Acrolein Decreases Endothelial Cell Migration and Insulin Sensitivity Through Induction of let-7a

Abstract

Acrolein is a major reactive component of vehicle exhaust, and cigarette and wood smoke. It is also present in several food substances and is generated endogenously during inflammation and lipid peroxidation. Although previous studies have shown that dietary or inhalation exposure to acrolein results in endothelial activation, platelet activation, and accelerated atherogenesis, the basis for these effects is unknown. Moreover, the effects of acrolein on microRNA (miRNA) have not been studied. Using AGILENT miRNA microarray high-throughput technology, we found that treatment of cultured human umbilical vein endothelial cells with acrolein led to a significant (>1.5-fold) upregulation of 12, and downregulation of 15, miRNAs. Among the miRNAs upregulated were members of the let-7 family and this upregulation was associated with decreased expression of their protein targets, β3 integrin, Cdc34, and K-Ras. Exposure to acrolein attenuated β3 integrin-dependent migration and reduced Akt phosphorylation in response to insulin. These effects of acrolein on endothelial cell migration and insulin signaling were reversed by expression of a let-7a inhibitor. Also, inhalation exposure of mice to acrolein (1 ppm x 6 h/day x 4 days) upregulated let-7a and led to a decrease in insulin-stimulated Akt phosphorylation in the aorta. These results suggest that acrolein exposure has broad effects on endothelial miRNA repertoire and that attenuation of endothelial cell migration and insulin signaling by acrolein is mediated in part by the upregulation of let-7a. This mechanism may be a significant feature of vascular injury caused by inflammation, oxidized lipids, and exposure to environmental pollutants.

Abbreviations

HNE

4-hydroxy-trans-2-nonenal

IPA

Ingenuity Pathway Analysis

LPS

lipopolysaccharide

PAPC

1-palmitoyl-2-arachidonoyl-sn-glycero-3-phosphocholine

POVPC

1-palmitoyl-2-(5-oxovaleryl)-sn-glycero-3-phosphocholine

HBSS

Hank's buffered salt solution

The endothelium is a critical regulator of vascular tone and thrombosis and overall cardiovascular homeostasis. It also plays an important role in regulating insulin signaling, metabolite flux, and angiogenesis (Aird, 2007; Bonetti et al., 2003; Giacco and Brownlee, 2010; Schmidt et al., 2007). Hence endothelial dysfunction is an early and robust indicator of cardiovascular disease (CVD) risk, and endothelial injury has been linked to insulin resistance, hypertension, impaired angiogenesis, and the development of atherosclerotic lesions. Although several CVD risk factors induce endothelial dysfunction, there is increasing recognition that exposure to toxic environmental pollutants or reactive endogenous metabolites contributes to CVD risk by inducing endothelial injury (Bhatnagar, 2004, 2006). Because of its unique interface between blood and smooth muscle cells, the endothelium is vulnerable to injury caused by blood-borne toxicants and endothelial dysfunction is a key feature of cardiovascular injury induced by exposure to endogenous toxicants and environmental pollutants. Nevertheless, the mechanisms by which such exposures cause endothelial injury and dysfunction remain unclear.

Of the several endogenous metabolites and environmental pollutants that cause endothelial injury, acrolein is one of the most ubiquitous and reactive toxicants (Feron et al., 1991). It is an abundant environmental pollutant generated during the combustion of organic material and, therefore, it is a major component of cigarette and wood smoke as well as automobile exhaust (Feron et al., 1991). Acrolein is also a natural constituent of several food substances; it is generated in high amount during high temperature cooking and frying and is also generated in vivo during the metabolism of several pharmaceuticals (Ludeman, 1999). In addition, acrolein is also a byproduct of myeloperoxidase present in neutrophils (Anderson et al., 1997) and a reactive end product of lipid peroxidation reactions (Uchida et al., 1998). Consequently, high concentrations of acrolein, or protein-acrolein adducts, are present at sites of injury (Vasilyev et al., 2005), inflammation (Anderson et al., 1997), and atherosclerotic lesions (Shao et al., 2005). Our previous studies have shown that acrolein exposure induces dyslipidemia (Conklin et al., 2010), vascular injury (Conklin et al., 2006), and endothelial dysfunction (Conklin et al., 2009), leading to the acceleration of atherogenesis (Srivastava et al., 2011), destabilization of atherosclerotic lesions (O'Toole et al., 2009), disruption of cardioprotective signaling (Wang et al., 2008), and the induction of dilated cardiomyopathy (Ismahil et al., 2011).

Acrolein is believed to exert effects on the proliferative, apoptotic, inflammatory, and signaling properties of multiple cell types through mechanisms that have been attributed, in part, to its adduction with proteins or nucleic acids, alteration of cation fluxes, depletion of intracellular glutathione, and the induction of endoplasmic reticulum and oxidative stress (Feron et al., 1991; Haberzettl et al., 2009; O'Toole et al., 2009). However, whether acrolein can influence microRNA (miRNA) expression remains unknown. Therefore, the present study was designed to investigate whether acrolein affects miRNA levels in endothelial cells and to assess the contribution of these changes to acrolein-induced endothelial injury. Numerous studies have characterized the miRNA repertoire of cells or tissues of distinct differentiation states or pathological conditions. In some cases, phenotypic properties characteristic of these conditions has been directly attributed to changes in individual miRNA species. Here we show that acrolein exposure alters the levels of several miRNAs in cultured human umbilical vein endothelial cells (HUVECs). We found significant changes in members of the let-7 family that mediate, in part, the deleterious effects of acrolein on endothelial migration and insulin signaling. These findings suggest that endothelial miRNA repertoire is sensitive to reactive aldehydes such as acrolein and that this sensitivity may be related to the mechanisms by which these toxicants contribute to early endothelial injury and increased CVD risk.

MATERIALS AND METHODS

Reagents

Acrolein was purchased from Sigma Chemical Co. (St. Louis, MO; no. 110221), whereas 4-hydroxy-trans-2-nonenal (HNE), 1-palmitoyl-2-(5-oxovaleryl)-sn-glycero-3-phosphocholine (POVPC), and oxidized 1-palmitoyl-2-arachidonoyl-sn-glycero-3-phosphocholine (PAPC) were prepared in the laboratory according to published procedures (Srivastava et al., 1998). TaqMan primers for individual miRNA species and rtPCR reagents were from Applied Biosystems (Carlsbad, CA), whereas let-7a and non-specific miRNA inhibitors were obtained from Dharmacon (Lafayette, CO). The β3 integrin antibody was from Becton Dickinson (San Jose, CA; no. 61140), whereas the K-Ras (no. sc-30) and Cdc34 (no. sc-5616) antibodies were purchased from Santa Cruz Biotechnologies (Santa Cruz, CA). Antibodies against phosphorylated Akt (p-Akt) (no. 9271), phosphorylated eNOS (p-eNOS) (no. 9570), pan Akt (no. 9272S), and pan eNOS (no. 9572S) were from Cell Signaling Technology (Danvers, MA), whereas the actin antibody was from Sigma Chemical Co. (no. A5060). Calcein AM was from Invitrogen (Carlsbad, CA; no. C-1430), whereas R&D Systems supplied the vitronectin (Minneapolis, MN; no. 2349-VN).

miRNA expression analysis

Total HUVEC RNA was isolated using the miRNeasy kit (Qiagen, Germantown, MD; no. 217004) and its quality was verified with an Agilent 2100 Bioanalyzer (Agilent, Santa Clara, CA). Total RNA (100 ng) from four biological replicates was used in miRNA microarray analysis by the Genomics Core at the University of Louisville. Briefly, an Agilent miRNA microarray slide (8 ×15K, CA G4470A), which contained 470 unique human miRNAs, was used to hybridize with Cyanine 3-pCP-labeled RNA. The slide was scanned with the aid of the Agilent DNA microarray scanner G2505B (Agilent). The one-color microarray image (*.tif) was extracted with the aid of Feature Extraction 7.0 (Agilent). The eight raw data files (*.txt) were imported into GeneSpring GX9.0 software (Agilent), normalized, and analyzed. Differentially expressed miRNAs were identified using the unpaired Student's t-test. The putative biological functions affected by differentially expressed miRNAs were analyzed with the aid of Ingenuity Pathway Analysis (IPA) software. Individual miRNA species were quantified by real-time PCR using specific TaqMan miRNA primers (Applied Biosystems) on iCycler (Biorad, Hercules, CA) or 7900HT Fast Real Time PCR (Applied Biosystems) instruments.

Cell and animal experiments

The HUVECs, from a mixed pool of donors, were obtained from Lonza (Walkersville, MD; no. CC-2519) and maintained in endothelial basal media (EBM) supplemented with the appropriate bullet kit (Lonza; no. CC-3124). Cells were used between passages 3 and 9. Before treatment, the cells were starved overnight in media containing 0.1% serum. The cells were then washed in Hank's buffered salt solution (HBSS: 20-mM Hepes, pH7.4, 135-mM NaCl, 5.4-mM KCl, 1-mM MgCl₂, 2-mM CaCl₂, 2-mM NaH₂PO₄, 5-mM glucose) and incubated with the appropriate stimulus (see Table 3) or HBSS as a control for 2 h. After washing, the cells were returned to normal media. RNA was isolated 4 h later to determine miRNA changes, whereas other cells were lysed 24 h later to determine changes in protein expression. Cell lysates were prepared in 1% Triton, 0.1% SDS, 500-mM NaCl, 50-mM Tris pH7.4, 10-mM MgCl₂, 0.5% sodium deoxycholate, and 1X protease inhibitors (Sigma; no. P8340) and total protein concentration measured (BioRad; no. 500-006). Levels of β3 integrin and Cdc34 were compared by Western blotting. Bands were quantified by densitometry and normalized to actin levels in the same blot. K-Ras was identified by first immunoprecipitating the protein from 150–300 μg of lysate using 1 μg of a specific antibody followed by standard Western blotting. miRNA inhibitors (50nM final) were transfected using the siPORT NeoFX reagent (Ambion, Austin, TX; no. AM4511) according to the manufacturer's instructions.

TABLE 3. Effects of oxidants and inflammatory stimuli on let-7a levels in HUVECs and mouse aorta.

Treatment	Fold change ± SE	n	p
10-μM acrolein	1.42 ± 0.18	13	0.02
10-μM HNE	0.97 ± 0.08	10	>0.05
25-μM HNE	0.66 ± 0.09	3	>0.05
50-μM HNE	1.01 ± 0.04	3	>0.05
10-μM POVPC	0.82 ± 0.08	3	>0.05
10-μM acrylic acid	0.51 ± 0.14	4	0.05
10-μM allyl alcohol	0.80 ± 0.27	5	>0.05
10-μg/ml ox PAPC	0.80 ± 0.001	3	0.001
25-μg/ml ox PAPC	0.84 ± 0.11	3	>0.05
50-μg/ml ox PAPC	0.59 ± 0.36	3	>0.05
200-μM H2O2	0.92 ± 0.12	9	>0.05
10-ng/ml TNF-α	0.97 ± 0.14	5	>0.05
1-μg/ml LPS	1.08 ± 0.14	5	>0.05
Inhaled acrolein
(a) Intact aorta	1.87 ± 0.25	25	0.04
(b) denuded aorta	1.26 ± 0.03	5	>0.05

Levels of let-7a were measured in HUVECs after a 2-h treatment with the listed stimuli and a 4-h recovery period. For measurements in the aorta, the mice were exposed to 1-ppm acrolein or filtered air for 4 days and changes in let-7a were measured immediately after the end of the last exposure. The miRNA levels were measured by rtPCR. Values are fold change ± SE versus control.

Male C57BL6 mice (n = 25) were obtained from Jackson Labs (Bar Harbor, ME) and were used at 12 weeks of age. The mice were exposed to 1-ppm volatile acrolein or high efficiency particulate arresting (HEPA)-filtered air for 6 h/day for 4 consecutive days as previously described (Wheat et al., 2011). Immediately after the last exposure, the mice were euthanized, the aortas extracted and total RNA was isolated with the miRNeasy kit. All animal procedures were approved by the University of Louisville IACUC and done according to the APS's Guiding Principles of Care and Use of Animals.

Migration

Haptotactic HUVEC migration was analyzed by a modified Transwell method. Control or acrolein-treated cells were grown in media containing 0.1% serum overnight and harvested with Cell Dissociation Buffer (Life Technologies, Carlsbad, CA; no. 13151-014). An aliquot of 2.5 × 10⁴ cells was allowed to migrate on vitronectin (Vn)-coated Transwells for 6 h and during the final 30 min, 5-μM calcein AM was added to the lower chamber. After washing the filters three times in PBS containing 0.2% Tween, cells remaining on the upper side of the filter were removed with a Q-tip and the fluorescence of calcein-stained cells on the bottom side quantified using a plate reader (ex. 485 nm, em. 530 nm). Results are expressed as a percentage of the fluorescence of an identical aliquot of input cells (2.5 × 10⁴ cells) and are normalized to the migration of mock-transfected and untreated cells.

Insulin signaling

HUVECs treated with or without acrolein were incubated in media containing 0.1% serum overnight. The cells were then treated with 2-nM calyculin A (Millipore, Billerica, MA; no. 208851) for 30 min prior to addition of 1-μM insulin (Humulin-RP, Eli-Lilly, Indianapolis, IN). After 60 min, the cells were lysed in a buffer consisting of 10% NP40, 0.25% sodium deoxycholate, 150-mM NaCl, 50-mM Tris pH 7.4, 1X protease inhibitor solution (Sigma), and 1X phosphatase inhibitor solution (Thermo Scientific, Rockford, IL; no. 1862495). Protein concentration was measured and 50 μg of lysate was used to detect p-Akt and total Akt by Western blotting. To determine aortic insulin signaling, tissues were excised and cleaned in cold PBS. For some experiments, the vessels were denuded with a single hair of moose mane (Natural Duhmom, White River Brand; BassPro Shops) that was inserted into the lumen of the aorta and vigorously rubbed against the endothelium for 10–15 strokes. This procedure causes functional impairment of the endothelium as tested by acetylcholine-induced relaxation of phenylephrine-pre-contracted aorta ex vivo (Cherng et al., 2009). The tissues were then incubated in HBSS for 1 h (5% CO₂, 37°C) followed by stimulation with 100-nM insulin for 15 min. Western blot analysis was performed using antibodies (Cell Signaling Technology) to detect p-Akt (Ser473), p-eNOS (Ser1177), Akt, eNOS, and actin.

Statistical analysis

Differentially expressed miRNAs that showed an absolute fold change of at least 1.5 between acrolein treatment and no treatment, and a p < 0.05 (n = 4) were identified and targeted for further study. Western blot quantification is presented as mean ± SEM. An unpaired t-test was used to compare levels between untreated and acrolein-treated samples. One-way ANOVA with Bonferroni's post hoc test was used for comparing multiple groups (SigmaStat; SPSS, Chicago, IL). p < 0.05 was considered significant.

RESULTS

Exposure to Acrolein Alters Endothelial Cell miRNA Content

Physiological or pathological changes in cellular miRNA levels regulate protein expression and influence cell function. Hence, we examined whether acrolein could affect vascular cell function by altering miRNA levels. To test this, we incubated HUVECs with 10-μM acrolein or HBSS for 2-h and isolated total cellular RNA after a 4-h recovery period. These RNAs were then subjected to miRNA array analysis using an Agilent human miRNA microarray platform that contained 470 unique miRNAs. Expression levels of the 15 most abundant miRNAs detected in HUVECs and their levels after acrolein exposure are shown in Figure 1. From this set, statistically significant changes (of at least 1.5-fold) were observed for miR-638, miR-370, miR-188, miR-565, and let-7a. Using this same criterion for analyzing all miRNA species detected in the array, we identified 12 miRNAs that were significantly upregulated by at least 1.5-fold over control cells and 15 miRNAs that were significantly downregulated by at least 1.5-fold (Table 1).

FIG. 1.

Acrolein-induced changes in HUVEC miRNA. Illustrated are the 15 most abundant miRNAs detected in untreated HUVECs and their corresponding levels after acrolein exposure as determined by microarray analysis. An asterisk denotes those species whose levels are changed by at least 1.5-fold and whose changes demonstrate statistical significance (p < 0.05).

TABLE 1. Acrolein-induced changes in miRNA levels in endothelial cells.

miRNA	Fold change	p	miRNA	Fold change	p
mir-493-5p	2.61	5.2E−06	mir-198	0.64	0.002
mir-191*	2.26	0.01	mir-452	0.62	0.002
mir-223	2.25	0.02	mir-370	0.62	0.002
mir-636	1.87	0.02	mir-770-5p	0.60	0.001
mir-218	1.58	0.01	mir-296	0.60	0.01
mir-629	1.52	0.04	mir-671	0.56	0.004
mir-98	1.52	0.03	mir-663	0.56	0.002
hsa-let-7a	1.50	0.04	mir-638	0.55	0.01
hsa-let-7d	1.50	0.02	mir-601	0.55	0.02
hsa-let-7e	1.50	0.04	mir-622	0.54	0.001
mir-301	1.50	0.02	mir-188	0.52	0.001
mir-224	1.50	0.01	mir-623	0.52	0.004
			mir-765	0.50	0.002
			mir-373*	0.50	0.01
			mir-565	0.22	3.4E−05

Cultured HUVECs were incubated with 10-μM acrolein in HBSS for 2-h and total RNA was isolated after a 4-h recovery in complete media. Relative levels of miRNA were detected using a microarray. The table shows individual miRNAs whose expression was significantly (p < 0.05) upregulated (left side) or downregulated (right side) by at least 1.5-fold.

To confirm the array data, changes in selected miRNAs from Table 1 were analyzed by rtPCR. These analyses were performed using RNA samples isolated from new HUVEC exposures. Appropriate changes in the most highly upregulated (miR-493-5p) and downregulated (mir-565) miRNAs from Table 1 were identified by this technique (Table 2). Acrolein-induced changes in mir-223 and let-7a were also confirmed by PCR (Table 2). To gain insight into functional consequences of acrolein-induced changes, target proteins and the biological pathways affected by these proteins were analyzed using the IPA software (Fig. 2). Prominent among the associated molecular and cellular functions affected were pathways involved in cell cycle, movement, growth, and proliferation.

TABLE 2. miRNA levels in acrolein-treated endothelial cells.

miRNA	Fold change (array)	Fold change ± SE (rtPCR)	p
mir-493-5p	2.61	1.69 ± 0.16	0.03
mir223	2.25	3.42 ± 1.16	0.02
let-7a	1.50	1.42 ± 0.18	0.02
mir-565	0.22	0.13 ± 0.04	<0.01

A quantitative analysis of four of the miRNAs listed in Table 1 was performed using RNA samples from additional HUVEC exposures (n ≥ 3). Listed are the levels as determined in the array experiment or by rtPCR (n ≥ 7 determinations) in the additional exposures.

FIG. 2.

Pathway analysis. Illustrated are selected, altered miRNAs (trapezoids) and their predicted targets which are involved in the regulation of cell morphology, cell cycle, and cell growth.

Acrolein-Induced Changes in Let-7 miRNAs and their Target Proteins

Based upon the IPA analysis, we focused on the let-7 family of miRNAs, which play a prominent role in cell growth and differentiation (Roush and Slack, 2008). Three members of this family (let-7a, let-7d, and let-7e) were among those upregulated miRNAs detected in the array experiment (Table 1). To confirm these results, we performed rtPCR in additional HUVEC exposures. As shown in Table 2, rtPCR results were in agreement with array data, showing a consistent increase in let-7a. To characterize this HUVEC response further, we examined whether the upregulation of let-7a was specific for acrolein or whether this was a more general consequence of exposure to other oxidized lipids or stress-related signals. The results of these experiments showed that the upregulation of let-7a was specific for acrolein, but not for the longer chain aldehydes HNE and POVPC (Table 3). The aldehydic group appeared to be essential as the acid or alcohol derivatives of acrolein (acrylic acid and allyl alcohol, respectively) did not increase let-7a (Table 3). Oxidized lipids (oxPAPC) as well as oxidative and inflammatory stimuli (H₂O₂, TNF-α, and LPS) were likewise ineffective in upregulating let-7a (Table 3). To determine if acrolein alters let-7a expression in vivo, we quantified its levels in the aortas of mice exposed to acrolein by inhalation. After 4 days of exposure (1 ppm x 6 h/day), we observed a 1.87 ± 0.25 (p = 0.04) fold increase in let-7a compared with aortic levels of mice exposed to filtered air (Table 3). When the aortas of acrolein-exposed mice were denuded of the endothelial layer prior to RNA isolation, the increase (1.26 ± 0.03-fold) in let-7a was less than that observed for the intact aorta. Levels of let-7a in the denuded aorta were not significantly different from control tissue (Table 3). This suggests that the increase in aortic let-7a is mostly due to increases in the endothelium.

Changes in the expression of miRNAs are accompanied by corresponding changes in the expression of their protein targets. Several predicted mRNA targets of the let-7 miRNAs have been verified by genetic, biochemical, and Western blotting approaches. Thus, we next determined whether acrolein exposure also downregulated the previously documented let-7 target proteins, β3 integrin (Muller and Bosserhoff, 2008), Cdc34 (Legesse-Miller et al., 2009), and K-Ras (He et al., 2010). We found that treatment with acrolein led to a significant decrease (p < 0.05) in the expression of β3 integrin (57 ± 8%), Cdc34 (73 ± 8%), and K-Ras (55 ± 4%) (Fig. 3), indicating that let-7 upregulation by acrolein results in the downregulation of its target proteins.

FIG. 3.

Acrolein-induced changes in protein expression. (A) Levels of β3 integrin and Cdc34 in lysates of mock-treated HUVECs (M) or those cells exposed to acrolein (A) were identified by Western blotting. Actin levels were used as a loading control. K-Ras levels in the same lysates were determined after immunoprecipitation and Western blotting. Illustrated are blots from a representative experiment. (B) Group data for β3 integrin, Cdc34, and K-Ras expression (n = 3; *p < 0.05).

Acrolein Inhibits Endothelial Cell Migration in a Let-7a-Dependent Manner

To elucidate the functional consequences of let-7a upregulation, we first examined the effects of acrolein on β3 integrin-dependent HUVEC migration. In these experiments, the matrix protein vitronectin was used as a substrate, as the migration of HUVECs on vitronectin is mediated by the αvβ3 integrin heterodimer (Leavesley et al., 1993). Cells incubated with 10-μM acrolein demonstrated decreased migration (38 ± 4% of control; p < 0.05) when compared with untreated cells and the extent of migration of acrolein-treated cells was indistinguishable from migration on bovine serum albumin (BSA), an irrelevant substrate (Fig. 4A). To verify that these effects were due to increases in let-7a, we transfected cells with a let-7a inhibitor prior to acrolein treatment. These transfectants exhibited near normal levels of migration after treatment with acrolein (87 ± 6% of control) (Fig. 4A). On the other hand, migration remained at basal levels (48 ± 14% of control; p < 0.05) when cells were transfected with a non-specific inhibitor prior to acrolein treatment (Fig. 4A). Importantly, the migratory capacity of treated cells was correlated with β3 integrin expression levels. Specifically, exposure to acrolein resulted in a loss of this integrin subunit compared with untreated cells (Fig. 4B), whereas β3 expression was retained in the let-7a inhibitor transfectants (Fig. 4B). In contrast, β3 expression was decreased in cells transfected with a non-specific miRNA inhibitor (Fig. 4B). These results suggest that the inhibition of cell migration by acrolein was mediated by let-7a.

FIG. 4.

Acrolein exposure attenuates HUVEC migration in a let-7a-dependent manner. (A) Mock transfectants and HUVECs transfected with 50-nM let-7a (let7a-I) or non-specific (NS-I) inhibitors as indicated were treated with or without acrolein and migration on vitronectin (Vn) or BSA quantified (*: p < 0.05 vs. mock; #: p < 0.05 vs. let-7a-I). Also Illustrated are Western blots for β3 integrin and actin in a representative experiment (B) and group data (C). (*: p < 0.05 vs. let-7a-I; n = 5).

Acrolein Induces Insulin Resistance in a Let-7a-Dependent Manner

Recent work has shown that let-7 miRNAs attenuate insulin-induced responses by targeting proteins in this signaling pathway (Frost and Olson, 2011; Zhu et al., 2011). To determine whether acrolein could impair aortic insulin sensitivity by upregulating let-7a, we initially examined levels of p-Akt and p-eNOS in the aortas of exposed animals. These proteins are downstream effectors of the insulin receptor and their phosphorylation is commonly used to index insulin sensitivity. We found that aortas from mice exposed to 1-ppm acrolein for 4 days and then stimulated with insulin ex vivo had reduced levels of p-Akt and p-eNOS when compared with aortas from animals inhaling filtered air (Fig. 5). To confirm let-7a dependence, we characterized these effects in HUVECs. These cells responded to insulin and demonstrated a 1.4 ± 0.2 (p < 0.05) fold increase in p-Akt compared with untreated cells (Figs. 6A and B). However, levels of p-Akt remained at baseline levels when HUVECs were treated with acrolein 24 h prior to insulin stimulation. This attenuation of insulin signaling was reversed in cells transfected with a let-7a inhibitor prior to acrolein treatment, but not in cells transfected with a non-specific inhibitor (Figs. 6A and B). In the absence of acrolein, mock transfectants or cells transfected with miRNA inhibitors displayed normal levels of insulin-stimulated Akt phosphorylation (Fig. 6B). These observations suggest that upregulation of let-7a by acrolein negatively regulates endothelial insulin signaling.

FIG. 5.

Acrolein exposure impairs aortic insulin signaling. Aortas isolated from mice exposed to air or acrolein (1 ppm, 6 h/day) for 4 days were variably stimulated ex vivo with insulin (+I: 100-nM insulin, 15 min). Homogenates were prepared and levels of phospho-Akt (A) and phospho-eNOS (B) were determined by Western blotting. Upper panels present representative blots, whereas lower panels present group data. n = 4–5; *: p < 0.05.

FIG. 6.

Acrolein exposure attenuates insulin signaling in a let-7a-dependent manner. Wild-type HUVECs, mock transfectants, and those transfected with either 50-nM let-7a (let7a-I) or non-specific (NS-I) inhibitors were treated with or without acrolein and variably stimulated with insulin as indicated. Levels of p-Akt and actin were then determined in lysates of these cells by Western blotting. Illustrated are representative blots (A) and normalized data plotted relative to untreated cells (B). (n = 5; *: p < 0.05 vs. let-7a-I transfectants).

DISCUSSION

The results of this study show that acrolein, an in vivo product of lipid peroxidation and myeloperoxidase, and a common environmental pollutant, alters the levels of specific miRNAs in endothelial cells. Among the upregulated miRNAs, significant changes were observed in let-7a, both in cultured endothelial cells and in the aortas of mice inhaling volatile acrolein. Moreover, let-7a upregulation was accompanied by corresponding changes in its protein targets, including the β3 integrin. Functionally, acrolein attenuated both β3 integrin-dependent migration and endothelial insulin signaling in a let-7a-dependent manner. These results suggest that cellular miRNA levels can be altered by acute exposure to toxicants such as acrolein. Consequent changes in protein expression and cell function could define a mechanism for endothelial dysfunction arising from continual endogenous or exogenous exposure to aldehydes, as in ongoing inflammatory conditions such as atherosclerosis.

Humans are exposed to acrolein from both endogenous and exogenous sources. Basal concentrations in healthy adults have been estimated to be between 1 and 2μM (Carmella et al., 2007), but these levels may increase due to ongoing disease or inflammation (Takamoto et al., 2004). In addition, humans could be exposed to acrolein from the environment. Acrolein is a ubiquitous environmental pollutant, which is present in high amounts (6–8 ppm) in exhaust gases from petrol or diesel engines (Smith et al., 2002). Acrolein is also a major (50–70 ppm) component of tobacco smoke (Dong and Moldoveanu, 2004). Indeed, the urine from smokers has been estimated to contain 2–4-μM acrolein. Thus, the concentrations of acrolein used in our in vitro studies are within the range achieved by endogenous inflammation and that encountered by smokers or those with occupations involving acute or chronic exposures to acrolein-containing pollutants (e.g., bus drivers, firefighters). Similarly, our mouse exposure levels are also within a range of potential environmental exposures. Based on our exposure protocol (1 ppm x 6 h/day) and a murine breathing rate of 25 ml/min, we estimate that about 21 μg of acrolein (assuming 100% uptake) would be absorbed in 2-ml blood. This equates to a total acrolein concentration of 178μM in the blood. However, it is likely that this simple calculation is an overestimation of free acrolein in the blood because this aldehyde is highly reactive and it is readily metabolized to less toxic metabolites. Thus, the effective concentration available to a given tissue (aorta) is likely much less than 178μM and more in accord with the levels used in our in vitro studies with HUVECs (10μM).

It has been shown that acrolein exposure induces dyslipidemia, accelerates atherogenesis, and destabilizes atherosclerotic lesions (Conklin et al., 2010; O'Toole et al., 2009; Srivastava et al., 2011). A central theme underlying these pathological changes is the presence of a damaged and dysfunctional endothelium. Indeed, our previous studies have shown that exposure to acrolein leads to endothelial dysfunction and deficits in acetylcholine-induced relaxation (Conklin et al., 2009). The results presented here suggest that acrolein also causes changes in the levels of several miRNAs and that this effect could mediate long-term changes in endothelial function. In particular, we found a significant increase in the levels of the let-7 miRNAs. The let-7 miRNAs comprise a highly conserved family and were among the first miRNAs identified (Roush and Slack, 2008). Studies with members of this family have identified their prominent role in stem cell differentiation, tissue development, and cell proliferation (Roush and Slack, 2008). In addition, several let-7 family members have been found to function as tumor suppressors and a decrease in their expression is associated with an increased incidence of cancer (Roush and Slack, 2008). Thus, the upregulation of let-7a, let-7d, and let-7e could broadly impact a range of endothelial functions and pathophysiological responses to endogenous or exogenous exposures.

Although our results showed that the upregulation of let-7a in response to acrolein exposure was rather modest (∼1.5-fold; Table 1), we found that acrolein strongly limited β3 integrin expression and β3 integrin-dependent HUVEC migration (Fig. 4). We speculate that the rather large functional effects of modest increases in let-7a may be due to the changes in other members of the let-7 family (let-7d, let-7e; Table 1) that were also upregulated by acrolein. Both let-7d and let-7e target the same mRNAs as let-7a. Thus, the robust impact of modest let-7a upregulation on β3 integrin expression and function could be due to the cumulative effect of these three let-7 miRNAs on β3 mRNA. The fact that β3 integrin expression did not return to baseline in the presence of the let-7a inhibitor (Fig. 4B) is consistent with the notion that other let-7 family members, or other gene expression regulators, are also impacted by acrolein. Nevertheless, expression of the let-7a inhibitor in the absence of acrolein treatment resulted in an approximate 60% increase over baseline migration (data not shown), confirming that β3 integrin is a valid target of let-7a. The fact that migration does not return to this heightened level in those samples treated with the let-7a inhibitor and acrolein is also consistent with the involvement of additional regulators or additional let-7 family members.

The results of cell migration studies reported here identify a potentially new role for the let-7 miRNAs in angiogenesis. The regulation of angiogenesis by miRNAs is complex, as individual miRNAs could have opposing effects (Landskroner-Eiger et al., 2013). Like the let-7 family, miR-221, miR-222, and miR-92a inhibit angiogenesis. On the other hand, miR-296, miR-126, miR-27b, miR-130a, and miR-210 appear to promote angiogenesis. As we did not observe changes in any of these additional miRNAs, it appears likely that any potential effects of acrolein on angiogenesis would be largely mediated by members of the let-7 family. The fact that acrolein can be generated in high amounts at sites of inflammation (Anderson et al., 1997) and vascular lesions (Shao et al., 2005) suggests that local let-7 elevations and β3 integrin downregulation could limit pro-angiogenic and vascular repair responses at these sites. Moreover, because acrolein can be generated by myeloperoxidase, this mechanism could also account for inflammation-dependent attenuation of left ventricular remodeling after ischemia/reperfusion injury (Vasilyev et al., 2005). This decrease in capacity to develop new vessels or repair existing, damaged vessels could exacerbate vascular disorders and contribute to the vascular pathology symptomatic of chronic acrolein exposure.

Recent studies have shown that exposure to traffic-related pollution (Thiering et al., 2013) or tobacco smoke (Thiering et al., 2011) leads to the development or exaggeration of systemic insulin resistance in humans. It has been suggested that these effects may be due to the induction of a pro-inflammatory state or changes in the metabolic capacity of adipose tissue (Mendez et al., 2013). Thus, the development of insulin resistance is largely viewed as a long-range systemic response to toxicant exposure. In contrast, our results suggest that acrolein-induced let-7a upregulation defines a novel, direct mechanism for the development of endothelial insulin resistance. Because during diet-induced obesity, the development of insulin resistance in the endothelium is thought to precede that in other organs (Kim et al., 2008), a let-7a-based mechanism of endothelial insulin resistance could be an important contributory factor to other inflammation-based mechanisms leading to systemic insulin resistance. Furthermore, because acrolein is also a significant component of several food substances, our results raise the interesting possibility that the increase in insulin resistance secondary to the consumption of acrolein-rich (e.g., deep-fried) foods may be related to changes in let-7 levels.

The possibility that exposure to soluble or volatile pollutants can affect tissue function by changing miRNA levels has only recently been recognized (Hou et al., 2011; Jardim, 2011). For exposure to air toxics, these changes have been typically observed in the tissue of initial exposure (lungs). For instance, exposure to diesel exhaust particles (Jardim et al., 2009) or cigarette smoke (Izzotti et al., 2009; Schembri et al., 2009) has been shown to alter miRNA levels both in the lungs and the airway epithelial cells. However, there is also evidence to suggest that airborne toxics can affect miRNA levels systemically. For example, exposure to metal-rich, airborne particulate matter or polycyclic aromatic hydrocarbons has been found to alter the expression of leukocyte and myocardial miRNAs (Bollati et al., 2010; Farraj et al., 2011). Our results support the notion that local exposure could have systemic effects as we detected let-7a changes in the aortas of mice inhaling volatile acrolein. Obscuring the interpretation of these earlier studies is the fact that these airborne exposures are of complex and variable composition and the identity of the causative agent(s) is unclear. In this respect, our study is unique in identifying a specific component (acrolein) that could contribute to the cardiovascular effects of exposure to complex pollutant mixtures. Furthermore, in previous studies, exposures that lasted several days or even longer were used. In such protocols, it is difficult to distinguish between the direct effects of exposure on miRNA expression and indirect effects that might be secondary to another systemic response. In contrast, our studies, which identified changes in HUVEC miRNA repertoire 6-h after initiation of exposure, suggest that acrolein can directly affect miRNA levels in the absence of systemic confounders.

Exposures to environmental pollutants or toxic chemicals could impact miRNA expression through genetic or epigenetic mechanisms and this is especially relevant given that many miRNA coding sequences lie within susceptible regions of the genome (Calin et al., 2004). However, the effects of acrolein exposure on miRNA expression are rapid (≤ 6-h), making it less likely that large-scale genetic alterations are involved. Moreover, it seems unlikely that acrolein could impart the same genetic changes on different members of the let-7 family which are encoded by different loci both in humans and mice. Alternatively, acrolein could impact miRNA expression through direct interactions. RNAs in general are sensitive to the binding of electrophiles (Nishimura et al., 2009). As a potent electrophile, acrolein could preferentially form a sequence-specific adduct with let-7a (and other miRNAs listed in Table 1). The miRNAs modified in this way may demonstrate enhanced or restricted access to processing by Dicer, thus affecting overall expression levels as recently proposed (Izzotti and Pulliero, 2014). Although this awaits further study, our findings clearly indicate that exposure to pollutants and toxicants can induce rapid changes in miRNA and target protein expression, which can profoundly affect endothelial cell function. Hence, reversal of such changes may be a gainful therapeutic approach to limit pathophysiological responses to toxic substances such as acrolein.

FUNDING

National Institutes of Health (ES019217, GM103492, HL55477, P20 GM103436, ES024030); University of Louisville Center for Environmental Genomics and Integrated Biology (NIEHS/NIH; P30 ES014443).