Epigenetic regulators of the revascularization response to chronic arterial occlusion

Abstract

Peripheral arterial disease (PAD) is the leading cause of lower limb amputation and estimated to affect over 202 million people worldwide. PAD is caused by atherosclerotic lesions that occlude large arteries in the lower limbs, leading to insufficient blood perfusion of distal tissues. Given the severity of this clinical problem, there has been long-standing interest in both understanding how chronic arterial occlusions affect muscle tissue and vasculature and identifying therapeutic approaches capable of restoring tissue composition and vascular function to a healthy state. To date, the most widely utilized animal model for performing such studies has been the ischaemic mouse hindlimb. Despite not being a model of PAD per se, the ischaemic hindlimb model does recapitulate several key aspects of PAD. Further, it has served as a valuable platform upon which we have built much of our understanding of how chronic arterial occlusions affect muscle tissue composition, muscle regeneration and angiogenesis, and collateral arteriogenesis. Recently, there has been a global surge in research aimed at understanding how gene expression is regulated by epigenetic factors (i.e. non-coding RNAs, histone post-translational modifications, and DNA methylation). Thus, perhaps not unexpectedly, many recent studies have identified essential roles for epigenetic factors in regulating key responses to chronic arterial occlusion(s). In this review, we summarize the mechanisms of action of these epigenetic regulators and highlight several recent studies investigating the role of said regulators in the context of hindlimb ischaemia. In addition, we focus on how these recent advances in our understanding of the role of epigenetics in regulating responses to chronic arterial occlusion(s) can inform future therapeutic applications to promote revascularization and perfusion recovery in the setting of PAD.

1. Peripheral arterial disease

Peripheral arterial disease (PAD) is a chronic condition caused by atherosclerotic blockage(s) of the arteries in the lower limbs.¹ Indeed, the majority of patients with symptomatic PAD are burdened with a complete occlusion in the iliac/femoral arterial system spanning between the aorta and the foot.¹ PAD is the leading cause of lower limb amputations in the USA,² and it is estimated that >200 million people worldwide, in high-, middle, and low-income countries, have PAD. Thus, PAD is truly a global problem.³ The underlying mechanisms for the functional limitations associated with PAD are haemodynamic in origin.⁴ Reduced blood flow to the lower extremities leads to an oxygen supply that is insufficient for meeting the metabolic demands of distal tissues. Moreover, PAD patients often exhibit endothelial dysfunction^5,6 and microvascular impairments (i.e. reduced capillarity⁷ and increased arteriolar rarefaction⁸) that further exacerbate distal tissue ischaemia and its pathological sequelae.

Surgical- and catheter-based revascularization methods are currently the preferred treatment for patients presenting with claudication.⁹ Such interventions can preserve life and limb in correctly chosen patients; however, many PAD patients are not amenable to surgical revascularization options. Further, many that do undergo surgery receive little to no long-term benefit.⁹ These disappointing outcomes have motivated new therapeutic strategies that seek to use biochemical, molecular, and cellular-based approaches to induce endogenous revascularization (i.e. angiogenesis and/or arteriogenesis) to restore lower limb perfusion. The stimulation of angiogenesis, or the growth of new capillaries from pre-existing vessels, is important as capillary density is reduced in PAD patients.^7,9,10 However, it is also imperative to restore the driving pressure to the distal tissue via the structural lumenal expansion of collateral arteries (i.e. arteriogenesis) bypassing the occlusion(s).^8,11,12 Nonetheless, large-scale therapeutic arteriogenesis clinical trials have been largely unsuccessful to date,^13–17 highlighting our incomplete understanding of both the pathophysiology of PAD and the underlying mechanisms governing revascularization responses to chronic arterial occlusion.

That said, it is certainly well-known that the pathophysiology of PAD is governed by both genetic and epigenetic mechanisms. Genetics clearly contribute to PAD development, though they do not fully explain PAD pathophysiology.^18–20 The strong association of ‘environmental’ risk factors (e.g. age, smoking, and diabetes) with PAD suggest a potential role for epigenetic regulation in PAD. Indeed altered histone modifications and DNA methylation have been associated with all three of these risk factors (aging,²¹ smoking,^22–25 and diabetes^26,27). Epigenetic regulation of cardiovascular disease,^28,29 vascular gene expression,³⁰ and biomechanical signaling^31–33 have been reviewed elsewhere and will not be covered here.

2. Revascularization response to hindlimb ischaemia

While there is no perfect animal model of PAD per se, the murine hindlimb ischaemia model does recapitulate many key features of PAD (i.e. angiogenesis, arteriogenesis, and altered skeletal muscle tissue composition). As such, the hindlimb ischaemia model is commonly employed to both investigate the underlying mechanisms governing revascularization responses to chronic arterial occlusion and facilitate the testing of new therapeutic revascularization approaches in vivo. The hindlimb ischaemia model typically involves surgically occluding the femoral and/or iliac artery on one side. In some versions of this model, a segment of the major artery is also resected. These interventions serve to reduce blood flow to distal tissue, with the severity of distal ischaemia depending upon the anatomical location of the occlusion/resection along the major artery. It should be noted that hindlimb ischaemia is most often induced in mice via an acute occlusion procedure, which obviously contrasts with the gradual chronic occlusion seen in most cases of PAD. Thus, this represents a weakness of the model that should be taken into consideration when extrapolating experimental results to human PAD. In addition to the obvious advantages associated with genetic manipulations, mice are usually chosen for this model due to their limited collateral network, which more closely resembles the human vascular system. In contrast, more robust collaterals are present in rats.³⁴ Following placement of an occlusion in the iliac and/or femoral artery, mice display a complex sequence of responses that are summarized in Figure 1. These responses include, but are not limited to, the expansion of capillary networks by angiogenesis in the ischaemic tissue of the lower limb and the arteriogenic lumenal growth and expansion of collateral arteries bypassing the occlusion^34–38 in the upper hindlimb. Angiogenesis and arteriogenesis are briefly summarized in the following sections.

Figure 1.

Overview of revascularization responses in the mouse hindlimb ischaemia model. (A) Relative anatomical positions of the ‘upper’ and ‘lower’ hindlimb regions. The ‘upper’ hindlimb contains a network of collateral arteries capable of bypassing an occlusion of the femoral artery. Skeletal muscle (pink oval) in the ‘lower’ hindlimb is supplied by feed arteries distal to the upper hindlimb collateral network. (B) Pre-occlusion conditions. Arrow size in the upper hindlimb reflects relative magnitude of collateral blood flow. (C) Conditions immediately after occluding a section of the femoral artery (white region). Collateral blood flow increases due to a pressure drop downstream of the occlusion. In turn, this leads to increased shear stress, which then initiates signalling for collateral arteriogenesis. Meanwhile, blood flow is reduced in the lower hindlimb, causing the muscle to become hypoxic. (D) Within hours to days of femoral artery occlusion, muscle in the lower hindlimb becomes necrotic and/or fibroadipose and atrophies. Microvessel rarefaction occurs. Hypoxia and inflammation begin to drive angiogenesis. (E) Depending on the severity of the ischaemic stimulus, arteriogenesis and angiogenesis serve to restore, or at least partially restore, lower hindlimb perfusion and muscle tissue composition.

2.1 Angiogenesis

Angiogenesis, a term which refers to the generation of new capillaries via endothelial sprouting from existing capillaries and small venules, can occur in response to both hypoxia and inflammation. In both human PAD and the murine hindlimb ischaemia model, skeletal muscle becomes ischaemic, hypoxic, and inflamed. This pathological state of sustained hypoxia stimulates the activity of hypoxia-inducible factor 1-alpha, which serves as a transcription factor to promote the activity of vascular endothelial growth factor (VEGF) and other pro-angiogenic proteins. In turn, enhanced growth factor expression triggers endothelial cell (EC) degradation of basement membrane and surrounding matrix, followed by EC migration, proliferation, and the formation of nascent capillary tubes with a new basement membrane. Many of these new capillaries will eventually be ‘pruned’ away, while others will enlarge and become stabilized by pericytes or smooth muscle cells (SMCs).³⁹ The growth of these new vessels provides additional perfusion to the ischaemic tissue beds; however, excessive apoptotic cell death in the hypoxic regions can counter the effects of hypoxia-mediated angiogenesis, resulting in reduced capillary density in the ischaemic muscle.^9,40 Exercise has been shown to promote angiogenesis in the ischaemic tissue, leading to improved functional performance of the muscle.¹⁰

2.2 Arteriogenesis

It has been known since the 1780s that collateral artery networks capable of compensating for major arterial occlusions⁴¹ are intrinsic to many tissues and organs. The arteriogenic growth of these endogenous collateral arteries that provide alternate paths for blood flow around arterial occlusions can prevent ischaemic injury. In humans, arteriogenic collateral growth has been shown to occur in the arterial system of the lower limbs, heart, and brain.^41–43 Altogether, arteriogenesis is a multifaceted, highly coordinated signalling cascade involving the recruitment, migration, and proliferation of multiple cell types, as well as re-organization of the extracellular matrix. The progression of this complex process can be broken down into three stages: initiation, growth, and maturation.⁴⁴

The key initiating stimulus for arteriogenesis is an increase in shear stress.^42,45 Upon occlusion of a major artery, the downstream vascular network experiences a drop in pressure. In turn, this generates a steep pressure gradient across collateral arteries that bypass the occlusion, resulting in increased blood flow and a concomitant increase in wall shear stress exerted on the endothelium.⁴⁴ ECs then sense and transduce these changes in wall shear stress into biophysical, biochemical, and gene signalling responses (i.e. mechanotransduction),⁴⁶ via mechanosensory complexes (e.g. CD31/VE-Cadherin/VEGFR2).^47,48

Altered shear stress activates the ECs lining collateral arteries, inducing an inflammatory response that begins the collateral growth phase of arteriogenesis.⁴⁵ Both in vivo and in vitro studies have shown that increased shear stress promotes the expression of a wide array of inflammatory cytokines, chemokines, and adhesion molecules.^49–51 Shear stress mediated up-regulation of adhesion molecules, particularly ICAM-1,⁵² is of particular importance as adhesion molecules are critical for collateral growth, primarily through their role in the recruitment of leucocytes.⁵³ Additionally, the up-regulation of a variety of chemokines (e.g. CCL2, GM-CSF, and CXCL1) helps to induce the recruitment of bone-marrow derived cell populations that propagate the growth process.^54–56 While multiple leucocyte populations play a role initiating collateral artery growth (see review by Meisner and Price⁴⁴), monocytes are the most widely studied. The recruitment of monocytes and subsequent pericollateral accumulation of macrophages has been shown to be necessary for collateral artery growth^53,57–64. Recruited monocytes and macrophages act as ‘point-sources’ of paracrine growth factors (e.g. INFβ,⁶⁵ TNFα,⁶⁶ TGFβ,⁶⁷ and FGF⁶⁸) that promote the proliferative activity of ECs, SMCs, and adventitial fibroblasts.^69–71 Additional growth factors are simultaneously released from an initial, coordinated breakdown of basement membrane components (e.g. desmin and laminins) and of the internal elastic lamina by matrix metalloproteases (e.g. MMP-2 and MMP-9).⁴⁴ Furthermore, this inflammatory signalling induces the differentiation of collateral artery SMCs from a contractile to a synthetic phenotype.⁴⁹

As vascular cells proliferate and collateral vessels grow, there is a corresponding expansion of the media and the formation of a neointima.⁴⁹ Over time, as lumenal diameter increases, the signalling environment shifts from pro-inflammatory to anti-inflammatory.⁵⁰ The number of perivascular macrophages begins to decrease, paralleling a reduction in vascular cell proliferation.^49,59 SMCs return to a contractile phenotype⁴⁹ and a mature basement membrane is re-established.^72,73 Ultimately, the largest, most developed collateral vessels tend to mature and stabilize, albeit at the expense of the smaller, less developed collateral vessels which undergo eventual regression.^49,74,75 Outward lumenal growth is hypothesized to continue until normalization to the original shear stress level (i.e. the shear stress ‘set-point’) has been achieved.^44,76–78 Indeed, there is considerable evidence that arteries adapt to chronic changes in blood flow by undergoing adjustments of their internal diameters.^45,78–83 Though its magnitude can vary in different vascular beds,^84,85 numerous studies have either experimentally or theoretically supported the concept of a shear stress set-point.^{78,80,82,86–91}

3. Epigenetics

Epigenetics is essentially a ‘bridge between genotype and phenotype’, referring to the modification of gene expression and phenotype in a sequence-independent manner.⁹² Epigenetic mechanisms, such as DNA methylation, histone modifications, and non-coding RNAs, can regulate gene expression by altering DNA accessibility and chromatin structure.⁹² The significance of epigenetics in vascular biology, as regulators of molecular signalling and as potential therapeutic targets to treat disease, is now well-recognized.^30,31,93

3.1 Histone modifications

Chromatin is a ‘beaded chain’ DNA-protein complex consisting of a string of nucleosomes (i.e. DNA nucleotides wrapped around an octamer of histone proteins) connected by linker DNA.⁹⁴ The higher order packing of chromatin can be dynamically regulated by post-translational modifications to histone proteins.⁹⁴ Typically, more open chromatin enables for transcription factor binding whereas more condensed chromatin represses gene expression.³⁰ There are numerous (>100) post-translational histone modifications, though the most well-characterized types are acetylation and lysine methylation.^93,95 Histone acetylation is governed by histone acetyltransferase (HAT) and deacetyltransferase (HDAC) activity.^93,96 Histone acetylation tends to correlate with chromatin accessibility and transcriptional activity.⁹⁵ Histone lysine methyltransferases and demethyltransferases regulate histone lysine methylation.⁹⁶ Importantly, histone lysine residues can be mono-, di-, or tri-methylated and both the specific lysine residue and number of methyl groups on that residue determine the specific effect on transcription.⁹³

3.2 DNA methylation

DNA methylation is the addition of a methyl group to a cytosine base pair at the 5ʹ carbon-position, (i.e. 5-methylcytosine), occurring most often at CpG dinucleotide sites.⁹³ Methylation occurs through the activity of DNA methyltransferases (DNMTs), including DNMT1, DNMT3a, and DNMT3b.^30,96,97 DNMT1 is thought to pre-dominate post-development, preferentially methylating hemi-methylated DNA, whereas DNMT3A and DNMT3B preferentially target fully unmethylated DNA during embryogenesis and gametogenesis.⁹⁷ DNMT1, however, is also capable of de novo methylation.⁹⁷ Demethylation can occur passively or actively, the latter thought to be mediated by the ten-eleven translocation enzyme family (TET1, TET2, TET3) and by oxidation of 5-methylcytosine to 5-hydroxymethylcytosin.^98,99 Approximately 40–60% of human genes are associated with dense regions of CpG sites with higher than expected CG content called CpG islands.^30,100,101 In healthy cells, CpG islands are typically hypo-methylated, thereby allowing for an open chromatin structure and transcriptional activity.¹⁰⁰ However, DNA methylation of gene promoter regions is associated with repression of gene expression.^102–104 Inhibition of gene expression is thought to be achieved by physically impeding transcription factor binding to the gene promoter and/or to methyl-CpG-binding domain proteins.³²

3.3 Non-coding RNA

In addition to post-translational DNA methylation and histone modifications, non-coding RNA can regulate gene expression. Non-coding RNA can be classified broadly as long non-coding RNAs (lncRNAs) and small non-coding RNAs [e.g. microRNAs (miRNAs)]. lncRNAs are arbitrarily defined as >200-nucleotide long, non-coding transcripts that may be 5ʹ-capped, spliced, or polyadenylated, and are typically not well-conserved.^93,105 lncRNAs are typically generated by the same transcriptional machinery as messenger RNAs (mRNAs).¹⁰⁶ Regulation of gene expression can be mediated by lncRNAs via RNA-, protein-, and DNA-binding interactions,¹⁰⁶ with several lncRNAs having been shown to regulate chromatin remodelling, splicing, and by acting as miRNA sponges.¹⁰⁵ In contrast to lncRNAs, small non-coding RNAs are well-conserved, <200-nucleotides long, and are endogenously processed by endonucleoases.¹⁰⁵ There are a number of different classes of small non-coding RNAs, including miRNAs, PIWI-interacting RNAs (piRNAs), short interfering RNAs (siRNAs), tRNA-derived RNA fragments (tRFs), and small nucleolar RNAs (snoRNAs). Of these, miRNAs have been the most extensively studied to date. miRNAs are ∼22 nucleotide non-coding RNAs that are endogenous regulators of gene expression.¹⁰⁷ Mature miRNA incorporates into the RNA-induced silencing complex and then binds to a target mRNA, usually in the 3ʹ untranslated region of the mRNA.¹⁰⁸ MicroRNA binding acts to suppress target gene expression by inhibiting mRNA translation to protein or by promoting mRNA degradation,¹⁰⁹ depending on miRNA-target complementarity.¹¹⁰

4. Epigenetic regulators of the response to hindlimb ischaemia

4.1 Histone post-translational modifications

Studies exploring the role of the epigenetic mechanisms of histone post-translational modifications and DNA methylation in experimental models of PAD are limited. Transcriptional co-activator p300-CBP-associated factor (PCAF) is a HAT that promotes histone (H3 and H4) acetylation and the transcriptional activation of multiple pro-inflammatory genes.¹¹¹ Both genetic and pharmacological inhibition of PCAF impairs blood flow recovery and arteriogenesis following hindlimb ischaemia via suppression of inflammatory signalling and leucocyte recruitment.¹¹¹ Additionally, though HDAC inhibition is a promising treatment for several ischaemic diseases, it has been shown to have variable effects on the response to hindlimb ischaemia.¹¹² To this end, in a femoral arterial excision model, male C57BL/6 mice treated with a class I specific HDAC inhibitor (MS275) exhibited significant muscle atrophy and increased fibrosis. In contrast, treatment with the class IIa specific HDAC inhibitor (MC1568) increased the number of regenerating muscle fibres, though it delayed their terminal differentiation. Moreover, the class IIa inhibitor also modestly increased the arteriolar density in ischaemic tissues, though neither the class I nor the class IIa HDAC inhibitor affected capillary density or reperfusion.¹¹² In another study,¹¹³ HDAC9 specific inhibition was shown to reduce EC tube formation, sprouting, and retinal vessel outgrowth. Defects in angiogenesis were rescued by both HDAC9 overexpression and inhibition of miR-17-20a. Additionally, perfusion recovery was reduced ∼50% in HDAC9 knockout mice compared with littermate controls following induction of hindlimb ischaemia. Interestingly, HDAC9^−/− mice appeared to exhibit sex-dependent responses, as the females trended towards improved perfusion recovery.¹¹³ Interestingly, ex vivo treatment of isolated endothelial progenitor cells with histone modifying drugs, such as the HDAC inhibitor trichostatin A or a combination treatment consisting of H3K27me3 (GSK-343) and an HDAC inhibitor panobinostat, have been shown to improve the efficacy of ECFCs in transiently improving limb perfusion following hindlimb ischaemia.^114,115

4.2 DNA methylation

Few previous studies to date have investigated the regulation of the ischaemic response by DNA methylation. Two of these studies focused on a conserved family of methyl-CpG-binding proteins (MBD1, MBD2, MBD3, MBD4, and MeCP2) which are responsible for reading alterations to the DNA methylome and for enacting subsequent changes in transcriptional activity.¹¹⁶ EC-specific MeCP2-null mice were protected against a TGF-β induced impairment in angiogenesis.¹¹⁷In vitro, knockdown of MBD2 significantly enhanced angiogenesis and provided protection against peroxide-induced apoptosis.¹¹⁸In vivo, Mbd2^−/− mice were protected against hindlimb ischaemia, demonstrating a significant improvement in perfusion recovery, along with increased capillary and arteriole densities following femoral arterial ligation (FAL), compared with littermate controls.¹¹⁸ MBD2 binds to hyper-methylated CpG regions in the gene promoters of eNOS and VEGFR2; therefore, loss of MBD2 increases eNOS and VEGFR2 expression.¹¹⁸ These studies demonstrate that DNA methylation is likely a crucial, though under-studied, mechanism regulating vascular growth following arterial occlusion.

Recently, DNA methylation has been shown to differentially regulate flow-mediated endothelial gene expression through DNMT1.^119–122 Furthermore, our group has identified a key linkage between DNMT1 and arteriogenesis in response to femoral artery ligation.¹²³ A schematic outline of this recent study is provided in Figure 2. In brief, we first demonstrated that collateral artery segments exposed to an increase in shear stress magnitude, without a change in flow direction, display limited arteriogenic capacity when compared with segments exposed to both increased shear stress magnitude and reversed flow direction. We then determined that these non-reversed flow collateral segments exhibit global DNA hyper-methylation in vivo. In vitro, ECs exposed to the non-reversed waveform exhibited increased DNMT1 expression, genome-wide hyper-methylation of significantly regulated gene promoters, and a DNMT1-dependent reduction in pro-arteriogenic monocyte adhesion. Furthermore in vivo study revealed that, in non-reversed flow collateral artery segments, DNMT1 inhibition rescued arteriogenic capacity and returned the elevated shear stress back to its original set point. Collectively, these results demonstrate that DNMT1-dependent DNA hyper-methylation constrains arteriogenesis by dampening EC mechanosensing, which effectively augments shear stress set-point.¹²³ The epigenetic regulation of shear stress set-point may therefore have a critical impact on both endogenous and therapeutic arteriogenesis in patients with arterial occlusive disease.

Figure 2.

Differential collateral growth responses due to flow reversal after femoral artery occlusion reveal a putative role for DNMT1-dependent EC hypermethylation in limiting arteriogenic capacity. (A) Flow patterns in upper hindlimb collaterals before (yellow arrows) and after (white arrows) femoral artery occlusion. Collateral artery segments exposed to reversed flow (orange annotations) exhibit amplified arteriogenesis when compared with segments exposed to non-reversed flow (blue annotations). (B) As evidenced by 5-hydroxymethylcytosin staining (5-mc; green) and other approaches, EC (CD31; magenta) DNA is hypo-methylated in reversed flow segments when compared to non-reversed flow segments. (C) Arteriogenic capacity, as delineated by steady-state collateral lumenal diameter, is limited in non-reversed collateral arteries with hyper-methylated EC DNA. Inhibiting DNMT1 reduces shear stress set-point, thereby releasing the limitation on arteriogenic capacity and allowing the collateral artery to grow to a larger diameter.

4.3 Non-coding RNA

Non-coding RNAs have been more widely studied in experimental chronic arterial occlusion models than any other epigenetic mechanism (Table 1). The vast majority of these studies have focused on miRNAs. In these studies, candidate miRNAs have been primarily identified by one of three methods, namely (i) investigation of miRNA expression levels in isolated cells, (ii) selection of miRNAs with functions known to be important for the response to hindlimb ischaemia, and (iii) by explicitly examining changes in miRNA expression associated with responses to hindlimb ischaemia in vivo.

Table 1.

Non-coding RNAs known to regulate vascular growth and perfusion recovery in the ischaemic mouse hindlimb model.

Non-coding RNA(s)	Ref.	IH affected response?			Comments/additional mechanisms
		AN	AR	PR
miR-17∼92	124	Y	Y	Y	Integrin subunits α5 dependent
miR-17∼92	125	U	Y	Y	miRNA-19 repression of WNT signalling
miR-223	126	Y	Y	Y	β1 integrin dependent
miR-503	127	Y	Y	Y	cdc25A mechanism
miR-146	128	N	Y	Y	Modulation of Irak2 and NOS3 expression
miR-199a	129	N	Y	Y	Pericollateral monocyte recruitment affected; postulated Cav1-dependent mechanism
miR-15a	131	Y	U	Y	FGF and VEGF dependent responses
14q32 miRs	132	Y	Y	Y	Studied miR-329, miR-487b, miR-494, and miR-495
miR-126	133	N	Y	N	Postulated role for Spred-1 and PIK3R2
miR126∼25	137	Y	U	Y	Bone marrow-derived stromal cell mechanism
miR-let-7g	147	Y	U	Y	Enhanced endothelial progenitor cell recruitment
miR-100	148	Y	Y	Y	Identified mTOR-dependent mechanism
miR-155	149	Y	Y	Y	Differential angiogenic and arteriogenic responses; SOCS-1 and AGTR1-dependent mechanism
miR-132/212	150	U	Y	Y	Ras-MAPK pathway mechanism; direct targeting of Spred1 and Rasa1
miR-93	156	Y	Y	Y	Polarization of macrophages to M2
miR-352	157	NA	Y	NA	AV shunt model; IGFR2/CI-M6P mechanism
miR-15b-15p	158	Y	Y	Y	AKT3 dependent mechanism
miR-150	159	Y	Y	Y	Src-mediated mechanism
MALAT1	160	Y	U	Y	Endothelial phenotypic switch to proliferation
MALAT1	164	Y	U	Y	VEGF-R2 dependent mechanism
MEG3	165	U	U	Y	MEG3 mechanism for reduced angiogenesis with aging

IH, ischaemic hindlimb; AN, angiogenesis and/or capillary density changes; AR, collateral arteriogenesis and/or small arteriole growth; mTOR, mammalian target of rapamycin; PR, perfusion recovery; NA, not applicable; Y, yes; N, no; U, unknown or not reported.

Several studies have identified relevant miRNAs by examining miRNA expression levels in isolated ECs. To this end, the miR-17∼92a cluster was found to be highly expressed in human umbilical vein endothelial cells (HUVECs) and to negatively regulate angiogenesis in vitro.¹²⁴ Furthermore, systemic delivery of miR-92a antagomirs enhanced perfusion recovery in C57BL/6 mice following hindlimb ischaemia.¹²⁴ In another study, endothelial-specific deletion of the miR-17∼92 cluster was also seen to increase foot perfusion 14 days following induction of hindlimb ischaemia, as well as arteriole density in ischaemic limbs.¹²⁵ Within this miR-17∼92 cluster, antagonism of miR-19a/b in aged mice improved the perfusion recovery after ischaemia by targeting key aspects of the WNT signalling pathway.¹²⁵ MicroRNA-223 was also found to be highly expressed in freshly isolated human, murine, and porcine ECs and to abrogate EC migration and sprouting in vitro, via regulation of β1 integrin.¹²⁶ Compared with littermate controls, miR-223^−/y mice exhibited enhanced perfusion recovery (nearly a two-fold), increased α-SMC⁺ staining in adductor muscles, and increased capillary density in gastrocnemius muscles following hindlimb ischaemia.¹²⁶ In ECs cultured under high-glucose and ischaemic conditions, expression of miR-503 was found to be up-regulated. Adenovirus mediated overexpression of a miR-503 decoy (i.e. reduced miR-503 expression) abrogated diabetes-induced impairment in post-ischaemic angiogenesis and blood flow recovery via a cdc25A mechanism.¹²⁷

Our group has recently investigated the roles of the mechanosensitive miRNAs miR-199a and miR-146 in perfusion recovery and collateral arteriogenesis following FAL.^128,129 A schematic representation of our approach to studying the role of miR-199a in the hindlimb ischaemia model is provided in Figure 3. First, we identified candidate regulators of arteriogenesis by comparing differential miRNA expression in ECs exposed to shear stress waveforms corresponding to moderate and amplified arteriogenesis responses in vivo. Among these candidate miRNAs, miR-199a and miR-146 were significantly down-regulated by the amplified arteriogenesis (reversed flow) waveform. In the mouse hindlimb ischaemia model, overexpression of miR-199a limited foot perfusion, arteriogenesis, and monocyte recruitment. In contrast, inhibition of miR-199a elicited complete foot perfusion recovery, substantially enhanced collateral arteriogenesis via increased pericollateral macrophage recruitment, and considerably improved gastrocnemius muscle tissue composition.¹²⁹In vivo inhibition of miR-146a also resulted in complete foot perfusion recovery in a hindlimb ischaemia model, and animals treated with the inhibitor displayed a marked enhancement in arteriogenesis (as measured by collateral vessel lumenal diameter) relative to scramble-treated controls.¹²⁸ Thus, miR-199a, miR-146, and other mechanosensitive miRNAs may represent new targets for the therapeutic stimulation of arteriogenesis and the treatment of PAD.

Figure 3.

Differential collateral growth responses due to flow reversal after femoral artery occlusion identify miR-199a-5p as a potent regulator of arteriogenesis. (A) Upper hindlimb collateral circulation visualized by ink-filling, with outsets showing amplified arteriogenesis in regions exposed to reversed flow direction (orange) when compared with non-reversed flow regions (blue). (B) Mapping reversed and non-reversed flow patterns onto HUVECS in vitro elicits cell repolarization. HUVECS exposed to reversed flow conditions exhibit a significant reduction in the expression of miR-199a-5p, a miRNA with several downstream targets known to be regulators of arteriogenesis. (C) Inhibiting miR-199a-5p via intramuscular injection of linked nucleic acid anti-miR-199a-5P into the upper hindlimb augments collateral arteriogenesis and accelerates perfusion recovery in the lower hindlimb, as assessed by laser Doppler perfusion imaging.

Other studies have sought to determine if miRNAs with known functions important for vascular growth also regulate the response to hindlimb ischaemia. To this end, miR-15a had been previously shown to be overexpressed in oxygen-glucose deprived cerebral ECs and to regulate Bcl-2.¹³⁰ Following hindlimb ischaemia, EC-specific miR-15a overexpressing transgenic mice exhibited reduced capillary density in ischaemic limbs and reduced perfusion recovery compared with littermate controls by suppression of FGF2 and VEGF.¹³¹ Another study identified candidate miRNAs bioinformatically through reverse target prediction of 197 genes known to be involved in neovascularization.¹³² This analysis identified 14q32 as an enriched cluster of miRNAs with targets involved in neovascularization.¹³² Systemic inhibition of four different 14q32 cluster miRNAs (miR-320, -487 b, -494, and -495) increased perfusion (∼25–40%), arteriolar diameter, and capillary densities following hindlimb ischaemia.¹³² Another highly abundant, EC-specific, miRNA that has been shown to have a role in vascular development is miR-126.^133–135 Following femoral artery occlusion in C57BL/6 mice, miR-126a [-3p] antagomirs impaired angiogenesis, both in vivo and in vitro, but had no effect on overall reperfusion in vivo.¹³³ However, overexpression of miR-126-3p through the use of a non-invasive ultrasound-targeted microbubble destruction method, increased perfusion recovery and microvessel density in a rat hindlimb ischaemia model.¹³⁶ The miR-126∼25 cluster, which includes three mature miRNAs (miR-106b, miR-93, and miR-25) and is a paralogue of the miR-17∼92 and miR-106a∼363 clusters, had been previously shown to be highly expressed in HUVECs and to promote tumour angiogenesis.¹³⁷ This cluster also appears to be necessary for normal reperfusion following femoral artery occlusion, as blood flow recovery was impaired in miR-126∼25 knockout mice compared with controls, but was rescued by miR-126∼25 overexpression.¹³⁷ The miRNA let-7 family had previously been shown to be highly expressed in a variety of cell types within the cardiovascular system,^138–140 and has been implicated in the regulation of cell proliferation,^141,142 tumour suppression,^141,143,144 and cardiovascular disease.^145–147 Indeed, an intramuscular injection of let-7g in a mouse hindlimb ischaemia model resulted in improved neovascularization, perfusion recovery, and endothelial progenitor cell recruitment relative to a control miRNA.¹⁴⁷

Finally, many recent studies have quantified altered miRNA expression in the ischaemic tissue following arterial occlusion in vivo to identify potential regulators of reperfusion. To this end, miR-100 was found to be down-regulated in the adductor muscle of C57BL/6 mice 3 days after induction of hindlimb ischaemia.¹⁴⁸ Subsequent inhibition of miR-100 enhanced perfusion recovery and capillary density through regulation of mammalian target of rapamycin signalling.¹⁴⁸ Similarly, miR-155 was also down-regulated during hindlimb ischaemia in C57BL/6 mice.¹⁴⁹ Inhibition of miR-155 increased EC proliferation and angiogenic tube formation in vitro and promoted angiogenesis in the ischaemic tissue of mice in vivo.¹⁴⁹ Despite promoting angiogenesis, miR-155 deficient mice exhibited attenuated blood flow recovery and leucocyte recruitment, thereby indirectly implying miR-155 impairs arteriogenesis, though collateral diameters were not explicitly reported.¹⁴⁹ Another study identified miR-132/212 to be significantly up-regulated in the adductor muscles at both 4 and 7 days after femoral arterial occlusion.¹⁵⁰ Perfusion recovery after arterial occlusion was slower in miR-132/212 knockout mice compared with wild-type mice.¹⁵⁰ Hazarika et al.¹⁵¹ incorporated a variation on this approach by comparing differential genome-wide miRNA expression between C57BL/6 and Balb/c mice, two strains known to exhibit widely different baseline vascular network structures and reperfusion capacities.^152–155 From this approach, they identified miR-93 and determined that it enhanced perfusion recovery and capillary density by modulation of multiple genes involved in the regulation of cell proliferation and apoptosis.¹⁵¹ Further study by this group has also revealed that miR-93 promotes an M2-like polarization of macrophages in vitro, even under conditions which traditionally induce an M1 polarization (i.e. hypoxia).¹⁵⁶ The delivery of miR-93 to miR-106b-93-25 cluster-deficient mice resulted in increased levels of M2-like macrophages in the ischaemic muscle, as well as improved angiogenesis, arteriogenesis, and perfusion recovery.¹⁵⁶ Guan et al. found miR-352 to be down-regulated in vasculature exposed to increased fluid shear stress via an arteriovenous shunt following FAL. Inhibition of miR-352 in vivo resulted in increased collateral vessel growth and perfusion, as well as increased expression of IGF2R, which may regulate autophagy during collateral vessel growth.¹⁵⁷ Zhu et al.¹⁵⁸ recently reported that miR-15b-5p can act as a potential circulating biomarker for poor collateralization in CAD patients and that miR-15b-5p overexpression has functional consequences as it impairs arteriogenesis and perfusion recovery via targeting AKT3. Desjarlais et al. identified miR-150 expression to be reduced in ischaemic muscle of hypercholesterolemic (ApoE^−/−) mice whereby subsequent miR-150 overexpression was sufficient for restoring perfusion, vascular density, and functional mobility via a Src-mediated mechanism.¹⁵⁹

Recently, candidate angiogenesis-regulating lncRNAs have been identified by characterizing the response of ECs to hypoxia in vitro.^160–163 Knockdown of several of these hypoxia-induced lncRNAs, including GATA-AS,¹⁶¹ H19,¹⁶² LNC00323,¹⁶³ and MIR503HG,¹⁶³ have been shown to regulate EC function in vitro. However, only two lncRNAs, metastasis associated lung adenocarcinoma transcript 1 (MALAT1) and maternally expressed 3 (MEG3), have been studied in response to arterial occlusion in vivo thus far. MALAT1 inhibition impaired blood flow recovery and reduced capillary density following hindlimb ischaemia in vivo.^160,164 In contrast, MEG3 inhibition was reported to improve blood flow recovery in aged mice in a follow-up study.¹⁶⁵ These studies represent only very recent discoveries of functional roles for these non-coding RNAs. We anticipate that the regulation of vascular growth by lncRNAs and other non-coding RNAs will likely be an area of intense research focus in the near future.

5. Perspectives for future studies and translation

PAD is a complex, chronic disease in which patients can display a wide variety of clinical outcomes, ranging from critical limb ischaemia to intermittent claudication and even ‘asymptomatic’ disease. While genetics clearly contribute towards the risk of developing PAD, they do not fully explain the variability in clinical outcomes.^18–20 Epigenetics may therefore help explain the remaining risk for PAD development and the variability in clinical outcomes. Additionally, epigenetic modifications could act as biomarkers or as new therapeutic targets. To this end, altered circulating miRNA expression^{127,166–168} and lncRNA expression¹⁶⁹ has been observed in PAD patients. Future studies are needed to more comprehensively determine the robustness of these biomarkers in large, diverse PAD patient populations. Additionally, while circulating miRNA expression could aid in diagnosis, disease monitoring, and outcome predictions, it does not necessitate a functional role for these miRNAs in disease pathophysiology.

In addition to acting as biomarkers, several miRNAs are both differentially expressed in PAD patients and functionally significant in the murine hindlimb ischaemia model, including miR-15a/b, -19a, 27 b, -93, -126, -199a, -210, -223, and -503. Of these, an antagomir-based approach appears best for miR-199a and miR-503, as antagonism improved revascularization in mice and these miRNAs were overexpressed in PAD patients. In contrast, overexpression via mimic, plasmid, etc., may be the best approach for miR-93, as miR-93 is pro-angiogenic in murine pre-clinical models^151,156 but has lower expression in the peripheral blood of PAD patients. Several of these miRs may actually be a ‘last-ditch effort’ to mount a healing response in PAD patients. To this end, overexpression of miR-210 and miR-27b improve revascularization in mice^170,171 but both are elevated in the plasma of PAD patients, with miR-27b expression even correlating with Fontaine stage.¹⁷² Similarly, miR-223 knockdown in a pre-clinical model improves perfusion recovery and capillary density¹²⁶ but plasma levels are decreased in PAD patients with type II diabetes.¹⁶⁷ Together, these data may suggest that the effects of these miRs are inhibited by co-factors (i.e. other miRs, lncRNAs, etc.) or are cell-specific.

In addition, epigenetic mechanisms may help to both explain the heterogeneity in responses to treatment (i.e. responders/non-responders) and provide an opportunity for personalized medicine. For instance, we have shown that regions within the same collateral arteries exhibit differential growth potential based on the extent of DNA methylation. By reducing DNA methylation with 5AZA, these collateral vessels were able to increase in diameter, even when 5AZA treatment was delayed 2-weeks after arterial occlusion.¹²³ Reduction of hyper-methylation may help to induce endogenous revascularization directly or could also ‘re-sensitize’ ECs to other pro-angiogenic/arteriogenic drugs in a combination treatment. Another example may be miR-15a, wherein overexpression has been shown to reduce capillary density, arteriogenesis, and perfusion recovery in murine pre-clinical models.^130,168 In one clinical study of CLI patients with concomitant diabetes, a higher plasma miR-15a expression level corresponded to an increased risk of both restenosis and amputation.¹⁶⁸ However, in another study, plasma miR-15a expression was reduced in PAD patients with concomitant diabetes, with lower miR-15a expression correlating to lower ABI.¹⁶⁷ This may highlight a need to both better understand the heterogeneity within the PAD patient population and tailor individualized treatments based on circulating biomarkers.

In moving towards epigenetic-based theragnostic or therapeutic approaches, it will be vital to test mechanism and function in relevant disease models. To this end, it will be necessary to determine temporal and cell-specific effects, such as with inducible, cell-specific, genetic models. This will be particularly true for non-coding RNA studies. Studies targeted towards histone modifications and DNA methylation currently utilize pan-regulators of an epigenetic mark that can widely alter epigenetic modification of numerous genes. Future work will need to develop site-specific, targeted approaches for performing rigorous gain/loss of function studies, perhaps using CRISPR-Cas-guided fusion proteins. Additionally, development of in vivo reporter systems, ideally allowing for repeated non-invasive imagining, will aid in providing time-dependent regulation information and improving both therapeutic dosing and delivery. Finally, we must also acknowledge the limitations of using such preclinical studies to predict outcomes in human PAD patients. Indeed, these preclinical models often (i) are performed using an acute surgical method of arterial occlusion, (ii) apply a therapeutic treatment at nearly the same time as the arterial occlusion, and (iii) utilize young, healthy mice, without other co-morbidities.^38,173

The recent work investigating the role of epigenetics in PAD has provided more mechanistic insight into the regulation of arteriogenesis and angiogenesis. Moreover, it has opened a new world of possible diagnostic and therapeutic options for PAD patients. Future work will need to address a number of questions to help realize the potential of targeting epigenetic mechanisms therapeutically. What is the cell specificity? Does the modification occur broadly across the epigenome or only at a single locus? How do epigenetic modifications change in patient subpopulations? What are the off-target effects and contraindications of modifying the epigenome? To address these questions, we will need to (i) better characterize the epigenome in PAD patient populations, (ii) develop more sophisticated tools to probe and monitor epigenetic modifications in a targeted manner, and (iii) invoke more clinically relevant disease models. Such models should consider common co-morbidities (e.g. diabetic, hypertensive, aged), as well as the timing of experimental revascularization therapies with respect to when patients would actually be seeking treatment (i.e. long after the occlusion has developed). By continuing to address these questions and expanding our knowledge, we may fully realize the power of targeting the epigenome for treating PAD.

Conflict of interest: none declared.

Funding

Supported by National Institutes of Health Grants R01EB020147 and R21EB024323 to R.J.P. J.L.H. was supported by UVA Cardiovascular Research Center Training Grant T32HLI007284 from the National Institutes of Health. C.M.G. was supported by American Heart Association Fellowship 18PRE34030022.