function, PTCH1 +/+ and PTCH1 −/− H1299 cells were established. Deletion of PTCH1 resulted in activation of Hh signaling, as confirmed by an increase in the ratio of GLI2FL/GLI2R and GLI luciferase activity in PTCH1 −/− cells, although no obvious difference in the ratio of GLI3FL/GLI3R was observed (Fig 4A and 4B, S7A Fig). With bixin treatment, the levels of NRF2 and NQO1 were up-regulated in both PTCH1 +/+ and PTCH1 −/− cells, while the level of Ac-Tub and ARL13B gradually decreased in both cell lines (Fig 4C, S7B Fig), indicating that deletion of PTCH1 had no significant effects on NRF2-mediated repression of primary ciliogenesis. Interestingly, the ratio of GLI2FL/GLI2R and GLI luciferase activity were reduced with bixin treatment in both PTCH1 +/+ and PTCH1 −/− cells, but the reduction was diminished in PTCH1 −/− cells (Fig 4D and 4E, S7C Fig). Furthermore, although PTCH1 deletion recovered ciliary entrance of SMO in cells treated with bixin ( Fig 4F), it had no detectable effects on the reduction of percentage of ciliated cells in response to bixin treatment ( Fig 4G). Collectively, these results demonstrate PTCH1 is only partially responsible for NRF2-mediated suppression of Hh signaling.

NRF2 inhibits primary ciliogenesis and Hh signaling by up-regulating p62 expression, increasing inclusion body formation, and suppressing ciliary entrance of BBS4
It has been reported that OFD1 accumulation at centriolar satellites due to dysregulation of autophagy results in defective ciliary recruitment of BBS4 and shorter/fewer primary cilia and that OFD1 depletion promotes cilia formation [39]. Therefore, the possible connection between NRF2 and OFD1 or BBS4 was investigated. In KEAP1 −/− cells, high NRF2 levels resulted in up-regulation of p62/SQSTM1 (Fig 5A), which is consistent with p62 being an NRF2-target gene [40,41]. The protein levels of OFD1, as well as microtubule-associated proteins 1A/1B light chain 3B (LC3)-II, an indicator of autophagosome number, were also increased ( Fig  Next, the importance of p62 in NRF2-mediated suppression of primary ciliogenesis and Hh signaling was tested. As shown in Fig 5D, p62 is required for the reduction of ciliated cells upon NRF2 up-regulation by bixin. Furthermore, bixin partially lost its effect on the ratio of GLIFL/GLIR (especially GLI3FL/GLI3R) and GLI luciferase activity in p62/SQSTM1 −/− cells (Fig 5E and 5F, S8E Fig).
Since p62 is a critical protein required for formation of protein aggregates, and increased levels of p62 can sequester its interacting partners (for example, LC3) into protein/ubiquitination (Ub)-containing inclusion bodies, a monomeric red fluorescent protein (mRFP)-green fluorescent protein (GFP)-LC3 reporter construct was utilized to test the requirement of PTCH1 and p62 in mediating inclusion body formation. Bixin treatment resulted in an accumulation of LC3 positive puncta, which are p62-, but not PTCH1-, dependent (S9A and S9B  Fig). Next, whether increased OFD1, which has also been shown to interact with LC3, was sequestering BBS4 into the p62-positive inclusion bodies was tested. Indirect IF revealed that bixin increased the colocalization of p62/OFD1 and p62/BBS4, as well as decreased the ciliary dehydrogenase; Hh, hedgehog; IF, immunofluorescence; KEAP1, Kelch-like ECH-associated protein 1; NQO1, NAD(P)H Quinone Dehydrogenase 1; NRF2, nuclear factor-erythroid 2-like 2; PTCH1, Patched 1; SMO, smoothened.

Simultaneous ablation of PTCH1 and p62 abolishes NRF2-mediated effects on both primary ciliogenesis and Hh signaling
Next, we tested whether depletion of PTCH1 and p62 is sufficient to block NRF2-mediated negative regulation on both Hh signaling and primary ciliogenesis. CRISPR knockout of p62/ SQSTM1 in PTCH1 −/− cells or knockout of PTCH1 in p62/SQSTM1 +/+ cells failed to generate any viable clones. Therefore, p62-small interfering RNA (siRNA) was used to transiently KD p62. The NRF2-mediated increase in the protein levels of LC3-I, LC3-II, and OFD1 was observed in WT and PTCH1 −/− cells, but not in PTCH1 −/− ; p62-KD cells (Fig 6A, S10A Fig). Importantly, the reduction in the ratio of GLI2FL/GLI2R and GLI3FL/GLI3R, as well as GLI luciferase activity, observed in WT and PTCH1 −/− cells in response to bixin was completely abolished in the PTCH1 −/− ;p62-KD cells (Fig 6B and 6C, S10B Fig). Similarly, NRF2-mediated reduction of ciliated cells was also lost in PTCH1 −/− ;p62-KD cells (Fig 6D). These results demonstrate that NRF2 negatively regulates Hh signaling and primary ciliogenesis through its target genes PTCH1 and p62.

NRF2 is required for HPI-4-mediated inhibition of ciliogenesis and Hh signaling
Hedgehog pathway inhibitor-4 (HPI-4, ciliobrevin A) is a small molecule that was discovered during a high-throughput screen for Hh pathway antagonists and later reported to disrupt ciliogenesis and inhibit Hh signaling [35,42]. Thus, we next aimed to determine whether HPI-4 effects were NRF2-dependent. HPI-4 induced NRF2 and NQO1 in a KEAP1-dependent manner (S11A Fig). In addition, HPI-4 activation of the NRF2 pathway required KEAP1-C151 (S11B Fig), indicating that HPI-4 is a canonical NRF2 inducer similar to bixin. Moreover, the effect of HPI-4 on inhibiting primary cilia and Hh signaling requires NRF2, as evidenced by the fact that HPI-4 significantly reduced percentage of ciliated cells in both Nrf2 +/+ MEF and NRF2 +/+ H838 cells; however, no significant difference in primary cilia formation was found when NRF2 −/− cells were treated with HPI-4, as detected by IF for Ac-Tub and ARL13B ( Fig  7A) or immunoblotting for Ac-Tub and ARL13B protein levels (Fig 7B, S12A and S12B Fig). HPI-4-mediated inhibition of Hh signaling is also NRF2-dependent because the ratio of GLIFL/GLIR and GLI luciferase activity were reduced only in NRF2 +/+ cells, not NRF2 −/− cells (Fig 7B, S12A-S12D Fig). Collectively, these results demonstrate that HPI-4 is a canonical immunoblot results is shown in S8A Fig. (B)  NRF2 inducer, and NRF2 activation is required for HPI-4-mediated suppression of primary ciliogenesis and Hh signaling.

Discussion
Tumor progression is associated with a shift from normal homeostasis to a protumorigenic phenotype centered on rapid proliferation, metabolism, and survival under harsh conditions. As such, transformation and subsequent cancer cell survival is often associated with hyper-or hypoactivation of key cell signaling cascades, whose controlled activation is integral to normal cellular function. A key example of this phenomenon is hyperactivation of the NRF2 signaling pathway. Controlled activation of NRF2, as a result of increased oxidative or xenobiotic stress,
https://doi.org/10.1371/journal.pbio.3000620.g007 NRF2 negatively regulates primary cilia and Hh signaling is critical in restoring the redox, proteostatic, and metabolic balance in the cell under stressed conditions. However, multiple cancer types have been found to have hyperactivation of the NRF2 response, conferring not only a growth and survival advantage over their noncancerous counterparts but also resistance to both chemo-and radiotherapies [6]. Thus, understanding the key signaling cascades linking hyperactivation of NRF2 to tumor formation is fundamental in cancer biology.
The Hh signaling pathway controls cell fate and self-renewal and plays a key role in mediating developmental processes. Uncontrolled activation of Hh signaling has been implicated in tumor initiation and progression. SMO and SUFU mutations have been reported in a variety of cancers. Therefore, inhibitors targeting the Hh signaling pathway have been developed, but so far, the clinical outcomes are not promising [43]. Hh signaling is intimately linked to the primary cilium, which is important in regulating cell proliferation, migration, and differentiation. Primary cilia function as a tumor suppressor organelle, and a reciprocal relationship between primary ciliogenesis and cell cycle progression has been reported [44]. In fact, many tumor types are associated with a loss of primary cilia, as well as aberrant Hh signaling [45,46]. Therefore, restoration of primary cilia and activation of Hh signaling should be a viable approach for cancer treatment.
In this study, we identified a previously unrecognized role of NRF2 in controlling two key cellular processes, primary ciliogenesis and the Hh signaling pathway, via two distinct mechanisms: (1) PTCH1, a critical negative regulator of Hh signaling, was found to contain a functional ARE, indicating that NRF2 can negatively control Hh signaling through transcriptional up-regulation of PTCH1; and (2) increased expression of p62/SQSTM1, a key autophagy adaptor protein that was previously identified as an NRF2-target gene, promotes aggregation and mislocalization of key proteins controlling ciliogenesis such as OFD1 and BBS4 (Fig 8). Importantly, only simultaneous ablation of both PTCH1 and p62, but neither one alone, was sufficient to prevent the observed suppressive effects of NRF2 on primary cilia and Hh signaling. Since both p62 and PTCH1 are NRF2 target genes, and many cancers have been associated with a loss of primary cilia and aberrant Hh signaling, our data reveal a mechanism by which hyperactivation of NRF2, as seen in lung cancer as well as head and neck cancer, promotes tumor progression via dysregulation of a fundamental cellular organelle: the primary cilium and its associated Hh signaling. This was also confirmed in human lung tumor tissues in which high NRF2 expression strongly correlates with high levels of PTCH1, implicating downregulation of Hh signaling as a possible mediator of NRF2's oncogenic effects.
The mechanistic details illustrated here that link NRF2 to primary cilia/Hh signaling are significant for designing cancer therapeutic drugs with a defined mode of action. A small molecule, HPI-4, was originally identified as an Hh pathway inhibitor that acts downstream of SMO in the Hh signaling cascade and interferes with GLI processing and stability [42]. Later on, HPI-4, renamed ciliobrevin A, was reported to disrupt primary cilium formation and GLIdependent Hh signaling by inhibiting cytoplasmic dynein, an AAA+ ATPase motor protein that is critical in regulating many cellular processes, including ciliary protein trafficking, mitotic spindle formation, and organelle transport through microtubule gliding [35]. Interestingly, we demonstrated here that HPI-4 is a canonical NRF2 inducer, inducing NRF2 in a KEAP1-C151-dependent manner. Moreover, HPI-4-mediated inhibition of primary ciliogenesis and Hh signaling required NRF2 because HPI-4 (ciliobrevin A) had no effect on the percentage of ciliated cells or Hh signaling in NRF2 −/− cells in our study. Therefore, activation of NRF2 by targeting KEAP1-C151 may be the major effect of HPI-4 on inhibition of primary ciliogenesis and Hh signaling.
There are a few studies that have reported the connection between NRF2 and general ciliation processes. For example, Jang and colleagues reported that increased primary ciliation can induce autophagy, which somehow results in the inactivation of NRF2 and subsequent promotion of neuroectoderm derivation in human embryonic stem cells [47]. Other studies have also indicated that cigarette-smoke-induced mucociliary clearance is preserved in mice treated with the chemical chaperone 4-phenylbutyric acid but that the protective effect was disrupted in mice lacking NRF2, suggesting a key role for NRF2-mediated protection against oxidative stress and altered proteostasis in maintaining proper mucociliary clearance [48]. Moreover, treatment with Manganese(III) (Mn(III)) tetrakis(1-methyl-4-pyridyl) porphyrin, an antioxidant, accelerated the normalization of cilia length concomitant with a decrease of oxidative stress and morphological recovery in the recovery process of damaged kidneys [49]. Therefore, both autophagy and oxidative stress could be the key factors linking NRF2 and ciliation. During the course of generating this publication, a study reporting the opposite of our findingsi.e., that NRF2 positively regulates primary cilia-was also published [50]. In fact, some of their findings support our own work, including the presence of an ARE in PTCH1, as well as NRF2 regulation of ciliogenesis occurring independently of Hh signaling. However, our data indicate negative regulation of ciliogenesis and Hh signaling by NRF2, which is fully supported by our results utilizing pharmacological and genetic manipulation of NRF2 in multiple lung cancer cell lines as well as MEFs. Therefore, it is clear that a relationship among NRF2, ciliogenesis, and Hh signaling exists and still needs further investigation.
Lastly, a better understanding of the crosstalk between NRF2 and primary cilia/Hh signaling not only opens new avenues for cancer therapeutic discovery but also has significant implications regarding pathologies other than cancer, including developmental disorders, in which improper function of these pathways plays a major role. Intriguingly, Nrf2 −/− mice develop normally, except that these mice are more sensitive to chemical carcinogens, whereas Keap1 −/− mice die shortly after birth because of hyperkeratosis of the forestomach and esophagus as a result of hyperactivation of the NRF2 pathway [51,52]. Thus, a more detailed analysis of NRF2 https://doi.org/10.1371/journal.pbio.3000620.g008 expression, ciliation, and the Hh signaling pathway at the different stages of life, from embryogenesis to adulthood, particularly with regards to how this crosstalk changes from ensuring proper development and orchestration of the antistress response to more malignant programming in certain disease contexts, is of critical importance.

Ethics statement
All mice were handled according to the Guide for the Care and Use of Laboratory Animals, and the protocols were approved by the University of Arizona Institutional Animal Care and Use Committee (approval number: 11-287). Our protocols comply with (1) the Animal Welfare Act/Animal Welfare Regulations (AWA/AWRs) and other applicable federal regulations such as GLP for covered species and activities; (2) The National Research Council Guide for the Care and Use of Laboratory Animals, 8th Edition (Guide), for all vertebrate animals used in biomedical research; and (3) The Guide for Care and Use of Agricultural Animals in Research and Teaching, 3rd Edition (Ag Guide), for production farm animal research and teaching. In addition, this study was approved by the Committee on the Ethics of Human Subject Research at the University of Arizona. In this study, the patients who donated their cancer tissues for our research have provided their written informed consent.

Construction of recombinant DNA molecules
The NRF2 overexpression vector was constructed by cloning a PCR-generated portion of the NRF2 coding sequence into the pCI vector (Promega, Madison, WI, USA). For the dual luciferase assay, the portion of the human PTCH1 promoter and mouse Ptch1 promoter containing the putative ARE sequence was then amplified by PCR, and the amplified fragments were cloned into the pGL4.22 vector (Promega). The pGL3-8XGliBS:Luc, ptf-LC3 (mRFP-GFP-LC3), pcDNA-KEAP1-WT, and pcDNA-KEAP1-C151S plasmids were generated as described previously [56,57].

Transfection of cDNA and luciferase reporter assay
Transfection of cDNA was performed using the Lipofectamine 3000 reagent (Invitrogen) according to the manufacturer's instructions. Luciferase activity was measured using the dual luciferase reporter assay system (Promega). For relative luciferase activity analysis, the value of Firefly-luciferase was normalized to the value of Renilla luciferase. The experiment was repeated 3 times. The data are expressed as means ± SD.

mRNA extraction and real-time qRT-PCR
Total mRNA was extracted using TRIzol (Invitrogen) according to the manufacturer's instructions. cDNA was then synthesized using 2 μg of mRNA and a Transcriptor first-strand cDNA synthesis kit (Promega). Real-time qPCR was then performed as previously described [55]. βactin was used for qPCR normalization, and all experiments were measured in triplicate. Primer sequences (5 0 -3 0 ) are as follows: Mouse-Ift20-Forward

WB, IF, and immunohistochemical (IHC) analysis
WB, IF, and IHC were performed as previously described [55,58,59]. Primary antibodies against NRF2 for IF, Cat# A10254 and A32731) and Alexa-Fluor-594-conjugated secondary antibody (1:2,000 for IF, Cat# A11037 and A32742) were obtained from Invitrogen. All of uncropped blots throughout the paper were shown in S1 Raw images.

Biotin-DNA pull-down
Biotin-DNA pull-down was performed as reported previously [60]. In brief, cells were lysed in RIPA buffer containing 1 mM DTT, 1 mM phenylmethylsulfonyl fluoride (PMSF), and 1% protease inhibitor cocktail (Sigma-Aldrich). The cell lysates were precleared with streptavidin beads and incubated with 2 μg biotinylated DNA probes that spanned the ARE-containing sequences in the promoter regions of both human PTCH1 and mouse Ptch1. The DNA-protein complexes were further pulled down by streptavidin beads, and complexes were washed 3 times, resolved on an SDS-PAGE gel, and subjected to immunoblot analysis. The sequences of the 41-bp biotinylated DNA probes used are as follows: Human WT PTCH1-ARE probe:

Statistical analysis
Results are presented as means ± SD for at least 3 independent experiments. Statistical analysis was performed using SPSS 17.0. Unpaired Student t tests were applied to compare the means for two groups. One-way ANOVA with Bonferroni's correction was used to compare the means of 3 or more groups. p < 0.05 was considered statistically significant. The values for the data used to create the graphs throughout the paper are shown in S1 Data.
Supporting information S1 Fig. NRF2 deletion enhances ciliogenesis and Hh signaling (related to Fig 1). (A-C) Relative quantification of immunoblot results in Fig 1A, 1B and 1E. (D) The normalized result of Fig 1F. The level of relative luciferase activity in all control groups (both NRF2 +/+ cells and NRF2 −/− cells) was considered as "1." Results are expressed as mean ± SD. A t test was used to compare the various groups, and p < 0.05 was considered statistically significant. � p < 0.05 compared between the two groups. Hh, hedgehog; NRF2, nuclear factor-erythroid 2-like 2 (PDF)  6). (A-B) Relative quantification of immunoblot results in Fig 6A and 6B. Results are expressed as mean ± SD. A t test was used to compare the various groups, and p < 0.05 was considered statistically significant. � p < 0.05 compared with the control group. Hh, hedgehog;