|Home | About | Journals | Submit | Contact Us | Français|
Heme oxygenase-1 (HO-1) enzyme plays a critical role in metabolizing the excess heme generated during hemolysis. Our previous studies suggested that during intravascular hemolysis the expression of HO-1 protein is not sufficient to reduce the oxidative burden of free heme in the vasculature. This led us to hypothesize that a post-translational mechanism of control exists for HO-1 expression. Micro-RNAs (miRNA) affect gene expression by post-transcriptional gene regulation of transcripts. We performed in silico analysis for the human HMOX1–3′ untranslated region (3′ UTR) and identified candidate miRNA binding sites. Two candidate miRNAs, miR-377 and miR-217, were cloned and co-transfected with a luciferase vector containing the human HMOX1-3′UTR region. The combination of miR-377 and miR-217 produced a 58% reduction in HMOX1–3′UTR luciferase reporter expression compared with controls. The same constructs were then used to assess how overexpression of miR-217 and miR-377 affected HO-1 levels after induction with hemin. Cells transfected with the combination of miR-377 and miR-217 exhibited no change in HMOX1 mRNA levels, but a significant reduction in HMOX1 (HO-1) protein expression and enzyme activity compared with non-transfected hemin-stimulated controls. Transfection with either miR-377 or miR-217 alone did not produce a significant decrease in HO-1 protein expression or enzyme activity. Knockdown of miR-217 and miR-377 in combination leads to up-regulation of HO-1 protein. Exposure to hemin induced a significant reduction in miR-217 expression and a trend toward decreased miR-377 expression in two different cells lines. In summary, these data suggests that the combination of miR-377 and miR-217 help regulate HO-1 protein expression in the presence of hemin.
During hemolysis, up to 20 μm of free heme can be released leading to the breakdown of heme detoxification systems and subsequent damage to lipids, proteins, and DNA primarily through the ability of heme to generate reactive oxygen species (1). Free heme is quite hydrophobic in nature and readily enters cell membranes and increases cellular susceptibility to oxidant-mediated killing (1). Heme oxygenase-1 (HO-1)2 is a 32-kDa microsomal/mitochondrial (2) enzyme, which oxidizes protoheme to biliverdin IXα in a three-step process, which requires oxygen and reducing equivalents from NADPH. In the process, this enzyme releases the antioxidant molecules carbon monoxide and biliverdin (2). Transcription of HO-1 is induced by a variety of agents, such as heme, oxidants, hypoxia, and cytokines and leads to induction of the enzyme and protection of tissues and cells against ischemia-reperfusion injury and oxidative stress (3, 4). As illustrated in patients and HMOX1−/− mice, HO-1 deficiency leads to oxidant-mediated injury, highlighting that the control and regulation of HO-1 expression is critical to protect cells from oxidative stress and damage (5,–7).
To date several hundred microRNA (miRNA) genes have been identified in the human genome and it is proposed that at least 50% of all protein-encoding genes are regulated by miRNA (8, 9). Mature miRNAs are ~21–22 nucleotides in size and affect post-translational expression of genes by interacting with complementary target sites within the 3′ untranslated region of the messenger RNA (mRNA) (8). The exact molecular mechanisms by which miRNAs mediate translational repression are still under intense study. However, the dogma is that most miRNAs control gene expression post-transcriptionally by regulating mRNA translation or stability in the cytoplasm (10).
Our lab and others have demonstrated that in sickle cell disease, HO-1 levels are elevated, particularly during hemolytic crisis (11,–13). However, exposure of animals or humans with elevated HO-1 expression to agents that are able to promote HO-1 transcription can lead to additional up-regulation of the protein, providing additional protection against oxidative stress and inflammation (11, 12, 14). These observations lead us to hypothesize that a post-translational mechanism may exist for HO-1 expression. Our results demonstrate that there are at least two mature miRNAs which interact with the HMOX1–3′UTR in regulating HO-1 protein expression and enzymatic activity.
BLOCK-iTTM Pol II miR RNAi Expression Vector Kit with EmGFP featuring an engineered pre-miRNA cloning site that is flanked on either side with sequences from murine miR-155 to allow proper processing of the engineered pre-miRNA sequence was used to create pcDNATM6.2-GW/miR-377, pcDNATM6.2-GW/miR-217, pcDNATM6.2-GW/scrmiR-377, and pcDNATM6.2-GW/scrmiR-217 constructs following the manufacturer's instructions (Invitrogen). Two complementary single-stranded DNA oligonucleotides encoding the miRNA of interest were designed for each miRNA as previously described (15) (Table 1). The sequences for the scrambled controls were generated using siRNA WizardTM version 3.1 (InvivoGen, San Diego, CA). The synthetic oligonucleotides (Table 1) were purchased (IDT, Coralville, IA) and annealed to generate double-stranded oligos. The double-stranded oligos were cloned into the linearized pcDNATM6.2-GW/±EmGFP-miR vector according to the manufacturer's directions. Highly purified DNA for transfections was isolated using the Wizard® Plus maxiprep system (Promega, Madison, WI).
A 3′ UTR Reporter Vector containing the 603-bp human HMOX1–3′UTR (chr22+:34114115- 34114805) was purchased from SwitchGear Genomics (Menlo Park, CA). This vector contains the human HMOX1–3′UTR fragment cloned into a multiple cloning site to produce a hybrid transcript containing the luciferase gene fused to the 3′ UTR of interest. Mutants of the HMOX1–3′UTR were generated using the QuikChange Lightning Multi Site-directed Mutagensis kit (Agilent Technologies, La Jolla, CA) using primers 5′-CCTTCAGCATCCTCAGTTCACTCAGCAGAGCCTGGAAGACA-3′ to mutate 3 base pairs within the miR-217 binding site and 5′-ACTCTGTTCCTGGCTCAGCCTAAAGTGCGGTATTTTTGTTGTGTTCTGTTG-3′ to mutate three base pairs within the miR-377 binding site. An empty luciferase reporter vector (pGL3, Promega) was used as a negative control. HEK 293 cells were cultured in an opaque collagen-coated 96-well plate and grown for 24 h in basal medium containing high glucose Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% heat-inactivated fetal bovine serum (FBS, Invitrogen), sodium bicarbonate, sodium pyruvate, penicillin G, streptomycin sulfate, and amphotericin B (Invitrogen). After 24 h the medium was changed to basal medium without antibiotics. Cells were transfected with 10–100 ng of reporter and miRNA expression constructs using DharmaFECT 2 (Dharmacon, ThermoFisher, Lafayette, CO) following the manufacturer's directions. After 24 h luciferase expression was analyzed following the manufacturer's protocol (Promega). Luciferase expression was detected on a Synergy HT Multi-Mode Microplate Reader (BioTek, Winooski, VT).
HEK 293 cells were cultured in basal media as described above. Primary human umbilical vein endothelial cells (HUVECs) were removed from human umbilical veins by exposure to dispase (0.2% for 20 h at 4 °C) and cultured in medium 199 (Invitrogen) containing 12.5% (v/v) FBS, supplemented with sodium pyruvate, penicillin G, streptomycin sulfate, amphotericin B, heparin (5 units/ml), l-glutamine, and EndoGro (Vec Tec, Albany, NY). HUVEC cells were used between passages 2 and 4 for all experiments. HEK 293 and HUVEC cells were treated with 10 μm hemin chloride (Calbiochem, EMD Chemicals, Gibbstown, MD) as previously described (16,–18). Prior to cell treatment, the cells were washed gently three times with Hanks' balanced salt solution to remove serum, and then treated with hemin. After 1 h the treatment medium was removed and replaced with growth medium. Cells were incubated for 4–24 h and collected for HO-1 enzyme activity, Western blot, or total RNA as described below.
Primary HUVEC (passage 1) were passaged in endothelial cell growth medium-2 (Lonza, Walkersville, MD). Two days later, nucleofection of HUVECs was performed according to the manufacturer's instructions (Amaxa/Lonza). Cells were resuspended in 0.1 μl of HUVEC Nucleofector® Solution (Lonza). and 0–20 nmol/liter of miRIDIAN miR-217 mimic (miR-217), 0–20 nmol/liter of miRIDIAN miR-377 mimic (miR-377), 20 nmol/liter of miRIDIAN miR-217 inhibitor, 20 nmol/liter of miRIDIAN miR-377 inhibitor, or 0–40 nmol/liter of control scramble sequences (Thermo Fischer Scientific/Dharmacon, Lafayette, CO) was added. Samples were transferred into certified cuvettes (Amaxa/Lonza) and transfected with program U-001. Fresh medium (500 μl) was added immediately after transfection to each cuvette, and the cells were plated and incubated at 37 °C until experiments were performed.
HO-1 activity was measured in freshly isolated microsomes collected from HEK 293 cells and sonicated for 10 s. Microsomes (2 mg) in 2 mm MgCl2, 0.1 m K2HPO4 buffer, pH 7.4, were added to the reaction mixture (400 μl, final) containing 2.5 μg of recombinant biliverdin reductase (Assay Designs), 2 mm glucose 6-phosphate, 0.2 units of glucose-6-phosphate dehydrogenase, 50 μm hemin chloride, and 0.8 mm NADPH (Calbiochem) for 1 h in the dark. The bilirubin formed was extracted into chloroform and the δA464–530 was measured (extinction coefficient, 40 mm−1 cm−1 for bilirubin). Heme oxygenase activity was calculated as picomole of bilirubin formed/mg of microsomal protein/h and normalized to the hemin-treated control to eliminate intra-assay variation.
Buffer A containing 0.1% Triton X-100 (Sigma), 300 mm NaCl, 1.5 mm MgCl2, 20 μm EDTA, 25 mm HEPES, pH 7.6, with the addition of protease and phosphatase inhibitors, 1 m dithiothreitol, 5 mm vanadate, 1 m β-glycerophosphate, 1 mg/ml of leupeptin, 100 mm phenylmethanesulfonyl fluoride in ethanol, and 1 μg/ml of protease inhibitor mixture (Sigma) was added to cell culture dishes and placed on a shaker at 4 °C for 15 min. A sterile cell scraper was used to scrap the cells off the plates and the lysate was transferred to a 1.5-ml microcentrifuge tube and spun at 4 °C for 15 min at 10,000 × g. The supernatant was then transferred to a separate tube and frozen at −80 °C until use.
Protein content of the supernatants was determined using Bio-Rad Protein Assay (Bio-Rad). An equal amount of protein was loaded per lane and subjected to electrophoresis on a 15% Tris-HCl gel (Bio-Rad). Afterward, the samples were transferred electrophoretically to PVDF membranes (Millipore), and immunoblotting of the protein homogenates was performed with rabbit polyclonal anti-HO-1 antiserum (Assay Designs, Ann Arbor, MI). Sites of primary antibody binding were visualized with alkaline phosphatase-conjugated goat anti-rabbit or rabbit anti-goat IgG (Santa Cruz Biotechnology, Santa Cruz, CA). The final detection of immunoreactive bands was performed using an ECF substrate (GE Healthcare) and visualized on a Storm Reader (GE Healthcare). All membranes were stripped using Restore® Stripping Buffer (Thermo Scientific), and reprobed with rabbit anti-GAPDH (Sigma). Densitometry of the bands was performed using Image J software (National Institutes of Health, Bethesda, MD).
TRIzol® reagent was added to cell culture as per the manufacturer's directions (Invitrogen) and stored at −80 °C for up to 1 month. Total RNA was isolated and washed with 100% ethanol to preserve small RNAs as previously described (19). The quality and quantity of total RNA were determined by UV spectrophotometer. Samples were stored at −80 °C until use.
An aliquot of 1 μg of total RNA from each sample was subjected to cDNA synthesis with a reverse transcriptase, a poly(A) polymerase, and tagged oligo(dT) primers in 20 μl of a reaction using for quantitative reverse-transcribed polymerase chain reaction (qRT-PCR) (miScript, Qiagen, Valencia, CA) according to the manufacturer's protocol. For miRNA expression the cDNA underwent qRT-PCR using a universal primer (miScript Universal Primer) and miRNA-specific primers (see Table 1) in an Mx3000P (Agilent, Stratagene, La Jolla, CA) to detect miRNA gene expression. Each qRT-PCR contained 2 μl of cDNA, 400 nm miRNA-specific primer, and 2× QuantiTect SYBR®Green QPCR master mix (Qiagen). The PCR conditions included activating the DNA polymerase at 95 °C for 15 min, followed by 40 cycles of three-step PCR (94 °C for 15 s, 55 °C for 30 s, and 70 °C for 30 s). For mRNA expression the cDNA underwent qRT-PCR for detection of mRNA gene expression. Each qRT-PCR contained 2 μl of cDNA, 400 nm of each sense and antisense primer, and 2× QuantiTect SYBR® Green QPCR master mix (Qiagen). PCR was carried out by activating the DNA polymerase at 95 °C for 5 min, followed by 40 cycles of two-step PCR (95 °C for 10 s and 60 °C for 30 s). To validate the efficiency of target and reference, serially diluted human umbilical vein cell cDNA samples were run on quantitative PCR side by side with experimental samples in each 96-well reaction plate. Melt curves for each primer set was run and verified. The cycle threshold (Ct) values from samples of each gene and the internal control U6 (Qiagen) or GAPDH were obtained and the relative quantification for each gene was calculated.
All statistical analyses were performed with SigmaStat 2.0 for Windows (SYSTAT Inc., Chicago, IL). Comparisons of multiple treatment groups to controls were made using one-way analysis of variance (ANOVA) with Tukey, Bonferroni, or Holm-Sidak post hoc analysis as indicated in the figure legends or under “Results.” Comparisons of single treatment groups to controls were made using Student's t test.
In silico analysis of the human HMOX1-3′ UTR began with surveys of several miRNA target-prediction sites, including miRBase (20), miRGator (21), and TargetScan Release 5.1 (22,–24). To investigate our hypothesis that miRNAs are involved with control of HO-1 protein expression we chose to validate two miRNAs, miR-217 and miR-377, which had low context scores in TargetScan and were also predicted by at least one other target-prediction site.
miR-217 is a miRNA located on chromosome 2p16.1 within intron 2 of the non-protein coding transcript AC011306.2 in close proximity to miR-216. The mature sequence (MIMAT000274) is predicted to interact with an 8-mer seed match with the HMOX1–3′ UTR from position 462–468 (Fig. 1). miR-377 is located within a miRNA cluster on chromosome 14q32.2. The stem-loop for miR-377 contains two different mature miRNA sequences, miR-377 (MIMAT0000730) from positions 45–66 and miR-377* (MIMAT0004689) from positions 7–28. miR-377 is predicted to interact with an 8-mer seed match with the HMOX1–3′ UTR from position 576 to 582 (Fig. 1). Synthetic nucleotides of the miRNAs were designed and used to generate overexpression constructs that feature an engineered pre-miRNA cloning site to allow proper processing of the engineered pre-miRNA sequence. Using these constructs as well as commercially purchased mimics we can overexpress miR-377 and miR-217 in vitro (supplemental Fig. S1).
A luciferase reporter containing the human HMOX1–3′ UTR was used to determine whether the candidate miRNAs interacted specifically with the HMOX1–3′UTR (Fig. 2). We assessed luciferase activity 24 h after transfection of HEK 293 cells with the reporter and miRNA overexpression constructs. In this assay system a decrease in luciferase expression indicates a specific miRNA-3′ UTR interaction. As Fig. 2 demonstrates, overexpression of miR-377, but not miR-217, resulted in a 37.5% decrease in luciferase expression (n = 5, p < 0.05). This result was not seen when the scrambled versions of miR-377 or miR-217 or a combination of scrmiR-217 and scrmiR-377 were expressed in the presence of the HMOX1–3′ UTR. To determine whether the overexpression constructs were interacting with their predicted binding sites, mutations in the miR-217 and miR-377 binding sites were made in the HMOX1–3′ UTR reporter vector. As demonstrated in Fig. 2, mutating either of these sites eliminated the reduction in luciferase expression observed in Fig. 2. Interestingly, when both miR-377 and miR-217 were transfected, there was a 58.45% decrease (n = 5, p < 0.001), even though luciferase expression was not significantly decreased with miR-217 expression alone (Fig. 2). This indicates that miR-377, and perhaps the combination of miR-377 and miR-217 are involved in the control of HO-1 expression.
Most mammalian cells induce HO-1 upon stimulation with hemin. We verified that HEK 293 cells increase HO-1 message and protein after stimulation with hemin in a time- and dose-dependent manner (data not shown). Therefore, this cell line was used to evaluate the extent to which miR-377 and miR-217 can affect HO-1 message and protein expression. We evaluated the levels of HO-1 mRNA in HEK 293 cells transfected with both miR-217 and miR-377 with and without the presence of hemin. There was no change in HO-1 mRNA expression (Fig. 3A) with the overexpression miRNA constructs, even in the presence of hemin. miRNAs interact with an mRNA after incorporation into a silencing complex and can specify the post-transcriptional repression of that protein-coding message by transcript destabilization, translational repression, or a combination (24). The results indicate that the miRNA-mRNA interaction between miR-217, miR-377, and HO-1 may lead to translation repression and not transcript destabilization.
We next assessed the effect of miR-377 and miR-217 expression on HO-1 protein levels. As lane 5 and the corresponding bar graph in Fig. 3B demonstrate, hemin-treated cells transfected with the combination of miR-217 and miR-377 show an almost 20% reduction in HO-1 protein expression compared with hemin-treated or hemin-treated scrambled miRNA-transfected cells (p < 0.05, n = 3 independent trials). In primary HUVEC cells the expression of a combination of miR-217 and miR-377 mimics lead to a 46.5% reduction in HO-1 protein expression compared with hemin-treated cells (p < 0.05, n = 3 independent trials). However, there was no significant change in HO-1 protein expression when either miR-377 or miR-217 was overexpressed alone in HEK 293 cells (supplemental Fig. S2), but there was a significant (p < 0.05) 52% reduction in HO-1 protein expression when the miR-217 mimic was overexpressed alone in primary HUVEC (Fig. 3C). The results suggest that the combination of miR-217 and miR-377 is necessary and sufficient to attenuate HO-1 protein expression in both transformed and primary cell lines.
HO-1 is the rate-limiting step in the catabolism of heme. Therefore, we were interested to see if the interaction with miRNAs would attenuate not only HO-1 protein expression, but also its enzymatic activity. Therefore, we performed HO-1 enzyme activity in HEK 293 cells which had been transfected with the HO-1 overexpression constructs. Consistent with our findings in Fig. 3, the HO-1 enzyme activity level was reduced by 48.8% (p < 0.05) in cells transfected with both miR-217 and miR-377 constructs compared with controls (Fig. 4). Overexpression of each miRNA alone was not sufficient to reduce enzyme activity. These results not only confirmed those from Western blots (Fig. 3), but also indicate that expression of miR-377 and miR-217 can attenuate HO-1 protein expression to a level that can significantly reduce enzyme activity.
To further elucidate if both miR-377 and miR-217 were important in attenuation of HO-1 protein expression, we performed experiments inhibiting miR-217, miR-377, and the combination of miR-217 and miR-377 using miRNA hairpin inhibitors. As demonstrated in Fig. 5, knockdown of both miR-217 and miR-377 in primary HUVEC leads to a 2-fold increase in HO-1 protein expression after hemin treatment (lanes 10–12) when compared with cells treated with hemin alone (lanes 2 and 3). Levels of HO-1 protein remain the same when either miR-217 or miR-377 are inhibited alone (Fig. 5). These results strongly support our hypothesis that miRNAs are involved in attenuating HO-1 protein expression and further suggest that factors that influence the expression of both miR-217 and miR-377 are important in regulating HO-1 protein expression.
HO-1 mRNA and protein are induced upon stimulation with hemin, so we also examined its effect on levels of miRNAs. We stimulated HEK 293 and primary HUVEC cells with 10 μm hemin for 1 h, followed by a 8–24-h incubation in basal medium. Assessment of miRNA expression via qRT-PCR showed that expression of miR-217 was significantly (p < 0.05) decreased in both HEK 293 (Fig. 6A) and HUVEC (Fig. 6B) cells after hemin treatment when compared with untreated control cells. HEK 293 displayed the decrease at 8 h post-treatment (Fig. 6A), whereas the HUVEC cells did not show a decrease in miR-217 levels until 24 h post-treatment (Fig. 6B). The levels of miR-377 also decreased in both cell lines (Fig. 6, A and B); however, the decrement was not significant compared with controls (p < 0.08 for HEK 293, p = 0.35 for HUVEC). Once again, the decrease in miR-377 expression was delayed to 24 h in HUVEC cells (Fig. 6B). A more extensive survey of the expression of some other predicted HO-1 3′ UTR candidate miRNAs in the same HEK 293 samples from Fig. 6A revealed that some, but not all candidate miRNAs responded in a similar fashion as miR-217 and miR-377 (Fig. 6C). The data suggested that in the presence of heme, a regulatory network of miRNAs may be involved in calibrating HO-1 protein levels to meet the physiological demands.
The results of our studies support our hypothesis that a post-translational mechanism may exist to attenuate HO-1 expression. We demonstrated that miRNAs interact directly with the HMOX1–3′ UTR. We further demonstrate that two miRNAs, miR-217 and miR-377, combine to attenuate HO-1 protein expression, which ultimately results in a significant reduction in HO-1 enzyme activity. In return, we demonstrate that knockdown of both miR-217 and miR-377 increases HO-1 protein expression. Additionally, we demonstrate that exposure to hemin influences the levels of some, but not all miRNAs involved with HO-1 expression. Based on these studies, we propose a model (Fig. 7) in which miRNAs may be serving as a rheostat to titrate the levels HO-1 expression to a physiologic set point. This model illustrates that a decrease in miR-217 and miR-377 leads to increased HO-1 protein (Fig. 7B), and overexpression of the same miRNAs leads to attenuation of protein expression (Fig. 7C).
In this work we selected two candidate miRNAs to validate based on their context score. Additionally, we chose to validate each miRNA separately and in combination. As recent publications have highlighted, many genes have multiple binding sites for a single miRNA within their 3′ UTR, but also binding sites for many other miRNAs (25, 26). Therefore, the combination of multiple miRNAs may determine the level of gene expression; in part, because the combination of multiple miRNAs may overcome the relative weakness of each separate miRNA:mRNA seed match (26). Therefore, as demonstrated in Fig. 2, a weak interaction between miR-217 and the HMOX1–3′ UTR may help strengthen the association that occurs between miR-377 and the HMOX1–3′ UTR. In addition, the expression of each miRNA is also controlled by various factors, such as levels of primary and precursor miRNA transcripts and cell type (39). Further validation of other predicted targets, such as the ones surveyed in Fig. 6, may yield a more complex system to coordinate HO-1 protein expression depending on the environment of the cell.
Interestingly, both miR-377 and miR-217 have been investigated for their role in stress responses during pathological conditions, such as diabetes and aging (27,–29). Using mouse models of diabetic nephropathy, Kato and associates (28) demonstrated that miR-217, and its clustered counterpart miR-216a, are induced by transforming growth factor-β (TGF-β), leading to an inhibition of PTEN, and ultimately activation of Akt. Wang and associates (29) also demonstrated that miR-377 plays a role in the pathogenesis of diabetic nephropathy in both human cell lines and mouse models through its ability to target and translationally repress superoxide dismutase 1 and 2 (SOD1, SOD2) and p21/Cdc42/Rac1-activated kinase 1 (PAK1) leading to increased fibronectin production (29). SOD2 is also known as mitochondrial superoxide dismutase or manganese SOD (MnSOD) and is responsible for reducing toxic reactive oxygen species in the mitochondria. Interestingly, carbon monoxide, a by-product of HO-1, has been shown to increase levels of MnSOD (30). Therefore, in accord with our hypothesis a decrease in miR-377 will lead to an up-regulation of both HO-1 and MnSOD, leading to an antioxidant effect. In our study, the effect of hemin on miR-377 and miR-217 was dependent on the cell line and time period evaluated (Fig. 5). Therefore, the balance between levels of miR-217 and miR-377, along with other potential miRNAs, may act like a rheostat that attenuates pro-apoptotic versus the antioxidant pathways in response to cellular stressors. Concordantly, this analysis also highlights the fact that miRNA:mRNA networks may differ between cell types and disease model systems.
Previous work investigating miRNAs and HO-1 has focused on the effects of miRNAs binding to Bach1, a potent repressor of HMOX1 transcription (31, 32). Bach-1 is a member of the bZIP transcription factor family that serves as a potent repressor of HMOX1 transcription through its higher binding affinity for multiple Maf recognition elements regions compared with nuclear factor erythroid 2-related factor 2 (Nrf2) (33,–37). In two recent papers, Bonkovsky and associates (31, 32) explored the role of miR-122 and miR-196 in regulating Bach1 in hepatocytes infected with hepatitis C virus. miR-122 is the most abundant miRNA transcript in the liver and necessary for hepatic accumulation of the hepatitis C virus (38). In fact, recent work using lock-nucleic acid antagonists of miR-122 in chimpanzee models of high cholesterol and hepatitis C virus demonstrated the efficacy and feasibility of miRNA-based therapeutics for human clinical trials (38, 39). miR-122 is predicted to bind with the Bach1 3′ UTR, which leads to a decrease in Bach1 protein and a subsequent increase in HO-1 expression. In a separate report linking miRNA levels and HO-1 expression, it was demonstrated that miR-196 interacts with Bach1 via two binding sites within the 3′ UTR, and that overexpression of miR-196 leads to repression of Bach1 protein and subsequent up-regulation of HMOX1 mRNA (31). When we surveyed for expression of mature miRNAs after hemin treatment we did find that in HEK 293 cells, but not HUVEC, treatment with hemin resulted in a decrease (p = 0.09) in miR-196 expression (data not shown). This interaction could lead to an up-regulation in Bach1 in an attempt to balance factors required for the control HO-1 expression in the presence of heme.
It has been reported that heme may play a significant role in miRNA processing. Mature miRNAs are cleaved from ~70-nucleotide hairpin structures, called precursor miRNAs (pre-miRNAs), in the cytoplasm by the enzyme dicer. Prior to this final step, pre-miRNAs are excised in the nucleus from a primary miRNA (pri-miRNA) transcript by the RNase III enzyme Drosha and its co-factor DiGeorge Critical Region 8 (DGCR8, also known as Pasha), which are necessary and sufficient for pri-mRNA processing (9, 40). DGCR8 is a double-stranded RNA-binding protein that may be the only member of the small non-coding RNA processing pathway specific for miRNAs (41). DGCR8 binds heme, which promotes the homodimerization of DGCR8, forming a complex containing one heme molecule per homodimer (9). The heme-free DGCR8 monomer is much less active than the heme-bound dimer, perhaps because the conserved cysteine residue that binds heme prevents autoinhibition of DGCR8. The ability of DGCR8 to function as a heme biosensor provides an elegant mechanism of linking heme levels to global protein synthesis and cellular differentiation. Rapid changes in the availability of heme may lead to significant changes in the ability of the body to process crucial miRNA molecules, such as those regulating a key heme catabolizing enzyme, HO-1. As Fig. 6 demonstrates, some, but not all, of the miRNAs predicted to interact with the HMOX1–3′ UTR were affected by heme. Additionally, the timing of this interaction also varied according to the cell type, which may reflect cell-specific copy number variations of the individual miRNAs (42).
A recent study on HO-1 expression in endothelial cells derived from different donors demonstrates a wide range of basal HO-1 levels in human populations. However, when the cells were stimulated with oxidized lipids, all cells reached a similar level of HO-1 mRNA, regardless of basal levels of HMOX1 transcript (43). We propose that miRNAs may be serving as a rheostat to titrate the levels of HO-1 expression to a physiologic set point (Fig. 7). In this work we have demonstrated that two miRNAs, miR-217 and miR-377, work together and in combination to attenuate HO-1 protein expression and enzyme activity, highlighting that miRNA interactions are involved in the control of HO-1 expression. Further work is warranted to clarify the role of miRNA-mRNA interactions during hemolysis and uncover new potential therapeutic modalities to modulate HO-1 expression.
We acknowledge the following individuals for their support and assistance during the completion of this project: Dr. John Belcher, Dr. Liming Milbauer, Dr. Yvonne Datta, Carol Bruzzone, Paul Marker, and Jean Herron.
*This work was supported, in whole or in part, by National Institutes of Health Grants F30AG030909 from the NIA (to J. D. B.) and R01 HL67367 and P01HL055552 from the NHLBI (to G. M. V. and R. P. Hebbel).
2The abbreviations used are: