ALDH1A3 Regulations of Matricellular Proteins Promote Vascular Smooth Muscle Cell Proliferation

Summary

Vascular smooth muscle cell (VSMC) proliferation promotes intimal hyperplasia (IH) in occluding vascular diseases. Here we identified a positive role of ALDH1A3 (an aldehyde dehydrogenase) in this pro-IH process. The expression of ALDH1A3, but not that of 18 other isoforms of the ALDH family, was substantially increased in cytokine-stimulated VSMCs. PDGF(BB) stimulated VSMC total ALDH activity and proliferation, whereas ALDH1A3 silencing abolished this effect. ALDH1A3 silencing also diminished the expression of two matricellular proteins (TNC1 and ESM1), revealing a previously unrecognized ALDH1A3 function. Loss-of-function experiments demonstrated that TNC1 and ESM1 mediated ALDH1A3's pro-proliferative function via activation of AKT/mTOR and/or MEK/ERK pathways. Furthermore, ALDH inhibition with disulfiram blocked VSMC proliferation/migration in vitro and decreased TNC1 and ESM1 and IH in angioplasty-injured rat carotid arteries. Thus, ALDH1A3 promotes VSMC proliferation at least partially through TNC1/ESM1 upregulation; dampening excessive ALDH1A3 activity represents a potential approach to IH mitigation.

Graphical Abstract

Highlights

• •The ALDH1A3 isoform promotes vascular smooth muscle cell proliferation
• •ALDH1A3's function is mediated by its upregulation of TNC1 and ESM1
• •The pan-ALDH inhibitor drug disulfiram mitigates intimal hyperplasia

Pathophysiology; Vascular Remodeling; Molecular Biology; Molecular Mechanism of Behavior

Introduction

Over one million vascular reconstructions are performed in the United States each year, including coronary artery angioplasty/stenting or bypass vein grafting, arteriovenous fistula for dialysis access, and allograft transplantation. Unfortunately, (neo)intimal hyperplasia (IH), which narrows the vascular lumen, can lead to short-term or long-term failure in all these treatments. It is well documented that IH is principally caused by a vascular smooth muscle cell (VSMC) transition from normal to proliferative/migratory and synthetic phenotypes (Alexander and Owens, 2012). Therefore, blocking this VSMC phenotypic transition is key to IH attenuation. There are clinically used IH-attenuating drugs, i.e., rapamycin and paclitaxel, both anti-proliferative agents. However, the efficacy of rapamycin is reduced under diabetic conditions (Woods, 2013). Moreover, both rapamycin and paclitaxel have been found to increase the incidence of endothelial dysfunction and thrombosis (Inoue et al., 2011). These limitations strongly urge research efforts for better understanding of IH biology and identification of unstudied molecular mechanisms for optimal intervention.

Traditionally deemed as merely participating in cellular metabolism, in the past decade metabolic enzymes have refreshed researchers' interest for their diverse roles. For example, metabolic reprogramming has profound effects on epigenetic mechanisms that modify chromatin and alter gene expression. Aldehyde dehydrogenases (ALDHs) are a family of enzymes consisting of 19 isoforms that have distinct and overlapping functions and substrate specificities (Marchitti et al., 2007). Recent evidence suggests a vascular protective role of ALDH2, an isoform primarily mitochondrial-localized (Daiber et al., 2009, Mali et al., 2016a, Mali et al., 2016b). However, whether or how the 19 ALDH isoforms differentially regulate VSMC pathobiology is poorly understood. ALDH1A3 has received recent attention as a marker of normal and cancer progenitors. It is found to play an important role in maintaining cell cycle and promoting proliferation in various cell types (Dey et al., 2015, Cui et al., 2015, Zhang et al., 2014, Hegab et al., 2014, Szabo et al., 2013, Zou et al., 2012). Whether this enzyme regulates the expression of extracellular matrix (ECM) proteins was not known.

Matricellular proteins belong to a distinct category of ECM proteins. Rather than serving as ECM structural elements, they can evoke intracellular signaling and cell state changes. As such, they typically act as mediators or modulators of environment-cell interactions (Imanaka-Yoshida, 2016). It is well known that elevated ECM production is an acquired phenotype of VSMCs that undergo pro-IH phenotypic changes when stimulated in a cytokine-rich environment (Imanaka-Yoshida, 2016, Young et al., 2015). Interestingly, recent literature indicates that ECM proteins, in particular matricellular proteins, reciprocally influence VSMC phenotypes via various signaling pathways (Imanaka-Yoshida, 2016, Tratwal et al., 2015). TNC1 is a typical matricellular protein implicated as important in VSMC and IH pathobiology (Ishigaki et al., 2011). A relatively less known matricellular protein, ESM1 (or endocan), was recently deemed a marker of activated endothelial cells and reported to promote IH (Balta et al., 2015). As the majority of studies on ESM1 have been focused on endothelial cells, its expression and functional regulation in VSMCs are little known. More broadly, the relationships of TNC1 and ESM1 with ALDH1A3 have not been previously explored.

We hypothesized that ALDH activity plays an important role in VSMC proliferation. Indeed, our data revealed that cytokine-stimulated ALDH activation was accompanied by VSMC proliferation, and the ALDH1A3 isoform almost entirely accounted for these changes. More significantly, we found that knocking down ALDH1A3 reduced mRNA and protein levels of TNC1 and ESM1, both pro-proliferative matricellular proteins, as indicated by our data from VSMCs. ALDH1A3 regulation of ECM proteins was unanticipated as it has not been previously reported in any cell type. Importantly, pretreatment with a clinically approved pan-ALDH inhibitor, disulfiram (DSF), reduced VSMC proliferation and migration in vitro and IH in a rat carotid artery balloon angioplasty model. This suggests that targeting elevated ALDH activity may provide an amenable option to mitigate IH.

Results

Perivascular Administration of Disulfiram Reduces IH in a Rat Angioplasty Model

DSF is a pan-ALDH inhibitor. It was initially used to treat alcoholism and inadvertently found to have anti-cancer effects (Triscott et al., 2015). Considering that many alcoholic patients and/or patients with cancer may simultaneously have cardiovascular disorders, it is important to gain knowledge as to how DSF influences these disease conditions. Atherosclerosis and recurrent diseases (e.g., restenosis) following its treatment comprise the majority of vascular diseases, and they share a common etiology, i.e., IH or proliferation of the neointima. We therefore tested the effect of DSF in a well-established model of IH induced by balloon angioplasty in rat carotid arteries (Clowes et al., 1994) via perivascular administration (Wang et al., 2015) (for all methods, please refer to Supplemental Information). Animals were sacrificed at 14 days post angioplasty, and carotid arteries were collected and used for H&E staining. IH was measured as the neointima versus media area ratio. As shown in Figure 1, DSF-treated arteries displayed markedly reduced IH than that of vehicle control. This IH-mitigating effect of DSF inspired us to further investigate the underlying molecular and cellular mechanisms.

Figure 1.

Perivascular Administration of DSF Reduces IH In Vivo in a Rat Angioplasty Model

Rat carotid artery balloon angioplasty was performed followed by periadventitial administration of disulfiram (DSF, 9 mg/rat) or vehicle control (equal amount of DMSO). The animals were euthanized 14 days later, and the arteries were collected for histological analyses.

Shown are representative H&E-stained carotid artery cross sections and quantification of the intima versus media area ratio (I/M, measure of IH). The data were generated by averaging six to eight sections from each animal. The averages from six animals in each treatment group were then averaged (see the scatterplot). *p < 0.05, n = 6 animals, Student's t test. A pair of arrows define the neointima thickness. A, adventitia; M, media; N, neointima. Scale bar: 100 μm.

Cytokine Stimulation Elevates ALDH Activity of VSMCs, which Is Mainly Accounted for by the ALDH1A3 Isoform

IH occurs primarily because VSMCs transition into a proliferative state (Alexander and Owens, 2012). ALDH activity has been linked to cell proliferation in cancer cells (Wang et al., 2015). We therefore explored a possible relationship between VSMC proliferation and ALDH activity. Cells were subjected to cytokine stimulation (PDGF-BB, IL-1β, TNFα, or TGFβ1) for 20 h and collected for ALDH activity assay via flow cytometry (Figure S1). The same experiment was repeated by using both human primary aortic smooth muscle cells (AoSMC) and a mouse smooth muscle cell line (MOVAS). As shown in Figures 2A and 2B, compared with solvent control, where the ALDH-active population was 5.55% and 0.91% in AoSMC and MOVAS, respectively, all cytokine treatments significantly increased ALDH-active cell populations.

Figure 2.

The ALDH1A3 Isoform Predominantly Accounts for Cytokine-Stimulated Increase of VSMC Total ALDH Activity

(A) Increase of ALDH activity in AoSMCs (human) in response to treatment with cytokines.

(B) Increase of ALDH activity in MOVAS (mouse) in response to treatment with cytokines.

(C) Effects of cytokines on mRNA levels of 19 ALDH isoforms.

(D) Effects of cytokines on protein levels of ALDH1A3.

Cultured AoSMCs and MOVAS cells were starved for 24 h and then treated with PDGF-BB (40 ng/mL), IL1β (10 ng/mL), TGFβ1 (20 ng/mL), TNFα (20 ng/mL), or solvent control (4 mM HCl/0.1% BSA) for another 24 h. ALDH activity was determined using fluorescence-activated cell sorting analysis (FACS, see Figure S1) and presented as the percentage of the fluorescently positive cell number versus the total cell number. The mRNA expression of each ALDH isoform was determined using real-time qPCR. For Western blot replicates, samples of equal amount of total protein were loaded, and the exposure time length was kept consistent for the same protein to be detected. The densitometry of ALDH1A3 was normalized to that of beta-actin. Data are expressed as mean ± SE derived from three independent experiments. Student's t test: *p < 0.05, **p < 0.01, ***p < 0.001, each compared with solvent control; r.u., relative unit.

We then performed assays to distinguish which isoform mainly accounted for the increased ALDH activity after cytokine stimulation. Interestingly, real-time qPCR assay indicated that, among all the 19 isoforms, only the mRNA of ALDH1A3 was significantly increased (∼1.5–3 fold) in AoSMCs by stimulation with either of the four cytokines (PDGF-BB, IL1β, TNFα, or TGFβ1) (Figure 2C). In contrast, the other isoforms showed minor or insignificant changes. Cytokine-stimulated upregulation of ALDH1A3 was confirmed at the protein level by western blot assay (Figure 2D).

ALDH1A3 Genetic Silencing Abolishes PDGF-Stimulated Increase of VSMC ALDH Activity and Proliferation

We next investigated the specific role of ALDH1A3 in VSMC proliferation. Hereafter we used PDGF-BB as a stimulant throughout the in vitro assays. We decided to use PDGF-BB for the following considerations: (1) PDGF-BB is a mitogen that most potently stimulates VSMC proliferation and is most widely used for this experimental purpose (Wang et al., 2015). (2) Although IL1β resulted in greater stimulation of total ALDH activity than PDGF-BB (Figures 2A and 2B), ALDH1A3 protein levels stimulated by these two cytokines were essentially equal (Figure 2D). As shown in Figure 3A, western blots verified high ALDH1A3 gene knockdown efficiency. Consistent with the results presented earlier (Figure 2), PDGF-BB treatment of AoSMCs elevated ALDH activity by ∼2.5 fold, whereas ALDH1A3 gene silencing abrogated this effect (Figure 3B). Furthermore, PDGF-BB stimulated AoSMC proliferation by more than 2-fold, whereas ALDH1A3 silencing nearly abolished this phenotype (Figure 3C).

Figure 3.

ALDH1A3 Knockdown Reduces ALDH Activity, VSMC Proliferation, and the Expression of ESM1 and TNC1

(A) Western blots indicating efficient ALDH1A3 knockdown.

(B) ALDH1A3 knockdown abolishes the PDGF stimulation of total ALDH activity.

(C) ALDH1A3 knockdown abolishes the PDGF stimulation of AoSMC proliferation (CellTiter-Glo assay).

(D) ALDH1A3 knockdown reduces protein and mRNA levels of ESM1 and TNC1.

(E and F) ESM1 or TNC1 knockdown inhibits AoSMC proliferation.

Human AoSMCs were cultured and treated with solvent control or PDGF-BB as described in Figure 1. For ALDH1A3 knockdown, shRNA was initially used (in A and C) and then siRNA was used throughout (B and D). ALDH activity was determined using fluorescence-activated cell sorting analysis (FACS, Figure S2) and presented as the percentage of fluorescently positive cells versus total cells. mRNA expression of individual ALDH isoform was determined using real-time qPCR. Values were expressed as means ± SE derived from three independent experiments. Student's t test: **p < 0.01, ***p < 0.001, each compared with scrambled control.

Combined, these results demonstrate that ALDH1A3 is the isoform that is principally responsible for the cytokine-stimulated increase of ALDH activity that drives the VSMC phenotype change.

ALDH1A3 Genetic Silencing Diminishes the Expression of ESM1 and TNC1

We then asked the question as to what are the downstream effectors mediating ALDH1A3's regulations that lead to VSMC proliferation. We performed microarray analysis (Figure S3) to identify the genes whose expression can be upregulated by PDGF-BB in an ALDH activity-sensitive manner. By comparing the top 20 PDGF-upregulated genes with the top 20 genes whose PDGF-stimulated expression was reduced by pan-ALDH inhibition (Figure S3), we found two genes, ESM1 and TNC1, that appeared in both top-20 gene lists. To validate these two candidate genes, we determined whether ALDH1A3 knockdown changes their mRNA and protein levels. As shown by the RT-qPCR data in Figure 3D, ALDH1A3 silencing with small interfering RNA (siRNA) nearly eliminated ESM1 and TNC1 mRNAs. Moreover, the western blot data indicate that ALDH1A3 protein knockdown substantially reduced TNC1 and ESM1 proteins as well. These data revealed that ALDH1A3 potently regulates the expression of two matricellular proteins, TNC1 and ESM1.

Although TNC1 production has been previously reported to affect VSMC phenotypes (Fischer, 2007), no such report exists in the literature regarding ESM1's function. Data in Figures 3E and 3F confirm that siRNAs targeting ESM1 or TNC1 were effective in reducing target protein expression. As expected, TNC1-specific silencing largely abolished PDGF-stimulated AoSMC proliferation, consistent with the reported TNC1 function (Ishigaki et al., 2011). However, ESM1-specific silencing also effectively blocked this VSMC phenotype change, which has not been previously reported. These results suggest that TNC1 and ESM1 both are regulated by ALDH1A3 and may mediate ALDH1A3's functional effect on VSMC proliferation.

Silencing of either ALDH1A3 or ESM1 or TNC1 Mitigates PDGF-Stimulated Activation of AKT/mTOR and/or MEK/ERK Pathways

TNC1 and ESM1 both are matricellular proteins, a special category among ECM proteins that are not major structural elements but play a variety of signaling roles (Balta et al., 2015, Fischer, 2007). We inferred that TNC1 and ESM1 might regulate intracellular signaling that influences VSMC phenotypes. The best-known cytokine-responsive intracellular signaling pathways that strongly influence VSMC phenotypes include the AKT/mTOR/RPS6 pathway and the MAPK pathway. We therefore determined how the activation (phosphorylation) status of those pathways is impacted by silencing ALDH1A3, TNC1, and ESM1 individually. As shown in Figures 4A and 4B, although ALDH1A3 silencing almost completely diminished the levels of phosphor-AKT, phosphor-mTOR, and phospho-RPS6, silencing of ESM1 or TNC1 also substantially reduced phosphorylation of these proteins. Interestingly, silencing either ALDH1A3 or TNC1 diminished protein phosphorylation in the MAPK pathway, including p-MEK, p-ERK, p-p38, and also CyclinD1, a key player in cell division and proliferation, whereas silencing ESM1 did not have an effect on this pathway (Figure 4C).

Figure 4.

Effects of Gene Silencing of ALDH1A3, ESM1, or TNC1 on the AKT/mTOR and MEK/ERK Signaling Pathways in AoSMCs

(A and B) Western blots of phospho-AKT, phospho-mTOR, and phospho-Ribosomal protein S6 (RPS6) in AoSMCs transfected with a scrambled siRNA (Scr.) or an siRNA to knock down ALDH1A3, ESM1, or TNC1.

(C and D) Western blots of p-MEK, p-p38, and CyclinD1 in AoSMCs transfected with a scrambled siRNA or an siRNA to knock down ALDH1A3, ESM1, or TNC1.

In both (A) and (B), bar graphs show the quantification of protein levels. For western blot replicates, samples of equal amount of total protein were loaded, and the exposure time was kept consistent for the same protein to be detected. Values are expressed as means ± SE derived from three independent experiments. Student's t test: *p < 0.05, ***p < 0.001, each compared with scrambled control. P, phosphoprotein; T, total protein.

These results indicate that, although ALDH1A3 is responsible for the activation of AKT/mTOR and MAPK pathways, ESM1 and TNC1, both downstream of ALDH1A3, differentially regulate these pathways. Thus, the ALDH1A3-TNC1/ESM1 axis and the downstream regulation of AKT and MAPK pathways may constitute a molecular mechanism that underlies VSMC proliferation induced by elevated ALDH activity.

ALDH Inhibitor DSF Largely Recapitulates the Inhibitory Effect of ALDH1A3 Silencing on PDGF-Stimulated VSMC Proliferation and Intracellular Signaling

Since we elucidated that ALDH1A3 was the isoform that was upregulated and responsible for elevated total ALDH activity and VSMC proliferation after PDGF stimulation (Figure 3), we reasoned that pharmacological ALDH inhibition using DSF should influence VSMC phenotypes and ALDH1A3 downstream signaling similar to that produced by genetic silencing of ALDH1A3. Indeed, pretreatment with DSF dose-dependently inhibited PDGF-stimulated proliferation in all three types of VSMCs (AoSMC, MOVAS, and rat primary SMCs) used for the assays (Figure 5A). DSF also blocked PDGF-stimulated migration of MOVAS and rat SMCs (Figure 5B). We did not observe increased apoptosis in DSF-treated cells (10 μM, MOVAS), indicating low cytotoxicity (Figure 5C). Interestingly, we found that DSF pretreatment preserved the protein level of smooth muscle α-actin (αSMA), which was otherwise reduced by PDGF-BB (Figure 5D). Since reduced αSMA marks a de-differentiated VSMC state, this result further highlights an important role for ALDH1A3 in VSMC phenotypic changes.

Figure 5.

Pan-ALDH Inhibitor DSF Reverses PDGF-stimulated VSMC Phenotypic Changes

AoSMCs, MOVAS cells, or rat SMCs were cultured, starved, and then treated with solvent control or PDGF-BB (as described in Figure 1), with or without pretreatment with 10 μM DSF for 3 h.

(A) DSF inhibits PDGF-stimulated cell proliferation (determined by CellTiter-Glo viability assay). Mean ± SD, n = 3.

(B) DSF inhibits PDGF-stimulated cell migration (scratch assay). Shown are calcein-stained cells 20 h after PDGF-BB stimulation.

(C) DSF does not induce MOVAS cell apoptosis (flow sorting).

(D) DSF inhibits PDGF-stimulated activation of AKT and MEK signaling pathways. Shown are representative blots from at least two similar experiments. P, phosphoprotein; T, total protein.

Next, we determined the effect of DSF on ALDH1A3-directed intracellular pathways identified previously. DSF preferentially binds ALDH1A3 (IC₅₀ = 0.15μM) over other isoforms. As shown in Figure 5D, PDGF-BB stimulation dramatically increased phosphorylation levels of AKT, mTOR, RPS6, MEK, and ERK in AoSMCs, whereas DSF pretreatment abolished the PDGF-stimulated activation of almost all these proteins. In addition, PDGF-BB upregulated c-Myc, CDK1, and cyclin D1, all well-established factors that drive cell division and proliferation, and pretreatment with DSF attenuated PDGF-stimulated production of these proteins. The experiments using MOVAS largely repeated the results obtained from AoSMCs, although varied degrees of DSF inhibitory effects occurred (Figure 5D).

These results are consistent with the inhibitory effect of DSF on IH (Figure 1) and support ALDH inhibition as an approach to blocking the VSMC phenotypic changes, which are known to play a central role in the development of IH.

Balloon Angioplasty Induces ALDH1A3 Upregulation and DSF Inhibits TNC1 and ESM1 Protein Expression in Injured Rat Carotid Arteries

Based on the in vitro results demonstrating PDGF-stimulated ALDH1A3 upregulation and the specific role for ALDH1A3 in mediating VSMC proliferation (Figures 2 and 3), we inferred that ALDH1A3 protein levels could increase in the artery wall when stimulated by injury. We therefore evaluated ALDH1A3 level changes via immunohistochemistry on cross sections of uninjured and balloon-injured rat common carotid arteries that were collected 14 days after balloon angioplasty. Interestingly, the data in Figure 6A showed a 5-fold increase of ALDH1A3 in injured arteries compared with uninjured control. Staining mainly occurred in the neointima lesion and also the medial layer where VSMCs predominately reside, whereas the adventitial layer (mainly fibroblasts) was little stained. We also observed upregulation of the ALDH1A3 protein in the injured artery wall, which provides a logical basis for the mitigation of IH by DSF treatment (Figure 1). Furthermore, our data in Figure 6B indicated that protein levels of TNC1 and ESM1, both increased in injured (mainly in the neointima) versus uninjured arteries, were substantially reduced by treatment with DSF. This result is consistent with the in vitro study indicating that ALDH1A3 positively regulates the expression of these two matricellular proteins (Figure 3).

Figure 6.

ALDH1A3 Upregulation in Injured Rat Carotid Arteries

Rat carotid artery balloon angioplasty and periadventitial administration of DSF were performed as described in Figure 1. Artery cross sections were immunostained for ALDH1A3 (A) and TNC1 and ESM1 (B). Negative control staining was performed with the same procedures except that primary antibody was omitted. Positive staining was measured as colorimetric intensity per image field and quantified by averaging six to eight sections from each animal, and the averages from all of the six animals in each treatment group were then averaged (see the scatter plots). Student's t test: *p < 0.05, ***p < 0.001. A, adventitia; M, media; N, neointima. Scale bar: 100 μm.

In summary, the results in this study rationalize an approach to IH suppression via blockade of excessive ALDH activity (or ALDH1A3 protein) following arterial injury.

Discussion

Although IH is widely recognized as a common cause of essentially all occluding vascular diseases, its underlying molecular mechanisms remain inadequately explored. Here, our study reveals a previously unidentified ALDH1A3-directed mechanism that drives the pro-IH phenotypic changes of VSMCs. The major findings are the following: (1) In the entire ALDH family, ALDH1A3 is the isoform that predominantly accounts for the cytokine-stimulated elevation of ALDH activity and proliferation of VSMCs; (2) ALDH1A3 regulates the expression of matricellular proteins (TNC1 and ESM1); (3) Reducing ALDH activity effectively blocks VSMC transition to a pro-IH state. Ultimately, inhibition of ALDH activity can be translated into an IH-mitigating effect in a rat model of restenosis.

Elevated ALDH activity has been linked to cancer cell proliferation and tumorigenesis (Muzio et al., 2012), whereas its contribution to VSMC phenotypic changes and IH remained unclear. Earlier studies showed that ALDH1A1 was mainly responsible for the high ALDH activity in cancer and normal stem cells (Zou et al., 2012). However, this conclusion was challenged by evidence supporting ALDH1A3 as the primary isoform responsible for elevated ALDH activity in some progenitor cell and primitive human mammary cell populations. This discrepancy suggests that differential contributions of ALDH isoforms to total ALDH activity depend on specific cell type/state or disease type and stage. Indeed, via a comprehensive experimental survey using VSMCs, we found that ALDH1A3 was the isoform that showed predominant mRNA and protein level changes in response to stimulation of a panel of cytokines. Importantly, we obtained compelling evidence that specifically silencing ALDH1A3 alone effectively abolished PDGF-stimulated elevation of total ALDH activity. The involvement of another isoform (ALDH2) in cardiovascular disease has been reported (Mali et al., 2016a). However, our data showed that unlike ALDH1A3, ALDH2 mRNA did not increase under PDGF stimulation, but rather slightly decreased albeit without statistical significance. We thus speculate that ALDH2 may contribute in a different way, e.g., to mitochondrial activities, but not necessarily to PDGF-stimulated VSMC proliferation.

Identifying ALDH1A3 as predominantly responsible for PDGF-stimulated VSMC proliferation enabled us to focus on this isoform and further investigate its regulatory mechanisms. We hence made an unexpected finding of two matricellular proteins (TNC1 and ESM1) as lead candidate effectors downstream of ALDH1A3. ALDH1A3 has been reported to influence various cellular functions, typically progenitor cell proliferation and migration (Puttini et al., 2018). However, to the best of our knowledge, there is no report on ALDH1A3-regulated expression of ECM proteins, including matricellular proteins. This regulation (initially identified via microarray) was verified by inducible ALDH1A3 silencing and Western blotting. Moreover, silencing ALDH1A3, TNC1, or ESM1 inhibited the AKT/mTOR pathway, placing TNC1/ESM1 and ALDH1A3 in the same signaling context. Further consistent evidence was obtained from in vivo studies, indicating that, ALDH1A3 expression increased in balloon-injured arteries (versus uninjured control), whereas ALDH inhibition with DSF reduced both TNC1 and ESM1 proteins. Considering possible drug off-target effects, a definitive in vivo result requires future experiments using VSMC-specific ALDH1A3 knockout animals.

The finding of TNC1/ESM1 regulation by ALDH1A3 is significant, particularly in view of that matricellular proteins emerge as a unique category of ECM components, whose signaling importance in defining cell differentiation and phenotype has become increasingly appreciated (Puttini et al., 2018). Growing evidence supports a pro-IH role for both TNC1 and ESM1. A few studies showed that increased TNC1 promotes VSMC proliferation and migration (Malabanan et al., 2012, Ishigaki et al., 2011, Sawada et al., 2007). ESM1 was initially identified as an endothelial cell-specific marker mainly produced by ECs. Its function in prompting endothelial inflammation was recently supported by studies of various tissues (Balta et al., 2015). However, ESM1-dependent regulation of VSMC phenotypes was not specifically examined. In this regard, our data have led to another finding in regard to the role of ESM1 in regulating the behaviors of VSMCs; specifically, PDGF-BB stimulated ESM1 expression in VSMCs, and ESM1 silencing inhibited PDGF-stimulated VSMC proliferation.

Taken together, our results and literature evidence unequivocally support a prominent role for both TNC1 and ESM1 in promoting VSMC proliferation, consistent with a scenario where both proteins function in an ALDH1A3-dominated signaling cascade. Indeed, silencing ALDH1A3 maintained inactive AKT/mTOR and MEK/ERK, both pathways with established importance in VSMC proliferation, and silencing TNC1 had a similar effect. Interestingly, silencing ESM1 effectively inhibited the AKT/mTOR pathway but had little effect on MEK/ERK, suggesting a differential ESM1 downstream signaling mechanism as compared with TNC1.

Although intriguing, the result that ALDH1A3 as a metabolic enzyme regulates the gene expression of two matricellular proteins (TNC1 and ESM1) is seemingly counter-intuitive; mechanisms of this regulation remain to be determined. Traces of evidence can be found in the literature. For example, acetyl-Co-A, which could result from ALDH1A3-catalyzed metabolism, is a major source of the acetyl group for histone acetylation (Javaid and Choi, 2017), and H3K27 acetylation opens the chromatin structure thereby promoting transcription. Although plausible, this scenario requires detailed experiments to verify. It is also worth noting that most of the top 20 genes (including TNC1 and ESM1) whose expression is inhibited by DSF pretreatment are matricellular proteins, e.g., PTX3, ASPN, CTGF, PDCD1LG2, CYR61, CLDN11, and HAPLN1 (Figures S3–S5). Further in-depth investigation is warranted to address whether the top “hits” together implicate ALDH1A3-regulated transcription programs for ECM homeostasis.

In summary, our results outline a signaling scheme where increased ALDH1A3 protein/activity upregulates TNC1 and ESM1, which, possibly in an autocrine fashion, activate AKT/mTOR and/or MEK/ERK pathways that drive VSMC proliferation.

Limitations of the Study

One of the limitations is that we used DSF, a pan-ALDH inhibitor, in the microarray experiment. We designed this experiment based on the following rationale: ALDH1A3 but not other isoforms was prominently upregulated under cytokine stimulation; DSF preferentially binds to the ALDH1-class isoforms; a short time (<3 h) needed for pretreatment with a small molecule inhibitor favors observation of direct (versus secondary) effects on gene expression. Future studies will follow to strengthen the microarray results by using ALDH1A3 knockout or knockdown cells. Along this line, although DSF is relatively selective to ALDH1 isoforms over other isoforms (e.g., ALDH2), an ALDH1A3-specific inhibitor (not yet available) is highly desirable as some ALDH isoforms are cytoprotective (Mali et al., 2016a). Considering diverse functions of the ALDH family members, dampening the excessive ALDH activity (elevated owing to injury) rather than eliminating it may provide a viable option for IH mitigation. Another limitation of this study is that, although the in vivo DSF inhibition of IH is consistent with the in vitro blockage of VSMC proliferation by ALDH1A3 silencing, the in vivo process is too complex to be fully explained by in vitro results. For example, IH involves not only VSMCs but also ECs and fibroblasts (in the adventitia) in the artery wall. On the other hand, ECs were mostly removed during balloon angioplasty, ALDH1A3 staining was very low in the adventitia layer relative to that in neointima and media layers, and DSF was locally administered around the carotid artery. It is thus reasonable to speculate that the IH-mitigating effect of DSF may have largely occurred via VSMCs in the media and neointima. Nevertheless, more research to delineate ALDH1A3's specific role in IH in vivo and to develop isoform-specific ALDH1A3 inhibitors should facilitate translational advancements.

Methods

All methods can be found in the accompanying Transparent Methods supplemental file.

Published: September 27, 2019