Bench-to-Bedside in Vascular Medicine: Optimizing the translational pipeline for patients with peripheral artery disease

Abstract

Peripheral arterial disease (PAD) is a growing worldwide problem with a wide spectrum of clinical severity and is projected to consume more than $21 billion per year in the United States alone¹. While vascular researchers have brought several therapies to the clinic in recent years^{2, 3}, few of these approaches have leveraged advances in high-throughput discovery screens, novel translational models, or innovative trial designs. In the following review, we discuss recent advances in unbiased genomics and broader “omics” technology platforms, along with preclinical vascular models designed to enhance our understanding of disease pathobiology and prioritize targets for additional investigation. Further, we summarize novel approaches to clinical studies in subjects with claudication and ischemic ulceration, with an emphasis on streamlining and accelerating bench-to-bedside translation. By providing a framework designed to enhance each aspect of future clinical development programs, we hope to enrich the pipeline of therapies that may prevent loss of life and limb for those suffering from PAD.

Introduction:

Over 200 million people worldwide have peripheral arterial disease (PAD), commonly defined as atherosclerosis involving arteries outside of the coronary circulation⁴. The high incidence of lower extremity PAD places an enormous burden on healthcare systems worldwide, necessitating novel strategies to curb disease progression⁵. Affected patients suffer from significant morbidities, including diminished blood supply to the limbs leading to claudication, critical limb ischemia, ulcer development, gangrene, and ultimately limb loss. While PAD treatments aimed at preventing progression of atherosclerosis through targeting dyslipidemia⁶, hypertension⁷ and dysregulated coagulation⁸ have saved lives and reduced systemic complications, identification of novel targets against disease sequelae are lacking. In addition, there likely remains a substantial group of patients whose susceptibility to atherosclerosis in their peripheral arteries is disproportionate to their risk factor burden in other vascular beds, and those who continue to rapidly progress despite best evidence-based management. To address these issues, it is of the utmost importance to develop therapies of direct relevance to patients suffering from PAD, including those with variable responses to therapy or clinical trajectories.

Here we explore the impact of strategic research collaboration across the whole translational pipeline to bring new medical solutions for patients with PAD. Building large cohorts of biobanked tissue and blood specimens will provide enhanced opportunity for discovering new mechanisms of disease susceptibility⁹, augmented by high throughput “omics” platforms, and massive advances in bioinformatics¹⁰. Ensuring that discoveries made through such approaches inform in vitro and in vivo models will be key to translating knowledge of disease-promoting pathways into new therapies for patients suffering from vascular disease. We begin by discussing advances and discoveries from such unbiased approaches, and opportunities for the future in regard to target identification and prioritization.

Discovery: Target Identification and Prioritization through Genomics

Unraveling the genetic architecture of vascular disease is fundamental in the effort to innovate and realize novel targets for therapeutic intervention, as well as markers of risk. Translational genomics is perhaps best exemplified by the discovery of PCSK9 as a molecular target to lower LDL cholesterol^11–13, paving the way to explore alternatives to statin therapy for atherosclerosis, a continued top cause of morbidity and mortality worldwide despite vast efforts to curb its deleterious effects^{14, 15}. As was the case with PCSK9, uncovering novel therapeutic targets depends upon utilization of next-generation genomics approaches to disentangle disease complexity and home in on key drivers of disease progression. Here we explore some of the genomics approaches used to study vascular disease in an effort to guide the translational process from target identification to clinical trials.

Genome wide association studies:

Deposition of genetic information into large biobanks and databases has enabled genome wide association studies (GWAS) that have richly enhanced our understanding of inherited variants which contribute to complex disease phenotypes. Notably, GWAS have highlighted biological processes in the vessel wall and enabled identification of 9p21 as an impactful genetic locus in both coronary artery disease^{16, 17} and PAD¹⁸. Importantly, this locus contains the cyclin-dependent kinase inhibitor CDKN2B which was functionally shown to be involved in PAD pathogenesis after Cdkn2b^−/− mice developed advanced hindlimb ischemia after femoral artery ligation¹⁹. The Million Veteran Program (MVP)²⁰, a database established in 2011 of genomic data from the Veterans Affairs Healthcare System, is one such archive used to explore PAD susceptibility loci²¹. The MVP cohort includes a multiethnic population that is particularly susceptible to PAD and its comorbid sequelae due to the higher prevalence of smoking²², diabetes and hypertension²³, making this an ideal group from which to identify genetic factors that increase PAD risk. Indeed, Klarin, et al utilized the MVP dataset and identified 19 total PAD loci that exceeded genome-wide significance²¹. Importantly, their analysis highlighted 11 genetic loci common to coronary, cerebral and peripheral vascular disease (LDLR, LPA, LPL, SORT1 – lipids; PTPN11 – hypertension; TCF7L2 – diabetes), as well as four genetic signals specific to PAD, including Factor V Leiden variant, F5p.R506Q (thrombosis) and CHRNA3 (smoking), suggesting that smoking and thrombosis may play an even greater role in PAD than in other arterial territories. These results also reinforce the use of lipid-lowering therapies to target disease burden across multiple vascular beds and supply the genetic rationale for targeting the coagulation cascade in PAD, as was done in VOYAGER trial². MVP and other large databases are invaluable resources for assessing novel disease gene variants and potential genetic targets for a range of vascular diseases.

Importantly, while GWAS studies in general may be limited by confounders such as misclassification of case versus control and/or inappropriate bundling of disease subtypes under a common diagnosis, these issues may be more pronounced in PAD given the fact that many patients with disease have atypical or no symptoms, or because those with confirmed PAD are presumed to have atherosclerosis when they may in fact have some other vasculopathy. In the MVP GWAS, these issues were mitigated by chart review of a representative cohort of cases and controls to verify ICD-9 coding that correlated with clinician verified PAD diagnosis (determined through clinic notes and other diagnostic criteria). Improvements in subphenotyping with inclusion of imaging or advanced analysis of plaque characterization and disease distribution may help stratify patients in a manner reflecting distinct biology, with the potential to identify loci for specific sub-type of PAD that may be diluted by grouping all PAD patients together.

Next generation “omics” and single-cell technology:

Implementation of next-generation sequencing has become an important resource in the vascular biologist’s toolkit and may help uncover mechanisms underlying varied response to standard-of-care treatments. For example, although endarterectomy may reduce the risk of stroke in patients with symptomatic stenosis of the internal carotid artery (ICA), restenosis has been found in a substantial subset of patients, including over 35% during a three-year follow-up in one study²⁴. This may be due to cellular heterogeneity within the plaque or the vascular wall that then influences inter-cellular behavior and disease progression. Understanding the cell types and differential gene expression patterns found in normal arterial tissue versus atherosclerotic plaques could help to refine our understanding of why some patients fail treatments such as endarterectomy or stenting. Similarly, although antiplatelet therapy is a mainstay for patients with PAD, the prevalence of antiplatelet resistance is comparable to rates seen in cardiovascular disease²⁵, underscoring the importance of delineating cellular mechanisms leading to such variations in treatment efficacy.

Bulk sequencing studies, while yielding important overall genomic insight, obfuscate the differences between cell types contributing to gene expression. Recent efforts have focused on interrogation of the genome and its accessibility at the single-cell level. Leveraging single-cell technology to examine cellular subpopulations and genomic differences presents an important opportunity to identify disease mechanism and exploit novel targets for therapy. For example, single-cell RNA sequencing has shed critical insight into cardiac vascular differentiation²⁶, endothelial specification²⁷, clonal smooth muscle cell expansion²⁸, and hematopoietic potential²⁹. In a murine hindlimb ischemia model a small group of fibroblasts that expressed some endothelial cell genes were further dissected and clustered by scRNAseq, revealing subsets of tissue fibroblasts that seem to contribute to angiogenesis under ischemic conditions³⁰ Many recent efforts have utilized single-cell RNA sequencing to interrogate the plaque microenvironment due to its complex cellular composition^31–36. We have shown transcriptional differences in carotid plaque samples between the most advanced portion of the atherosclerotic core and a region directly proximal to this location, suggesting a possible anatomic framework for atherogenesis and the possibility of site-specific disruption of disease progression³⁶. While the majority of these studies utilize more easily obtainable tissues such as plaque from carotid endarterectomy, there is an opportunity to explore other vascular beds subjected to variable hemodynamic stressors which may shed light on site-specific genomic differences in plaque burden. For example, atherosclerotic plaque may be obtained from femoral endarterectomies for dissociation and sequencing, or alternatively full thickness artery may be harvested from lower extremity revascularization for critical limb threatening ischemia (CLTI). Transcriptome comparisons may be made between the various vascular beds sampled, leading to potentially novel and significant genomic insight into anatomic disease burden.

Refined examination of intercellular communication and cellular plasticity, both well-appreciated phenomena in vascular biology, further underscores the utility of next generation sequencing for molecular target identification. For example, vascular smooth muscle cells (VSMCs) have been shown to stimulate platelets through CLEC-2, possibly facilitating thrombus formation after plaque erosion³⁷. Delineating which cell subtypes are responsible through single-cell analyses may allow for more targeted perturbation of this process, especially as cell-specific nanotherapies evolve³⁸. Furthermore, the phenotypic landscape of VSMCs is understood to be quite dynamic as these cells often transition to multiple alternative phenotypes in the atherosclerotic environment^{39, 40}. VSMCs participate in phenotype switching between contractile and synthetic subtypes, and they contribute to the development of extracellular matrix producing cells⁴⁰. Advances in sequencing-based lineage tracing may facilitate mapping of clonal histories of VSMCs and other vascular cell types⁴¹, in addition to improved resolution of clonal substructure as seen in cells within the tumor microenvironemnt⁴². Additionally, many disease promoting processes are mediated through molecular crosstalk between VSMCs⁴³, but this type of crosstalk is not limited to signaling between the same cell types. Atherogenesis involves coordinated genomic and molecular communication between the plethora of immune cell subtypes and other cellular components of the arterial wall^{44, 45}. Single-cell technology offers an exciting opportunity to begin decoding this intercellular communication, while also incorporating gene network analysis which, by integration with single-cell RNA sequencing, may shed insight into the underlying biology of groups of cells^{36, 46}.

Beyond the transcriptome, detecting chromatin accessibility with single-cell ATACseq can unveil layers of biological complexity underlying the transcriptional differences observed between cells⁴⁷. Because the majority of complex disease genetic variation is non-coding⁴⁸, and functional non-coding gene elements are involved in epigenetic regulation of gene expression, single-cell ATACseq has been used for chromatin-mapping to facilitate analysis of cis- and trans- regulatory elements⁴⁹. Integration with single-cell RNA seq data further allows for refined cell subpopulation analysis through reconstruction of chromatin accessibility profiles of cell types defined by RNA profiles^{50, 51}. This has enabled high-resolution examination of cardiac progenitor cell fates⁵², with emerging opportunity to employ this technology to assess chromatin accessibility in the peripheral vasculature.

DNA sequence variations, alternative splicing of mRNA, post-translational modifications, phosphorylation and proteolytic cleavage collectively contribute to the diversity of the human proteome far beyond the known 20,000 protein-coding genes^53–55. CITE-seq allows for simultaneous analysis of the transcriptome and proteome via exploitation of DNA barcodes conjugated to antibodies to enable cell surface protein detection^{56, 57}. It has been used in conjunction with single-cell RNA sequencing for immunophenotyping of carotid plaques in order to delineate T-cell and macrophage subsets in symptomatic versus asymptomatic carotid plaques⁵⁸. Emerging single-cell mass cytometry methods will allow analysis of proteins and their interactions and degradation⁵⁹, bringing a unique tool for evaluation of protein signaling dynamics in the vasculature.

Single-cell analysis of the genome, epigenome, transcriptome and proteome has reshaped the ability to harness genetic variation and heterogeneity to identify and exploit novel therapeutic targets. Once identified through genomics approaches, these targets can then undergo functional analysis through in vitro and in vivo model system assays for feasibility and proof-of-concept before reaching clinical trials.

Multi-omic characterization of large, well-phenotyped clinical cohorts to discover new mechanisms and markers:

Understanding the complex interplay between traditional, environmental and host genome risk factors in determining the susceptibility to PAD requires consideration of biological networks. Tremendous advances have been made in the high throughput measurement and unbiased analysis of the “metabolome”, “proteome”, “lipidome” and the “transcriptome”¹⁰. However, to make the most of the power of these technologies, large cohorts with excellent clinical and image phenotyping are required. Application of network analytics and machine learning algorithms have the potential to uncover biological signaling pathways and mechanisms that do not rely on prior understanding. From a pragmatic perspective, greater access to tissue and atherosclerotic plaque from surgical procedures offers considerable advantages compared to similar approaches applied to the coronary or intracerebral disease process. Strategic leadership has led to the establishment of a number of large collaborative bioresources, including the Munich Vascular Biobank with biomaterial from over 1500 high-grade carotid artery stenoses; >700 PAD, and >480 aortic aneurysm cases⁶⁰; and the Biobank of Karolinska Endarterectomies (BiKE) with >200 samples⁶¹. The concomitant collection of peripheral blood at the time of surgery provides the opportunity to identify a circulating biosignature of particular plaque phenotype- and particular vulnerability⁶². The UK Biobank provides additional opportunities based on large scale imaging (100,000 participants) of carotid plaques⁶³.

Such approaches may also unravel some of the mysteries driving an individual’s differential susceptibility to atherosclerosis in different vascular beds (e.g. coronary versus aortic or femoral). Furthermore, the cataloging of samples based on gender and ethnicity, as well as stratification according to specific phenotypes of atherosclerosis and vascular disease (e.g. different patterns of calcification, tortuosity, morphology, histology, inflammation, neovascularization)⁶⁰, may be combined with multi-omic analysis in large cohorts may provide insights to therapeutic targets relevant for specific groups.

“The formation of large biorepositories present important opportunities for precision medicine and genomics, but include challenges with infrastructure, access to comparison tissue, and sample annotation, all of which are areas of active research for improvement⁶⁴.

Preclinical validation: Translational models to unravel mechanisms and test new therapies

The technologies employed to develop new drugs and products once definitive targets have been identified are rapidly expanding, and approaches such as in silico screening, fragment-based drug discovery and tagged libraries have transformed the pathways available for modern drug discovery⁶⁵. However, collaborative efforts and strategic leadership are required to ensure that the mechanisms directly relevant to unmet need in human disease are considered in the choice of in vitro and in vivo models to test such compounds. This will allow the more efficient and coordinated translation of new, urgently needed therapies.

In vitro and ex vivo vascular models:

Before advancing to in vivo animal studies, investigators commonly deploy an array of pharmacological studies which rely heavily on in vitro and ex vivo cellular models. This allows for screening of chemical compound libraries and specific novel candidate drugs or products to refine and optimize putative translational targets. Because the PAD patient commonly experiences ischemia due to insufficient blood flow, a longstanding goal for the vascular research community is to develop therapies that stimulate the sprouting of new blood vessels (angiogenesis) and/or trigger the maturation of pre-existing collateral networks (arteriogenesis). There are several well-established monoculture models that have been used to specifically examine the potential beneficial effects on angiogenesis as it relates to PAD. Here we discuss both the long-standing monolayer culture approaches, as well as the more recent approaches involving 3D spheroids, co-culture, and microphysiological systems.

In vitro functional assays provide a valuable tool for assessing the effect of angiogenesis. They utilize basic cell culture techniques which can be quickly and easily performed and analyzed quantitatively. They are essential for investigating the biochemical, cellular, and molecular mechanisms occurring in angiogenesis because individual components of angiogenesis can be assessed in isolation. Endothelial cells (ECs) undergo multiple processes during angiogenesis including cell proliferation, migration, sprouting, branching and tubule formation, as well as complex cell-cell interactions. In addition, ECs differ depending on their vascular bed and vessel type – the most commonly used EC type in angiogenic assays include HUVECs, HAECs and HMECs⁶⁶.

Proliferation –

The process of EC proliferation can be quantified using several different methods in vitro. The simplest method for assessing EC proliferation is counting cells. This can be achieved using the common hemocytometer and a viability dye, such as trypan blue, or through the use of automated cell counters. Colormetric assays like the MTT or the WST-1 assay are widely used to assess the natural by-product of cellular division⁶⁷. Essentially these assays determine cell viability more so than proliferation⁶⁸, however they are commonly used as methods to quantify EC proliferation in angiogenesis. Another potentially more accurate method used to assess proliferation includes quantifying DNA synthesis via BrdU incorporation⁶⁹ or Click-iT via EdU incorporation⁷⁰.

Migration –

After proliferation, ECs migrate away from their origin. All angiogenic models involve the sprouting of ECs from a monolayer, and the movement of ECs through the basement membrane is often referred to as “sprouting” and occurs due to nearby chemical signals that either promote or prevent angiogenic activity. Assessing the migratory ability of ECs can be achieved with a number of different assays each with their own advantages and disadvantages. Wound healing or the ‘scratch’ assay is the simplest and quickest assay to perform. ECs grown to a confluent monolayer are ‘wounded’ commonly with a pipet tip or other similar hard objects to create a denuded zone⁷¹. In response to this damage, ECs begin to proliferate and migrate into the denuded zone in order to heal the wound. The clear disadvantage of this assay is the reproducibility with issues in confluency of cells, consistency of the scratch and difficulty in accurate quantitation⁷¹. Transfilter assays or Boyden chamber assays are generally used to study migration of ECs towards a stimulus (commonly VEGF) and have the advantage of distinguishing between specific directional migration instead of random cell movement⁶⁹. ECs are seeded into the upper chamber with the stimulus in the lower chamber in culture medium. The migration of ECs through the filter can take anywhere from several hours to 48 hours and can be quantified by cell staining or counting of cells attached to the filter. The main weakness of this assay is the inability to observe real-time migration. New automated technologies are now able to overcome this limitation with the use of systems such as xCELLigence (Roche)⁷² and also the electric cell-substrate impedance sensing (ECIS) instrument (Applied Biophysics)⁷³. These systems allow real-time monitoring of cell proliferation and migration when cells are grown in special electrode covered chambers. The resistance generated by the cell membrane increases the electrical potential between the electrodes and this allows monitoring of the cell behavior⁶⁶.

Tube Formation –

Tubulogenesis, more commonly termed tube formation, is the connection of the new blood vessels to preexisting vessels beds. Modelling this process in vitro is a cornerstone of angiogenic research. Using ECs grown in culture and then placed on commercially available Matrigel®, this assay is a two-dimensional imitation of extracellular matrix. The Matrigel acts as the extracellular matrix containing cytokines and growth factors allowing the ECs to form tubule-like structures⁶⁷. Tube formation can be observed with imaging techniques and can take anywhere from 3–24 hours to provide reliable data. The results obtained can be highly variable and comparisons in the literature depend widely on cell type, quantity of Matrigel used, and endpoint analyzed (tubule number, tubule length or branch points)⁶⁹. Furthermore, different methods of analysis allow for increased variability, whether these data points are measured manually, which is more arduous and time-consuming, or using automated image analysis programs, which require screening to eliminate artifacts but can quantify at a faster rate. This 2D model has since been expanded to 3D assays to more closely mimic the angiogenic sprouting of ECs observed in vivo⁶⁹. Simple conversions of this assay include using thicker basement matrix so ECs can be seeded at differing levels to observe tubule formation both vertically and horizontally. More complex versions include scaffold-free cell sheet stacking, cell aggregate microtissue assays and microcarrier spherical scaffold assays^{66, 74}. All of these have the advantage of recapitulating all the stages of in vivo angiogenesis however the technology required for visualization and analysis can be costly and difficult to interpret. 3D models have also expanded to include co-culture- incorporating not just endothelial cells, but also vascular smooth muscle cells and fibroblasts^{19, 75}. Such approaches may enhance the modelling of in vivo human disease, and have the capacity to support early drug screening, potentially reducing the number of animal studies required. However, it is essential we improve the uniformity of organoid models at the necessary scale. Proof of translatability is still a major issue, similar to all preclinical models in the PAD space.

Embryoid body vasculogenesis assay –

Embryonic stem cells (ES cells) have been highlighted as promising cell sources for the study of differentiation towards the endothelial lineage and vasculogenesis^76–78 since they can go through the majority stages of angiogenesis observed in vivo⁷⁹. This assay requires ES cells to be initially differentiated into embryoid bodies (EBs) which are then placed into a 3D collagen matrix for the EB primary vascular structures to extend and invade the collagen matrix, leading to sprouting of ECs similar to the angiogenic process⁸⁰. Using cell imaging, this assay allows for qualitative and quantitative analysis of multiple steps in angiogenesis and makes it possible to test many different therapeutic agents⁸¹. While the assay is comprehensive it is time consuming both at the experimental and analytical stages⁸². Additionally, the use of ES cells in biomedical research has been hindered by ethical issues since extraction of these cells disrupts the human embryo^{83, 84}. Even with several studies demonstrating effective use of these cells in the study of angiogenesis, ES cells are not generally considered an ethical source of endothelial cells⁸⁰.

iPSCs – “disease in a dish” –

Overcoming the ethical issues with the use of ES cells, a breakthrough discovery in cellular biology was the discovery of induced pluripotent stem cells (iPSC) made by Takahashi in 2006⁸⁵. With advances in cellular reprogramming, iPSCs are derived from adult somatic cells⁸⁶ and blood^{87, 88}, into a stem-cell like state by the introduction of the common transcription factors OCT4, SOX2, KLF4 and c-MYC (Yamanaka factors) via viral transduction, protein and microRNA transduction, or by chemical/small molecule-based reprogramming strategies⁸⁹. The first use of iPSC in the field of vascular biology occurred in 2009, when Taura et al. demonstrated the ability to yield endothelial cells from iPSC (iPSC-EC), which when matured display phenotypic properties highly similar to primary endothelial cells^{90, 91}. More recently, iPSC-ECs have shown therapeutic potential by enhancing angiogenesis in mouse models with improvements in blood reperfusion and enhanced wound closure^{92, 93}. A major advantage of iPSCs is that they can be easily generated from patients with or without disease for use in angiogenic assays⁹⁴. Blood vessel development can be modelled in 3D in vitro assays using iPSC-ECs cultured within engineered platforms that mimic the 3D microenvironment⁹⁵. Hydrogel scaffolds, either on standard well plates or within a passive pumping microfluidic device have offered a more defined platform for vascular-based assays^{96, 97}. Major advances in genome editing techniques with the use of clustered regularly interspaced palindromic repeats/Cas9 (CRISPR/Cas9) allows causality of genetic variants to be explored in a rigorous manner in patient-derived iPSCs and derived vascular cells⁹⁸ with full characterization of cell signaling and molecular profiling. As the application of iPSC-derived vascular cells and models evolve, it will be critical for the field to develop standardized protocols^{86, 99}.

Aortic Ring Assays –

Beyond the standard assays outlined above, investigators commonly employ organ explant assays, which appear to synergistically assess almost all steps of physiological angiogenesis¹⁰⁰. Aortic ring assays, often termed “ex vivo” models, have been developed to be the in vitro mimic of in vivo angiogenesis with the ability to measure angiogenic sprouting, outward growth, and (to an extent) stabilization of new blood vessels from explanted segments of vasculature^{101, 102}. Vessel rings are excised, cleaned, cut into ~1mm sections, immersed in collagen, fibrin or Matrigel and cultured for vessel outgrowth. After a few days, the outgrowth can be measured using staining and microscopy, then compared to samples that have been exposed to potential angiogenic agents. The advantages of the aortic ring assay are numerous including the capture of multiple cell organization with additional paracrine support and extracellular matrix components^{69, 100}. The downside of this assay is reproducibility due to imprecise segments and the absence of blood flow effects.

Vascular organoids (scaffold free) –

The assembly of different cell types into 3D spheroid co-cultures called ‘vascular organoids’ are being described as a bridge in the gap between traditional 2D in vitro cell culture and the complexity of animal models. These transformational 3D model systems provide a more physiologically relevant assay encompassing complex cell-cell interactions which enables the deposition of their own extracellular matrix^{103, 104}. The advent of hybrid spheroids using more than one cell type provide useful models for angiogenesis assays with the formation of tubular-like structures^{105, 106}. Multiple different techniques including the hanging-drop, the use of low adhesion plates or self-organising vessel-like structures have been utilized to assess angiogenesis. While these methods can be relatively cost-effective they are time-consuming and require advanced imaging capabilities for analysis¹⁰⁷. More recently, these methods utilize the addition of iPSC-ECs with multiple other cell types in order to increase the ability for cells to reorganise and form complex vascularized networks^{108, 109}. Studies have shown that after implantation of iPSC-EC organoids into the chick chorioallantoic membrane assay, the generated vessels connect with the host circulation and vessel structure is preserved¹⁰⁹. Many aspects of vessel formation within the vascular organoid can be detected including branching, an extracellular matrix, cell-cell interaction, microvesicle release and responsiveness to pro- and antiangiogenic stimuli¹¹⁰.

Vascular scaffolds –

In addition to the more commonly used scaffold of Matrigel to assess EC interactions in angiogenic-based assays, other more complex vascular networks are being bioengineered to incorporate the missing in vivo elements. Electrospinning and 3D bioprinting are being utilized to address the need for a more complete microenvironment including cell organization, spatial control and localized release of growth factors in response to cues from the extracellular matrix and localized blood flow. Electrospinning can produce nanofiber-based synthetic vascular networks with similar dimensions to naturally occurring extracellular matrix (50–500nm) and provides an excellent platform for the adhesion of ECs in a monolayer. A limitation in this methodology however, is that nanofibers create a physical barrier for cells and this limits the ability to mimic the permeability of the endothelial monolayer¹¹¹. 3D bioprinting is an emerging approach to generate scaffolds with hydrogel which can be embedded with microchannels to study angiogenesis^{112, 113}. Using either direct or indirect printing methods, cells, biomaterials, and growth factors can be combined to produce complex shaped constructs with defined micron-sized channels and pore sizes that are capable of guiding angiogenesis. Both electrospinning and 3D printing techniques have various advantages and are frequently used in tissue engineering applications. The 3D bioprinting technique allows controlling the production of a large number of scaffolds with precise measurements relatively quickly while electrospinning builds scaffolds with a wide range of properties to mimic blood vessel structure including composition, diameter, thickness, porosity, and degradation rates¹¹³.

Nanoparticles –

The testing of targeted angiogenic pathways has been significantly enhanced by technological advances in nanotechnology. More efficacious administration of angiogenesis modulators using nanoparticles, including metallic, metal oxide, glass-ceramic and polymeric nanoparticles, has allowed for treatments with a longer half-life and more selective targeting of the vasculature^114–117. Some of the first studies used albumin and gelatin nanoparticles containing a DNA plasmid to trap excess VEGF and reduce angiogenesis¹¹⁸, while alternative applications were designed to increase angiogenesis and enhance revascularization^118–120. Clinical trials involving VEGF therapy have been unsuccessful at improving peak walking time, ankle-brachial index, or quality of life^{121, 122}, thus underscoring the need to evaluate each step of the translational cascade described.

Other platforms include single- and multi-walled nanotubes containing micro-RNA oligonucleotides and VEGF-targeted siRNA, able to regulate EC proliferation and prevent angiogenesis^{123, 124}. Graphene oxide nanoparticles structures have demonstrated increased proliferation and migration of cells to promote wound healing and nano diamonds can increase bioavailability of angiogenic drugs for improved delivery to tissue^{125, 126}. Among metallic options, gold nanoparticles are being recognized for their angiogenic potential¹¹⁷. They are composed of an inorganic core containing gold encircled by an organic monolayer and have displayed numerous anti-angiogenic effects by suppressing activation of VEGFR2, Tie2R, FGFR and many of their downstream signaling pathways^{114, 127, 128}. While the majority of these applications have been targeted for tumor-related angiogenesis and therefore focus on anti-angiogenic properties, there are some applications that have demonstrated pro-angiogenic effects¹²⁹, with increased cell survival and proliferation of ECs and increased vessel-like structures, particularly in wound healing^{130, 131}. Nanoparticle characterization is increasingly variable and therefore difficult to compare across multiple studies. Given the missing details in some publications there has been a call for “minimum information reporting” with regards to the details in the study of nanomaterials and their biological interactions¹³². Obtaining a standard in the field will improve the growth of available and effective nanoparticles for use in angiogenesis.

Specific vascular targeting nanoparticles continue to be developed in an effort to advance precision therapy for atherosclerosis. Lesional macrophage-specific single walled nanotubes (SWNTs) are shown to accumulate in the atherosclerotic plaque and, with the appropriately coupled therapeutic, reactivate efferocytosis in order to reduce plaque burden with minimal off-target cytotoxicity³⁸. Targeting of vascular inflammation using leukosomes, a biomimetic nanoparticle that integrates leukocyte-derived membrane proteins into the phospholipid bilayer, has been shown to enhance delivery of therapeutic to inflamed endothelium at sites of plaque development¹³³. Continued efforts to refine cell-type specific therapeutic delivery will likely be critical for future vascular therapies.

Pre-clinical in vivo models:

Once a novel target gene or therapeutic candidate has been identified, the preceding in vitro assays are commonly used to determine which leads should be prioritized for additional in vivo testing. In the field of vascular medicine, investigators tend to be broadly focused on determining whether a given pathway may regulate new blood vessel formation, peripheral atherosclerosis, ischemic tissue wound healing, and/or thrombosis¹³⁴. In the following section, we provide a general overview of the in vivo mouse models available to test these processes, including the strengths and limitations of each approach, as well as emerging techniques that may allow more sophisticated analyses of the biology underlying peripheral vascular disease.

Hind limb ischemia models:

As discussed previously, mitigation of ischemic complications in PAD through stimulation of angiogenesis and arteriogenesis are longstanding goals in vascular research. These processes can be assessed several ways (including with retinal angiogenesis or Matrigel implantation assays) but are most commonly tested in the murine hindlimb ischemia (HLI) model^{135, 136}. This model relies on the interruption of lower extremity blood flow, most often induced by the ligation of the femoral artery (which can model claudication given that collaterals downstream of the profunda femoral artery are left intact) or via the simultaneous ligation and excision of the femoral artery (which can model critical limb ischemia given that the collateral circulation is also perturbed with this approach). Limb perfusion is then quantified by noninvasive laser Doppler imaging, where the flux of blood (normalized to the non-ischemic limb) can be quantified during recovery or in response to therapeutic intervention. Other reproducible readouts include the use of microCT angiography, digital necrosis scoring, and blinded histological quantification of neovessel density¹⁹. Studies in large animal models of PAD have had success in mimicking the ischemic myopathy seen in ischemic human limbs by using endovascular techniques and occluding inline flow through the external iliac artery and impeding the rapid arterial collateralization through the use of a covered stent¹³⁷.

While the HLI model is commonly used, it does have several limitations that warrant discussion¹³⁶. First, these studies are most often performed in young, non-atherosclerotic mice without any of the comorbidities typically encountered in humans with PAD. While some have advocated for the use of diabetic (e.g. db/db) or dyslipidemic (e.g. apoE^−/−) animals, it is important to note that the acute cessation of flow in an otherwise healthy artery is perhaps more reflective of acute limb ischemia (ALI) than the gradual reduction in perfusion that is observed in vasculopaths who experience reductions in their ankle brachial indices (ABI) over a period of months to years. Models that utilize an ameroid constrictor device (which can gradually occlude the vessel) have been reported and are being studied as a potentially more relevant simulator of human vascular disease¹³⁸. Second, it is critical to recognize that there are profound differences across mouse strains which can influence the interpretation of results with this model¹³⁹. For example, the widely used C57Bl/6 strain has robust collateral reserve and is resistant to critical limb ischemia, while the BALB/c strain is highly susceptible to digital necrosis and even limb loss after femoral ligation. Thus, investigators must consider whether they are interested in modelling claudication or CLI before generating a new genetic knockout model or embarking on a therapeutic intervention study. While multiple modalities can be used to quantify limb perfusion, each has its own intrinsic limitations, as previously summarized by Lotfi and colleagues¹³⁶. In the particular case of laser Doppler imaging, factors that can alter vasoconstriction or vasomotor tone (e.g. depth of sedation, room temperature, animal sex, etc.) should be controlled to the best extent possible¹⁴⁰. Finally, studies exploring the cell-cell interactions occurring during limb perfusion are providing mechanistic insights into vascular repair. More than simply an EC proliferative response, the involvement of VSMCs, macrophages and pericytes are contributing at a molecular level to participate in tissue perfusion.

Peripheral plaque vulnerability models:

Given that most PAD is atherothrombotic in nature¹⁴¹, the translational vascular research community also employs a wide array of mouse models that develop atherosclerotic plaques in the peripheral vasculature, including the aorta and brachiocephalic artery. Traditional dyslipidemia models (e.g. LDLR^−/− and apoE^−/− mice fed a high fat ‘Western’ diet) can be used to reproducibly quantify plaque burden (using lipid stains such as Oil Red O) and lesion composition (with histological analysis of necrotic core size, foam cell burden, and fibrous cap thickness). However, most commonly used murine atherosclerosis models are not susceptible to spontaneous plaque rupture, and do not consistently demonstrate the intraplaque hemorrhage and superimposed luminal thrombosis encountered in subjects with PAD. Accordingly, a variety of models that induce mechanical alterations in blood flow to promote endothelial dysfunction and plaque-destabilization have now been described. These methods (which include interposition grafting, perivascular device placement, arterial ligations, or fistula formation) were recently reviewed by Winkel and colleagues¹⁴², including their predicted impact on shear stress and how those changes can be used to model various aspects of peripheral vascular disease¹⁴³.

Amongst these models, the application of an extrinsic silastic collar to the carotid artery (with¹⁴⁴ or without^{145, 146} angiotensin infusion) seems to have emerged as the approach which most faithfully recapitulates the human condition¹⁴⁷. While technical variations in the approach exist¹⁴⁸, these so-called ‘cuff’ or ‘cast’ models promote oscillatory and/or reduced shear stress in the branch vessels of the aortic arch, thus leading to the generation of advanced and destabilized carotid lesions. These plaques demonstrate a high burden of macrophage infiltrates, a reduction in collagen and smooth muscle cell (SMC) content, thinning of the fibrous cap, and intraplaque hemorrhage¹⁴⁸. Such vulnerability models can be deployed in mice with indelible cell-specific lineage tracers, which are recognized as being highly informative given the increasingly appreciated role for phenotype switching, cellular plasticity, and clonal expansion during atherogenesis^149–151. Moreover, our group recently demonstrated that cast-induced changes in lesion vulnerability can be quantified non-invasively (using 18F-FDG-PET CT scanning), and that this imaging modality is sensitive enough to detect changes in plaque stability after drug treatment over time¹⁵². Accordingly, these peripheral plaque rupture models may prove useful for investigators studying new translational therapies.

Ischemic wound healing models:

Due to poor circulation and prolonged tissue ischemia, many patients with PAD develop chronic wounds or ulcers on their lower limbs. These can be very difficult to heal and often become infected and require amputation of the affected foot or leg^{153, 154}. In addition to the health burden, the economic consequences are substantial. The complications associated with infection and amputation also lead to increased mortality in these patients, particularly in diabetic patients¹⁵³. Wound healing is a process not able to be replicated in cell models and has primarily been limited to large animal models such as swine, to ensure a comparable process to humans¹⁵⁵. In the last two decades however, murine in vivo wound healing experiments have become effective models for studying human wound healing¹⁵⁶. Three reproducible murine wound healing models that recapitulate the human wound healing process have been described – the splinted excisional wound¹⁵⁷, ischemia-reperfusion model¹⁵⁸ and the ischemic flap model¹⁵⁶. Excisional wounds on the dorsal surface are the most commonly used wound healing model, however in mice wound contraction is a significant limitation in assessing wound closure. The use of splints therefore increases the relevancy of this model to human wound healing¹⁵⁹. Silicone splints are fixed to the wounds to prevent contraction and allow the wound to heal through tissue formation and reepithelialization^{155, 157}. Despite this, it still remains a great challenge to replicate wound healing as murine models present different anatomy to humans, particularly for the use of rodent skin. In a recent systematic review¹⁶⁰, animal models of ischemic wound healing were investigated with only three studies utilizing more clinically relevant wounds created on the paws of mice after undergoing hindlimb ischemia^161–163. Some recommendations to help improve this model includes the use of aged mice to mimic human diseases, improving bias and standardizing assessments for the ideal ischemic ulcer model in addition to using a more stable induction of ischemia¹³⁸.

From bench to bedside: Accelerating translation through early phase clinical trials

While there have been substantial improvements in preclinical models of PAD, the disease complexity has contributed to disappointing results in the clinical translation of therapies showing potential at earlier stages. Collaboration at the interface of the preclinical and clinical efforts is required to improve the prioritization of the most promising novel therapies and to strategize how to overcome developmental hurdles. More efficient early phase studies focused on biologically relevant endpoints in more precisely stratified patients will be key. In addition to traditional and novel systemically administered therapies, the ‘peripheral’ and accessible nature of PAD should also make innovative approaches to local delivery of novel therapies feasible. This may involve stents or bioresorbable scaffolds, or topical nanoparticle delivery, for example to ischemic ulcers. Outlined below are concepts that may be useful in accelerating the translation of discoveries from bench to bedside.

Precision stratification of patients based on biology:

There remains a considerable degree of disease heterogeneity in PAD that directly relates to underlying biological mechanisms, as outlined above. A key consequence of such heterogeneity is dilution of the average therapeutic effect in trials, thus necessitating larger, longer and more expensive clinical development programs. Approaches described in this review, including genomics, molecular phenotyping and improved imaging techniques will provide new risk stratification tools with better opportunity to ‘divide by biology’. While pursuing broadly generalizable therapies with that can be used by all vascular patients obviously remains a key goal, testing new treatments in the groups where the therapy makes the most biological ‘sense’ may help reduce the size of studies and facilitate the demonstration of the impact of the intervention.

Identification of those with greatest unmet need:

A current unmet need in the management of cardiovascular disease is the ability to address the substantial number of individuals who develop disease or progress despite best practice treatment. While there is an obvious group of patients who develop disease due to a history of smoking and/or mixed adherence to recommended treatments, there are also many patients whose disease is driven by unknown biological susceptibility. While at a community and policy level it makes sense to invest in the widespread use of cost-effective and evidence-based care, recognizing that there are subgroups of individuals who demonstrate rapid disease progression, similar to cancer patients with metastasis on therapy, is key. Emerging machine learning and neural networking approaches are beginning to allow for highly accurate sub-phenotyping of those most susceptible to disease, or those least likely to respond to conventional therapy¹⁶⁴. These clinical and genetic enrichment-based approaches will enhance discovery of novel biological mechanisms that are not addressed by existing agents, as well as improve our ability to rapidly test new therapies in early phase trials due to higher event rates, and smaller sample size requirements.

Enhancing our focus on biologically relevant intermediate endpoints:

Efficient translational pipelines require consideration of staged regulatory approaches. Vascular research has set itself an extremely high bar, often using Major Adverse Cardiovascular Events (MACE) or Major Adverse Limb Events (MALE) as the primary endpoints in late-stage clinical trials. Using traditional study design, and minimally stratified recruitment approaches, this high bar drives the need for massive recruitment requirements and high study costs. As a result, many important trials are slower than expected, fail to fully enroll, or run out of funding while waiting for events to accrue¹⁶⁵. Working with regulatory authorities to derive and agree upon biologically meaningful intermediate endpoints with a direct role in disease progression and/or patient reported outcomes is therefore of critical importance. In the case of oncology, tumor size and its impact on quality of life are relatively obvious measures. In the case of PAD, consideration of vascular function, tissue perfusion, lesion “vulnerability” or degree of plaque inflammation may all be considered. In addition, rigorous definition and consensus around tissue specific endpoints may be a unique opportunity for PAD within the Cardiovascular field. This relates to our ability to access relevant tissue much more easily than cardiologists- such as at the time of carotid endarterectomy, debridement or amputation. The field could consider more strategic ways of using this in design of clinical trials- unlocking the possibility of novel biologically relevant intermediate endpoints for new therapies. Further opportunities for biologically relevant endpoints with potential for mechanistic insights include measures of vascular function (e.g. EndoPAT¹⁶⁶) or vascular inflammation and plaque burden (FDG/PET CT¹⁶⁷). Remembering our goal of not only helping patients live longer but also feel better, disease-specific endpoints of relevance to the PAD patient should also be more actively integrated into clinical development programs, including improvements in claudication symptoms or the ability to walk more freely at home (quantified via mobile health devices).

Future Directions:

Vascular biology research continues to build upon important discoveries by embracing the translational approaches described in this review. Next generation genomics should not be considered an adjunct, but rather a critical component of vascular studies used to elucidate mechanisms of disease at the cell-specific level. Delineation of disease process in immune cells versus cells of the vascular wall, for example, may inform future therapeutic direction. Correlation and integration of this data with population-wide GWAS studies may form the foundation upon which novel therapies are pursued in clinical trials, which clearly can be enhanced and streamlined. It is also critical to utilize and improve upon existing animal models of disease. Combining animal models, sophisticated in vitro techniques, and advanced genomics will ultimately bring important discoveries in vascular research closer to translational potential. Consideration and consensus on the use of new biologically meaningful intermediate endpoints in PAD disease progression will be crucial for future clinical trials. Finally, as investigators continue to make discoveries, new challenges emerge about how discovery turns into useful therapy for clinical practice. One consideration is that while the vasculature serves as a conduit for systemic therapeutic delivery, how are blood vessels themselves targeted in a specific and meaningful manner? Based on the disease process being studied, investigators should continue to assess whether targeting disease systemically or at its nidus would provide the most benefit. Along these lines, refinement of tissue and/or cell-specific therapeutic delivery systems will impact vascular research that relies on a targeted approach for disease amelioration. Collectively, the challenges and opportunities outlined in this review ensure an exciting future for translational vascular biology, hopefully bringing enhanced vascular care closer to patients who increasingly need it.

Figure 1.

Schematic demonstrating existing and evolving state-of-the-art components to enhance bench to bedside translation in vascular research.

Figure 2.

Opportunities for novel target identification and prioritization in large, well-phenotyped clinical cohorts integrating molecular and histological characterization with GWAS and single cell multi-“omic” technologies.

Figure 3.

Schematic summary of preclinical cellular and animal models relevant to PAD. In vitro models include angiogenesis assays, and patient-derived co-cultured organoids with the opportunity for CRISPR gene editing. In vivo models include hind limb ischemia and wound-healing in mice and abdominal aneurysm model established in pig models.

Figure 4.

Schematic clinical trial concept with example intermediate biologically and clinically relevant endpoints.

Non-standard Abbreviations and Acronyms:

PAD

Peripheral arterial disease

GWAS

Genome wide association studies

VSMC

Vascular smooth muscle cell

EC

Endothelial cell

ATAC-seq

Assay for transposase-accessible chromatin-sequencing

CITE-seq

Cellular indexing of transcriptomes and epitopes by sequencing

iPSC

induced pluripotent stem cell

SWNT

Single walled nanotube

MACE

Major adverse cardiac events

MALE

Major adverse limb events