Abstract

   Idiopathic pulmonary fibrosis (IPF) is a progressive and irreversible
   disease characterized by an increase in differentiation of fibroblasts
   to myofibroblasts and excessive accumulation of extracellular matrix in
   lung tissue. Pharmacological activation of NRF2 has proved to be a
   valuable antifibrotic approach, however the detailed mechanisms of how
   NRF2 mediates antifibrotic function remain unclear. In this study, we
   found that the antifibrotic function of sulforaphane (SFN), an NRF2
   activator, was largely dependent on LOC344887, a long noncoding RNA.
   Two functional AREs were identified in both the promoter and intron 1
   of LOC344887, which defines LOC344887 as a novel anti-fibrotic NRF2
   target gene. RNA-seq analysis revealed that LOC344887 controls genes
   and signaling pathways associated with fibrogenesis. Deletion or
   downregulation of LOC344887 enhanced expression of CDH2/N-cadherin, as
   well as a number of other fibrotic genes and blunted the antifibrotic
   effects of SFN. Furthermore, LOC344887-mediated downregulation of
   fibrotic genes may involve the PI3K-AKT signaling pathway, as
   pharmacologic inhibition of PI3K activity blocked the effects of
   LOC344887 knockdown. Our findings demonstrate that NRF2-mediated
   LOC344887 upregulation contributes to the antifibrotic potential of SFN
   by repressing the expression of CDH2 and other fibrotic genes,
   providing novel insight into how NRF2 controls the regulatory networks
   of IPF. This study provides a better understanding of the molecular
   mechanisms of NRF2 activators against pulmonary fibrosis and presents a
   novel therapeutic axis for prevention and intervention of
   fibrosis-related diseases.

   Keywords: NRF2, LOC344887, CDH2, PI3K-AKT, Extracellular matrix,
   Pulmonary fibrosis

Graphical abstract

   [41]Image 1
   [42]Open in a new tab

1. Introduction

   Idiopathic pulmonary fibrosis (IPF) is a chronic lung disease caused by
   different factors, such as environmental pollutants, drug side effects,
   and viral or fungal infection [[43][1], [44][2], [45][3]]. In fact,
   pulmonary fibrosis is also regarded as one of the major thoracic
   complications in patients with coronavirus disease 2019 (COVID-19)
   [[46]4,[47]5]. Pulmonary fibrosis is characterized by activation of
   lung fibroblasts, excessive accumulation of extracellular matrix (ECM)
   in lung tissues, and progressive lung scarring. In recent years, the
   incidence of pulmonary fibrosis has increased sharply [[48]2], but
   effective treatments for fibrotic diseases are limited. The current
   treatments for pulmonary fibrosis, immunosuppressants (e.g.
   cyclophosphamide) and corticosteroids (e.g. dexamethasone), are limited
   in their potential because of severe side effects and low efficacy
   [[49]6,[50]7]. Nintedanib and pirfenidone, two drugs recently approved
   by the FDA for the treatment of pulmonary fibrosis, can slow down the
   progression of lung fibrosis and preserve lung function, however,
   neither of those drugs are able to reverse fibrosis. Furthermore, these
   new drugs display severe side effects, causing injury of many organs
   including the liver and skin [[51][8], [52][9], [53][10], [54][11]].
   Therefore, understanding the molecular events underlying the
   pathogenesis of IPF, coupled with developing novel and effective means
   of prevention/intervention that target early stages of fibrogenesis
   remain an important area of research.

   Nuclear factor erythroid 2 (NFE2)-related factor 2 (NRF2) is an
   important transcription factor that regulates cellular antioxidant and
   anti-inflammatory responses by controlling expression of its target
   genes containing antioxidant response elements (AREs) in their
   regulatory regions. In response to oxidants and electrophiles, NRF2
   forms heterodimeric complexes with small musculoaponeurotic
   fibrosarcoma (sMAF) protein and binds to AREs to transcriptionally
   activate its target genes and restore cellular homeostasis [[55][12],
   [56][13], [57][14]]. The role of NRF2 in pulmonary fibrosis has been
   investigated, and those studies indicate that NRF2 protects against the
   development of pulmonary fibrosis, indicating NRF2 activators could be
   used to inhibit fibrogenesis [[58][15], [59][16], [60][17]]. For
   instance, sulforaphane (SFN), a well-studied NRF2 activator, attenuates
   bleomycin-induced pulmonary fibrosis by effectively suppressing the
   epithelial-mesenchymal transition (EMT) process [[61]15], an effect
   that was also observed when NRF2 was activated by another inducer
   artemisitene [[62]17]. In addition, trans-4,4′-dihydroxystilbene
   significantly ameliorates cigarette smoke-induced pulmonary fibrosis
   via NRF2 activation [[63]18]. Therefore, NRF2 activators hold promising
   potential to be developed into preventive agents against pulmonary
   fibrosis. The current mechanistic action of NRF2 activators against
   pulmonary fibrosis are thought to derive from its mediation of the
   antioxidant response and inhibition of EMT, which is essential for the
   overexpression of ECM genes [[64]15,[65]19,[66]20]. However, the
   detailed molecular mechanisms linking NRF2 activation to the reduction
   of fibrogenesis are missing and warrant investigation.

   Long noncoding RNAs (lncRNAs) have attracted widespread attention in
   various fields. LncRNAs are non-coding transcripts longer than 200
   nucleotides involved in various biological processes, such as cell
   proliferation and cell fate decisions. More importantly, aberrant
   IncRNA expression has been associated with developmental disorders and
   the pathogenesis of many human diseases [[67]21,[68]22]. LncRNAs have
   also been demonstrated to play a role in pulmonary fibrosis. Knockdown
   of lncRNA H19 was able to attenuate pulmonary fibrosis in vitro and in
   vivo, through modulation of the miR-140-TGF-β/Smad 3 signaling pathway
   [[69]23]. LncRNA ZEB1-AS1 was reported to promote lung fibrosis by
   enhancing ZEB1-mediated EMT [[70]24]. Another study demonstrated that
   lncRNA FENDRR also plays a role in mitigating pulmonary fibrosis
   [[71]25]. Therefore, modulating lncRNAs has been considered as a novel
   therapeutic strategy for treating pulmonary fibrosis as well as other
   respiratory diseases. In this study, we report that a lncRNA, NmrA like
   redox sensor 2, pseudogene (NMRAL2P/LOC344887), is a novel
   anti-fibrotic NRF2 target gene that is largely responsible for the
   previously observed antifibrotic function of NRF2. Knockdown or
   knockout of LOC344887 was associated with enhanced fibrosis, including
   increased expression of N-cadherin and ECM via the PI3K-AKT pathway.

2. Materials and methods

2.1. Chemicals, reagents, and antibodies

   Sulforaphane (S6317), as well as anti-mouse IgG (HRP) (1:3000, A-9044)
   and anti-rabbit IgG (HRP) (1:3000, A-0545) antibodies were purchased
   from Sigma-Aldrich. LY294002 (ab120243) was purchased from Abcam. The
   rabbit normal IgG and antibodies against NRF2 (1:1000, sc-13032), sMAF
   (1:1000, sc-22831), NQO1 (1:1000, sc-32793), FN1 (1:1000, sc-8422),
   COL1A1 (1:1000, sc-293,182), α-SMA (1:1000, sc-53142), N-Cadherin
   (1:500, sc-7939) and GAPDH (1:3000, sc-32233) were purchased from Santa
   Cruz Biotechnology. The antibody against E-Cadherin (1:1000, 610,181)
   was purchased from BD Biosciences. Antibodies against AKT (1:2000,
   4691S) and phospho-AKT (Ser473) (1:1000, 4060S) were purchased from
   Cell Signaling Technology. TGF-β (100–21) was purchased from PeproTech.

2.2. Cell culture

   Human non-small cell lung cancer cell lines, NCI–H1299 and A549, human
   lung bronchus epithelial cell line, BEAS-2B, and human lung fibroblast
   cell line, HFL1, were purchased from American Type Culture Collection
   (ATCC). A549, H1299 and BEAS-2B cells were grown in DMEM medium
   supplemented with 10% FBS (Gibco, 26,140–079) and 1%
   penicillin/streptomycin. HFL1 cells were cultured using DMEM:F12 Medium
   supplemented with 10% FBS and 1% penicillin/streptomycin. All cells
   were cultured in a humidified incubator at 37 °C with 5% CO[2], and the
   medium was changed every two days.

2.3. Generation of gene knockout cell lines

   LOC344887 knockout cells were generated using CRISPR-Cas9-mediated
   homologous recombination as reported previously [[72]26,[73]27]. In
   brief, one single guide RNA (sgRNA) sequence was used to target the
   sequence around the insert point, and the sgRNA sequence is as follows:
   5′-TCGTCCAGAGCTAGCTGATG-3′. The puromycin N-acetyltransferase (pac)
   plus a combined poly(A) and a transcriptional pause site (TPS), were
   inserted into the third intron (between +15,770 and + 15,771, [74]Fig.
   2A). The positive cells were selected by puromycin treatment (3 μg/ml,
   Gibco), and the result was further confirmed by qPCR.

Fig. 2.

   [75]Fig. 2
   [76]Open in a new tab

   LOC344887 knockout enhances the expression of fibrotic genes. (A–B) A
   schematic diagram of generation of a LOC344887 knockout A549 cell line
   using CRISPR/Cas9 by insertion of the pac gene (for puromycin
   resistance), followed by a combined poly(A) and a transcriptional pause
   sites (A). Two primer pairs, one located upstream (primer pair 1), the
   other downstream (primer pair 2), of the insertion site (A) were used
   for qRT-PCR analysis of the mRNA level of LOC344887 (B). (C–F)
   LOC344887 WT and KO cells were subjected to RNA-seq analysis. A volcano
   plot showing differentially expressed genes in WT vs. KO cells (C). GO
   (D) and KEGG pathway enrichment (E) analyses were performed to analyze
   the enrichment of genes or pathways. Heat map showing the differential
   expression of fibrotic genes in WT vs. KO cells (F). *: P < 0.05
   between two different groups.

   NRF2 knockout cells were generated using CRISPR-Cas9 mediated gene
   editing as per our previous studies [[77]28,[78]29]. A pair of sgRNA
   sequences (sgRNA-A: 5′-TATTTGACTTCAGTCAGCGA-3′; sgRNA-B:
   5′-TAGTTGTAACTGAGCGAAAA-3′) were used to target coding sequences, and
   successful knockout was confirmed using immunoblot.

2.4. Affymetrix human transcriptome array (HTA) and RNA seq analysis

   Total RNAs were extracted using TRIzol (Invitrogen) and further
   purified using RNeasy MinElute Cleanup Kit (Qiagen). RNA samples were
   sent to the UCLA Clinical Microarray Core for Affymetix HTA or RNA seq
   analysis. Briefly, in Affymetix HTA assay, the samples were first
   fragmented and labeled with GeneChip WT Terminal Labeling Kit
   (Affymetrix), and the ssDNA was further hybridized onto Gene Chip Human
   Gene 1.0 ST Arrays (Affymetrix) according to the manufacturer's
   instructions. Finally the arrays were analyzed using Agilent Microarray
   System. For RNA seq analysis, the rRNA in each sample was depleted
   using Ribo-Zero Gold rRNA Removal Kit (Illumina) and further prepared
   for sequencing using the Ultra Directional RNA library Prep Kit (New
   England Biolabs). Then, the samples were run on a HiSeq 3000 Illumina
   Sequencing Platform as 50-bp single-end read runs. Herein, the results
   were sorted and indexed for quicker access using SAMtools, and the
   reading number of each gene was annotated with a custom in-house
   generated script.

2.5. Real-time quantitative reverse transcription PCR (qRT-PCR)

   Total mRNA was extracted using TRIzol (Invitrogen), and 2 μg of RNA was
   used to synthesize cDNA as reported previously [[79]30,[80]31]. Herein,
   ACTB was used for all of qPCR normalization, and all experiments were
   performed in triplicate. The primer sequences (5′–3′) are as follows:

   LOC344887-Primer 1-F 5′-CGTGAAAGCCTCTGATGGAGA-3′

   LOC344887-Primer 1-R 5′-AGACGGCTGCTCCAATATCA-3′

   LOC344887-Primer 2-F 5′-GGTTTTGGGGAAAGAAGTCC-3′

   LOC344887-Primer 2-R 5′-CGACACATATTGGCTATTTCCTT-3′

   COL1A1-F 5′-GAGGGCCAAGACGAAGACATC-3′

   COL1A1-R 5′-CAGATCACGTCATCGCACAAC-3′

   FN1–F 5′-GAGAATAAGCTGTACCATCGCAA-3′

   FN1-R 5′-CGACCACATAGGAAGTCCCAG-3′

   ACTA2-F 5′-CTATGAGGGCTATGCCTTGCC-3′

   ACTA2-R 5′-GCTCAGCAGTAGTAACGAAGGA-3′

   CDH1-F 5′-CGAGAGCTACACGTTCACGG-3′

   CDH1-R 5′-GGGTGTCGAGGGAAAAATAGG-3′

   CDH2-F 5′-AGCCAACCTTAACTGAGGAGT-3′

   CDH2-R 5′-GGCAAGTTGATTGGAGGGATG-3′

   NQO1-F 5′-ATGTATGACAAAGGACCCTTCC-3′

   NQO1-R 5′-TCCCTTGCAGAGAGTACATGG-3′

   ACTB-F 5′-CCCAGAGCAAGAGAGG-3′

   ACTB-R 5′-GTCCAGACGCAGGATG-3′

2.6. Luciferase reporter gene assay

   Both wild type and mutated 41-bp putative ARE-containing sequences
   ([81]Fig. 1A) were cloned into the pGL4.22-luciferase vector, and
   transfected using Lipofectamine 3000 (Invitrogen) according to the
   manufacturer's instructions. In addition, 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 as reported
   previously [[82]28,[83]29].

Fig. 1.

   [84]Fig. 1
   [85]Open in a new tab

   LOC344887 is an NRF2 target gene. (A) The location and sequences of two
   putative AREs in LOC344887. MU indicates the highlighted nucleotides in
   the putative ARE were mutated. (B–C) H1299 and HFL1 cells transfected
   with ARE-firefly luciferase or TK-renilla luciferase vectors for 24 h
   were left untreated or treated with SFN (5 μM) for 16 h. The cells were
   harvested for immunoblot (B) and dual luciferase assay (C). (D–E)
   Biotinylated DNA probes (41bp) containing the LOC344887-ARE1-WT/MU (D)
   or LOC344887-ARE2-WT/MU (E) were incubated with lysates from NRF2^+/+
   or NRF2^−/− A549 cells. The DNA-bound proteins were pulled down using
   streptavidin beads and measured by immunoblot analysis (D–E). (F)
   Chip-qPCR using A549 NRF2^+/+ cells. Rabbit IgG was used as a negative
   control. (G–H) mRNA levels of LOC344887 and NQO1 were compared in
   NRF2^+/+ vs. NRF2^−/− cell lines (G), or in untreated vs SFN-treated
   NRF2^+/+ cells. *: P﹤0.05 between two different groups.

2.7. Small interfering RNA (siRNA) transfection

   Human LOC344887 siRNA (#5: SI04712344 and #8: SI05178838), human CDH2
   siRNA (#5: SI02663927, #6: SI02757335, #10: SI04434612 and #11:
   SI04434619) and non-targeted control siRNA (SI03650318) were purchased
   from Qiagen. siRNA constructs were transfected using Hiperfect reagent
   (Qiagen) as per the manufacturer's protocol. Briefly,
   1 × 10^5 cells/well were seeded into a 12-well plate and transfected
   with different siRNAs (10 nM) for 72 h. Finally, the knockdown
   efficiency was evaluated via qPCR or immunoblot.

2.8. Immunoblot analysis

   In our study, all of the cell samples were harvested using 1X Sample
   buffer (100 mM DTT, 2% SDS, 10% glycerol, 0.1% bromophenol blue and
   50 mM Tris-HCl pH = 8) and resolved by SDS-PAGE. The samples were
   further transferred onto a nitrocellulose membrane and blocked with 5%
   nonfat dried milk diluted in 1X PBS for 1 h. The membrane was then
   incubated with primary antibody overnight at 4 °C, followed by
   incubation with secondary antibody for 1 h at room temperature.
   Finally, the protein bands were visualized using SuperSignal West Femto
   Maximum Sensitivity Substrate (Thermo Fisher Scientific) on the C600
   Imaging System (Azure Biosystems). The results in different groups were
   quantified using ImageJ 1. x (National Institutes of Health).

2.9. Chromatin immunoprecipitation assay (ChIP)-qPCR

   The CHIP-qPCR assay (EZ-CHIPTM, Merck, Germany) was performed as per
   our previous study [[86]29]. For qPCR detection, 1 μL of purified DNA
   was amplified using primers specific for the sequence flanking the AREs
   located in the LOC344887 promoter and first intron. For qPCR, the
   following primers were used:

   LOC344887-ARE1-F 5′-CTCAAGCCTGGGTAACAAGAAC-3′

   LOC344887-ARE1-R 5′-AACCCATTCCAATTACCACGCC-3′

   LOC344887-ARE2-F 5′-TGAGGCAGGAGAATGGCGTG-3′

   LOC344887-ARE2-R 5′-CTCACAAAAGTGAGAACGGTGC-3′

2.10. Biotin-DNA pull-down

   Biotin-DNA pull-down was performed as reported previously
   [[87]28,[88]29]. Briefly, BEAS-2B cells were lysed using RIPA buffer
   containing 1 mM DTT (Golden Biotech), 1% protease inhibitor cocktail
   (Sigma-Aldrich) and 1 mM phenylmethylsulfonyl fluoride (PMSF, Thermo
   Fisher). Then the samples were precleared with streptavidin beads and
   further incubated with 2 μg of the biotinylated DNA probes overnight.
   The sequences of the different DNA probes are shown in [89]Fig. 1A.
   Finally, the DNA–protein complexes were pulled down using streptavidin
   beads and subjected to immunoblot analysis.

2.11. Statistical analysis

   The results a presented as means ± SD. Statistical analysis was
   performed using SPSS 17.0. Unpaired Student's t-tests were used to
   compare the means of two groups. One-way ANOVA with Bonferroni's
   correction was used to analyze the means of three or more groups.
   P < 0.05 was considered statistically significant.

3. Results

3.1. LOC344887 is an NRF2 target gene

   To determine what lncRNAs might be regulated by NRF2 in lung cells,
   analysis of a Human Transcriptome Array comparing all transcripts in
   BEAS-2B NRF2^+/+ vs. BEAS-2B NRF2^−/− cell lines was performed.
   Interestingly, array analysis indicated a significant reduction in the
   expression of lncRNA (LOC344887) in NRF2 knockout cells compared to the
   wild type (data not shown). Computational analysis identified two
   putative AREs in the promoter region (^−82GTGACACTGCA^−72) and first
   intron (^+3842GTGACAGAGCG^+3852) of LOC344887. To verify the
   authenticity of these two putative AREs, both wild type (WT) and
   mutated (MU) ARE-containing 41-bp sequences ([90]Fig. 1A) were cloned
   into the pGL4.22-luciferase vector. Human H1299 lung adenocarcinoma
   cells and HFL1 lung fibroblasts were transfected with the different
   ARE-reporter containing plasmids, and treated with DMSO (vehicle
   control) or SFN, an NRF2 activator, for 16 h. Immunoblot analysis
   indicated the induction of NRF2 by SFN ([91]Fig. 1B), and relative
   luciferase activity was only increased in response to SFN in cells
   transfected with reporter plasmids containing ARE1/2-WT, but not
   ARE1/2-MU ([92]Fig. 1C). Furthermore, NRF2-sMAF binding to both ARE1
   ([93]Fig. 1D) and ARE2 ([94]Fig. 1E) was confirmed using biotinylated
   41-bp ARE-containing DNA probes, as ARE1-WT ([95]Fig. 1D) and ARE2-WT
   ([96]Fig. 1E) were only able to pull down NRF2 and sMAF in NRF2^+/+
   cells, but not in NRF2^−/− cells ([97]Fig. 1D and E). Endogenous
   binding of NRF2 to ARE1 and ARE2 of LOC344887 was further confirmed
   using CHIP-qPCR ([98]Fig. 1F). In addition, the transcript level of
   LOC344887 was much lower in NRF2^−/− cells, compared to NRF2^+/+ cells
   ([99]Fig. 1G), and was enhanced significantly by SFN in both H1299 and
   HFL1 wild type cell lines ([100]Fig. 1H). Importantly, the expression
   pattern of LOC344887 following loss of NRF2, or upon SFN treatment was
   similar to NQO1, a well-established NRF2 target gene ([101]Fig. 1G and
   H). Collectively, these results demonstrate that there are two
   functional AREs located in the promoter and intron 1 region of
   LOC344887, which defines LOC344887 as an NRF2 target gene in lung
   cells.

3.2. LOC344887 knockout enhances the expression of fibrotic genes

   A LOC344887 knockout (KO) A549 cell line was established using the
   CRISPR/Cas9 technique to insert a DNA sequence encoding the pac gene
   (for puromycin resistance), followed by a combined poly(A) and a
   transcriptional pause site (TPS) ([102]Fig. 2A). Two primer pairs, one
   targeting upstream (Primer 1), and the other downstream (Primer 2), of
   the pac gene insertion site ([103]Fig. 2A) were used for qRT-PCR
   analysis. It was confirmed that transcription of LOC344887 downstream
   of the insertion site (downstream of exon 3) did not occur in
   LOC344887-KO cells ([104]Fig. 2B). RNA-seq analysis was performed to
   compare genome-wide mRNA expression changes in LOC344887-WT and
   LOC344887-KO cells. The results showed that 733 genes were upregulated
   and 316 genes were downregulated in LOC344887-KO compared with the
   LOC344887-WT cell line ([105]Fig. 2C). Gene Ontology (GO) ([106]Fig.
   2D) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway
   enrichment ([107]Fig. 2E) analyses indicated that deletion of LOC344887
   resulted in overexpression of numerous genes involved in fibrogenic
   processes, including reorganization and production of ECM, ECM-receptor
   interactions, and dysregulation of the actin cytoskeleton ([108]Fig. 2D
   and E). The KEGG pathway mapping indicated that the PI3K-AKT signaling
   pathway might be associated with the enhanced fibrogenic process
   ([109]Fig. 2E). A heat map of transcripts that were significantly
   increased in LOC344887 knockout cells indicated enhanced expression of
   a number of genes involved in the fibrogenic process, such as ACTA2
   (encoding alpha smooth muscle actin, a-SMA), COL1A1 (encoding the major
   component of type I collagen), and FN1 (encoding fibronectin, a
   high-molecular weight extracellular matrix protein that binds to
   integrins), among others ([110]Fig. 2F).

3.3. The antifibrotic function of SFN largely depends on LOC344887

   To confirm the function of LOC344887 in controlling pulmonary fibrosis,
   LOC344887-siRNA was used to knockdown the LOC344887 transcript in HFL1
   cells. Consistent with the RNA-seq data shown in [111]Fig. 2, LOC344887
   knockdown enhanced the expression of ACTA2/α-SMA, COL1A1, and FN1 at
   both the mRNA and protein level ([112]Fig. 3A–C). Next, the
   contribution of LOC344887 on the previously reported therapeutic
   effects of SFN against pulmonary fibrosis was evaluated. To simulate
   fibrosis in vitro, the fibrotic activator TGF-β, was used. While TGF-β
   activated the expression of the fibrotic gene (FN1), it did not affect
   LOC344887 expression ([113]Figs. S1A–S1B). In addition, SFN induced
   NRF2 expression and suppressed the expression of ACTA2/α-SMA, COL1A1
   and FN1 in both basal and TGF-β-stimulated conditions. However,
   LOC344887 knockdown significantly limited the effect of SFN on the
   reduced expression of these proteins in both basal and TGF-β-stimulated
   conditions ([114]Fig. 3D and E). It is worth mentioning that while the
   effect of SFN on the expression of these fibrogenic genes was much
   weaker in LOC344887 knockdown than Ctrl group, the induction of NQO1 by
   SFN was not affected by LOC344887 knockdown ([115]Fig. 3D and E).

Fig. 3.

   [116]Fig. 3
   [117]Open in a new tab

   The antifibrotic function of SFN largely depends on LOC344887. (A–C)
   HFL1 cells transfected with siRNA against LOC344887 (10 nM) for 72 h
   were harvested for qRT-PCR (A), and immunoblot analysis (B,
   quantification in C). (D–E) HFL1 cells transfected with LOC-siRNA-#5 or
   Ctrl-siRNA (10 nM) for 72 h were treated with SFN (5 μM) or TGF-β
   (10 ng/ml) for 16 h. The cells were harvested for immunoblot analysis
   (D, quantification in E). *P < 0.05 between different groups.

3.4. LOC344887 knockdown increases the expression of CDH2/N-cadherin

   Another gene whose expression is upregulated by the knockout of
   LOC344887 is CDH2, a gene that encodes the transmembrane cell adhesion
   protein N-Cadherin (N-Cad) ([118]Fig. 4A). SiRNA mediated knockdown of
   LOC344887 in HFL1 cells confirmed the transcriptional upregulation of
   CDH2, but not CDH1 ([119]Fig. 4B). The protein level of N-Cad was also
   enhanced when LOC344887 was knocked down ([120]Fig. 4C and D). Since
   N-Cad has been reported to be associated with EMT, a process known to
   contribute to lung fibrosis [[121]32,[122]33], the importance of N-Cad
   in fibrotic gene expression was tested. Knockdown of CDH2 resulted in a
   decreased expression of ACTA2/α-SMA, COL1A1, and FN1 ([123]Fig. 4E and
   F). More importantly, the effect of LOC344887 on the expression of
   these genes could be abolished by CDH2 knockdown ([124]Fig. 4G and H).
   Taken together, LOC344887 may modulate the expression of ACTA2/α-SMA,
   COL1A1, and FN1 through altering CDH2 gene expression.

Fig. 4.

   [125]Fig. 4
   [126]Open in a new tab

   LOC344887 knockdown increases the expression of CDH2/N-Cadherin. (A) A
   heat map showing the expression of CDH1 and CDH2 in A549 LOC344887-WT
   and LOC344887-KO cells. (B–D) HFL1 cells transfected with
   LOC-siRNA-#5/#8 (10 nM) for 72 h were harvested for qRT-PCR (B) and
   immunoblot analysis (C, quantification in D). Herein, AML12, a liver
   epithelial cell line, was used as a positive control for E-Cad
   detection. (E–F) HFL1 cells transfected with CDH2-siRNA-#5/#6/#10/#11
   (10 nM) for 72 h were harvested for immunoblot analysis (E,
   quantification in F). (G–H) HFL1 cells co-transfected with LOC-siRNA-#5
   and CDH2-siRNA-#5 (10 nM) for 72 h were harvested for immunoblot
   analysis (G, quantification in H).

3.5. LOC344887 regulates the expression of CDH2/N-cad via the PI3K-AKT
signaling pathway

   KEGG pathway enrichment analysis indicated that the pathway most
   affected by LOC344887 knockout is the PI3K-AKT signaling pathway
   ([127]Fig. 2E). Indeed, LOC344887 knockout in A549 cells ([128]Fig. 5A
   and B) and knockdown in HFL1 cells ([129]Fig. 5C and D) increased the
   level of phospho-AKT (Ser473), without altering the total level of AKT.
   To further confirm the requirement of PI3K-AKT signaling in mediating
   the antifibrogenic effects of loss of LOC344887, LY294002, a PI3K
   inhibitor was used. LY294002 not only decreased the level of
   phospho-AKT, N-Cad, FN1, COL1A1, and ACTA2/α-SMA, but also completely
   abolished the effect of LOC344887 in increasing these proteins
   ([130]Fig. 5E and F). Thus, PI3K-AKT signaling appears to play a key
   role in the mediating the effect of LOC344887 on the expression of
   CDH2, FN1, COL1A1, and ACTA2/α-SMA.

Fig. 5.

   [131]Fig. 5
   [132]Open in a new tab

   LOC344887 negatively regulates the expression of CDH2/N-Cad via the
   PI3K-AKT signaling pathway. (A–B) A549 LOC344887-WT and LOC344887-KO
   cells were harvested for immunoblot analysis (A, quantification in B).
   (C–D) HFL1 cells transfected with LOC-siRNA-#5/#8 (10 nM) for 72 h were
   harvested for immunoblot analysis (C, quantification in D). (E–F) HFL1
   cells transfected with LOC-siRNA-#5 (10 nM) and treated with LY294002
   (30 μM) for 72 h were harvested for immunoblot analysis (E,
   quantification in F).

4. Discussion

   IPF is a progressive and irreversible lung disease with an average life
   expectancy of only 3–5 years after diagnosis in untreated patients
   [[133]34,[134]35]. The current drugs used to treat pulmonary fibrosis
   slow down the progression of the disease and preserve lung function,
   but do not reverse or halt its progression. In addition, patients must
   balance the limited lung preserving effects and toxic side-effects
   associated with current drug regimes. Therefore, the development of
   better antifibrotic therapies, especially targeting early stages of
   pulmonary fibrosis, represents an unmet medical need. Targeting
   activation of the NRF2 pathway for IPF prevention or intervention has
   been investigated, and the therapeutic potential of NRF2 activators
   against pulmonary fibrosis has been demonstrated [[135][15], [136][16],
   [137][17], [138][18],[139]20]. Also, the NRF2 activator SFN has
   recently been used in a phase II clinical trial to treat chronic
   obstructive pulmonary disease ([140]NCT01335971).

   Mechanistically, inhibition of EMT and TGF-β1/Smad signaling has been
   considered the major mechanism for the antifibrotic function of NRF2
   [[141]15,[142]18,[143]20]. In addition, treatment with SFN has been
   shown to suppress TGF-β1-induced EMT via decreased SNAIL protein
   levels, with SNAIL knockdown also abolishing the ability of SFN to
   reduce EMT and fibrosis [[144]36]. These studies clearly indicate that
   NRF2 negatively regulates fibrogenic gene expression; however, which
   set of NRF2 targets is primarily responsible for reduced fibrogenesis
   is unclear. For example, it is unknown if the observed inhibition of
   EMT, compromised TGF-β1/Smad signaling, and decreased SNAIL expression
   in response to NRF2 activation is driven by antioxidant or
   anti-inflammatory NRF2 target genes. Interestingly, our study
   identified LOC344887 as an NRF2 target gene that was largely
   responsible for SFN-mediated antifibrotic activity in vitro. LOC344887
   inhibited the expression of CDH2/N-cad via PI3K-AKT signaling pathway,
   which is essential for the progression of fibrosis, i.e. excess
   production of FN1, COL1A1, and ACTA2/α-SMA. This work provides a novel
   molecular insight into the antifibrotic function of NRF2 and highlights
   the importance of a specific NRF2 target gene, LOC344887, in
   suppressing pulmonary fibrosis.

   LncRNAs have emerged as key regulators of many cellular processes,
   including chromatin remodeling, scaffolding, transcriptional and
   translational regulation, splicing modulation, and RNA degradation
   [[145]37]. Proper function of lncRNAs has been demonstrated as a key
   mechanism controlling cellular homeostasis and is critical in the
   development of organisms [[146]37]. Dysregulation of lncRNAs has been
   associated with many human diseases [[147]37]. In this study, we
   demonstrated that a lncRNA, LOC344887, is an anti-fibrotic NRF2 target
   gene with two functional AREs ([148]Fig. 1) that plays an important
   role in mediating fibrosis based on RNA-seq data analysis ([149]Fig.
   2). Functional analysis confirmed that LOC344887 negatively regulates
   fibrogenic gene expression ([150]Fig. 3, [151]Fig. 4). Importantly, the
   effects of SFN in reducing the expression of FN1, COL1A1, and
   ACTA2/α-SMA was largely abolished when LOC344887 was knocked down by
   siRNA ([152]Fig. 3), indicating the significant contribution of this
   newly-discovered NRF2 target gene in controlling the fibrogenic
   process. At this point, it remains unclear if the antifibrogenic
   function of LOC344887 involves the modulation of antioxidant or
   anti-inflammatory pathways; however, the role of the NRF2-LOC344887
   axis in mediating at least some aspects of the fibrotic stress response
   is clear. Another important finding of this study is that LOC344887
   regulates the expression of these fibrotic genes via inactivation of
   PI3K-AKT signaling ([153]Fig. 5). It is reasonable to assume an
   indirect effect of LOC344887 on the PI3K-AKT pathways, considering the
   wide spectrum of roles that lncRNAs play in regulating cellular
   signaling networks. The detailed molecular events that link LOC344887
   upregulation to dephosphorylation of AKT, thus blunting PI3K-AKT
   signaling, warrant further investigation.

   The role of LOC344887 in cancer development has been investigated
   previously. The expression of LOC344887 is upregulated in
   hepatocellular carcinoma compared with normal tissue [[154]38]. In
   other studies, LOC344887 might function as an oncogene in gallbladder
   cancer and colon cancer, as high expression of LOC344887 is associated
   with larger tumor size and the downregulation of LOC344887 inhibited
   cell proliferation in gallbladder cancer cells and colon cancer cells
   [[155]39,[156]40]. These observations are in line with the “dark side”
   of NRF2 in cancer, i.e. certain cancer cells have constitutively high
   NRF2 levels, which increases growth rate and leads to resistance to
   anti-cancer drug treatment. Clinically, high expression of NRF2 in
   patient tumors is strongly correlated with a poor prognosis. NRF2 is
   highly expressed in many cancer types including: lung, head and neck,
   and bladder, due to somatic mutations in KEAP1 or NRF2 [[157]41].
   Furthermore, KEAP1 is mutated in lung cancer as frequently (>30%) as
   the tumor suppressor gene TP53 [[158]12,[159]42,[160]43]. It would be
   interesting to dissect the degree of contribution of LOC344887 to the
   observed NRF2-dependent resistance and high proliferation in these
   cancer cells.

   Recent recognition that a large fraction of the noncoding genome that
   was previously considered “junk” DNA actually encodes many lncRNAs,
   coupled with an increasing understanding of the diverse roles of
   IncRNAs in human disease, has created exciting new therapeutic
   opportunities [[161]37]. Our study demonstrating that the antifibrotic
   activity of NRF2 is largely derived from this new NRF2 target gene
   LOC344887, provided an important new avenue for developing nucleic
   acid-based drugs to suppress fibrogenesis at early stages. However,
   studying LOC344887 in vivo has proved to be a challenge because
   LOC344887 is only found in the human genome, and a similar gene has not
   yet been identified in mice. If the mouse counterpart of the LOC344887
   gene cannot be identified, a humanized mouse model might be desirable
   to study the in vivo function of LOC344887 in the future. At this
   point, we do not know if LOC344887 is expressed in a tissue-specific
   manner. If LOC344887 is also expressed in liver tissues, it would be
   interesting to investigate if LOC344887 also negatively regulates liver
   fibrosis. In addition, our study also indicates an effect of LOC344887
   on the PI3K-AKT signaling pathway. Therefore, other diseases related
   with PI3K-AKT signaling, such as ischemia/reperfusion injury [[162]44]
   and diabetes [[163]45], could also be affected by LOC344887, along with
   cancer and fibrosis.

5. Conclusion

   In this study, we have identified the lncRNA LOC344887 as an NRF2
   target gene with two functional AREs in the promoter and the first
   intron. GO annotation and KEGG pathway enrichment implied an
   antifibrotic function of LOC344887. Using lung fibroblasts, we
   demonstrated that the antifibrotic function of NRF2 was largely
   dependent on LOC344887, since deletion or downregulation of LOC344887
   enhanced expression of CDH2/N-cadherin and fibrotic genes and blunted
   the antifibrotic effects of SFN. Furthermore, the LOC344887-mediated
   downregulation of CDH2/N-cadherin and fibrotic genes was connected to
   the PI3K-AKT signaling pathway, as pharmacologic inhibition of PI3K
   activity blocked the effects of LOC344887 knockdown. Although the
   detailed mechanism and function of NRF2-LOC344887 signaling in
   pulmonary fibrosis requires further investigation, our study
   illustrates the distinct function of this axis in suppressing fibrosis,
   and opens a new therapeutic avenue for treating human pulmonary
   fibrosis or other fibrotic diseases.

Declaration of competing interest

   The authors declare that they have no known competing financial
   interests or personal relationships that could have appeared to
   influence the work reported in this paper.

Acknowledgments