DCAF1-targeting microRNA-3175 activates Nrf2 signaling and inhibits dexamethasone-induced oxidative injury in human osteoblasts

Activation of nuclear-factor-E2-related factor 2 (Nrf2) signaling can protect human osteoblasts from dexamethasone-induced oxidative injury. DDB1 and CUL4 associated factor 1 (DCAF1) is a novel ubiquitin E3 ligase for Nrf2 protein degradation. We identified a novel DCAF1-targeting miRNA, miR-3175. RNA pull-down, Argonaute 2 RNA-immunoprecipitation, and RNA fluorescent in situ hybridization results confirmed a direct binding between miR-3175 and DCAF1 mRNA in primary human osteoblasts. DCAF1 3′-untranslated region luciferase activity and its expression were significantly decreased after miR-3175 overexpression but were augmented with miR-3175 inhibition in human osteoblasts and hFOB1.19 osteoblastic cells. miR-3175 overexpression activated Nrf2 signaling, causing Nrf2 protein stabilization, antioxidant response (ARE) activity increase, and transcription activation of Nrf2-dependent genes in human osteoblasts and hFOB1.19 cells. Furthermore, dexamethasone-induced oxidative injury and apoptosis were largely attenuated by miR-3175 overexpression in human osteoblasts and hFOB1.19 cells. Importantly, shRNA-induced silencing or CRISPR/Cas9-mediated Nrf2 knockout abolished miR-3175 overexpression-induced osteoblast cytoprotection against dexamethasone. Conversely, DFAC1 knockout, by the CRISPR/Cas9 method, activated the Nrf2 cascade and inhibited dexamethasone-induced cytotoxicity in hFOB1.19 cells. Importantly, miR-3175 expression was decreased in necrotic femoral head tissues of dexamethasone-taking patients, where DCAF1 mRNA was upregulated. Together, silencing DCAF1 by miR-3175 activated Nrf2 signaling to inhibit dexamethasone-induced oxidative injury and apoptosis in human osteoblasts.


INTRODUCTION
Dexamethasone (DEX) is commonly utilized in patients suffering from inflammatory and autoimmune diseases [1]. Yet, long-term or high-dose use of DEX would lead to direct cytotoxicity to osteoblasts, serving as one main cause of osteoporosis and/or osteonecrosis [2,3]. Our group and others have been exploring the underlying pathological mechanisms of DEX-caused osteoblast injury [4][5][6][7][8][9].
MicroRNAs (miRNAs) are small (~22 nucleotides long) and conserved noncoding RNAs (ncRNAs), regulating gene expression by directly binding to the 3′-untranslated regions (3′-UTRs) of target mRNAs [18,19]. They can lead to translation inhibition and/ or degradation of targeted mRNAs [18,19]. miRNAs can activate the Nrf2 signaling cascade via silencing their target genes [20][21][22], thereby protecting osteoblasts from DEX. A recent study using the RNAi screening method has discovered the ubiquitin E3 ligase DDB1 and CUL4 associated factor 1 (DCAF1) as a noncanonical regulator of Nrf2 stability and signaling [23]. There is a direct interaction between Nrf2 and DCAF1, the latter can promote Nrf2 ubiquitination and degradation [23]. In the present study, we identified a novel DCAF1-targeting miRNA, microRNA-3175 (miR-3175). Our results showed that DCAF1 silencing by miR-3175 activated Nrf2 signaling to ameliorate DEX-induced oxidative injury and apoptosis in human osteoblasts.

Cell culture
The primary human osteoblasts were provided by Dr. Ji's group at Nanjing Medical University [20,21]. The human osteoblasts were differentiated and cultured as previously described [20,21], and were used at passage-3 to passage-12. The hFOB1.19 osteoblastic cells were provided by Dr. Ji as well and were cultured in an FBS-containing DMEM medium. Mycoplasma/ microbial contamination examination was routinely performed. STR profiling, population doubling time, and morphology were monitored regularly to verify the genotype. The protocols of the current study were approved by the Ethics Committee of Hunan Provincial People's Hospital, according to the principles of the Declaration of Helsinki.

Quantitative real-time polymerase chain reaction (qRT-PCR)
TRIzol reagents were added to extract total RNA, which was reversetranscribed to cDNA. qRT-PCR assays were carried out through an SYBR Green PCR kit (Applied Biosystems, Shanghai, China) under the ABI-7900HT Real-Time PCR System (Applied Biosystems). A 2 -ΔΔ Ct method was utilized for data quantification. GAPDH was always examined as the internal control and the reference gene. For detection of miR-3175 expression, a PrimeScript miRNA qRT-PCR Kit (Takara, Tokyo, Japan) was utilized, with U6 RNA examined as the internal control. The mRNA primers for the Keap1-Nrf2 cascade genes were provided by Dr. Liu at Jiangsu University [24]. Other verified and specific primers were synthesized by Genechem (Shanghai, China).

Forced overexpression or inhibition of miR-3175
The miR-3175 precursor sequence ("pre-miR-3175", Genechem) or the antisense sequence ("antagomR-3175") was inserted into a GV-369 lentiviral construct (Genechem). The construct, along with the lentivirus Helper plasmids (psPAX2 and pMD2.G), were co-transfected to HEK-293T cells. Virus in culture supernatants were enriched and filtered and was added to the primary human osteoblasts or hFOB1.19 osteoblastic cells (cultured in polybrene-containing medium). Puromycin was added to selected stable cells. Expression of mature miR-3175 in stable cells was verified by qRT-PCR assays. The lentiviral microRNA control construct ("Vec") or the microRNA anti-sense control ("antaC") were utilized as controls.

RNA pull-down
A Pierce Magnetic RNA pull-down Kit [25,26] was utilized for RNA pulldown assays. In short, the biotinylated miR-3175 mimic (Genechem) or control mimic (Genechem) were transfected to primary human osteoblasts for 24 h by Lipofectamine 3000 (Invitrogen, Shanghai, China). Cell lysates (600 μg protein lysates per treatment) were incubated with streptavidincoated magnetic beads to pull down the biotin-captured RNA complex [25]. Expression of miR-3175-associated DCAF1 mRNA was tested by qRT-PCR assay, with its level normalized to the "Input" control.

RNA fluorescent in situ hybridization (FISH)
Fluorescent in situ hybridization (FISH) kit (RiboBio, Guangzhou, China) was employed for RNA-FISH experiments. Briefly, the FITC (green fluorescence)labeled miR-3175 probe and the Cy3 (red fluorescence)-labeled DCAF1 mRNA probe were cotransduced to primary human osteoblasts (at 37°C for 48 h). Cells were rinsed and observed under a fluorescence microscope (Leica, Shanghai, China).

RNA-immunoprecipitation (RNA-IP)
Lysates from the primary human osteoblasts (800 μg lysates per treatment) were precleared and incubated with magnetic beads conjugated with the anti-Argonaute 2 antibody (anti-Ago2, Santa Cruz Biotech). After 12 h, beads were washed and incubated with Proteinase K. Thereafter, qRT-PCR was performed to test the purified RNAs (including miR-3175 and DCAF1 mRNA), and their levels normalized to the "Input" control.

ARE luciferase reporter assay
Primary human osteoblasts or hFOB1.19 osteoblastic cells were seeded into six-well plates at 60-70% confluence and were transfected with the ARE-inducible firefly luciferase vector (from Dr. Jiang at Nanjing Medical University [27]). Cells were subjected to applied genetic modifications, and cell lysates were examined with the ARE luciferase reporter assay under a luminescence machine [27]. DCAF1 3′-UTR luciferase activity assay A pGL4.13 (luc2/SV40) construct encoding DCAF1 3′-UTR was synthesized by Genechem (Shanghai, China). The construct was transfected to primary human osteoblasts or hFOB1.19 osteoblastic cells by Lipofectamine 3000 (Invitrogen, Shanghai, China) together with the Renillaluciferase reporter vector and pRL-SV40 (from Dr. Jiang at Nanjing Medical University [27]). DCAF1 3′-UTR luciferase reporter activity in human osteoblasts with applied genetic modifications was tested by a Promega kit [28].

UTR-null DCAF1 expression
The GV-369 lentiviral construct encoding the 3′-UTR-null DCAF1 was provided by Genechem (Shanghai, China) and was sequence-verified. The primary human osteoblasts were seeded into six-well plates at 60% confluence and were transduced with construct via Lipofectamine 3000. Stable cells were selected by puromycin. The expression of DCAF1 was verified by a Western blotting assay.

Western blotting
Briefly, cell lysates were extracted, quantified (30 μg per treatment in each lane), and electro-transferred to 10% SDS-PAGE gels. The lysate proteins were then transferred to PVDF blots. After blocking, the blots were incubated with designated primary and corresponding secondary antibodies. To visualize signaling of antigen-antibody binding, the enhanced chemiluminescence (ECL) reagents (Sigma) were applied. An ImageJ software (downloaded from the NIH website) was used for data quantification.

Single-stranded DNA (ssDNA) ELISA
Human osteoblasts or hFOB1.19 osteoblastic cells were seeded into 96-well plates (at 4500 cells per well). Following treatment, a ssDNA ELISA kit (Roche Diagnostics, Shanghai, China) was utilized to quantify ssDNA contents. The ssDNA ELISA absorbance was tested at 450 nm in each well.

Reactive oxygen species (ROS) detection
Human osteoblasts or hFOB1.19 osteoblastic cells were seeded into sixwell plates (at 100,000 cells per well). Following treatment, cells were stained with 10 μM of CellROX (Invitrogen, Shanghai, China). After 30 min, cells were washed and CellROX red fluorescence intensity was tested under a Fluoroskan Ascent fluorescence spectrofluorometer. CellROX fluorescence images were presented as well.

Other cell functional assays
The protocols of other cell functional assays, including CCK-8 assaying of cell viability, caspase-3 activity assay, and thiobarbituric acid reactants (TBAR) activity assaying of lipid peroxidation, as well as nuclear TUNEL staining and Annexin V FACS, were described in detail in our previous studies and elsewhere [17,27,30,31].

Nrf2 silencing
To hFOB1.19 osteoblastic cells, the Nrf2 shRNA lentiviral particles (Santa Cruz Biotech, Santa Cruz, CA) were added. Stable cells were then selected by puromycin. Nrf2 silencing was verified by qRT-PCR assays.

Human tissues
As described in ref. [30], human necrotic femoral head tissues and matched surrounding normal femoral head tissues were from a total of 20 (n = 20) written-informed consent DEX-treated patients. All patients underwent femoral head resection. miR-3175 and DCAF1 mRNA expression in fresh tissues was examined by qRT-PCR assays. Protocols were in according to the principles of the Declaration of Helsinki, with approval from the Ethics Committee of Hunan Provincial People's Hospital.

Statistical analyses
Data were presented as mean ± standard deviation (SD). For comparison of multiple groups, statistical differences were analyzed by one-way ANOVA (multiple comparisons) with post hoc Bonferroni test (SPSS version 21.0). The unpaired t-test (Excel, 2007) was utilized to compare the significance between two treatment groups. P < 0.05 was considered statistically significant.

RESULTS
microRNA-3175 binds to and silences DCAF1 in human osteoblasts miRNA binds through complementary base pairing to the 3′-UTR of target mRNA, causing its translational blockage and/or degradation [32,33]. Computational prediction tools provide a rapid method to identify the putative miRNAs that could bind directly to target mRNA [34]. The miRNA database TargetScan (V7.2) was consulted to explore possible miRNAs targeting 3′-UTR of DCAF1. Other miRNA databases, including miRBase, miRNAmap, and miRTarbase, were searched as well to verify the retrieved miRNAs. Three candidate DCAF1-targeting miRNAs were retrieved, miR-876-3p, miR-3175, and miR-1236-3p. The context score percentage was >98% and the context ++ score was < −0.4 [34]. Each of the three miRNA mimics was individually transfected to primary human osteoblasts, their efficiency on DCAF1 silencing was examined. Bioinformatics studies and preexperimental results identified that miR-3175 potentially targets DCAF1's 3′-UTR (at position 828-835) (Fig. 1A). The binding context score percentage of miR-3175-DCAF1 3′-UTR is 98% and the context ++ score at −0.45 (TargetScan V7.2 [34], Fig. 1A) [34]. Argonaute 2 (Ago2) is a primary component of RNA-induced silencing complex (RISC) that facilitates miRNA binding to its target mRNA. It also promotes the cleavage of the target mRNAs by its endonuclease activity [32,33]. Ago2 RNA immunoprecipitation (RNA-IP) and subsequent expression analysis of Ago2immunoprecipitated RNA allows to identify mRNA transcripts enriched in the IP fraction as miRNA targets [32,33]. As shown in Fig. 1B, the endogenous DCAF1 mRNA and miR-3175 both coimmunoprecipitated with the anti-Ago2 antibody in human osteoblasts. The nonspecific anti-IgG antibody did not associate DCAF1 mRNA and miR-3175 (Fig. 1B). The RNA pull-down assay results, Fig. 1C, demonstrated that the biotinylated miR-3175 could pull down endogenous DCAF1 mRNA in human osteoblasts (Fig. 1C). Moreover, RNA-FISH results showed that miR-3175 (in green fluorescence) co-localized with DCAF1 mRNA (in red fluorescence) mainly in the cytosol of human osteoblasts (Fig. 1D). Thus, RNA-IP, RNA pull-down, and RNA-FISH assays implied that miR-3175 can directly bind to DCAF1 mRNA in primary human osteoblasts.

DISCUSSION
The function and the potential targets of miR-3175 are still largely unknown. Qi et al. reported that miR-3175 acted as a tumorsuppressive miRNA and was downregulated in human glioma. Transfection of miR-3175 mimics induced proliferation inhibition and apoptosis in glioma cells possibly by inhibiting the PI3K-Akt pathway [35]. Xia et al. found that LncRNA LINC00520 sponged miR-3175 to promote lung cancer cell growth [36]. Han et al. however reported cancer-promoting activity of miR-3175, as it silenced the tumor suppressor HOXB1 in human glioma [37]. miR-3175 downregulation inhibited cell proliferation and invasion and promoted apoptosis in glioma cells [37]. Zhong et al., also reported that miR-3175 silenced Smad7 to promote epithelialmesenchymal transition (EMT) in human conjunctiva and pterygium [38].
The results of the present study indicated that miR-3175 is a novel DCAF1-targeting miRNA in human osteoblasts. RNA pull-down, Fig. 2 microRNA-3175 activates Nrf2 signaling in human osteoblasts. The primary human osteoblasts were stably transduced with a lentiviral construct encoding the miR-3175 precursor ("OE-miR-3175-L1/L2", two lines) or the empty vector ("Vec"), listed mRNAs were tested by qRT-PCR assays (A and E). Expression of the listed proteins in cytosol lysates and nuclear lysates was tested by Western blotting assays (B, C, and F), with the relative ARE activity tested as well (D); The primary human osteoblasts were transfected with the applied miR-3175 mimic (wildtype/WT or mutants, listed in G) or nonsense control miR mimic ("miR-C"), 500 nM for 48 h, the DCAF1 3′-UTR luciferase activity (H), DCAF1 mRNA (I), and listed proteins (J) expression were tested, with the relative ARE activity examined as well (K). OE-miR-3175-L1 human osteoblasts were further transduced with or without a UTR-depleted DCAF1 expression construct ("UTR-null-DCAF1"), control human osteoblasts were with the empty vector ("Vec"); Expression of listed proteins was shown (L); The relative ARE activity (M) and miR-3175 expression levels (N) were presented. The hFOB1.19 osteoblastic cells, stably expressing a lentiviral construct encoding the miR-3175 precursor ("OE-miR-3175") or empty vector ("Vec"), were established; Expression of listed proteins (O) and mRNAs (P and Q) were tested by Western blotting and qRT-PCR assays, respectively. The expression of listed proteins was quantified and normalized to the loading control. Data were presented as mean ± standard deviation (SD, n = 5). *P < 0.05 versus "Vec" cells (D, E, and Q); *P < 0.05 versus "miR-C" cells (H, I, and K); # P < 0.05 (M); "n. s." stands for nonstatistical difference (A, H, K, N, and P). Experiments were repeated five times, with similar results obtained. osteoblastic cells (E and F), stably expressing the lentiviral construct encoding the miR-3175 precursor ("OE-miR-3175") or empty vector ("Vec"), were established and treated with dexamethasone (DEX, 2 μM) or vehicle control ("Veh"); Cells were further cultured for 16 h; Cellular ROS contents, lipid peroxidation intensity, mitochondrial depolarization, and DNA damage were tested by CellROX staining (A and E), TBAR activity (B), JC-1 monomers intensity (C and F), and single-strand DNA (ssDNA) ELISA (D) assays, respectively, and results quantified. Data were presented as mean ± standard deviation (SD, n = 5). *P < 0.05 versus "Veh" treatment in "Vec" cells. # P < 0.05 versus "DEX" treatment in "Vec" cells. Experiments were repeated five times, with similar results obtained. Scale bar = 100 μm (A).
Ago2 RNA-IP, and RNA-FISH experiment results all supported a direct binding between miR-3175 and DCAF1 mRNA in primary human osteoblasts. In human osteoblasts and hFOB1.19 osteoblastic cells, forced overexpression of miR-3175 inhibited DCAF1 3′-UTR luciferase activity and downregulated DCAF1 expression. Conversely, miR-3175 silencing by antagomiR-3175 increased DCAF1 expression. Importantly, the mutant miR-3175 mimics, containing mutations at the proposed binding sites to DCAF1 3′-UTR, failed to alter DCAF1 expression in human osteoblasts.
We further showed that silencing of DCAF1 by miR-3175 activated Nrf2 signaling in human osteoblasts and hFOB1.19 cells, causing Nrf2 protein stabilization, ARE activity increase, transcriptional activation of Nrf2-dependent genes (HO1, NQO1, and GCLC). Yet transfection of the two mutant miR-3175 mimics failed to activate Nrf2 signaling in human osteoblasts. Moreover, restoring DCAF1 expression, by the UTR-null-DCAF1 construct, abolished miR-3175 overexpressioninduced Nrf2 cascade activation in human osteoblasts. These results supported that DCAF1 silencing is the key mechanism responsible for Nrf2 cascade activation by miR-3175 (Fig. 6F).
Different miRNAs could protect osteoblasts/osteoblastic cells from DEX-induced oxidative injury and cell death. For example, Zhao et al. reported that Keap1 silencing by miR-200a activated Nrf2 signaling to protect osteoblastic cells from DEX [20]. In human osteoblasts miR-19a silenced tuberous sclerosis complex 1 (TSC1) to activate the mTORC1-dependent Nrf2 signaling cascade, protecting human osteoblasts from DEX-induced oxidative injury and cell death [21]. Inhibition of miR-107, a CAB39 (calciumbinding protein 39)-targeting microRNA, activated AMPKdependent Nrf2 signaling cascade to protect human osteoblasts from DEX [22]. stably expressing a lentiviral construct encoding the miR-3175 precursor ("OE-miR-3175") or empty vector ("Vec"), were established and treated with dexamethasone (DEX, 2 μM) or vehicle control ("Veh"); Cells were further cultured for applied time periods; Cell viability was tested by CCK-8 assays (A and F); The caspase-3 activity (B and G) and caspase-3-PARP cleavages (C) were examined. Cell apoptosis was tested by recording TUNEL-positive nuclei ratio (D and H) and Annexin V percentage (E and I), with results quantified. The nuclear TUNEL ratio, % vs. DAPI, from at least 1500 cells in five random views per treatment was calculated. Data were presented as mean ± standard deviation (SD, n = 5). *P < 0.05 versus "Veh" treatment in "Vec" cells. # P < 0.05 versus "DEX" treatment in "Vec" cells. Experiments were repeated five times, with similar results obtained. Scale bar = 100 μm (D).
We found that activation of Nrf2 signaling by miR-3175 potently inhibited DEX-induced oxidative injury in human osteoblasts. DEXinduced ROS production, lipid peroxidation, mitochondrial depolarization, and DNA damage were all largely attenuated in miR-3175-overexpressed human osteoblasts. Furthermore, forced overexpression of miR-3175 largely inhibited DEX-induced cytotoxicity and apoptosis in human osteoblasts. Importantly, we found that activation of the Nrf2 cascade was required for miR-3175-induced osteoblast cytoprotection against DEX. In hFOB1.19 cells, Nrf2 shRNA or KO abolished miR-3175 overexpressioninduced osteoblast cytoprotection. Conversely, DFAC1 KO activated the Nrf2 cascade and inhibited DEX-induced cytotoxicity in hFOB1.19 cells. Significantly, miR-3175 overexpression failed to further increase Nrf2 cascade activation nor inhibiting DEXinduced cytotoxicity in DCAF1 KO hFOB1.19 cells. These results clearly demonstrated that DCAF1 silencing by miR-3175 activated Nrf2 signaling to inhibit DEX-induced oxidative injury and death in human osteoblasts (Fig. 6E).