The hypoxia-inducible miR-429 regulates hypoxia-inducible factor-1α expression in human endothelial cells through a negative feedback loop

Abstract

Hypoxia-inducible factors (HIFs) 1 and 2 are dimeric α/β transcription factors that regulate cellular responses to low oxygen. HIF-1 is induced first, whereas HIF-2 is associated with chronic hypoxia. To determine how HIF1A mRNA, the inducible subunit of HIF-1, is regulated during hypoxia, we followed HIF1A mRNA levels in primary HUVECs over 24 hours using quantitative PCR. HIF1A and VEGF A (VEGFA) mRNA, a transcriptional target of HIF-1, increased ∼2.5- and 8-fold at 2–4 hours, respectively. To determine how the mRNAs were regulated, we identified a microRNA (miRNA), miR-429, that destabilized HIF1A message and decreased VEGFA mRNA by inhibiting HIF1A. Target protector analysis, which interferes with miRNA-mRNA complex formation, confirmed that miR-429 targeted HIF1A message. Desferoxamine treatment, which inhibits the hydroxylases that promote HIF-1α protein degradation, stabilized HIF-1 activity during normoxic conditions and elevated miR-429 levels, demonstrating that HIF-1 promotes miR-429 expression. RNA-sequencing-based transcriptome analysis indicated that inhibition of miRNA-429 in HUVECs up-regulated 209 mRNAs, a number of which regulate angiogenesis. The results demonstrate that HIF-1 is in a negative regulatory loop with miR-429, that miR-429 attenuates HIF-1 activity by decreasing HIF1A message during the early stages of hypoxia before HIF-2 is activated, and this regulatory network helps explain the HIF-1 transition to HIF-2 during chronic hypoxia in endothelial cells.—Bartoszewska, S., Kochan, K., Piotrowski, A., Kamysz, W., Ochocka, R. J., Collawn, J. F., Bartoszewski, R. The hypoxia-inducible miR-429 regulates hypoxia hypoxia-inducible factor-1α expression in human endothelial cells through a negative feedback loop.

Cardiovascular diseases are often associated with ischemic events that lead to a decrease in oxygen and nutrient delivery to the tissues. The lowered oxygen tension in the tissues induces a hypoxic adaptive response that enables cells to recover from this cellular insult. One mechanism for reestablishing cellular and tissue homeostasis is through the induction of angiogenesis and increasing oxygen delivery to the tissues (1). If the hypoxic conditions persist over extended periods of time, then the cells can undergo programmed cell death (2). Understanding the cellular pathways that mediate recovery from hypoxia is therefore critical for developing novel therapeutic approaches for cardiovascular diseases.

The master regulator of oxygen homeostasis that controls angiogenesis during hypoxia is hypoxia-inducible factor-1α (HIF1A) (3). HIF-1α protein expression is induced during hypoxia and associates with a stable, constitutively expressed HIF-1β subunit (also called aryl hydrocarbon receptor nuclear translocator) in a complex referred to as HIF-1 (4). HIF-1 expression is responsible for transcriptional activation of >200 genes by binding to hypoxia response elements (HREs) in the target gene promoter regions (5). HIF-1α expression and functions are tightly regulated through changes in oxygen tension. When cells and tissues return to normoxic conditions, HIF-1α is posttranslationally modified by 2 hydroxylase enzymes: proline-hydroxylase-2 (PHD2), and factor-inhibiting hypoxia-inducible factor-1(FIH-1; also called HIF-1α subunit inhibitor or HIF-1AN). PHD2 hydroxylation leads to polyubiquitination by the von Hippel-Lindau (pVHL) ubiquitin E3 ligase complex (6). The second hydroxylase, FIH-1, regulates the transcriptional activity of the heterodimeric complex by binding to HIF-1α and pVHL and inhibiting the transactivation domains of HIF-1α (7). During normoxic conditions, therefore, HIF-1 levels are low (8).

HIF-1α protein stability and function are maintained during low oxygen tension because PHD2 and FIH-1 are inactive (9). One key target gene of HIF-1 during hypoxia is the VEGF A (VEGFA) gene (10), which along with its receptors, are critical factors that promote angiogenesis by recruiting endothelial cells and stimulating their proliferation (11). Although HIF-1 regulation during normoxic conditions is carefully controlled by the hydroxylases described above, how HIF-1 activity and angiogenesis are controlled during hypoxia is less understood. Recent studies, however, indicate that a number of microRNAs (miRNAs) also play critical roles in angiogenesis and are induced during hypoxia [reviewed in (12–14)]. miRNAs are small noncoding RNAs that regulate mRNA stability or translation (15). These hypoxia-inducible miRNAs are referred to as hypoxamiRs, and a number of these that affect HIF-1α expression have been identified [reviewed in (12, 14)]. Conversely, HIF-1 promotes the expression of several hypoxamiRs including miR-210 in tumor cells (16) and miR-155 in intestinal epithelial cells (17).

To examine angiogenesis regulation in endothelial cells, we found that one miRNA of the miR-200 family, miR-429, was up-regulated during hypoxia in primary HUVECs. Furthermore, we demonstrate that miR-429 is up-regulated by HIF-1, and HIF1A message levels are negatively regulated by miR-429, establishing a negative regulatory feedback loop. This regulatory loop provides an important mechanism for regulating HIF-1 activity during extended periods of hypoxia.

MATERIALS AND METHODS

Cell lines and culture conditions

HUVECs were obtained from American Type Culture Collection (ATCC; Manassas, VA, USA) (www.atcc.org). HUVECs were maintained until passage 6 in EGM-2 Bulletkit medium (Lonza Group, Basel, Switzerland). HeLa S3 cells were obtained from ATCC and cultured in DMEM (Sigma-Aldrich, St. Louis, MO, USA) with 10% fetal bovine serum and 2 mM L-glutamine. Cells were split either into 6-well plates or 10 cm dishes and allowed to grow to 70–80% confluence prior to the start of the experiments.

Induction of hypoxia

Hypoxia was induced in a Binder CO₂/O₂ incubator (Tuttlingen, Germany) for hypoxia research. Briefly, cells were cultured in 10 cm dishes at 1% O₂ for the time periods specified. Control cells were maintained in normoxia in the same incubator and harvested at the specified times. Chemical HIF-1α stabilization was achieved with 200 µM CoCl₂ for 12 h or 100 µM deferoxamine mesylate salt (DFO) for 6 h. Both CoCl₂ and DFO were purchased from Sigma-Aldrich.

Isolation of RNA and miRNA

Total RNA containing the miRNA fraction was isolated using the miRNeasy Kit (Qiagen, Germantown, MD, USA) according to the manufacturer’s protocol. RNA concentrations were calculated based on the absorbance at 260 nm. RNA samples were stored at −70°C.

Measurement of mRNA and miRNA levels using quantitative real-time PCR

We used TaqMan One-Step RT-PCR Master Mix Reagents (Applied Biosystems, Foster City, CA, USA) as described previously (18–20) using the manufacturer’s protocol (relative quantification; Applied Biosystems StepOnePlus Real-Time PCR System). The relative expressions were calculated using the comparative relative standard curve method (21). Because hypoxia affects the expression levels for a number of genes that are often used as quantitative real-time PCR relative controls including glyceraldehyde 3-phosphate dehydrogenase and ACTB (22), we used 18S rRNA as the relative control for our studies. We also validated this relative control against another housekeeping gene, TATA-binding protein (TBP). As a relative control for miRNA quantification, we validated and used RNU44 and RNU48. Applied Biosystems TaqMan probe identification (id) numbers used were Hs99999901_s1 (18S), Hs4332659_m1 (TBP), Hs00153153_m1 (HIF1A), Hs00215495_m1 (HIF1AN), Hs00254392_m1 (EGLN1), Hs00900055_m1 (VEGFA), Hs00918445_g1 (CITED2), Hs00914223_m1 (EP300), Hs01026149_m1 (EPAS1), 001094 (RNU44), 001006 (RNU48), 001024 (hsa-miR-429), 002251 (hsa-miR-200b), and 000502 (hsa-miR-200a).

miRNA analogs and target protector transfections

miR-429 mimic (assay id: MC10221) and antagomir (assay id: MH10221) were purchased from Ambion/Life Technologies (Carlsbad, CA, USA). HUVECs were transfected in 6-well plates using the liposome Lipofectamine RNAiMax (Invitrogen, Life Technologies, Carlsbad, CA, USA) according to the manufacturer’s protocol. miR-429 mimic and antagomir were used at final concentrations of 10 and 20 nM, respectively. The transfected cells were cultured for 2 d prior to further analysis. The degree of miRNA over-expression or knockdown was determined by quantitative real-time PCR. Target protectors (TPs) were purchased from Qiagen (miScript Target Protector) and directed against HIF1A (TP HIF1A, 5′-ACATAAATAATAATGCTTTGCCAGCAGTAC-3′) and VEGFA (TP VEGFA, 5′-ATTAAAGAGTAGGGTTTTTTTTCAGTATTCTTGGTTAATA-3′). TPs were used at a final concentration of 600 nM. cel-miR-67 was used as a control because it has no homology to any known mammalian miRNA (Ambion assay id: MC22484). As an additional control for the transfection experiments, Ambion small interfering RNA (siRNA) Negative Control 1 [4390843; Invitrogen (the siRNA sequence is Ambion proprietary information)] was used as well.

RNA sequencing

HUVECs (passage 3) were transfected with miR-429 antagomir or with control cel-miR-67 as described above. At 72 h after transfection, following total RNA isolation (with miRNA fraction), samples were validated with quantitative real-time PCR for miR-429 inhibition, prior to further analysis. Following rRNA depletion, the remaining RNA fraction was used for library construction and subjected to 100 bp paired-end sequencing on an Illumina HiSeq 2000 instrument (San Diego, CA, USA). Sequencing reads were aligned to the human reference genome assembly (hg19) using TopHat (23). Transcript assembly and estimation of the relative abundance and tests for differential expression were carried out with Cufflinks and Cuffdiff (24). The resulting data were validated with quantitative real-time PCR.

Western blots

Cells were lysed in RIPA buffer [150 mM NaCl, 1% NP-40, 0.5% sodium deoxycholate, 0.1% SDS, and 50 mM Tris- HCl (pH 8.0)] supplemented with protease Inhibitor Complete Mini (Roche, Basel, Switzerland) on ice for 15 minutes. The cell lysates were rotated at 4°C for 30 min, and the insoluble material was removed by centrifugation at 15,000 relative centrifugal force for 15 minutes. Protein concentrations were determined by Bio-Rad Protein Assay (Hercules, CA, USA) using bovine serum albumin (BSA) as a standard. Following the normalization of protein concentrations, lysates were mixed with an equal volume of 2× Laemmli sample buffer and incubated for 5 minutes at 95°C prior to separation by SDS-PAGE on stain-free TGX gradient gels (Bio-Rad). Following SDS-PAGE, the proteins were transferred to PVDF membranes (300 mA for 90 minutes at 4°C). The membranes were then blocked with BSA (Sigma-Aldrich) proteins dissolved in PBS/Tween 20 (3% BSA and 0.5% Tween 20 for 1–2 hours), followed by immunoblotting with the primary antibody specified for each experiment: HIF-1α (1:150, ab16066; Abcam); VEGFA (1:250, ab51745; Abcam); PHD2 (1:800, ab133630; Abcam); β-actin (1:1000, ab1801; Abcam); and FIH-1 (1:5000, ab92498; Abcam). After the washing steps, the membranes were incubated with goat anti-rabbit IgG (H&L) or with goat anti-mouse IgG (H&L) horseradish peroxidase–conjugated secondary antibodies (Bio-Rad) and detected using ECL (Amresco, Solon, OH, USA). Densitometry was performed using Image Lab software, v. 4.1 (Bio-Rad).

Statistical analysis

Results were expressed as the mean ± sd. Statistical significance among means was determined using the Student’s t test (2 samples: paired and unpaired). Pearson product-moment correlation tests (25) were performed with SigmaPlot software (Systat Software Inc., San Jose, CA, USA).

RESULTS

mRNA expression of HIF1A, PHD2, FIH1, and VEGFA dynamically changes during hypoxia in endothelial cells

The dynamic mRNA changes during hypoxia were monitored during a time course by measuring the mRNA changes in HIF1A, HIF-1α’s inhibitors (PHD2 and FIH1), and a transcriptional target of HIF-1α: VEGFA. The results indicate that HIF1A mRNA is rapidly up-regulated after hypoxia induction and reaches a maximum level in 2 hours (Fig. 1A). After 12 hours of hypoxia, HIF1A mRNA levels decrease to below background levels. After 24 hours of hypoxia, the levels of HIF1A are essentially undetectable. The 2 regulators of HIF-1α stability and activity, PHD2 and FIH1, also appear to have the similar kinetics of up-regulation, with PHD2 mRNA levels peaking at 4 h (Fig. 1B, C). Given the dramatic up-regulation of HIF-1α, we next examined the mRNA levels of an HIF-1 transcriptional target that is critical for angiogenesis: VEGFA. As shown in Fig. 1D, VEGFA mRNA levels were rapidly activated during hypoxia and reached maximum mRNA levels at 4 hours postinduction. Interestingly, there appeared to be biphasic peaks of mRNA levels, with peaks at 4 and 20 hours after the induction of hypoxia. The other mRNAs displayed a similar biphasic pattern, although to a much less-pronounced degree. All 4 of the mRNA levels followed very similar time courses and peaked at 2–4 hours, suggesting that their regulatory mechanisms during hypoxia were very similar.

Figure 1.

Hypoxia induces dynamic changes of expression profiles of HIF1A (A), FIH1 (B), PHD2 (C), and VEGFA (D) in HUVECs. The mRNA levels were monitored in quantitative real-time PCR experiments. The results from 3 independent experiments (n = 32) are plotted normalized to 18S rRNA levels and expressed as a fold change over the normoxic control. Error bars represent sd. *P < 0.05.

Given the biphasic nature of the mRNA responses, particularly with regard to the VEGFA mRNA responses, which were very elevated at 20 hours, and given that the HIF-1α levels were low at 20 hours, we next examined HIF-2α, which has been reported to promote the chronic response to hypoxia. The time course analysis for HIF2A (also called EPAS1) mRNA shown in Supplemental Fig. 1 indicates that HIF2A mRNA levels are not elevated until 12–16 hours, suggesting that the second phase of the VEGFA elevation is due to HIF-2 activity at the later time points.

To evaluate the consequences of hypoxia on the protein expression levels of these HIFs, we analyzed the protein levels by Western blot. As shown in Fig. 2A, the protein level increases of HIF-1α closely follow the mRNA profile. The maximum peak of protein expression is at 4 hours, which is a 2-hour lag compared to the mRNA profile. The protein level is back to normoxic levels by 12 hours. Interestingly, FIH-1, PHD2, and VEGFA protein levels do not follow the same time course as HIF-1α (Fig. 2A). FIH-1 and PHD2 remain fairly constant throughout the hypoxia time course, whereas VEGFA levels show a biphasic response, with an increase in protein levels at 4 and 16 h after hypoxia induction, further suggesting that HIF-1α and HIF-2α are responsible for this response. These data indicate that HIF-1α and VEGFA protein levels are fairly consistent with the mRNA induction, whereas FIH-1 and PHD2 remain at a fairly constant level during the time course. Given that the HIF1A mRNA and protein levels were induced and then rapidly decreased during hypoxia, and both regulators are inactive during hypoxia and their protein levels changed very little during the time course, we next examined the role of miRNAs in regulating HIF1A mRNA levels.

Figure 2.

Hypoxia induces dynamic changes of protein levels of HIF-1α (A), FIH-1 (B), PHD2 (C), and VEGFA (D) in HUVECs. The bar graphs below show the relative protein amounts at each time point. The protein levels were detected with SDS-PAGE and Western blot and related to β-actin levels. There were 2 individual samples (4 µg total protein per lane) tested for each time point, and the experiments were repeated twice. The protein levels (bar graphs) are normalized to normoxic control. *P < 0.05.

miR-200b and miR-429 are up-regulated during hypoxia

In our analysis of miRNAs, we first focused on the miR-200 family members based on their known elevations during hypoxia (26–29). miR-429, miR-200b, and miR-200a are clustered on chromosome 1 (chr1), whereas miR-429 and miR-200b are particularly interesting because they have been identified as hypoxamiRs (16). Furthermore, miR-429 is elevated in head and neck squamous cell carcinoma (29) and focal cerebral ischemia in mice (27). Hypoxia in human dermal-derived microvascular endothelial cells, however, resulted in reduction of miR-200b (28).

To establish whether any of these family members were up-regulated during hypoxia in human endothelial cells, we examined the HUVEC samples used for the mRNA analyses and monitored the same time course as shown in Fig. 1 for the miRNA expression profiles. The results shown in Fig. 3A indicate that miR-200b was elevated ∼4-fold at 4 hours, remained elevated for 20 h, and returned to normoxic levels by 24 hours. miR-200a, on the other hand, was barely elevated during the time course (Fig. 3B), whereas miR-429 was elevated ∼13-fold at 4 hours and then rapidly declined by 12 h (Fig. 3C). Interestingly, the profile of the miR-429 time course was strikingly similar to the HIF1A mRNA profile (Fig. 1A), suggesting that miR-429 and HIF1A might be transcriptionally regulated by similar elements.

Figure 3.

Hypoxia-induced changes in expression profiles of miR-200b (A), miR-200a (B), and miR-429 (C) in HUVECs. The miRNA levels were monitored in quantitative real-time PCR experiments. The results from 3 independent experiments (n = 32) are plotted normalized to RNU44/RNU48 and expressed as a fold change over the normoxic control. *P < 0.05.

miR-429 is induced by HIF-1

To test this hypothesis, we used the miRGen analysis of the miR200b-miR200a-miR429 cluster region but could not identify any specific transcription factors potentially shared by this region and the HIF1A promoter that would facilitate miR-429 up-regulation during hypoxia (30). We next analyzed the miR200b-miR200a-miR429 cluster region for consensus HRE sequences [G/C/T ACGTGC (G/C)] and identified 2 putative HRE motifs at positions [+]1167559 and [+]1168416 (both on chr1; Fig. 4A). The first putative HRE is located between the miR-200b and miR-200a start sites, whereas the second one is between the miR-200a and miR-429 start sites. Interestingly, during the hypoxia time course, only miR-429 correlated with HIF1A mRNA expression, whereas miR-200a and miR-200b exhibited unrelated expression profiles. The Pearson product-moment correlation coefficients and P values for miR-200b-HIF1A, miR-200a-HIF1A, and miR-429-HIF1A were 0.0258 (P = 0.924; n = 32), 0.0290 (P = 0.915; n = 32), and 0.659 (P = 0.00547; n = 32), respectively. This suggested that the putative HRE directly preceding miR-429 could be transcriptionally active, whereas the other is probably not (Fig. 4A).

Figure 4.

HIF-1 induces miR-429 expression. A) The localization of putative HRE sequences in miR200b-miR200a-miR429 cluster (numbering is based on human National Center for Biotechnology Information genomic sequence). B) HUVECs were treated with hypoxia mimetics [100 µM DFO (white) for 6 hours, and 200 µM CoCl₂ (gray) for 12 hours], and miRNA levels were monitored in quantitative real-time PCR experiments. ctrl, control. The miRNA level results from 3 independent experiments (n = 10) are plotted normalized to RNU44 levels and expressed as a fold change over the untreated control. *P < 0.05.

To test this correlation, we chemically enhanced HIF-1α stability and followed the changes in the miRNA levels (Fig. 4B). We treated the HUVECs with the chemical hypoxia mimetics desferoxamine (DFO) and CoCl₂ (100 µM for 6 hours and 200 µM for 12 hours, respectively) that inhibit proline hydroxylase activity during normoxia. These treatments stabilize the HIF-1α subunit and thereby induce HIF-1 transcriptional activity (31, 32). Under these conditions, only miR-429 was significantly elevated (Fig. 4B), suggesting that only the HRE at position chr1: [+]1168416 preceding miR-429 is an HIF-1 transcriptional target.

Identification of miR-429 targets

PHD2 was previously identified as a target of miR-429 in mice (27), and the human mRNA contains a potential 3′ UTR-binding sequence for miR-429. Furthermore, our expression analysis indicated that there is a very strong correlation between miR-429 and PHD2 expression profiles during hypoxia (Supplemental Fig. 2). Therefore, we examined miR-429 over-expression and inhibition effects on PHD2 mRNA levels in HUVECs under normoxic and hypoxic conditions (Fig. 5A). Although the PHD2 mRNA levels did go up slightly upon hypoxia induction, neither expression of the miR-429 mimic (Mimic 429) nor antagomir (Antagomir 429) had any effect on the PHD2 mRNA levels under any condition, indicating that miR-429 does not regulate PHD2 mRNA levels in HUVECs. A similar analysis on FIH1 (HIF1AN) mRNA, which also contained a potential miR-429 targeting sequence, revealed that FIH1 was not regulated by miR-429 either (Fig. 5B) because the mimic and antagomir had similar effects. One interesting effect was that the mimic and antagomir 429 under normoxic conditions both significantly elevated FIH1 mRNA levels, presumably through different indirect effects. Transfection control experiments indicated that the expression of mimic and antagomir was effective under the experimental conditions (Fig. 5C), supporting the view that miRNA-429 does not regulate the mRNA expression levels of either PHD2 or FIH1 in HUVECs.

Figure 5.

miR-429 does not alter the expression of PHD2 (A) and FIH1 (B). HUVECs were transfected with miR-429 mimic or antagomir, and the mRNA levels were monitored in quantitative real-time PCR experiments in normoxic conditions as well as after 4 and 8 h of exposure to hypoxia. PHD2 mRNA and FIH1 levels from 3 independent experiments (n = 12) are plotted normalized to 18S rRNA levels and expressed as a fold change over the transfection control. The predicted target sites of miR-429 in PHD2 and FIH1 3′ UTRs are shown on the right. miR-429 levels for mimic or antagomir transfection were monitored for each experiment, as shown in (C). *P < 0.05.

miR-429 regulates HIF1A mRNA stability

Given that the miR-429 and HIF1A levels were elevated simultaneously during hypoxia induction, and the 3′ UTR of HIF1A has a potential miR-429 targeting sequence, we next tested whether miR-429 affected HIF1A mRNA and protein levels by transfecting the HUVECs under both normoxic and hypoxic conditions. The quantitative real-time PCR studies confirmed that inhibition of miR-429 with antagomir resulted in the accumulation of HIF1A mRNA, whereas miR-429 mimic transfection reduced the levels of both HIF1A mRNA and protein (Fig. 6A, B). The miR-429 effects on HIF1A were detectable even under normoxic conditions.

Figure 6.

A) miR-429 specifically modulates expression of HIF1A mRNA under normoxia and hypoxia. HUVECs were transfected with miR-429 mimic or antagomir, and the mRNA levels were monitored in quantitative real-time PCR experiments in normoxic conditions and after 4 and 8 hours of hypoxia. HIF1A mRNA levels from 3 independent experiments (n = 12) are plotted normalized to 18S rRNA levels and expressed as a fold change over the transfection control. *P < 0.05. B) The corresponding changes of HIF-1α protein levels were detected with SDS-PAGE and Western blot and related to total protein levels. There were 2 individual samples (3 µg total protein per lane) tested for each treatment, and the experiments were repeated twice. Mimic 429 (m429) and antagomir 429 (ant429) for each condition are shown. miR-429 levels were monitored for each experiment. C) The predicted target site of miR-429 in HIF1A 3′ UTRs is shown, and the minimum free energy (mfe) for miR-429-HIF1A mRNA was calculated with RNAhybrid (33) (miR-429 is shown in red; HIF1A target mRNA sequence is in green).

Using the RNAhybrid software (33), we determined the putative binding sequence in the 3′ UTR of HIF1A mRNA (Fig. 6C). In order to determine if miR-429 directly interacts with the HIF1A target sequence, we used a specific TP to test if the predicted binding site in the HIF1A mRNA was correct. TPs are modified RNAs complementary to miRNA target sequences that bind to the specific mRNA target sequence and block miRNA binding and thus prevent the formation of the miRNA-mRNA complex (34). As shown in Fig. 7, the HIF1A TP reverses the miR-429 mimic inhibitor effects on both mRNA and HIF-1α protein levels, suggesting that the predicted target sequence identified is correct. Furthermore, the results from Figs. 6 and 7 suggest that miR-429 limits HIF-1α expression during both normoxic and hypoxic conditions.

Figure 7.

A) miR-429 binds to predicted target sequence at HIF1A 3′ UTR. HUVECs were transfected with HIF1A target sequence-specific TP and/or miR-429 analog (Mimic 429). The HIF1A mRNA levels were monitored in quantitative real-time PCR experiments. The HIF1A mRNA level results from 3 independent experiments (n = 12) are plotted normalized to 18S rRNA levels and expressed as a fold change over the transfection control. *P < 0.05. B) The corresponding changes of HIF-1α protein levels were detected with SDS-PAGE and Western blot and related to β-actin levels. There were 2 individual samples (3 µg total protein per lane) tested for each treatment, and the experiments were repeated twice.

miR-429 indirectly affects VEGFA mRNA levels through its actions on HIF1A

HIF-1α is a transcriptional activator of VEGFA expression during hypoxia (35). Therefore, miR-429 could regulate VEGFA expression directly through binding to its 3′ UTR of the VEGFA mRNA, indirectly through direct regulation of HIF1A mRNA levels, or through both mechanisms. In order to determine which of these possibilities is true, we used a TP that specifically binds to the putative miR-429 target sequence in the VEGFA 3′ UTR [Fig. 8A, right panel; Target scan, v. 6.2 (36)]. As shown in Fig. 8A, miR-429 (Mimic 429) modulated VEGFA mRNA levels only during hypoxia with the largest effect at 4 hours. During normoxic conditions, when HIF-1α activity is limited, miR-429 had little effect on the VEGFA mRNA levels. In this case, only the antagomir (Antagomir 429) had a significant effect. Furthermore, as shown in Fig. 8B, the effects of the Mimic 429 were not blocked in the presence of VEGFA-specific TP (VEGFA TP). However, in the presence of HIF1A-specific TP (HIF1A TP), the VEGFA mRNA levels were elevated, even in the presence of Mimic 429 (Fig. 8B). As an additional control, we confirmed that the VEGFA TP had no effect on HIF1A mRNA levels (Fig. 8C). Thus, the effect of miR-429 on VEGFA mRNA expression results from the modulation of HIF1A mRNA levels rather than any direct effect.

Figure 8.

A) miR-429 modulates VEGFA expression indirectly through changing HIF1A levels. HUVECs were transfected with miR-429 analog or inhibitor, and the mRNA levels were monitored in quantitative real-time PCR experiments in normoxic conditions and after 4 and 8 hours of hypoxia. The mRNA level results from 3 independent experiments (n = 12) are normalized to 18S rRNA levels and expressed as a fold change over the transfection control. The predicted target site of miR-429 in VEGFA 3′ UTRs is shown on the right. The VEGFA (B) and HIF1A (C) mRNA levels were monitored in quantitative real-time PCR experiments following transfection of HUVECs with VEGFA TP or HIF1A TP and/or miR-429 analog (Mimic 429). The mRNA level results from 3 independent experiments (n = 12) are normalized to 18S rRNA levels and expressed as a fold change over the transfection control. *P < 0.05.

RNA-sequencing analysis of miR-429 targets

Given that miR-429 was dramatically up-regulated during hypoxia, and a number of other targets are up-regulated either directly or indirectly through miR-429 actions, we analyzed miR-429 antagomir effects on the mRNA profiles using a genome-wide transcriptome-sequencing [RNA-sequencing (RNA-seq)] analysis on HUVECs under normoxic conditions. Our goal was to determine the effects of this particular miRNA without the complicating effects of hypoxia and HIF-1α expression. We limited our RNA-seq studies to the effects of reducing endogenous miR-429 levels because exaggerated miRNA over-expression can potentially saturate RNA-induced silencing complexes by displacing other endogenous miRNAs and, therefore, cause low-affinity target sites to appear functionally important (37). Furthermore, in order to decrease false-positives, we narrowed our analysis only to these mRNAs that displayed at least 2-fold log₂ change, were statistically significant (q value <0.01), and belonged to abundant transcripts (normalized relative expression value >1). As summarized in Supplemental Table 1, reduction of miR-429 resulted in the up-regulation of 209 mRNAs, of which 166 had predicted miR-429 binding sites. We narrowed our focus to 18 genes involved in cellular response to hypoxia, and these are summarized in Table 1. All 18 of the genes in Table 1 were verified by quantitative real-time PCR. Of those genes, 6 [HMOX1, ADAMTS1, PMAIP1 (NOXA), KITLG, ESM1, and BMI1] are transcriptionally induced by HIF-1 (38–40, 43, 47, 48). ZEB1 and ZEB2 are direct targets of miR-429 (46). Furthermore, EP300, CITED2, and SIRT1 affect HIF-1 transcriptional activity (42, 45, 50), whereas UBE2D1, SPRY2, and BMP2 reduce HIF-1α stability (44, 51, 53), and ROCK2 increases stability (49). PTHLH has proangiogenic activity, whereas HOXA5 has the opposite function (41, 52). Induction of approximately one-third of the identified genes could potentially result from increased HIF-1 activity because no significant change of HIF2A and HIF3A transcript amounts was detected in RNA-seq experiments (Supplemental Table 1). Quantitative real-time PCR validation of RNA-seq data for both miR-429 mimic and antagomir under both normoxic and hypoxic conditions allowed us to eliminate EP300 and CITED2 as direct targets of miR-429 (Supplemental Fig. 3). Interestingly, the RNA-seq data indicated that VEGFA mRNA levels were elevated during miR-429 reduction, but the magnitude of this change was below the cutoff criteria (>2-fold log₂ change; q value <0.01).

TABLE 1.

Genes related to the HIF pathway and angiogenesis that were induced in HUVECs following miR-429 reduction

Gene	Locus	Control (value 1)	Antagomir 429 (value 2)	log2 (fold change)	P value	q value	mir-429 target site	Ref.
HMOX1	chr22: 35777059–35790207	30.9509	2300	6.21551	0.000257836	0.00524572	a	(38)
ADAMTS1	chr21: 28208605–28217728	1.19891	28.3802	4.56509	3.98 × 10−11	1.01 × 10−8	a	(39)
PMAIP1	chr18: 57567191–57571538	15.9333	207.216	3.70102	5.39 × 10−7	4.00 × 10−5	a	(40)
PTHLH	chr12: 28111016–28124916	1.96175	22.435	3.51554	5.52 × 10−12	2.06 × 10−9	a	(41)
CITED2	chr6: 139693396–139695785	13.5785	152.088	3.48552	2.14 × 10−6	0.00012595	a	(42)
KITLG	chr12: 88886569–88974250	2.41353	22.7088	3.23404	1.41 × 10−9	2.55 × 10−7	a	(43)
BMP2	chr20: 6748744–6760910	1.99326	16.2363	3.02602	3.49 × 10−6	0.000184069	a	(44)
SIRT1	chr10: 69644426–69678147	1.21114	8.65789	2.83765	3.21 × 10−6	0.0001708	a	(45)
ZEB1	chr10: 31608100–31818742	2.85333	20.1471	2.81985	0	0	a	(46)
BMI1	chr10: 22605311–22620414	4.38754	26.3866	2.58832	4.06 × 10−5	0.00136947	a	(47)
ESM1	chr5: 54273694–54281414	16.6654	92.909	2.47896	2.19 × 10−5	0.000839745	a	(48)
ROCK2	chr2: 11321777–11484711	1.51722	8.29206	2.4503	0.000105096	0.00274185	a	(49)
HIF1A	chr14: 62162118–62214977	8.41649	43.3998	2.3664	0.000483861	0.00832814	a	(3)
EP300	chr22: 41488613–41576081	3.14163	15.881	2.33771	0.000428375	0.00759795	a	(50)
ZEB2	chr2: 145141941–145277958	1.56902	7.69442	2.29395	9.84 × 10−6	0.000437459	a	(46)
SPRY2	chr13: 80910111–80915086	9.85041	47.7042	2.27586	0.000341079	0.0064662	a	(51)
HOXA5	chr7: 27179982–27196296	5.76847	27.3309	2.24427	0.000501449	0.00847293	a	(52)
UBE2D1	chr10: 60094738–60130513	4.97327	20.367	2.03397	0.000506271	0.00851449	a	(53)

Only genes represented by abundant transcripts (normalized relative expression value >1) were considered (for values 1 and 2). The significance was described by a log₂ fold change >2 and q (P value corrected for false discovery rate) value <0.01. Ref., reference; chr, chromosome.

^aThe presence of potential miR-429 target sequences [as predicted in a bioinformatics approach (33, 36, 55)].

DISCUSSION

The cellular response to hypoxia is a stress response that permits cells to effectively counteract stresses and survive. During hypoxia, angiogenesis is induced to supply adequate oxygen levels to the tissues. Regulation of this pathway is critical for attenuating the stress response once the stressor is gone, and when normoxia returns, HIF-1α protein levels are decreased by the actions of PHD2 and FIH-1. During extended periods of hypoxia, however, these 2 proteins are not active. The focus of this study was to examine how this process is regulated in primary human endothelial cells during extended periods of hypoxia. We focused on HIF-1α because the HIF-2α effect appeared at a later stage of the response in HUVECs, and this type of delayed response has been described previously as a “HIF switch” in which HIF-1 drives the initial cellular response to hypoxia, whereas HIF-2 drives the later stages [reviewed in (54)].

Our approach was to examine key proteins involved in the hypoxic response that included HIF-1α, PHD2, FIH-1, and VEGFA. All 4 of their mRNAs were dramatically up-regulated during hypoxia and displayed similar kinetics. Despite the rapid mRNA level increases of all 4, only the HIF-1α protein time course mirrored its mRNA profile. Interestingly, the VEGFA protein levels were most elevated after 16 h of hypoxia. Our analysis of the HIF-2α response during hypoxia in these cells provides a potential explanation for the biphasic response of VEGFA mRNA seen here.

To determine the role of miRNAs in this process, we focused on the 200 family of miRNAs that includes miR-200b, miR-200a, and miR-429, which constitute a single locus on chr1. We chose this family of miRNAs because our bioinformatic analyses indicated that the miRNA 200 family could potentially recognize binding sites in the 3′ UTR of a number of hypoxia-related mRNAs including HIF1A, FIH1, PHD2, and VEGFA (33, 36, 55). Our experimental analysis indicated that miR-429 was the most elevated of the 3 miRNAs during hypoxia, and the only one to be elevated during desferoxamine or CoCl₂ treatments, 2 methods that inhibit proline hydroxylase activity during normoxia, and thus stabilize the HIF-1α subunit (31, 32). Although this treatment could stabilize any of the α forms of HIF, only the HIF1A message is elevated with the same kinetics as miR-429 in the HUVECs, supporting the view that HIF-1 is driving the expression of miR-429 in these cells during hypoxia.

Our miRNA analysis indicated that miR-429’s only direct target among HIF1A, PHD2, FIH1, and VEGFA was HIF1A itself, indicating that miR-429 was a major regulatory factor for HIF1A expression during hypoxia. The HIF-1 complex is a critical transcription factor and functions both as a proadaptive and proapoptotic factor (56). Upon oxygen deprivation, the HIF-1 complex translocates to the nucleus and activates a number of genes that restores energy and oxygen homeostasis by increasing anaerobic energy production and improving tissue oxygenation by stimulating angiogenesis, vasodilation, and erythropoiesis (57, 58). If proper oxygen tension is not restored, the HIF-1 complex can induce apoptosis through stabilization and binding of the product of the tumor suppressor gene p53, as well as inducing expression of the proapoptotic protein, BCL2/adenovirus E1B 19 kDa protein-interacting protein 3 (59). Furthermore, HIF-1–regulated genes also play roles in tumor progression (60). Indeed, the disturbance of HIF-1 equilibrium and related angiogenesis can contribute to many human diseases including various cancers and cardiovascular disorders (9, 14, 60, 61).

HIF-1 activity is tightly regulated through the oxygen-dependent degradation of the HIF-1α subunit (62). Despite the central importance of hydroxylases in regulating HIF-1α, many additional factors including cytokines, growth factors, and histone deacetylase inhibitors have also been shown to control HIF-1α expression, even under normoxic conditions (63–65). However, mechanisms controlling HIF1A mRNA levels and their role in cellular response during hypoxia have not been well understood. Although hypoxia-induced changes in mRNA expression have been extensively studied in endothelial cells, little is known on the dynamics of that process, particularly during extended periods of hypoxia.

During the first 2–6 hours of hypoxia in HUVECs, HIF1A mRNA accumulates along with VEGFA, PHD2, and FIH1. Although the 2 hydroxylases are inactive during hypoxia, this induction allows for the rapid neutralization of HIF-1α protein upon restoration of oxygen homeostasis. After 8 hours of hypoxia, HIF1A and FIH1 mRNA expression patterns are back to normoxic levels, whereas PHD2 and VEGFA are still elevated even after 20 hours. PHD2 expression is stimulated by HIF-1 (66), and the time course seen in Fig. 1 is certainly consistent with that. Although it has been shown that transcription of both HIF1A and FIH1 is enhanced in an oxygen-independent mechanism involving PKC (9, 67), the mechanism underlying FIH1 transcriptional regulation during hypoxia requires further study. The observed reduction of accumulated HIF1A mRNA may prevent the adverse long-term HIF-1 effects. Thus, action of an miRNA that reduces HIF1A mRNA stability would explain the observed rapid reduction in HIF1A mRNA levels during hypoxia. A recent study suggested that miRNA effects during extended hypoxia may be limited given that Dicer is down-regulated (61), and this, along with the rapid decline in HIF1A levels, could partially explain the rapid decline in miR-429 during the time course.

These studies also highlight the known HIF-1–HIF-2 switch that occurs during extended periods of hypoxia (54). We followed a known target of HIF-1 and HIF-2, VEGFA (68), and found a biphasic response in the elevation of VEGFA mRNA that was consistent with the HIF1A and HIF2A mRNA elevations at 2 and 12–16 h of hypoxia, respectively. Regulation in this switch between HIF-1 protein activity and HIF-2 is modulated in part by the hypoxia-associated factor that promotes HIF-1α ubiquitination and degradation, and also promotes HIF-2α transactivation (69). Our results reveal that the HIF1A-HIF2A mRNA switch is driven transcriptionally (Fig. 1 and Supplemental Fig. 1) and then posttranscriptionally by the hypoxamiR miR-429 (Fig. 6A) that rapidly targets the HIF1A mRNA during an early stage of hypoxia in endothelial cells.

A number of different hypoxamiRs that affect HIF1A expression either positively or negatively have been identified including miR-20a, miR-20b, miR-130a, miR-130b, miR-155, miR-199a, miR-200b, miR-200c, miR-210, miR-424, and miR-429 in the present study (12, 14). The only other miRNA that has been identified in a negative regulatory loop with HIF-1α is miR-155 in gut epithelial cells (17).

This regulatory loop with miR-429 operates only during periods of hypoxia when both HIF1A and miR-429 are induced almost simultaneously (Fig. 9), and this provides an explanation for the HIF-1–HIF-2 switch model that has been proposed suggesting that HIF-1 drives the initial hypoxic response, whereas HIF-2 drives the chronic hypoxic response (54). Interestingly, we found that the HIF1A-miR-429 regulatory loop also appears to be operational in HeLa cells, where miR-429 regulated HIF1A and VEGFA mRNA levels, and HIF-1 activation induced miR-429 expression (Supplemental Fig. 4). This suggests that the mechanism of miR-429-HIF1A regulation that we describe in HUVECs is also active in human cancer cells, which creates novel opportunities in cancer therapeutics.

Figure 9.

Model of negative feedback loop between HIF1A and miR-429. During hypoxia, HIF1A mRNA is induced, whereas HIF-1α protein is stabilized, allowing for the for-mation and accumulation of active HIF-1 complexes. The HIF-1 is translocated to the nucleus in order to regulate expression of its target genes. HIF-1 binds to the HRE sequence located in the miR200b-miR200a-miR429 cluster and induces miR-429 expression and, thus, increases the cytosolic levels of this miRNA. miR-429 binds in cytosol to the target sequence located at the 3′ UTR of HIF1A mRNA, which leads to decreased HIF-1 activity.

Glossary

BSA

bovine serum albumin

chr1

chromosome 1

DFO

deferoxamine mesylate salt

FIH-1

factor-inhibiting hypoxia-inducible factor-1

HIF

hypoxia-inducible factor

HRE

hypoxia response element

hypoxamiR

hypoxia-inducible microRNA

id

identification

miRNA

microRNA

PHD2

proline-hydroxylase-2

pVHL

polyubiquitination by the von Hippel-Lindau

RNAseq

RNA sequencing

siRNA

small interfering RNA

TBP

TATA-binding protein

TP

target protector

VEGFA

VEGF A