Optimization of Biotinylated RNA or DNA Pull-Down Assays for Detection of Binding Proteins: Examples of IRP1, IRP2, HuR, AUF1, and Nrf2

Investigation of RNA- and DNA-binding proteins to a defined regulatory sequence, such as an AU-rich RNA and a DNA enhancer element, is important for understanding gene regulation through their interactions. For in vitro binding studies, an electrophoretic mobility shift assay (EMSA) was widely used in the past. In line with the trend toward using non-radioactive materials in various bioassays, end-labeled biotinylated RNA and DNA oligonucleotides can be more practical probes to study protein–RNA and protein–DNA interactions; thereby, the binding complexes can be pulled down with streptavidin-conjugated resins and identified by Western blotting. However, setting up RNA and DNA pull-down assays with biotinylated probes in optimum protein binding conditions remains challenging. Here, we demonstrate the step-by step optimization of pull-down for IRP (iron-responsive-element-binding protein) with a 5′-biotinylated stem-loop IRE (iron-responsive element) RNA, HuR, and AUF1 with an AU-rich RNA element and Nrf2 binding to an antioxidant-responsive element (ARE) enhancer in the human ferritin H gene. This study was designed to address key technical questions in RNA and DNA pull-down assays: (1) how much RNA and DNA probes we should use; (2) what binding buffer and cell lysis buffer we can use; (3) how to verify the specific interaction; (4) what streptavidin resin (agarose or magnetic beads) works; and (5) what Western blotting results we can expect from varying to optimum conditions. We anticipate that our optimized pull-down conditions can be applicable to other RNA- and DNA-binding proteins along with emerging non-coding small RNA-binding proteins for their in vitro characterization.


Introduction
Understanding the regulatory mechanisms of gene expression is vital for deciphering cellular responses to a plethora of external cues and physiological conditions. Gene expression is regulated at the transcriptional, post-transcriptional, and translational levels as a result of changes in interactions between cis-acting regulatory elements of DNA and RNA with specific trans-acting proteins [1][2][3]. These core mechanisms of gene expression are also subject to epigenetic regulation, such as DNA methylation and histone modifications as well as modulations by micro RNAs (miRNAs), long non-coding RNAs (lncRNAs), and circular RNAs (circRNAs) [4][5][6]. In the past, the initial in vitro characterization of interactions between defined cis-acting elements and their binding proteins was carried out using electrophoretic mobility shift assays (EMSAs), in which 32 P-end-labeled RNA or DNA harboring the defined sequence of interest was incubated with cell lysates, giving rise to the retarded migration of the radiolabeled RNA or DNA probe in non-denatured polyacrylamide gel electrophoresis upon the binding of proteins to the probe [7]. Along with the development of the non-radioactive detection of molecular interactions in bioluminescence resonance energy transfer (BRET) [8], force spectroscopy (FS) [9], molecular recognition imaging (MRI) [10], fluorescence cross-correlation spectroscopy (FCCS) [11],

The IRP-IRE Pull-Down Set-Up and Comparison between Streptavidin-Agarose and Magnetic Beads
The binding of IRP1 and IRP2 to IRE has been assessed in vitro traditionally by EMSA using a 32 P end-labeled IRE RNA probe [39] and also recently probed with a fluorescentlabeled IRE RNA [13]. The identification of binding proteins in EMSA usually needs further characterization through a specific antibody-mediated inhibition or retardation of the protein binding complex [34], while RNA pull-down assays allow the identification of targeted proteins in the binding complex by Western blotting [14][15][16] or more unbiased proteomic approaches [40,41]. To optimize and establish the radioisotope-free and userfriendly IRP-IRE binding assay, we attempted to set up an RNA pull-down assay using 5 -biotinylated 32nt human ferritin H IRE as a probe (wild-type and mutant IRE RNA oligonucleotides, Sigma-Aldrich, St. Louis, MO, USA, Figure 1A). First, as magnetic beads are much easier for washing and the subsequent recovery of precipitates, we compared streptavidin magnetic beads to high-capacity streptavidin agarose resin. An amount of 2 µg of the biotinylated IRE RNA (wt) was incubated with whole-cell lysates (WCL) from SW480 human colon carcinoma cells in magnetic beads (MBs) binding buffer (for MBs) or binding buffer A (for agarose) at room temperature for 1 h, followed by a further 1 h incubation with 2.5-20 µL suspensions of streptavidin magnetic beads or 20 µL of streptavidin agarose ( Figure 1B). Both beads were pre-washed once with modified RIPA buffer (see washing buffer in table in Section 4) and resuspended in each binding buffer prior to the incubation. The pull-down samples were subjected to IRP2 Western blotting. More details on the pulldown procedures and reagents can be found in the Materials and Methods section, each figure legend, and table in Section 4. Unexpectedly, we experienced very poor pull-down of the IRP2-IRE complex with streptavidin-magnetic beads (2.5-20 µL) in the MB binding buffer recommended by NEB (20 mM Tris-HCl pH 7.5, 0.5 M NaCl, 1 mM EDTA) while 20 µL streptavidin-agarose in the binding buffer A (20 mM Tris, pH 7.4, 300 mM KCl, 1.5 mM MgCl 2 , 0.2 mM EDTA, 0.5 mM PMSF) pulled down the IRP2-IRE complex much more efficiently ( Figure 1B).
We tested the possibility that the failure of IRP2-IRE pull-down with streptavidin magnetic beads might be due to the binding buffer difference. The magnetic beads failed to pull down the IRP2-IRE binding complex in the MB buffer again ( Figure 1C). Streptavidinagarose in the MB buffer also lost the pull-down capability compared to the same amount (20 µL) of streptavidin-agarose in binding buffer A ( Figure 1C). If the MB buffer would be the general problem for the pull-down of the IRP-IRE binding complex, we anticipated that the streptavidin-magnetic beads in buffer A or buffer C we used for EMSA may improve the pull-down efficiency. However, none of these buffers improved pull-down with streptavidin-magnetic beads even when twice the amount of magnetic beads (40 µL) was used ( Figure 1D, lanes a-f). In contrast, streptavidin-agarose (20 µL) in buffer A as well as buffer C precipitated the IRP2-IRE binding complex ( Figure 1D, lanes g and h). As another streptavidin-magnetic beads in buffer A or C failed to pull down the IRP2-IRE complex (see Dynabeads M280 data in Figure 2D), we concluded that our experimental conditions were not optimum for the magnetic beads we tested in this IRE pull-down application (see Section 3). We therefore focused on high-capacity streptavidin-agarose (ThermoFisher Scientific, Waltham, MA, USA) for the further optimization of RNA and DNA pull-down assays. Figure 1. The IRP-IRE pull-down assays and comparison between streptavidin-agarose and -magnetic beads. (A) Expression of the human ferritin H gene is regulated at least three cis-acting elements. The wild type (wt) and mutant (mt) IRE (iron responsive element) RNA sequences used for pull-down probes are shown. They were 5'-end biotinylated [Btn]. (B) 500μg of SW480 WCLs (60 μL in RIPA buffer) and 2 μg (equal to ~200 pmol) of biotinylated wt IRE RNA oligonucleotide were incubated in 140 μL of binding buffer A at room temperature for 1 h with constant rotation. 0-20 μL of streptavidin magnetic beads (NEB) or 20 μL of high capacity streptavidin agarose (ThermoFisher Scientific) were washed once with washing buffer, resuspended in 50 μL of binding buffer A, added to the IRE probe/WCLs mixture, and further incubated for 1 hr. The magnetic beads were collected using a magnetic stand, and agarose resins were precipitated by micro-centrifugation at 5000 rpm for 0.5 min. They were washed twice with 1ml of washing buffer. 12 μL of 2xSDS-PAGE sample buffer was added to precipitated resins, vortexed briefly, and heated at 95 °C for 10 min. After brief spinning, the samples were loaded on 10% SDS-PAGE gel and subjected to IRP2 western blotting. WCLs of HEK293 cells transfected with empty vector, IRP2, or IRP1 expression plasmid was loaded to verify the specificity of anti-IRP2 antibody (B-D,F). (C) lanes a-e: 500 μg of SW480 WCLs (50 μL in IP lysis buffer) and 2 μg of biotinylated wt IRE RNA oligonucleotide were incubated in 150 μL of magnetic beads binding buffer (MB, lanes a-d) or binding buffer A (lane e), followed by incubation and pull-down with 0-20 uL of pre-washed streptavidin magnetic beads (lanes a-d) or 20 μL of high capacity streptavidin agarose (lane e). lanes f, g: 500 μg of SW480 WCLs (60 μL in RIPA buffer) and 2 μg of biotinylated wt IRE RNA oligonucleotide were incubated in 140 μL of magnetic beads binding buffer (MB, lane f) or binding buffer A (lane g), pull-down with 20 μL of high capacity streptavidin agarose (lane g), and IRP2 western blotting. (D) 500μg of SW480 WCLs (50 μL in IP lysis buffer) and 2 μg of biotinylated wt IRE RNA oligonucleotide were incubated in 150 μL of magnetic beads binding buffer (MB, lanes a and d), binding buffer A (lanes b, e, and g), or binding buffer C (lanes c, f, and h) followed by pull-down with 20 μL or 40 μL of pre-washed streptavidin magnetic beads (lanes a-f) or 20 μL of high capacity streptavidin agarose (lane g and h), and IRP2 western blotting. (E) 250μg of K562 WCLs (20 μL in IP lysis buffer) and 2 μg of biotinylated wt IRE RNA oligonucleotide were incubated in 180 μL of binding buffer A, and pull-down with 0-30 uL of pre-washed high capacity streptavidin agarose and IRP2 western blotting. (F) 500μg of SW480 WCLs (130 μL in RIPA buffer) and 2 μg of biotinylated wt or mt IRE RNA oligonucleotide were incubated in 370 μL of binding buffer A (total 500 μL) in the absence of presence of 2 μg and 8 μg of non-biotinylated wt and mt RNA oligonucleotide competitors. The binding complex was pulled down with 20 μL of pre-washed high capacity streptavidin agarose and IRP2 western blotting. All experiments were repeated 2-3 times and the representative western blots are shown. Figure 1. The IRP-IRE pull-down assays and comparison between streptavidin-agarose and -magnetic beads. (A) Expression of the human ferritin H gene is regulated at least three cis-acting elements. The wild type (wt) and mutant (mt) IRE (iron responsive element) RNA sequences used for pull-down probes are shown. They were 5'-end biotinylated [Btn]. (B) 500 µg of SW480 WCLs (60 µL in RIPA buffer) and 2 µg (equal to~200 pmol) of biotinylated wt IRE RNA oligonucleotide were incubated in 140 µL of binding buffer A at room temperature for 1 h with constant rotation. 0-20 µL of streptavidin magnetic beads (NEB) or 20 µL of high capacity streptavidin agarose (ThermoFisher Scientific) were washed once with washing buffer, resuspended in 50 µL of binding buffer A, added to the IRE probe/WCLs mixture, and further incubated for 1 hr. The magnetic beads were collected using a magnetic stand, and agarose resins were precipitated by micro-centrifugation at 5000 rpm for 0.5 min. They were washed twice with 1 mL of washing buffer. 12 µL of 2xSDS-PAGE sample buffer was added to precipitated resins, vortexed briefly, and heated at 95 • C for 10 min. After brief spinning, the samples were loaded on 10% SDS-PAGE gel and subjected to IRP2 western blotting. WCLs of HEK293 cells transfected with empty vector, IRP2, or IRP1 expression plasmid was loaded to verify the specificity of anti-IRP2 antibody (B-D,F). (C) lanes a-e: 500 µg of SW480 WCLs (50 µL in IP lysis buffer) and 2 µg of biotinylated wt IRE RNA oligonucleotide were incubated in 150 µL of magnetic beads binding buffer (MB, lanes a-d) or binding buffer A (lane e), followed by incubation and pull-down with 0-20 uL of pre-washed streptavidin magnetic beads (lanes a-d) or 20 µL of high capacity streptavidin agarose (lane e). lanes f, g: 500 µg of SW480 WCLs (60 µL in RIPA buffer) and 2 µg of biotinylated wt IRE RNA oligonucleotide were incubated in 140 µL of magnetic beads binding buffer (MB, lane f) or binding buffer A (lane g), pull-down with 20 µL of high capacity streptavidin agarose (lane g), and IRP2 western blotting. (D) 500 µg of SW480 WCLs (50 µL in IP lysis buffer) and 2 µg of biotinylated wt IRE RNA oligonucleotide were incubated in 150 µL of magnetic beads binding buffer (MB, lanes a and d), binding buffer A (lanes b, e, and g), or binding buffer C (lanes c, f, and h) followed by pull-down with 20 µL or 40 µL of pre-washed streptavidin magnetic beads (lanes a-f) or 20 µL of high capacity streptavidin agarose (lane g and h), and IRP2 western blotting. (E) 250 µg of K562 WCLs (20 µL in IP lysis buffer) and 2 µg of biotinylated wt IRE RNA oligonucleotide were incubated in 180 µL of binding buffer A, and pull-down with 0-30 uL of pre-washed high capacity streptavidin agarose and IRP2 western blotting. (F) 500 µg of SW480 WCLs (130 µL in RIPA buffer) and 2 µg of biotinylated wt or mt IRE RNA oligonucleotide were incubated in 370 µL of binding buffer A (total 500 µL) in the absence of presence of 2 µg and 8 µg of non-biotinylated wt and mt RNA oligonucleotide competitors. The binding complex was pulled down with 20 µL of pre-washed high capacity streptavidin agarose and IRP2 western blotting. All experiments were repeated 2-3 times and the representative western blots are shown. buffer A, followed by pull-down with 20 μL of high-capacity streptavidin agarose shown in Figure 2A, 1 μg of the biotinylated IRE probe was sufficient for the bindin IRP2 as well as IRP1, and the increase in the probe up to 6 μg had no additional pull-do effects ( Figure 2A). We also observed the marginal effect by switching binding buffe to buffer C with 4 μg of the IRE probe ( Figure 2A). Similar results were obtained w 500 μg of K562 WCL was used ( Figure 2B). Based on the results in Figure 2A,B, we u 2 μg (200 pmol) of a biotinylated RNA probe for the rest of the experiments.  An amount of 20 µL of high-capacity streptavidin-agarose (binding capacity: >10 µg biotinylated BSA/µL resin) was used for 500 µg of WCLs to pull down the IRP2-IRE complex in Figure 1B-D. However, to assess the appropriate amount of streptavidin-agarose for IRE pull-down assays, we tested varying amounts (5-30 µL) of high-capacity streptavidinagarose to precipitate the IRP2-IRE complex. The IRP2 Western blot in Figure 1E showed that even a 5 µL suspension of high-capacity streptavidin-agarose was sufficient and equivalent to a 20 or 30 µL suspension to pull down the complex. As the packed precipitate of a 5 µL suspension of streptavidin agarose was too small to be easily handled during wash without loss of resins, we decided to use 20 µL of streptavidin-agarose for the rest of the pull-down assays.
The next question was whether this IRP2-IRE interaction is specific or not. To address this important issue, we used a mutated IRE RNA ( Figure 1A) as a probe and a competitor as well. The biotinylated wtIRE reproducibly precipitated IRP2 whereas the mutant IRE failed to pull down IRP2 ( Figure 1F). Furthermore, IPR2 bound to biotinylated wtIRE was competed out with non-biotinylated wtIRE RNA in a 4-fold excess condition, while the non-biotinylated mutant IRE failed in the same competition condition ( Figure 1F). These results suggest that our pull-down set-up detects the sequence-specific interaction between the biotinylated IRE probe and IRP2.
The next important questions were (1) what amount of biotinylated IRE RNA is sufficient and optimum to pull down the IRP-IRE complex and (2) whether this pulldown assay is semi-quantitative. To address the first question, we tested 1, 2, 4, and 6 µg (100-600 pmol) of biotinylated wtIRE probe for 500 µg of SW480 WCL in the 200 µL binding buffer A, followed by pull-down with 20 µL of high-capacity streptavidin agarose. As shown in Figure 2A, 1 µg of the biotinylated IRE probe was sufficient for the binding of IRP2 as well as IRP1, and the increase in the probe up to 6 µg had no additional pull-down effects ( Figure 2A). We also observed the marginal effect by switching binding buffer A to buffer C with 4 µg of the IRE probe ( Figure 2A). Similar results were obtained when 500 µg of K562 WCL was used ( Figure 2B). Based on the results in Figure 2A,B, we used 2 µg (200 pmol) of a biotinylated RNA probe for the rest of the experiments.
To address the second question of whether the pull-down assay is semi-quantitative or not, we incubated 100, 250, and 500 µg (each duplicate) of WCLs from K562 and SW480 cells with 2 µg of a biotinylated IRE probe for 1 h followed by another 1 h incubation with 20 µL of high-capacity streptavidin agarose to pull down the IRE binding complex. As shown in Figure 2C, we observed that increasing amounts of input WCLs gave rise to a more intensified IRP2 band. IRP1 also showed a similar trend but the IRP1 bands from 250-100 µg of K562 WCLs were too weak to be visualized ( Figure 2C). Taken together, we concluded that our RNA pull-down assay using 2 µg of the biotinylated IRE probe and 20 µL of high-capacity streptavidin agarose is semi-quantitative enough to measure the binding of IRP2 and IRP1 to IRE.
To corroborate our conclusion in the IRE-IRP pull-down set-up, we prepared WCLs from K562 cells treated with 100 µM FAC (ferric ammonium citrate) or 25 µM iron chelator DFO (deferoxiamine mesylate) for 24 h and tested whether we could detect changes in the binding of IRP2 and IRP1 to the IRE under high-and low-iron conditions. We also tested another streptavidin-agarose (Invitrogen 15492-050, Waltham, MA, USA, equivalent to Life Technologies SA100-04 that binds 3-8 µg of biotinylated IgG per µL suspension) and another type of magnetic bead, Dynabeads M-280 (Invitrogen). As shown in Figure 2D, IRP2 binding to IRE was decreased following FAC treatment while it increased after iron chelator DFO treatment. The increased IRP1 binding to IRE was also observed in DFO-treated cells whereas the effect of FAC was marginal. Like NEB magnetic beads ( Figure 1B-D), we failed to pull down the IRP-IRE complex with 30 µL of Dynabeads M-280 ( Figure 2D).

Detection of Binding Proteins to the 3 -UTR AU-Rich Element in the Human Ferritin H mRNA
Little is known about binding proteins to the 3 -UTR of ferritin mRNA, partly due to the significant roles of the translational regulation via the IRP-IRE interaction in ferritin expression [42][43][44] along with the transcriptional regulation through the antioxidantresponsive element (ARE) [23,34,45]. It was shown by EMSA that U-rich sequences in the 3 -UTR of the human ferritin H mRNA bind unknown protein(s) that seemingly play a role in the post-transcriptional regulation of ferritin H in PMA-treated human monocytic THP1 cells [32]. We also reported that the expression of ferritin H is regulated via mRNA stability in response to calcium elevation [33]. There are at least four putative AU-rich elements in the human ferritin H mRNA (NM_002032), which harbor the core AU-rich sequence U(U/A)(U/A)UUU(U/A)(U/A)U [46]. We used one of them, located at nt1026 to nt1053 in NM_002032, as a biotinylated RNA probe ( Figure 3A) for pull-down assays. Since AU-rich RNA sequences are known to bind several binding proteins including HuR and AUF1 [47], we applied our IRP-IRE pull-down conditions for pull-down assays of these RNA-binding proteins. Indeed, the AU-rich element in the ferritin H mRNA bound HuR and AUF1 in three human cell lines ( Figure 3B). This interaction seemed to be specific to the AU-rich sequence because the pull-down of HuR was failed when the mutant probe with the impaired AU-rich sequence (6Us to 6Cs, Figure 3A) was used as a probe ( Figure 3C). We concluded that our RNA pull-down conditions can be used not only for IRPs but also AU-rich element-binding proteins, and that the expression of ferritin H may be subject to mRNA stability regulation through AU-rich elements and their binding proteins including HuR and AUF1. The location and RNA sequences for one of putative AU-rich elements studied in pull-down assays. The RNA probes were 5'-end biotinylated and the mutated AU-rich sequence is also shown. (B) 75μg of WCLs from HepG2, HaCaT, and HEK293 cells were incubated with 4 μg of the 5'-biotinylated wt AU-rich RNA probe and 20 μL of high capacity streptavidin agarose all together for 3 h at room temperature. 20 μg of WCLs were also loaded for input proteins. Western blotting for HuR was done first (12.5% acrylamide gel), followed by incubation with anti-AUF1 antibody. As indicated, some HuR bands came back on the AUF1 western blot. (C) 200 μg of WCLs from HEK293 and K562 cells, and 500 μg of the cytoplasmic fraction from HEK293 cells were incubated with 4 μg of the 5'-biotinylated wt or mt AU-rich RNA probe and 20 μL of high capacity streptavidin agarose all together followed by pull-down and HuR western blotting. 30 μg of the WCLs and cytoplasm were loaded for input of HuR. All experiments were repeated 2-3 times and the representative results are shown.

DNA Pull-Down for Detection of Nrf2 Bound to the Ferritin H Antioxidant-Responsive Element
In addition to the post-transcriptional regulation through these cis-acting RNA elements, both human and mouse ferritin H genes are transcriptionally regulated via the antioxidant-responsive element (ARE) enhancer [23,45], to which Nrf2 and other b-zip family members bind under chemical and oxidative stress conditions [24,25,34]. Our previous in vitro characterization of the ferritin H ARE-binding proteins was carried out by EMSA [34]. In this study, we applied the RNA pull-down procedures to DNA pull-down assays for the detection of Nrf2 bound to the ferritin H ARE. As shown in Figure 4A, the human ferritin H ARE is composed of a 22 bp AP1-like element (ARE1) and 23 bp AP1/NFE2 element (ARE2) separated by a 20 bp spacer [23,34]. Nrf2 was detected by Western blotting in 20 μg of the cytoplasmic and nuclear fractions of K562 cells treated with 25 μM sodium arsenite (NaAsO2) or 25 μM t-BHQ (tert-butyl hydroquinone) for 14 h ( Figure 4B, input). To assess the binding of the activated Nrf2 to the AREs, 4 μg of annealed 5′-biotinylated sense and antisense strand oligonucleotides of ARE1 and ARE2 (2 μg each of the strand) were incubated with 100 μg of the K562 cytoplasmic and nuclear fractions. In this binding reaction, we simultaneously added all together with 30 μL of streptavidin-agarose (Invitrogen) in 200 μL of PBS (phosphate-buffered saline) as a binding buffer and incubated it with constant rotation at room temperature for 2 h. After spinning down the streptavidin-agarose and wash with PBS, ARE-binding proteins were The location and RNA sequences for one of putative AU-rich elements studied in pull-down assays. The RNA probes were 5'-end biotinylated and the mutated AU-rich sequence is also shown. (B) 75 µg of WCLs from HepG2, HaCaT, and HEK293 cells were incubated with 4 µg of the 5'-biotinylated wt AU-rich RNA probe and 20 µL of high capacity streptavidin agarose all together for 3 h at room temperature. 20 µg of WCLs were also loaded for input proteins. Western blotting for HuR was done first (12.5% acrylamide gel), followed by incubation with anti-AUF1 antibody. As indicated, some HuR bands came back on the AUF1 western blot. (C) 200 µg of WCLs from HEK293 and K562 cells, and 500 µg of the cytoplasmic fraction from HEK293 cells were incubated with 4 µg of the 5'-biotinylated wt or mt AU-rich RNA probe and 20 µL of high capacity streptavidin agarose all together followed by pull-down and HuR western blotting. 30 µg of the WCLs and cytoplasm were loaded for input of HuR. All experiments were repeated 2-3 times and the representative results are shown.

DNA Pull-Down for Detection of Nrf2 Bound to the Ferritin H Antioxidant-Responsive Element
In addition to the post-transcriptional regulation through these cis-acting RNA elements, both human and mouse ferritin H genes are transcriptionally regulated via the antioxidant-responsive element (ARE) enhancer [23,45], to which Nrf2 and other b-zip family members bind under chemical and oxidative stress conditions [24,25,34]. Our previous in vitro characterization of the ferritin H ARE-binding proteins was carried out by EMSA [34]. In this study, we applied the RNA pull-down procedures to DNA pull-down assays for the detection of Nrf2 bound to the ferritin H ARE. As shown in Figure 4A, the human ferritin H ARE is composed of a 22 bp AP1-like element (ARE1) and 23 bp AP1/NFE2 element (ARE2) separated by a 20 bp spacer [23,34]. Nrf2 was detected by Western blotting in 20 µg of the cytoplasmic and nuclear fractions of K562 cells treated with 25 µM sodium arsenite (NaAsO 2 ) or 25 µM t-BHQ (tert-butyl hydroquinone) for 14 h (Figure 4B, input). To assess the binding of the activated Nrf2 to the AREs, 4 µg of annealed 5 -biotinylated sense and antisense strand oligonucleotides of ARE1 and ARE2 (2 µg each of the strand) were incubated with 100 µg of the K562 cytoplasmic and nuclear fractions. In this binding reaction, we simultaneously added all together with 30 µL of streptavidin-agarose (Invitrogen) in 200 µL of PBS (phosphate-buffered saline) as a binding buffer and incubated it with constant rotation at room temperature for 2 h. After spinning down the streptavidin-agarose and wash with PBS, ARE-binding proteins were eluted into 10 µL of 2xSDS-PAGE sample buffer by heating at 90 • C for 5 min and subjected to Western blotting for the detection of Nrf2. Consistently, this pull-down assay showed that both arsenic and t-BHQ treatments increased the binding of the nuclear Nrf2 to the ARE1 and ARE2 ( Figure 4B). Not only PBS as a binding buffer but also buffer C, used in our EMSA [34] and the IRE-IRP pull-down in Figure 2A,B, worked well in the ARE1, ARE2, and complete ARE (65 bp) pull-down assays ( Figure 4C). eluted into 10 μL of 2xSDS-PAGE sample buffer by heating at 90 °C for 5 min and subjected to Western blotting for the detection of Nrf2. Consistently, this pull-down assay showed that both arsenic and t-BHQ treatments increased the binding of the nuclear Nrf2 to the ARE1 and ARE2 ( Figure 4B). Not only PBS as a binding buffer but also buffer C, used in our EMSA [34] and the IRE-IRP pull-down in Figure 2A,B, worked well in the ARE1, ARE2, and complete ARE (65 bp) pull-down assays ( Figure 4C).

Discussion
We demonstrated here how to set up biotinylated IRE and AU-rich RNA pull-down assays for IRP2, IRP1, HuR, and AUF1 in addition to biotinylated ARE DNA pull-down for Nrf2. In the IRE-IRP pull-down assay, we used 2 μg (200 pmol) of a 5′-biotinylated IRE RNA probe for 500 μg of SW480 or K562 WCL in a total 200 μL of binding buffer A or C, along with incubation with constant rotation for 1 h at room temperature, followed by another 1 h incubation with 20 μL of pre-washed high-capacity streptavidin agarose

Discussion
We demonstrated here how to set up biotinylated IRE and AU-rich RNA pull-down assays for IRP2, IRP1, HuR, and AUF1 in addition to biotinylated ARE DNA pull-down for Nrf2. In the IRE-IRP pull-down assay, we used 2 µg (200 pmol) of a 5 -biotinylated IRE RNA probe for 500 µg of SW480 or K562 WCL in a total 200 µL of binding buffer A or C, along with incubation with constant rotation for 1 h at room temperature, fol-lowed by another 1 h incubation with 20 µL of pre-washed high-capacity streptavidin agarose (binding capacity: >10 µg biotinylated BSA/µL resin, ThermoFisher Scientific). Alternatively, 30 µL of streptavidin agarose (binding capacity: 3-8 µg of biotinylated IgG per µL suspension, Invitrogen) worked well to pull down the biotinylated IRE-IRP binding complex. We noticed no clear difference in pull-down efficiency between binding buffer A and buffer C ( Figures 1D and 2A and the buffer compositions in table in Section 4). PBS was recommended for a binding buffer of high-capacity streptavidin agarose in batch analyses (ThermoFisher Scientific, also suggested to add 0.1 % SDS, 1% NP40, or 0.5% sodium deoxycholate to reduce non-specific binding). Although we did not try to use PBS for our RNA pull-down assays, note that PBS worked nicely in biotinylated DNA pull-down assays for the detection of Nrf2 binding to ARE (Figure 4). The successive incubation for 1 h each of a biotinylated RNA or DNA probe with WCL followed by the addition of streptavidin agarose may not be necessary because mixing the biotinylated probe, WCL, and streptavidin agarose all together for 1-2 h successfully pulled down the binding complexes of IRP-IRE ( Figure 2D), HuR and AUF1 with an AU-rich RNA ( Figure 3B,C), and Nrf2-ARE ( Figure 4B,C). Furthermore, RNase and protease inhibitors, such as 10 mM of ribonucleotide vanadyl complex (NEB S1402S) and 1× protease inhibitor cocktail set I (Millipore Calbiochem 539131), can be added to the binding buffer; however, we did not observe the merit during 1-2 h incubation at room temperature for the pulldown of IRPs, HuR, AUF1, and Nrf2 as shown in this study. If the degradation of proteins would be an issue, we may be able to shorten the incubation time to 15-30 min as we did in EMSA [23,48]. We also noticed that there is no clear difference in the pull-down efficiency of WCLs prepared in IP lysis buffer or RIPA buffer (see figure legends for Figure 1). However, we recommend that the volume of WCL should not exceed~20% (40 µL) of the final binding reaction volume (usually 200 µL in binding buffer A or C). If necessary due to the lower protein concentration of WCLs, we recommend just to increase the total binding reaction volume from 200 µL to 500 µL without increasing other components as the biotinylated RNA probe and streptavidin agarose are still sufficient ( Figures 1E and 2A,B).
To confirm the specificity of IRE-IRP interactions, we tested a 5 -biotinylated mutant IRE probe for binding to IRPs, in addition to a pull-down competition assay by adding 4-fold in excess of non-biotinylated wt and mutant IRE competitors ( Figure 1F). Like EMSA, the inclusion of non-biotinylated wt and mutant competitors is important for the verification of their specific interactions between a probe and binding proteins in the pull-down assays.
We intended that the conditions optimized in this work can be used for RNA and DNA pull-down assays and detection by Western blotting to test (1) whether the binding protein predicted from the defined nucleotide sequences is present or not in the binding complex and (2) whether the protein binding is increased or decreased according to gene expression changes in the same experimental conditions. Protein purities in the pull-down samples are therefore not critical issues for such targeted pull-down assays detected by Western blotting with a specific antibody. However, we also tested the effect of pre-clearing WCLs with a biotinylated mutant IRE and streptavidin-agarose on the protein precipitates subsequently pulled down with a biotinylated wild-type IRE RNA probe. The results showed that the pull-down with a biotinylated IRE RNA probe per se (no pre-clear) can remove the majority of non-interacting proteins (compare lane 1 to 1/25 of input WCL in the Supplemental Figure S1). Pre-clearing WCLs with a biotinylated mutant IRE or/and streptavidin agarose did not improve the efficiency of IRP2 pull-down and the purity of the pull-down protein samples (Supplemental Figure S1).
We are not sure about the reason for the inability of the magnetic beads to pull down the IRP-IRE binding complex in our assays; however, some key procedures including the biotinylation of an RNA probe and/or binding conditions of the probe with magnetic beads and WCLs may need to be further optimized or modified. For instance, the Magnetic RNA-Protein Pull-Down Kit (ThermoFisher Pierce) uses a 3 -desthiobiotinylated RNA probe (RNA incubated with biotinylated cytidine bisphosphate and T4 RNA ligase) and incubates first with magnetic beads in RNA capture buffer (20 mM Tris pH 7.5, 1 M NaCl, and 1 mM EDTA), followed by incubation with WCL in protein-RNA binding buffer (20 mM Tris pH 7.5, 50 mM NaCl, 2 mM MgCl 2 , 0.1% Tween 20). As desthiobiotin binds streptavidin reversibly, HuR bound to an AU-rich RNA was gently eluted with 4 mM biotin-containing elution buffer [49]. Furthermore, it was reported that there are significant variations in the binding capacity of streptavidin magnetic beads not only from different vendors but also intralot numbers from the same vendor [50]. In addition, binding capacities of various streptavidin beads including those we used in this study were assayed for different proteins (not biotinylated RNA or DNA) [50]. This may make it more difficult to find the optimum of streptavidin beads when they are not working. Of note, using much more magnetic beads (100 µL of 10 mg/mL Dynabeads M-280) might be necessary for pull-down assays as recommended [51]; however, 100 µL of magnetic beads per sample is not practical for many laboratories due to the cost of magnetic beads per sample.
A growing number of studies on interactions between non-coding RNAs and macromolecules such as RNA, DNA, and proteins have been reported, and representative interactions with lncRNAs (long non-coding RNAs) are summarized by Kazimierczyk et.al [4]. For the characterization of proteins that specifically interact with lncRNA, mass spectrometry was employed after the pull-down of a biotinylated lncRNA and protein binding complex with streptavidin magnetic beads. For instance, using the Magnetic RNA-Protein Pull-Down Kit, the lncRNAs TPA [41], XIST [52], and RP3-326I-13.1 (PINCR) [53] were in vitro transcribed and desthiobiotin-labeled for pull-down with streptavidin magnetic beads and subjected to the identification of several interacting proteins such as HSP90B by mass spectrometry [41,53]. Similarly, ADAR1 (adenosine deaminase RNA specific 1) was identified as a binding protein to the immunosuppressive lncRNA LINC00624 that was in vitro transcribed in the presence of biotin-16-UTP and pulled down with Dynabeads C1 and mass spectrometry [40]. The LINC00624 binding to ADAR1 stabilized the ADAR1 protein that ultimately promotes tumor progression [40]. Proteins interacting with circular RNAs (circRNAs) produced during RNA splicing were also characterized by pull-down with a biotinylated probe and streptavidin magnetic beads [54].
Accumulating evidence has indicated that non-AGO-family RNA-binding proteins directly interact with specific microRNAs and thereby play important regulatory roles in gene expression and cell physiology [36] including cancer [37]. For instance, HuR was shown to block the miR-21-mediated translational repression of the tumor suppressor programmed cell death 4 (PDCD4) through two mechanisms: binding competition with miR-21 on the 3 -UTR of PDCD4 mRNA and direct interaction with miR-21 thereby working as a miR-21 sponge [55]. HuR was also shown to interact with miR-122 and facilitates the extracellular vesicle-mediated export of miR-122 in human liver cells under starvation [56]. The AUF1 p37 isoform was shown to bind miRNA let-7b and promote loading let-7b onto AGO2 for the enhanced repression of target mRNA [57]. In these studies, HuR, AUF1, and miRNA interactions were assessed or confirmed by EMSA [55][56][57]. We think that our biotinylated AU-rich RNA pull-down conditions for HuR and AUF1 can be applicable for biotinylated miRNA pull-down assays to assess miRNA interactions with HuR, AUF1, and other RNA-binding proteins.
In addition to biotinylation, various methods for RNA labeling to investigate RNAprotein interactions were summarized by Gemmill et al. [58] and further characterizations of RNA-protein interactions including RIP (RNA immunoprecipitation) and CLIP (crosslinking and immunoprecipitation) were reviewed by Barra and Leucci [59].
Conclusions and Future Perspectives: We determined the optimal condition for an IRP-IRE pull-down assay, in which 2 µg (200 pmol) of a 5 -biotinylated IRE RNA probe was used for 500 µg of WCL (either prepared in IP lysis buffer or RIPA) in a total of 200 µL of binding buffer A or binding buffer C. The incubation with constant rotation for 1 h at room temperature was followed by another 1 h incubation with 20 µL of pre-washed high-capacity streptavidin agarose or 30 µL of streptavidin agarose. Mixing all together and incubation for 1-2 h also worked well. To confirm the specificity of IRE-IRP interaction, the inclusion of 2 µg of a 5 -biotinylated mutant IRE probe compared to wt IRE probe in pull-down assays is important along with a competition assay by adding >4-fold in excess of non-biotinylated wt and mutant IRE competitors. This pull-down assay condition was applied for the detection of HuR and AUF1 bound to an AU-rich RNA element as well as Nrf2 bound to an ARE double-strand DNA enhancer element. The pull-down assays optimized in this work will save a fair amount of time for researchers who will need to establish and perform RNA and DNA pull-down assays for their initial characterization of binding proteins anticipated from their RNA and DNA sequences. We expect that this pull-down condition will also help or facilitate the characterization of macromolecules interacting with small non-coding RNAs that regulate the expression of target genes.

Cell Culture, Chemicals, and Buffer
All human cell lines (SW480, K562, HepG2, HaCaT, HEK293, Jurkat, and HL60) were cultured in a humidified 37 • C CO 2 incubator (5% CO 2 , MCO-17AIC, Sanyo, Osaka, Japan). The culture media and chemicals used in this work are listed in Table 1 along with suppliers, catalog numbers, and lot numbers if applicable. Ferric ammonium citrate (FAC) and deferoxamine mesylate (DFO) were dissolved in purified water (Millipore Z00QSV001, Milli-Q ® system, MilliporeSigma, Burlington, MA, USA) at 100 mM and 25 mM, respectively. Sodium arsenite (NaAsO 2 ) was dissolved in Milli-Q ® water at 10 mM. t-BHQ was freshly prepared, first dissolved in DMSO at 1 M and further diluted to 10 mM with Milli-Q ® water.

RNA and DNA Oligonucleotides
All biotinylated and non-biotinylated oligonucleotides were synthesized by Sigma-Aldrich (St. Louis, MO, USA). Desalt grade RNA oligonucleotides at 0.2 µmol scale yielded approximately 500-700 µg. We experienced that the synthesis scale increased to 1.0 µmol did not increase the yield of IRE RNA oligonucleotides. Desalt-grade DNA oligonucleotides at a 0.05 µmol scale yielded 200-500 µg. All RNA and DNA oligonucleotides were dissolved at 1 µg/µL in TE (10 mM Tris, pH 7.4 and 1 mM EDTA). 32nt IRE RNA oligos at 1 µg/µL is approximately 100 µM. DNA oligonucleotides were annealed by mixing 100 µL each of 1 µg/µL sense and antisense 5 -biotinylated ARE oligonucleotides together with 20 µL of NEBuffer 3 (50 mM Tris-HCl, pH 7.9, 10 mM MgCl 2 , 100 mM NaCl, 1 mM DTT, New England Biolabs, Ipswich, MA, USA) in a microcentrifuge tube, followed by incubation in 300-400 mL of microwaved 90 • C water in a beaker and leaving it at room temperature until gradually cooling down to room temperature. The sequences of human ferritin H RNA and DNA oligonucleotides used in this study (biotinylation at the 5 -end) are as follows:

RNA and DNA Pull-Down Procedure
For the preparation of whole-cell lysates (WCLs) from SW480 and other adherent cells, subconfluent to confluent cells were washed with phosphate-buffered saline (PBS: 137 mM NaCl, 27 mM KCl, 15 mM KH 2 PO 4 , 81 mM Na 2 HPO 4 ) and lysed in IP lysis buffer (25 mM Tris, pH 7.4, 150 mM NaCl, 1 mM EDTA, 1% NP40, and 5% glycerol) or RIPA buffer (25 mM Tris, pH 7.4, 150 mM NaCl, 1% NP40, 0.5% sodium deoxycholate, and 0.1% SDS). There was no difference in our results of RNA-pull down assays when WCL was prepared in IP lysis buffer or RIPA buffer. K562 and other suspension cells (0.5-1 × 10 6 cells/mL) were centrifuged at 1000 rpm, and cell pellets were washed with PBS and lysed in IP lysis buffer. Protein concentrations in WCLs were 5-25 µg/µL, measured with a protein assay dye reagent (#5000006, BIO-RAD, Hercules, CA, USA). Except for the optimization of input probes and the assessment of semi-quantitative pull-down assays in Figures 1 and 2, 500 µg of WCLs were incubated (rotated) at room temperature for 1 h with 2 µg of a biotinylated RNA probe in 200 µL of binding buffer A or binding buffer C (20 mM Hepes, pH 7.4, 100 mM KCl, 0.5 mM EDTA, 1.5 mM MgCl 2 , 20% glycerol, 1 mM DTT). There was no difference in the IRP pull-down efficiency incubated in binding buffer A or C (Figure 2A,B). In the IRE competition assay ( Figure 1F), 2 µg (probe:competitor = 1:1) or 8 µg (probe:competitor = 1:4) of non-biotinylated wt or mt IRE oligonucleotide was added to the binding reaction. After the 1 h incubation for biotinylated RNA probe and protein interaction, 20 µL of high-capacity streptavidin-agarose (ThermoFisher Scientific, Waltham, MA, USA) or streptavidin agarose (Invitrogen, Waltham, MA, USA) was added to the binding reaction and further incubated for 1 h at room temperature. The binding reaction was terminated by centrifugation at 5000 rpm for 0.5 min and washing the resins with 1 mL of washing buffer twice. After the complete removal of the washing buffer, 12 µL of a 2xSDS-PAGE sample buffer (63 mM Tris, pH 6.8, 25% Glycerol, 2% SDS, 0.01% bromophenol blue, and 5% β-mercaptoethanol) was added to the resins, vortexed briefly, and heated at 95 • C for 10 min in a heating block dry bath (11-718-2, Fisher Scientific, Hampton, NH, USA) containing water.
For the ARE DNA pull-down assays, 50-75 µg of cytoplasmic and nuclear fractions was incubated (rotated) at room temperature for 1-2 h with 4 µg of annealed 5 -biotinylated ARE DNA in 200 µL of binding buffer C. The ARE DNA pull-down assay for the detection of Nrf2 was also performed in PBS as binding buffer ( Figure 4B). The rest of the procedure is the same as the RNA pull-down assays using streptavidin agarose resins.
In IRE-IRP RNA pull-down assays for testing streptavidin magnetic beads (NEB, Ipswich, MA, USA) in Figure 1, binding buffer A, C, or NEB magnetic beads binding buffer was used as indicated. Buffer C was used for the binding buffer during the incubation with Dynabeads M-280 in Figure 2D. After the binding step, NEB magnetic beads and Dynabeads M-280 were magnetized and washed twice with NEB magnetic beads buffer and Dynabeads buffer, respectively. After the complete removal of the washing buffer, 12-15 µL of a 2xSDS-PAGE sample buffer was added to the beads, briefly vortexed, and heated at 95 • C for 10 min for the elution of the protein binding complex.

Western Blotting
After briefly spinning down heated samples in the microcentrifuge tubes, binding complexes eluted in the 2xSDS-PAGE sample buffer were loaded on a 10% acrylamide SDS-PAGE (10% acrylamide, 0.3% bisacrylamide, 375 mM Tris, pH 8.8, 0.1% SDS) minigel (4 × 2.9 inches, 0.75 mM thickness), along with protein size markers (Precision Plus Protein Standards, BIO-RAD, Hercules, CA, USA). The stacking gel was 5% acrylamide, 0.1% bisacrylamide, 125 mM Tris, pH 6.8, and 0.1% SDS. The electrophoresis was run at 15 mA per mini-gel for approximately 1. . Incubation with a primary antibody was conducted at 4 • C overnight on a rocker platform (Speci-Mix, Barnstead Thermolyne, Dubuque, IA, USA), followed by washing with 0.1% Tween20/TBS (15 min, 3 times) and incubation with a secondary antibody at room temperature for 1.5 h. All primary and secondary antibodies were diluted with the same blocking solution (either 1% BSA or 5% skim milk in Tween20/TBS). The PVDF membrane was washed three times, incubated with ECL reagents (Clarity, BIO-RAD, Hercules, CA, USA), and immediately exposed to X-ray films (ProSignal Blotting Film 30-810 L, Genesee Scientific, San Diego, CA, USA). For incubation of the same membrane with another primary antibody, the membrane was soaked in the stripping solution (1.5% glycine, 0.1% SDS, 1% Tween 20, pH 2.2 adjusted with HCl) for 1 h on a rocker and the incubation was repeated with primary and secondary antibodies.
The HRP-conjugated secondary antibodies used in this study were either anti-rabbit IgG (AP132P, 500 µg/mL, MilliporeSigma, Burlington, MA, USA) or anti-mouse IgG (7076S, Cell Signaling Technology, Danvers, MA, USA) at 5000-fold dilution with the same blocking solution as used for the primary antibody.

Isolation of Cytoplasmic and Nuclear Fractions
For the DNA pull-down assay to detect Nrf2 binding to the ARE, we used cytoplasmic and nuclear fractions isolated from K562 cells. After treatment with NaAsO 2 or t-BHQ, cells were washed with PBS, resuspended in 200-500 µL hypotonic buffer (20 mM Tris, pH 7.4, 10 mM NaCl, 3 mM MgCl 2 ), and incubated for 15 min on ice. The 1/20 volume (e.g., 25 µL for 500 µL cell suspension) of 10% NP40 was added, vortexed for 10 s at a high setting, and the suspension was centrifuged at 12,000 rpm for 30 s (MX-160 microcentrifuge, TOMY TECH/AMUZA, San Diego, CA, USA). The supernatant was transferred to a microcentrifuge tube (cytoplasmic fraction). The nuclear pellet was suspended in 50-200 µL of the nuclear extraction buffer (100 mM Tris, pH 7.4, 2 mM Na 3 VO 4 , 100 mM NaCl, 1% Triton X-100, 1 mM EDTA, 10% glycerol, 1 mM EGTA, 0.1% SDS, 1 mM NaF, 0.5% deoxycholate, 20 mM Na 4 P 2 O 7 , and 1× proteinase inhibitor cocktail), vortexed for 10 s at high setting, and incubated on ice for more than 30 min. The nuclear pellet was vortexed for 30 s at a high setting (making bubbles is no problem) and centrifuged at 12,000 rpm for 10 min at 4 • C. The supernatant (nuclear fraction) was transferred to a new microcentrifuge tube. Protein concentrations in cytoplasmic and nuclear fractions were measured with the BIO-RAD protein assay dye reagent, usually yielding 5-15 µg/µL.
Funding: This work was supported in part by P30ES025128 from the National Institute of Environmental Health Sciences to the Center for Human Health and the Environment (CHHE). The content is solely the responsibility of the author and does not necessarily represent the official views of the National Institutes of Health.

Institutional Review Board Statement: Not applicable.
Informed Consent Statement: Not applicable. Data Availability Statement: All data sources presented in this work are directed at the corresponding author Y. Tsuji.