Micro- and nanoscale biophysical cues for cardiovascular disease therapy

Abstract

After cardiovascular injury, numerous pathological processes adversely impact the homeostatic function of cardiomyocyte, macrophage, fibroblast, endothelial cell, and vascular smooth muscle cell populations. Subsequent malfunctioning of these cells may further contribute to cardiovascular disease onset and progression. By modulating cellular responses after injury, it is possible to create local environments that promote wound healing and tissue repair mechanisms. The extracellular matrix continuously provides these mechanosensitive cell types with physical cues spanning the micro- and nanoscale to influence behaviors such as adhesion, morphology, and phenotype. It is therefore becoming increasingly compelling to harness these cell-substrate interactions to elicit more native cell behaviors that impede cardiovascular disease progression and enhance regenerative potential. This review discusses recent in vitro and preclinical work that have demonstrated the therapeutic implications of micro- and nanoscale biophysical cues on cell types adversely affected in cardiovascular diseases— cardiomyocytes, macrophages, fibroblasts, endothelial cells, and vascular smooth muscle cells.

Graphical Abstract

Micro- and nanoscale topographies that mimic biophysical cues observed in native tissue architectures can be used to modulate pathological behaviors of adversely impacted cell types involved in cardiovascular diseases— cardiomyocytes, macrophages, fibroblasts, endothelial cells, and vascular smooth muscle cells— to promote wound healing and tissue repair mechanisms.

1. Introduction

Cardiovascular diseases (CVDs) remain the leading cause of death worldwide and continue to place debilitating health and economic burdens on global communities today. Current projections anticipate worsening conditions worldwide with CVDs estimated to cause 22.2 million deaths and cost over $1 trillion by 2030.^1,2 CVDs refer to a range of conditions that affect proper functioning of the heart and blood vessels. These conditions include ischemic heart disease, cerebrovascular disease, and coronary artery disease which often result from atherosclerosis and represent the largest proportion of CVDs.³ Other conditions such as congenital heart disease, rheumatic heart diseases, cardiomyopathies, and cardiac arrythmias are also included.³ In many of these pathologies, the onset of cardiovascular tissue damage and dysfunction triggers wound repair mechanisms that lead to stark changes in resident cell behaviors as well as in the native cellular microenvironment.

Several physiological events and processes that are important when discussing cellular response in CVDs include cell death⁴, inflammation⁵, hypertrophy⁶, fibrosis^7,8, and impaired endothelial function⁹ and vascularization.¹⁰ Many of these reparative responses function as compensatory mechanisms tasked with salvaging and preventing further insult. Even so, morbidity and mortality of CVDs worsen when these wound healing responses become pathologic and ultimately contribute to disease progression. In the case of myocardial infarction (MI), profound cardiomyocyte death triggers a robust inflammatory response and fibrosis results in order to maintain the structural integrity of the weakened myocardium. Cardiomyocyte hypertrophy in non-infarcted myocardium may also occur to compensate for lost myocyte volume due to cell death.¹¹ However, problems arise when deposited granulation tissue disrupts myocyte contractility and endothelial cell (EC) proliferation is insufficient to compensate for enlarged myocytes: this results in an increased oxygen diffusion distance and an exacerbated disease state.^12–14 Further, in atherosclerosis, endothelial dysfunction and persistent inflammation of the arterial wall triggered by accumulation of lipoproteins in the arterial intima result in fibroblast and vascular smooth muscle (VSMC) proliferation, driving the formation of atherosclerotic plaques and onset of disease.¹⁵

Owing to these altered cell phenotypes, it is therefore important to dictate and control cellular responses after cardiovascular injury in order to provide more favorable microenvironments that promote wound healing resolution and tissue repair mechanisms. As such, CVD therapies should focus on modulating the function of cell types adversely impacted during CVD pathologies, such as cardiomyocytes, macrophages, fibroblasts, ECs, and VSMCs. Many strategies have explored regulating cell behavior and mitigating CVD processes through the use of biologics, pharmacological agents, and cell-based therapies. However, these interventions have had limited success in promoting long-term cardiovascular repair, with reasons including inefficient delivery and short half-life of therapeutic factors,^16,17 as well as poor engraftment, survival, and differentiation of transplanted cells.^18,19 Therefore, technologies that provide continuous cues to recapitulate native tissue architectures and appropriately guide cell behavior may yield higher therapeutic success for the treatment of CVDs.

It is well established that many cell types are mechanosensitive and that their biological behaviors vary in response to physical cues sensed in their microenvironment.^20,21 In the human body, the extracellular matrix (ECM) is a unique environmental niche composed of fibrillar and laminar structural components that provide micro- and nanoscale biophysical cues to resident cells. In the myocardium, linear bundles of collagen (120 – 150 nm in diameter) connect adjacent myocytes and networks of coiled perimysial fibers (1 – 10 μm in diameter) are present throughout the muscle and ventricular wall.^22,23 The ECM surrounding cardiac cells exhibits average pores sizes of 21.4 μm as seen in decellularized porcine myocardium.²⁴ Further, topographies present within vascular basement membranes consist of pores and fibers that span the submicron (100 – 1000 nm) and nanoscale (1 – 100 nm) range.²⁵ Physical cues such as these can influence and dictate cellular behavior like adhesion, orientation, morphology, and phenotype through a phenomenon called “contact guidance”.^26,27 As such, it has become increasingly compelling to harness these micro- and nanoscale cell-substrate interactions to elicit more native, non-diseased cell behaviors that impede progression of CVD pathological processes and enhance regenerative potential.

This review seeks to highlight and discuss recent in vitro work that has investigated the therapeutic implications of micro- and nanoscale topography at the cellular level and highlight preclinical studies that have utilized biophysical cues as therapeutic strategies in CVD models. In particular, this review focuses on studies regarding cell types involved in CVDs— cardiomyocytes, macrophages, fibroblasts, ECs, and VSMCs. A full discussion regarding methods of micro- and nanofabrication and the mechanisms by which these topographies modulate cell behavior is beyond the scope of this review and can be found elsewhere.^20,28–30 Further, while an understanding of how biophysical cues alter platelet interactions and influence material hemocompatibility demands consideration when developing materials-based strategies for CVDs, this topic will not be covered in this review and readers are directed elsewhere.^31,32 Lastly, while the term “biophysical cues” is inclusive of characteristics ranging from stiffness, geometry, externally applied forces, material chemistry, and surface wettability, this review will focus on the effect of geometry-based topographic cues on cardiovascular cell types.

2. Micro- and nanoscale biophysical cues to modulate cellular function

2.1. Cardiomyocytes

The native myocardial architecture features directionally dependent behavior in both mechanical and electrical properties, known as cardiac anisotropy. During therapeutic intervention, it has become increasingly apparent that inability to ensure appropriate cardiomyocyte orientation, differentiation, and intercellular connections post-transplantation could have dire consequences, such as poor host-graft interactions and lethal arrythmias.^33,34 As such, recent efforts have successfully designed substrates that provide directional cues to developing cells in vitro that yield cardiomyocytes with better morphological, conductive, and mechanical properties compared to counterparts grown in the absence of these features. Understanding how biophysical cues can be harnessed to increase cardiomyocyte maturation will inform clinical strategies that exogenously deliver stem cells with better host-integration and function and those that aim to restore normative function to endogenous cardiomyocytes during CVDs. This section will discuss in vitro findings pertaining to how physical cues have promoted more mature cardiomyocyte phenotypes by modulating alignment, morphology, gene and protein expression, and contractile behavior of cells. In vivo studies that have explored using topographic cues to improve disease outcomes by augmenting cardiomyocyte maturation and function in CVD preclinical models will be discussed. Additionally, implications for biophysical cues in drug discovery and screening for CVDs will be explored.

2.1.1. In vitro cardiomyocyte responses to micro- and nanotopography

In the native myocardium, regional cardiomyocyte alignment leads to global arrangements of cardiac muscle fibers which can effectively coordinate anisotropic myocardial contractions and electrical signal propagation. Adult cardiomyocytes are generally characterized as having high aspect ratios (~7:1),^35,36 containing parallel arrays of laterally registered sarcomeres,³⁷ and having end-to-end polarization of gap junction proteins^38,39. Studies culturing cells on anisotropic micro- and nanotopographies have demonstrated that biophysical cues can dictate numerous aspects of cardiomyocyte behaviors that may have important implications for translational impact (Table 1). Through dynamic organization of focal adhesions, cardiomyocytes are capable of interacting with and responding to physical cues on substrata.⁴⁰ However, topographic modulation is dependent on intact cellular mechano-signaling capabilities⁴⁰ and continuous presence of patterned cues⁴¹ as otherwise, the observed effects are negated. Various studies have demonstrated increased alignment, aspect ratio, and elongated, cross-striated sarcomere structures in cardiomyocytes when cultured on patterned surfaces compared to unpatterned surfaces (Table 1). Interrogation of geometric constraints on cardiomyocyte behavior indicate that alignment may be more dependent on feature width than aspect ratio⁴² and that there are ideal size regimes that promote morphological indices such as increased cell areas, elongation, and sarcomere lengths in cardiomyocytes (Fig. 1A).⁴³ Sarcomere length is also an important indicator of cardiac maturity as it directly correlates with the amount of force that can be generated from cell contractions.⁴⁴ Adult cardiomyocyte sarcomeres range from 1.8 – 2.1 μm in length,³⁵ whereas immature human induced pluripotent stem cell-derived cardiomyocytes (hiPSC-CMs) have lengths of ~1.65 μm.⁴⁵

Table 1.

In vitro cardiomyocyte response to micro- and nanotopography

Feature Geometry	Feature Size [a]	Substrate Material [b]	Cell Type [c]	Morphology & Cell Behavior [d]	Expression & Function [e]	Ref.
aligned fibers	diameter: 50, 100, 200 μm	PCL/SF/CNT	NRVMs	increased alignment, elongation, α-actinin coverage, Cx43 expression, and sarcomere length on 50 μm aligned fibers	increased beat frequency on 50 μm aligned fibers	(Wu et al., 2017)
	diameter: 5, 10 μm spacing: 25 – 75 μm	ORMOCER®	hiPSC-CMs (WT, LQT3)	for WT— highest alignment on 75 μm spacing, increased 3D tissue formation on 50 μm spacing, and Cx43 end-to-end polarization observed	Higher MCV, MRV, and BD for WT and LQT3 cells on 5 μm vs. 10 μm fibers, LQT3 cells more susceptible to drug- induced cardiotoxicity when grown on 3D matrices with 5 μm fibers	(Ma et al., 2014)
	diameter: 7 μm	PCL	hiPSC-CMs	increased alignment and sarcomere length on aligned vs. random fiber scaffolds	increased expression of cardiac maturationrelated genes (MYH6, MYH7, TNNT2, CASQ2) on aligned scaffolds, increased MCV for aligned scaffolds observed only with ES	(Wanjare et al., 2017)
	diameter distribution: 500 – 2000 nm	PLGA	hiPSC-CMs	increased sarcomere alignment on aligned vs. random fiber scaffolds and flat control	increased expression of cardiac maturationrelated genes (ACTN2, TNNT2, TNNI3, MYH7, MYL2, PLN) and anisotropic propagation of electrical signals on aligned vs. random fiber scaffolds and flat control	(Li et al., 2017)
	diameter distribution: 230 – 2200 nm average diameter: 700 nm	PCL	hiPSC-CMs	increased alignment, elongation, and cell area on aligned vs. random fiber scaffolds	increased expression of CASQ2 & KCNJ2 on aligned vs. random fiber scaffolds, aligned and random fiber scaffolds had slower calcium transient kinetics vs. TCP	(Han et al., 2016)
	diameter: 320 nm	PAN	NRVMs	increased alignment and cardiomyocyte length on aligned vs. random fiber scaffolds, Cx43 end-to-end polarization observed on aligned scaffolds	higher Cx43 expression, beating rate, and beating amplitude on aligned vs. random fiber scaffolds	(Lin et al., 2014)
grooves	width: 10 – 50 μm depth: 3 μm	PGS	NRVMs	increased alignment on grooves (highest on 20 μm width) vs. unpatterned, cross-striated sarcomeres formed perpendicular to cell elongation	increased coordinated beating and synchronized calcium transients on grooves (highest for 20 μm width) vs. unpatterned	(Zhu et al., 2018)
	width/ridge: 350 – 2000 nm depth: 500 nm	PUA	hiPSC-CMs	increased cell area (600 – 1200 nm), increased cell perimeter (600 – 900 nm), highest elongation (700 – 1000 nm), highest alignment (800 nm), increased sarcomere length (750 – 1000 nm)	N/A	(Carson et al., 2016)
	width/ridge: 800 nm depth: 600 nm	AMSF & AMSF + PPy	hESC-CMs	increased alignment, sarcomere length, z-band width, and Cx43 end-to-end polarization on grooves vs. unpatterned	increased expression of cardiac maturationrelated genes (MYH7, GJA1*, SCN5A) on grooves vs. unpatterned *only AMSF substrate	(Tsui et al., 2018)
hexagon patterns	length: 400 μm θ: 60°	mPCL	hiPSC-CMs	enhanced sarcomere density, alignment, and length on hexagon vs. control rectangle patterns	increased expression of calcium handling genes (ATP2A2, KCNJ2, GJA1, and PPARGC1A) and contraction rate on hexagon vs. control rectangle patterns	(Castilho et al., 2018)
	length: 250 μm width: 30 μm spacing: 10 μm θ: 28° depth: 10 μm	PDMS	hiPSC-CMs	increased sarcomere alignment on hexagon patterns vs. unpatterned	increased expression of cardiac maturationrelated genes (MYH6, MYH7, TNNI3, KCNE1, KCNH2 and RYR2) on hexagon patterns but no significant differences in contractility and Ca2+ handling between hexagon patterns and unpatterned	(C. Xu et al., 2017)
	width: 200 μm	GelMA/PEGDA	hiPSC-CMs	with dual MS— increased sarcomere density & gap junctions on hexagon and control mesh patterns	with dual MS— increased MYL2 & RYR2 gene expression and contraction rate on hexagon vs. control mesh patterns	(Cui et al., 2020)
island patterns	AR: 1:1 – 11:1 width: 15 – 115 μm	matrigel/Fn/ glass	hESC-CMs	cell alignment dependent on width (increased for 30 – 80 μm widths) not AR	no trend observed between pattern width and Ca2+ propagation rate	(Salick et al., 2014)
pillars with grooves	pillar diameter: 16 μm pillar length: 48 μm groove width/ridge: 1.5 μm groove depth: 0.5 μm	PDMS	NRVMs	increased alignment on pillars with grooves vs. unpatterned pillars	increased contraction force and higher sensitivity to drug-induced toxicity on pillars with grooves vs. unpatterned pillars	(Oyunbaatar et al., 2016)
pillars	diameter: 120 – 200, 200 – 280, 280 – 360 nm	PS	Flk1+ MPCs	increased colony number, colony area coverage, and colony perimeter (highest on 280 – 360 nm) on pillars vs. unpatterned	increased expression of cardiac markers (MESP1, MYH6, TNNT2, TNNI3, and NKX2.5) on 200 – 280 nm patterns vs. unpatterned	(Seo et al., 2017)
pores	length: 600 (AR: 1.732), 1200 (AR: 1.997) μm	PDMS	NRVMs	increased cell & ECM alignment for 1200 μm pores	higher LCV, CV anisotropy ratio, & twitch force for 1200 μm pores	(Bian et al., 2014)
wrinkles	wavelength: 50 – 3000 nm	PS	hESC-CMs	increased alignment on wrinkles vs. grooves and flat substrates	greater sensitivity to drug-induced toxicity on wrinkles vs. grooves and flat substrates	(Chen et al., 2014)

^[a]AR = aspect ratio

^[b]AMSF = acid-modified silk fibroin, CNT = carbon nanotubes, Fn = fibronectin, GelMA = gelatin methacrylate, mPCL = medical grade polycaprolactone, PAN = polyacrylonitrile, PCL = polycaprolactone, PDMS = polydimethylsiloxane, PEGDA = polyethylene glycol diacrylate, PGS = poly(glycerol sebacate), PLGA = poly(D,L-lactic-co-glycolic acid), PPy = poly(pyrrole), PS = polystyrene, PUA = polyurethane acrylate, SF = silk fibroin

^[c]Flk1⁺ MPCs = fetal liver kinase 1-positive mesodermal precursor cells, hESC-CMs = human embryonic stem cell-derived cardiomyocytes, hiPSC-CMs = human induced pluripotent stem cell-derived cardiomyocytes, LQT3 = long QT syndrome type 3, NRVMs = neonatal rat ventricular myocytes, WT = wildtype

^[d]MS = mechanical stimulation

^[e]BD = beat duration, CV = conduction velocity, ES = electrical stimulation, LCV = longitudinal conduction velocity, MCV = maximum contraction velocity, MRV = maximum relaxation velocity, TCP = tissue culture plastic

Figure 1. Directional biophysical cues improve cardiomyocyte maturation and function in vitro and in vivo.

(A) Sarcomere development of hiPSC-CMs is enhanced on 800 nm wide patterns (α-actinin = red, f-actin = green, nuclei = blue, scale bars = 20 μm) (Carson et al. 2016). (B) Nanogroove patterns increase Cx43 expression and polarization (α-actinin = red, Cx43 = green, nuclei = blue) (scale bars = 25 μm) (Tsui et al. 2018). (C) Masson’s trichome staining of each treatment group post-MI corroborate data showing cell-seeded aligned patches improved LVEF and decreased infarct size after eight weeks post-MI (scale bars = 2 mm) (Lin et al. 2014). (D) Cells on aligned nanofibers have higher and more defined expression of cardiac markers compared to those on random nanofibers (top, left) (scale bars = 50 μm). Anisotropic propagation of electrical signals (bottom) and improved cardiac performance (top, right) was observed for aligned nanofiber scaffolds (Li et al. 2017).

Cx43 is a key gap junction protein important for normal electrophysiological function and is preferentially located on the axial ends of mature cardiomyocytes to enable the generation of high contractile forces and rapid conduction between adjacent cells.^38,39 Many studies have reported higher Cx43 expression^46–49 and characteristic end-to-end polarization of Cx43 on patterned substrates (Fig. 1B)^47,48,50, both of which are indicative of mature contractile function. The ability to control Cx43 end-to-end polarization will be crucial to ensuring efficient electrical activation and potential propagation, as lateral Cx43 distribution often impedes electrical conduction and is frequently observed in CVDs.⁵¹ Studies have also successfully utilized a combination of conductive material,⁴⁸ electrical stimulation,⁵² or mechanical stimulation⁵³ and physical cues to produce more mature cardiomyocyte phenotypes with increased expression of gap junction proteins, enhanced sarcomere density, and larger Z-band widths than patterns alone. The ability for topographical cues to increase expression of cardiac markers relating to early regulators of cardiomyogenesis and cardiac maturity when used alone^48,54–56 as well as in conjunction with directed differentiation and reprogramming protocols^57–59 have been documented. Morez et al. showed that parallel microgrooves enhanced lentivirus-mediated direct reprogramming of cells via topographical regulation of their epigenetic landscape.⁵⁸ Further, differentiation of hiPSC-CMs cultured on aligned poly lactic-co-glycolic acid (PLGA) fibrous scaffolds was significantly enhanced, as indicated by increased expression of genes encoding structural proteins alpha actinin), troponin T, and troponin I and Ca²⁺ handling proteins like phospholamban and ryanodine receptor 2.^59,60 In addition to physical cues being able to enhance cardiac maturity at the transcriptional and translational level, these effects have been reported to increase functional characteristics such as contractile and electrophysiological capability of cardiomyocytes (Table 1). Several studies have reported enhanced calcium transient synchronicity⁶¹ and contraction force⁶² of cardiomyocytes in the presence of microgrooves likely due to increased cell alignment and overall cell connectivity. Further, alignment and elongation of cells have been linked to enhanced contractile function. Cells on aligned scaffolds had significantly increased maximum contraction velocity,⁵⁶ beating rate,⁴⁷ and beating amplitude⁴⁷ compared to those on random scaffolds. Moreover, cells on tissue patches with higher aspect ratio pores and thus, more elongated morphologies, produced higher conduction velocity anisotropy ratios and twitch forces than cells on lower aspect ratio pores.⁶³

2.1.2. Implications of micro- and nanotopography in CVD therapeutic strategies via cardiomyocyte modulation

While cell-based therapies for CVDs have generated a lot of interest, barriers to therapeutic success remain due to challenges in successful integration and appropriate function of delivered cells post-transplantation.^18,19,64 Progenitor cell-seeded material scaffolds have represented an exciting avenue for mitigating issues relating to insufficient cell retention in target tissues.^65,66 Further, material-based solutions are amenable to surface modifications and can deliver continuous cues to promote mature cardiomyocyte phenotypes in transplanted cells. Several labs have interrogated the effects of cell-seeded directional scaffolds on CVD outcomes. Lin et al. assessed the effects of cell-seeded aligned- and random-oriented polyacrylonitrile nanofibrous patches on cardiac function and electrophysiology in rodent models of permanent-ligation MI.⁴⁷ Aligned cell-seeded patches yielded high survival of implanted cardiomyocytes that were also aligned with host cardiomyocytes and resulted in improved left ventricle ejection fraction (LVEF), mitigated ventricular dilation, and reduced infarct size 2 months post-MI (Fig. 1C). Interestingly, cell-seeded random patches resulted in deteriorated cardiac function and maladaptive cardiac remodeling similar to that of the MI control group, indicating the severity of poor host-graft interactions due to inappropriate cues. Aligned cell-seeded patches also improved post-MI electroconductivity and had high expression and polarization of Cx43, whereas cell-seeded random patches resulted in slower conductivities with Cx43 distribution around the nucleus.⁴⁷ In another study, Li et al. utilized cell-seeded aligned PLGA nanofibers and characterized preclinical therapeutic performance on permanent ligation post-MI outcomes. After 4 weeks, improved LVEF, increased fractional shortening, and decreased left ventricle (LV) end-systolic diameter was observed in cellular scaffolds versus acellular control scaffolds, demonstrating favorable ventricular remodeling outcomes (Fig. 1D).⁶⁰

Beyond the use of directionally-aligned fiber scaffolds, a few groups have utilized unique micropattern designs, such as hexagon and honeycomb-like shapes, to produce cardiac patches that better recapitulate the architectural and mechanical properties of native myocardial tissue.^53,67 Cui et al. investigated the use of stretchable hexagon patterned cardiac patches for the treatment of murine models of chronic MI with ischemia/reperfusion (I/R) injury.⁵³ Four months after epicardial implantation, while cellular cardiac patches demonstrated high engraftment rates and strong host-integration, implanted human induced pluripotent stem cell-derived endothelial cell (hiPSC-CMs) did not exhibit organized sarcomere structures, indicating immature phenotypes. Moreover, while both cellular and acellular cardiac patch groups demonstrated increased cardiac performance compared to MI control, their performance was equivalent. These results indicate that the mechanical support these patches offered was the primary mechanism for preventing maladaptive LV remodeling instead of enhanced cardiomyogenesis of implanted cells.⁵³ Kapnisi et al. explored the use of auxetic micropatterning to generate conductive cardiac patches that were capable of expanding in multiple directions to better accommodate the mechanical stress of the heart. In rodent models of permanent-ligation MI, no differences in fractional shortening, end diastolic volume, and end systolic volume were found between auxetic patch and MI control groups. However, higher LV mass in the control MI group indicated that the auxetic patch may have exerted some beneficial effects on cardiomyocytes by reducing mechanical stress experienced through bulking effects as opposed to topographical regulation of cells. To confirm this hypothesis, in-depth analysis of cardiomyocyte size and gene expression would be required to assess the effect of auxetic patches on hypertrophy.⁶⁷ Other work has also utilized electrically-conductive carbon nanotube and nanofibers to augment electrophysiological therapeutic outcomes in infarcted myocardium.^68,69

Another interesting avenue for biophysical cues to help progress work for viable CVD therapies is during drug screening and discovery. Over 30% of drug withdrawals from various markets between 1990 and 2006 were related to cardiotoxicity concerns, making cardiotoxicity determination a critical task during drug discovery.⁷⁰ Therefore, the use of physiologically representative in vitro models to identify therapeutic candidates can greatly minimize potential failures and health risks during clinical trials and expedite time spent during CVD drug development. Several published works have demonstrated that cells cultured on scaffolds with directional cues respond more robustly to drug challenges than cells on conventional culture systems and represent better alternatives to use during drug discovery.^59,60,62,71 To highlight the impact of more accurate culture models, one study analyzed a series of drugs that failed or were withdrawn from the market due to induction of torsades de pointes (Tdp), a potentially lethal form of arrhythmia. They found that an in vitro microgroove culture system was able to capture the torsadogenicity of these drugs whereas conventional culture systems had failed in detecting their cardiotoxicity prior to clinical trials.⁷² Further, biophysical cues can also enable the visualization of more physiologically relevant and differential responses of diseased and healthy cardiomyocyte phenotypes in vitro during drug challenges, such as in the case of patient-specific long QT syndrome type 3 (LQT3) and Duchenne Muscular Dystrophy (DMD). Ma et al. found that on smaller diameter fibers, LQT3-iPS-CMs were more susceptible to drug-induced cardiac arrythmias when compared to 2D pharmaceutical industry standard surfaces, potentially leading to high rates of false-negative readouts during toxicity testing if only traditional 2D cell culture systems are used for drug screening.⁵⁰ Another group found that myotubes derived from healthy and varying DMD severities had substantial differences in alignment when presented with topographic substrates— a finding that could provide an inexpensive phenotypic readout during high-throughput screening of DMD drugs.⁷³

2.2. Macrophages

Macrophages provide the first line of defense and serve as key regulators of homeostasis upon invasion by foreign bodies, during infection, and after injury. Generally, macrophage polarization exists on a spectrum between two activation states— classically activated, “pro-inflammatory” (M1) or alternatively activated, “anti-inflammatory” (M2).⁷⁴ In CVDs such as MI and atherosclerosis, persistent inflammation due to continued presence of M1-like macrophages that secrete pro-inflammatory cytokines to recruit other immune cells greatly contributes to disease progression.⁵ Additionally, the increasing number of procedures involving permanent cardiac implantable electronic devices highlights the need for strategies that attenuate device foreign body response (FBR), which often leads to device complications and failure.^75,76 After implantation, macrophages are among the first cells to interact with the foreign material, and their subsequent activation can lead to further leukocyte recruitment, pro-inflammatory cytokine secretion, and foreign-body giant cell formation. Consequences of these processes include enhanced material degradation, reduced macrophage bactericidal capacity, and increased risk of device infection and failure.⁷⁷ As such, controlling these macrophage-induced inflammatory responses by modulating macrophage activation has gained tremendous interest. Anti-inflammatory agents have already shown beneficial outcomes in attenuating cardiovascular injury^78,79 and there is emerging evidence demonstrating that physical cues can regulate macrophage polarization.⁸⁰ This section will explore reports of how topographic features influence macrophage adhesion, morphology, gene and protein expression, phagocytosis, migration, and cytokine/chemokine release. Additionally, in vivo studies that have investigated topography-induced modulation of macrophage phenotype in CVD models will be discussed.

2.2.1. In vitro macrophage responses to micro- and nanotopography

Macrophages are mechanosensitive and utilize dynamic focal complexes known as podosomes to transduce mechanical cues in their environment.⁸¹ These adhesions can result in integrin clustering and focal adhesion kinase (FAK) activation leading to downstream activation of Rho family of GTPases, key regulators of actin cytoskeletal dynamics. Subsequent changes in actin stress fiber assembly as directed by Rho GTPases, ultimately, influence macrophage morphology.⁸² McWhorter et al. have elaborated on the intricate link between macrophage shape and functional behavior, noting that actin cytoskeletal contractility is a key mediator in shape-induced macrophage polarization.^80,83 Implications of this work emphasize the ability of topography to reliably elicit M1-like or M2-like macrophage phenotypes. Numerous investigations into how micro- and nanoscale topographies modulate macrophage morphology and behavior have been summarized (Table 2).

Table 2.

In vitro macrophage response to micro- and nanotopography

Feature Geometry	Feature Size [a]	Substrate Material [b]	Cell Type [c]	Morphology & Cell Behavior [d]	Expression & Function [e]	Ref.
aligned fibers	diameter: 6.27 μm	PCL	RAW264. 7	increased elongation while no change in cell area on aligned vs. random fiber scaffolds	decreased NOS2 and TNF gene expression, increased ARG1 and Ym1 gene expression, increased M2/M1 ratio, and increased IL-10 secretion on aligned vs. random fiber scaffolds	(Gao et al., 2020)
	diameter: 700 nm	PCL	THP-1	increased elongation on aligned vs. random fiber scaffolds	decreased IL1B gene expression while no differences in CCR7 and MRC1 expression or IL-6, TNF-α, IL- 1β, and IFN-γ secretion on aligned vs. random fiber scaffolds	(Schoenenberger et al., 2020)
circle patterns	area: ~200 μm2	PDMS	murine BMDMs & RAW264. 7	LPS-activated cells showed reduced nuclear area, nuclear volume, F-actin levels but increased chromatin compaction on patterned vs. unpatterned, no change in elongation observed between groups	LPS- activated cells have decreased gene expression of late LPS-activated markers (NOS2, IL1B, CXCL9, and IL6), HDAC3 levels, cytokine secretion (TNF-α, IL- 6, and IL- 12), and PI on patterned vs. unpatterned	(Jain & Vogel, 2018)
fibers	diameter: 0.69, 5.59 μm	PCL	RAW264. 7	cells had deeper infiltration on thick fiber scaffolds	increased gene expression of M1- related IL6, TNF, and MIF-1 on thin fiber scaffolds while increased gene expression of M2- related ARG1 and FIZZ1 on thick fiber scaffolds	(Z. Wang et al., 2014)
grooves	width: 0.15 – 50 μm pitch: (2 x width) μm depth: 0.8 – 1.3 μm	Ti	murine BMDMs	highest elongation on 400 – 500 nm width grooves, no observed differences in cell area across groove widths	ARG1 expression was dependent on groove dimensions (highest on 400 nm – 5 μm grooves) while iNOS expression remained low across all groove widths, 5 μm grooves yielded highest secretion of IL-10	(Luu et al., 2015)
	width: 2 – 40 μm ridge: 10 μm depth: 11 μm	gelatin	THP-1	PMA-activated cells showed increased cell area over time on grooves vs. unpatterned	20 μm grooves led to increased IL-1β, TNF- α, IL-12, and IL-10 secretion in PMA/M1activated cells vs. unpatterned	(Dollinger et al., 2017)
	width: 5 – 30 μm depth: 4 μm	PEG-modified AuNRs/PCL	murine BMDMs	greatest elongation on 10 and 20 μm width grooves	after NIR- stimulation for “groove” surface— increased ARG1 gene expression, decreased iNOS gene expression, increased IL-10, IL1-β, TNF-α, and CCL1 secretion vs. before NIR- stimulation “flat” surface	(Zheng et al., 2019)
	width: 20 μm ridge: 10 μm depth: 20 μm	GelMA	human MDMs	no significant changes in cell area, perimeter, cytoplasm:nucleu s ratio, and elongation between pillars, grooves, and unpatterned substrates	with LPS- stimulation — grooves reduced TNF-α secretion vs. unpatterned; increased average number of zymosan particles per phagocytic cell on grooves vs. pillars	(Singh et al., 2017)
pillars	height: 20 μm diameter: 20 μm spacing: 5 μm	GelMA	human MDMs	no significant changes in cell area, perimeter, cytoplasm:nucleu s ratio, and elongation between pillars, grooves, and unpatterned substrates	with LPS- stimulation — pillars reduced TNF-α secretion vs. unpatterned, trend towards decreased PI for pillars vs. grooves and unpatterned substrates	(Singh et al., 2017)
	diameter: 55 nm	Pt-based BMG (Pt57.5Cu14.7Ni5.3P22.5)	murine BMDMs	For M0 and IFN-γ/LPS-activated cells— decreased cell area and perimeter but no change in elongation on pillars vs. flat substrate	with IFN-γ/LPS- stimulation — decreased IL-1α, TNF- α, IL-6, IL- 12, CCL2, CCL4, CCL5, and CXCL1 secretion, increased phagocytosis on pillars vs. flat substrate	(Shayan et al., 2018)
pits	diameter: 3 – 20 μm inter- element distance: 6 – 20 μm elemental cell: SQ, HEX	biocellulose (Acetobacter Xylinum)	THP-1	75% reduction in adhesion on patterned vs. unpatterned surface	decreased gene expression of CCL17 on patterned vs. unpatterned surfaces	(Robotti et al., 2018)
textures	roughness: ~3.58 μm	Ti (hydrophobic, hydrophilic)	murine BMDMs	decreased adhesion and elongation on all rough vs. smooth surfaces	decreased M1-related IL-1β, IL-6, and TNF-α secretion and increased M2-related IL-4 and IL- 10 secretion for hydrophilic rough surfaces vs. hydrophobic smooth surfaces	(Hotchkiss et al., 2016)
	roughness: 0.10, 1.2 μm Str: 0.03, 0.08 μm	epoxy	RAW264. 7	no change in adhesion but increased elongation, alignment, and multinucleated cell number on grooves vs. smooth surfaces	increased gene expression of macrophage attractant chemokines (CCL2, CCL3, CCL4, CCL5, CCL7) and higher secretion of CCL4 on grooves vs. smooth surfaces	(Moon et al., 2016)
	roughness: 50 – 150 nm	CNT/mPCU	J774A.1	increased adhesion and proliferation on CNT/mPCU vs. mPCU substrates, decreased migration distance and velocity on rougher surfaces	decreased nitrite, TNF-α, and IL-1β secretion on CNT/mPCU vs. mPCU substrates	(Khang, 2015)
tubes	diameter: 100 nm	TiO2	murine BMDMs	decreased elongation and cell area on tubes vs. flat surfaces	increased CD86 and iNOS protein expression, decreased ARG1 protein expression, and increased CCR7 and IL1B gene expression on nanotube vs. flat surfaces	(He et al., 2020)
	diameter: 50, 70 nm	TiO2	murine BMDMs & RAW264. 7	decreased proliferation on 70 nm nanotube vs. 50 nm tubes and rough surfaces	N/A	(Gulati et al., 2018)
wrinkles	major depth: 1 μm major wavelength : 3 – 5 μm minor depth: 300 nm minor wavelength : 300 nm	PE	murine BMDMs	increased elongation on 1D wrinkles vs. 2D wrinkles & flat surfaces	increased ARG1 protein expression, increased IL-10 secretion, and decreased TNF-α secretion on 1D wrinkles vs. flat surfaces	(T. Wang et al., 2016)

^[a]HEX = center to center distance in a hexagonal array, SQ = center to center distance in a square array, S_tr: texture aspect ratio (S_tr = 1, isotropic S_tr = 0, anisotropic

^[b]AuNRs = gold nanorods, BMG = bulk metallic glass, CNT = carbon nanotubes, GelMA = gelatin methacrylate, mPCU = medical grade polycarbonate urethane, PCL = polycaprolactone, PDMS = polydimethylsiloxane, PE = polyethylene, PEG = poly(ethylene glycol), Pt = platinum, Ti = titanium

^[c]BMDMs = bone marrow-derived macrophages, MDMs = monocyte-derived macrophages

^[d]IFN-γ = interferon gamma, LPS = lipopolysaccharide, PMA = phorbol-12-myristate-13-acetate

^[e]NIR = near-infrared, PI = phagocytic index

Several studies have used micro- and nanoscale features to modulate macrophage adhesion,^84–86 proliferation,^86–88 and migration⁸⁶ on substrates. Interestingly, a high-throughput screening of diverse micropatterns identified a link between increased macrophage attachment and M2 phenotype.⁸⁹ Further, topographical influence of macrophages into more elongated morphologies have been found to accompany M2-like phenotypes, as characterized by increased gene expression of arginase-1 (ARG1) and secretion of interleukin 10 (IL-10), an anti-inflammatory cytokine (Fig. 2A).^88,90–92 Macrophages cultured on larger diameter polycaprolactone (PCL) fibrous scaffolds showed M2 polarization as evidenced by increased ARG1, resistin-like molecule alpha (FIZZ1), and matrix metalloproteinase 2 (MMP2) expression (Fig. 2B).⁹³ Meanwhile, smaller diameter PCL scaffolds showed M1 polarization as indicated by elevated interleukin 6 (IL6/IL-6), tumor necrosis factor (TNF-α), and macrophage migration inhibitory factor 1 (MIF-1) expression (Fig. 2B).⁹³ Work done by Jain et al. further elaborated on the link between macrophage morphology and activation by showing a correlation between cell spreading area, M1 activation, and total F-actin levels.⁹⁴ Micropatterning adhesive areas of 200 μm² confined lipopolysaccharide (LPS)-stimulated bone-marrow derived macrophages and resulted in lower expression of the late LPS-activated transcriptional markers IL6, C-X-C motif chemokine ligand 9 (CXCL9), interleukin 1 beta (IL1B/IL-1β), and nitric oxide synthase (NOS2) compared to unconfined cells (Fig. 2C).⁹⁴ Not only have topographic cues been shown to be potent mediators of macrophage activation state, but they can also promote⁹⁵ or inhibit^94,96 phagocytic behavior.

Figure 2. Topographic cues are potent modulators of macrophage polarization in vitro and in vivo.

(A) Macrophage elongation, gene expression, and secretion profile is modulated by culture on multiscale wrinkle features (scale bars = 50 μm) (Wang et al. 2016). (B) Low diameter fiber scaffolds (left, bottom) promote M1 polarization (middle, top) while larger diameter fiber scaffolds (left, top) promote M2 polarization (middle, bottom). Thick fiber grafts promote M2 macrophage infiltration (CD206 = green, nuclei = blue) between day 7 (right, top) and day 100 (right, bottom) (scale bars = 200 μm). (Wang et al. 2014). (C) Macrophages restricted to adhesive islands (CC) had reduced gene expression levels of late LPS-responsive gene clusters compared to unrestricted cells (UC) (Jain et al. 2018). (D) Distinct macrophage localization profiles between grafts were observed. Grafts with larger pores (PLA-PLCL) had reduced neointimal calcification compared to grafts with smaller pores (PLA-nano) (scale bars = 200 μm) (Tara et al. 2014).

Several studies have also investigated the effect of topography on macrophage fusion since the formation of foreign body giant cells is a hallmark of FBR against implants. The use of micropatterned islands was shown to reduce both macrophage adhesion and formation of multinucleated cells,⁹⁷ while higher surface roughness led to increased expression of proteins involved in the recruitment and fusion of macrophages that correlated with increased number of multinucleated cells.⁹⁸ Material surface interactions may play a role in some observed macrophage responses to topographies as macrophage fusion requires sufficient surface adsorption of proteins^99,100; perhaps enhanced protein adsorption on grooved surfaces compared to smooth may account for this discrepancy.⁹⁸ Further, it was shown that anti-inflammatory macrophage phenotypes were further increased on surfaces that were hydrophilic when roughness was held equivalent.⁸⁵ Beyond geometry and surface interactions, pattern density and area of topographies can also influence macrophage phenotype; M1 phenotypes were found to be more prevalent on wider, more disperse micropillar features compared to smaller, more dense micropillars which promoted the M2 phenotype. Lastly, the biochemical microenvironment may also alter topography-mediated polarization of macrophages as pre-differentiated M1/M2 macrophages and co-stimulated macrophages in M1/M2 media had very different cytokine/chemokine secretion profiles when cultured on microgrooves.¹⁰¹

Since macrophage activation exists on a continuum, oftentimes observed macrophage expression and secretion profiles do not always fit the distinct “M1” or “M2” profile.⁷⁴ Even in studies discussed here, biophysical cues have yielded complex phenotypic profiles that cannot be characterized as strictly “pro-inflammatory” or “anti-inflammatory”. Dollinger et al. noted that phorbol-12-myristate-13-acetate (PMA)/M1-activated THP-1 macrophages on 20 μm width gelatin grooves yielded increased IL-1β, TNF-α, interleukin 12 (IL-12), and IL-10 secretion vs. unpatterned surfaces— a distribution of factors that are attributed to both M1 (IL-1β, TNF-α, IL-12) and M2 (IL-10) phenotypes.¹⁰¹ As such, efforts are being made to look beyond the traditional analysis of pro-inflammatory/anti-inflammatory gene expression. Beyond the conventional factors associated with M1 and M2 phenotypes, studies have shown that topographies elicit significant expression changes in genes related to primary metabolic processes such as transcription, translation, protein trafficking, DNA repair, and cell survival when compared to unpatterned.⁹⁶

2.2.2. Implications of micro- and nanotopography in CVD therapeutic strategies via macrophage modulation

Much of our understanding of topography-induced modulation of macrophages in CVDs comes from biomaterials work developing vascular grafts. Numerous studies have leveraged material-cell interactions by tuning pore size of vascular grafts to maximize cell infiltration and vascularization while also documenting macrophage response and therapeutic outcomes.^93,102,103 Vascular grafts made of PCL microfiber scaffolds with 5 – 6 μm diameters and 30 – 40 μm pores increased polarization and recruitment of anti-inflammatory polarized macrophages. In vivo implantation of grafts was assessed over 100 days and showed that these scaffolds induced a large number of CD206⁺ M2 macrophages to infiltrate the graft wall (Fig. 2B), further promoting vascularization and regeneration of the tunic media compared to smaller diameter grafts with correspondingly smaller pore sizes.⁹³ Another study demonstrated that polylactic acid (PLA)-based nanofiber grafts with large pores (~30 μm) induced well-organized neointima and negligible neointimal calcification compared to grafts with smaller pores (~0.7 μm) after 12 months.¹⁰² Interestingly, there were regional differences of macrophage localization with higher macrophage presence in the neointima of smaller pore grafts compared to larger pores (Fig. 2D). As pro-inflammatory macrophage response is known to be a contributor to early stages of atherosclerotic intimal calcification,¹⁰⁴ it was speculated that the macrophages present in the neointima of smaller pore grafts led to the observed calcification in these grafts, however analysis of macrophage phenotypes was not performed in the present study.¹⁰² Three-layer PCL grafts with a dense inner and outer layer (pore size of 3 – 4 μm) and a loose middle layer (pore size of ~26 μm) were used to promote vascular regeneration in a rabbit carotid artery model. Results showed 6x greater macrophage infiltration in three-layer PCL grafts versus dense monolayer grafts, with a majority of macrophages localizing to the loose middle layer. Significantly higher monocyte chemoattractant protein-1 (MCP-1) expression and vascular endothelial growth factor (VEGF) were detected in the middle layer as well, perhaps indicating increased M1 macrophage presence. While further investigation into macrophage phenotype switching needs to be conducted, the infiltrating macrophages may have played an important role in enhancing vascular smooth muscle cell migration, tunica media regeneration, and vascularization compared to monolayer grafts.¹⁰³

Other work with vascular grafts has investigated the use of directional cues to promote more native vascular regeneration.^105,106 Small diameter vascular grafts with oriented microchannels were found to promote the ratio of M2 to M1 macrophages at 4 and 12 weeks as compared to unaligned grafts (Fig. 2E).¹⁰⁵ Another study showed that while both M1 and M2 macrophage phenotypes were observed in circumferentially aligned bi-layered microfibrous PCL scaffolds 18 months post-transplantation, their distribution was different: M2 phenotypes localized to the outermost layer of the graft whereas M1 phenotypes were found within the graft wall.¹⁰⁶ These studies raise interesting observations where perhaps differential localization of polarized macrophages may serve to enhance therapeutic success of vascular grafts as both phenotypes contribute in different ways to tissue remodeling. While anti-inflammatory macrophage responses are pivotal in limiting FBR against implants, pro-inflammatory macrophage responses have been shown to promote vascularization by increasing expression of genes associated with sprouting angiogenesis in endothelial cells and secreting potent pro-angiogenic factors.^107,108 Establishing competent vascular networks inside grafts is crucial for enabling sufficient oxygen and nutrient exchange, proper integration with host, and improved tissue regeneration. Clearly both M1 and M2 macrophage populations play intricate and coordinated roles in cardiovascular protection and as such, the situations where these different macrophage phenotypes may be beneficial are likely nuanced and context dependent.

Beyond exploring applications for vascular grafts, scientists are also interested in identifying viable surface modifications to limit maladaptive FBR against medical implants like cardiovascular implantable electronic devices. Robotti et al. developed a micropatterned biocellulose membrane that served as a wrap to protect implanted pacemakers from fibrosis.¹⁰⁹ Explants from chronic minipig models after 3 and 12 months demonstrated that cellulose wrapped pacemakers had greatly reduced fibrous capsule formation and showed no signs of degradation or disruption to the devices compared to native pacemakers. Further, leads in native pacemakers were found strongly adhered to fibrous capsules and therefore dissection and severing of the leads was necessary in order to extract the devices. Interestingly, histological analysis revealed that the inflammatory response between wrapped and native pacemakers had similar morphological characteristics in terms of inflammatory cell composition and blood vessel presence, despite the pronounced reduction in fibrotic capsule formation in wrapped pacemakers. While analysis into markers of inflammation and macrophage phenotypes was not performed in this study, further investigations could point to a more comprehensive picture into topographical modulation of FBR.¹⁰⁹

2.3. Fibroblasts

Fibroblasts make up the largest cell population in the heart and are responsible for coordinating numerous processes that maintain normal cardiac function. They coordinate synthesis and degradation of ECM, cell-cell signaling between cardiomyocytes, and cytokine and growth factor secretion.¹¹⁰ Upon cardiovascular injury, fibroblasts become effector cells and transform into their activated myofibroblast phenotype under the influence of pro-inflammatory signals secreted during the wound healing response. Myofibroblasts predominantly function to increase ECM deposition to provide provisional matrices that stabilize the site of injury; in MI, this compensatory fibrosis maintains the mechanical integrity and pressure generating ability of the myocardium, reducing likelihood of myocardial dysfunction or rupture.¹¹¹ Issues arise when persistent inflammatory signals result in rampant fibrosis that may lead to chronic deposition of stiff, dense scar tissue that impairs contractility and perfusion mechanisms within the heart.¹¹¹ Mechanical cues have been reported to modulate fibroblast phenotype,^112,113 and as such, ongoing efforts aim to leverage fibroblast mechanosensing to yield quiescent fibroblast phenotypes that mitigate fibrotic response in CVDs and minimize scar tissue formation around implanted devices. This section will discuss findings related to how these features affect fibroblast adhesion, proliferation, morphology, gene and protein expression, migration, and secretion profiles. Additionally, in vivo studies that have investigated the ability of topographic cues to improve therapeutic outcomes in CVDs by inhibiting fibrotic responses will be discussed.

2.3.1. In vitro fibroblast responses to micro- and nanotopography

Fibroblasts are sensitive to biophysical cues in their environment. The transduction of these signals by focal adhesions subsequently affect contractility mechanisms and heavily dictate their resulting behaviors.²⁰ Recent studies that have reported topography-induced modulation of fibroblast morphology and behavior are briefly summarized (Table 3). During wound repair, fibroblasts can be activated by pro-fibrotic signals to increase proliferation and initiate differentiation into myofibroblasts. The use of biophysical cues have been documented to modulate fibroblast adhesion⁸⁴ and proliferation,^97,114 reporting unique size regimes that can promote or inhibit these behaviors. Identifying length scales that reliably dampen proliferative behaviors can prove essential to strategies aiming to mitigate rampant fibrotic response.

Table 3.

In vitro fibroblast response to micro- and nanotopography

Feature Geometry	Feature Size [a]	Substrate Material [b]	Cell Type [c]	Morphology & Cell Behavior [d]	Expression & Function	Ref.
discrete cubes	width: 15 μm height: 15 μm length: 15 μm	PEG	NIH 3T3	cells extend projections to bind to or envelop cubes	dose-dependent trend towards decreased MMP2, SRF, YAP1, and TAZ gene expression with cubes vs. untreated	(Pinney et al., 2014)
discrete rods	width: 15 μm height: 15 μm length: 100 μm	PEG	NRVFs & NIH 3T3	cells tend to cluster on/near rods	dose-dependent decrease in MMP2, SRF, YAP1, and TAZ gene expression with rods vs. untreated	(Pinney et al., 2014)
	width: 15 μm height: 15 μm length: 100 μm	HA	NIH 3T3	cell form distinct focal adhesions to rods, increased proliferation with rods vs. untreated	dose-dependent decrease in ACTA2, COL1, TGFB1, SMAD3, and MMP2 gene expression with rods vs. untreated	(Le et al., 2018)
fibers	diameter: 1 μm height: 6, 16 μm	PP	NIH 3T3	decreased prominent stress fibers, elongation, pMLC staining, and SMAD2/3 nuclear localization on fibers vs. unpatterned	decreased ACTA2 and COL1 gene expression on 16 μm fibers vs. unpatterned, decreased COL3, TGFB1, TGFBR2, and SMAD3 gene expression on 6 and 16 μm fibers vs. unpatterned	(Allen et al., 2016)
grooves	width/ridge: 100 μm depth: 3 μm	PDMS/PAA	NIH 3T3	increased elongation and decreased proliferation on grooves vs. bare PDMS	decreased α-SMA expression vs. bare PDMS and PAAcoated PDMS	(J. S. Lee et al., 2019)
	width: 5 – 50 μm ridge: 10, 50 μm depth: 5 – 15 μm	PEG	L929	preferential migration along patterns on 10 μm grooves, random migration on 50 μm grooves, migration speed increased with groove width	N/A	(De Vicente & Lensen, 2016)
	width: 10 μm ridge: 5 μm depth: 5 μm	SiO2/ZrO2	HGFs	decreased spreading, increased elongation, and no change in proliferation on grooves vs. flat control	no difference in FSP1 or COL1 gene expression on grooves vs. flat control	(Laranjeira et al., 2014)
island patterns	length: 2 – 20 μm width: 2 μm spacing: 5 μm	Fn or Col or sNAG/silicone	primary RSFs	greatest adhesion on 4 × 2 μm patterns vs. full protein/sNAG coated control	with and without TGF-β1stimulation— decreased number of α-SMA expressing cells on 4 × 2 μm patterns vs. full protein/sNAG coated control	(Majd et al., 2015)
pillars	height: 5 μm diameter: 5 μm spacing: 10 μm	SiO2/ZrO2	HGFs	decreased spreading and increased proliferation on pillars vs. grooves and flat control	no difference in FSP1, increased COL1 gene expression on pillars vs. grooves and flat control	(Laranjeira et al., 2014)
pits	diameter: 3 – 20 μm inter- element distance: 6 – 20 μm elemental cell: SQ, HEX	PDMS & biocellulose (Acetobacter Xylinum)	HDFs	feature-dependent adhesion, elongation, and cell area on patterned vs. flat PDMS, reduced adhesion on patterned vs. flat biocellulose	N/A	(Robotti et al., 2018)
	diameter: 120 nm depth: 100 nm spacing: 300 nm elemental cell: SQ, HEX, NSQ	PC	NIH 3T3	no difference in cell area, nucleus area, actin intensity, and focal adhesion area on patterns vs. flat	increased pFAK/FAK and TGFB1I1 gene expression on NSQ vs. flat, increased ELN gene expression on HEX vs. flat, and increased TAZ nuclear localization and COL3 gene expression on SQ, HEX, and NSQ vs. flat	(Cutiongco et al., 2020)
pores	diameter: 200 – 400 nm spacing: 480 nm	PS	NIH 3T3	cells unable to form adhesions inside pores, increased migration distance and speed with larger pore sizes vs. flat, decreased FA number and size with larger pore sizes vs. flat	increased pFAK/FAK ratio on 400 nm pores vs. flat	(Lim et al., 2018)
	diameter: 100 μm height: 3 μm	PDMS/PAA	NIH 3T3	decreased proliferation on pores vs. bare PDMS	decreased α-SMA expression vs. bare PDMS and PAAcoated PDMS	(J. S. Lee et al., 2019)
stripe patterns	width: 5 – 50 μm	Si/PVA	AG01523	highest alignment in 20 and 22.5 μm stripes, increased elongation along 20 – 25 μm stripes, increased proliferation on 22.5 μm stripes vs. unpatterned Si	N/A	(Bourkoula et al., 2019)
textures	roughness (Sa): 484– 522.7 nm	PDMS	BDFs	increased adhesion, increased proliferation, and decreased apoptosis on textured PDMS surfaces vs. smooth and textured silicone implant surfaces	decreased TNF and IL8 gene expression on textured PDMS surfaces vs. smooth silicone implant surfaces after 24hrs, decreased TGFB1 and HSP60 gene expression on textured PDMS surfaces vs. smooth silicone implant surfaces after 1 week, decreased IL- 1β, TGF-β1, IL-6, IL-8, IFN-γ, TNF-α, and GM-CSF secretion, and increased IL-10 secretion on textured PDMS surfaces vs. smooth silicone implant surfaces after 1 week	(Kyle et al., 2015)
tubes	diameter: 70 – 90 nm height: 200 – 250 nm	TiO2	HGFs	increased proliferation and number of extended lamellipodia on tubes vs. unpatterned	increased ITGA3, ITGB1, FN1, and VCL gene expression and increased COL1 secretion on tubes vs. unpatterned	(R. Xu et al., 2018)
wrinkles	amplitude: 0.14 – 3 μm wavelength: 2 – 8 μm	PDMS	L929	faster migration when scratch was made perpendicular to wrinkle features, greater migration on lower amplitude features when wavelength constant, greater migration on larger wavelength features when amplitude constant, decreased cell area and feature-dependent reduction in FA area per cell on wrinkles vs. flat	N/A	(Ge et al., 2020)

^[a]HEX = center to center distance in a hexagonal array, NSQ = center to center distance offset from 300nm by 50 nm in both x and y directions, S_a = arithmetical mean height, SQ = center to center distance in a square array

^[b]Col = collagen, Fn = fibronectin, HA = hyaluronic acid, PAA = polyacrylic acid, PC = polycarbonate, PDMS = polydimethylsiloxane, PEG = polyethylene glycol, PP = polypropylene, PS = polystyrene, PVA = polyvinyl alcohol, Si = silicon, sNAG = poly N-acetyl glucosamine

^[c]BDFs = breast-derived fibroblasts, HDFs = human dermal fibroblasts, HGFs = human gingival fibroblasts, NRVFs = neonatal rat ventricular fibroblasts, RSFs = rat subcutaneous fibroblasts

^[d]FA = focal adhesion

Myofibroblasts exhibit high expression of alpha-smooth muscle actin (ACTA2/α-SMA) in their stress fibers to augment contractile activity and aid in the generation of tissue contractures.¹¹⁵ Further, myofibroblast characteristics include upregulated ECM protein secretion, matrix metalloproteinases (MMPs) synthesis, and cytokine production to promote ECM remodeling and fibrosis.¹¹⁵ Many reports have discussed interesting relationships between focal adhesion-dependent cytoskeletal tension, nucleus remodeling, and transcriptional activity.^114,116,117 This therefore opens up the possibility for geometric control of gene and protein expression profiles of fibroblast populations to discourage myofibroblast phenotypes. The formation of focal adhesions on substrates subsequently leads to cytoskeletal tension. Generated local tensile stresses governed by substrate properties (i.e. geometry, surface area, and aspect ratio) have been reported to affect cytoskeletal organization, remodeling of the nucleus, and chromatin compaction.¹¹⁶ Another study showcased that changes in cell-geometry and nuclear morphology from underlying substrates can lead to reorientation of chromosome territories and result in differential transcriptional activity.¹¹⁷ Cell proliferation and collagen secretion were found to strongly correlate with nuclear volume, highlighting the nucleus as a critical mechanosensor for fibroblast behavior (Fig. 3A).¹¹⁴ Moreover, this same study showed that nuclear volume was effectively modulated by the height of nanotopography features through the rearrangement of focal adhesions, demonstrating the ability for topographies to influence cell behavior through nuclear remodeling.¹¹⁴ These findings are in line with observations that have demonstrated how topography can physically influence fibroblast gene and protein expression (Table 3) and, of clinical relevance, how physical cues can suppress fibrotic activation as evidenced by decreased expression of ACTA2, collagen I (Col1A1), matrix metalloproteinase 2 (MMP2), and tafazzin (TAZ)^97,118–121 and reduced secretion of common pro-fibrotic cytokines.¹²¹ Regular arrays of 4 × 2 μm adhesive islands allowed for fibroblast attachment and proliferation while simultaneously suppressing α-SMA organization into contractile fibers in the presence of TGF-β1, a potent pro-fibrotic cytokine (Fig. 3B).⁹⁷ Further, discrete microscale polymeric structures have also shown to have success in decreasing gene expression of markers of the fibrotic phenotype in both 2D and 3D cultures regardless of material used.^118,119 However, not all variations of topographic patterns may result in attenuation of myofibroblast phenotypes. Cutiongco et al. recently developed a model to relate the effects of cell morphology to gene expression induced by nanotopography. Interestingly, they found that fibroblasts showed increased expression of pathogenic fibrosis markers on nanotopographic substrates compared to unpatterned surfaces. Interestingly, they did not observe any similarities in fibrotic gene expression profiles by nanotopographies induction versus biochemical fibrotic induction, indicating unique fibrosis signatures being elicited by these nanoscale physical patterns.¹²²

Figure 3. Micro- and nanoscale topographies modulate fibrotic responses of fibroblasts in vitro and in vivo.

(A) Anisotropic and isotropic nanotopographies (left) show height-dependent changes in nuclear volume (right, top) and proliferation (right, bottom) (scale bars = 1 μm) (Wang et al. 2016). (B) 4 × 2 μm micropatterns of varying protein adhesive islands (left) decrease α-SMA organization into contractile fibers in the presence of TGF-β1 (right) (scale bars = 100 μm) (Majd et al. 2015). (C) Fibroblasts actively engage with HA microrods (left, top) and exhibit reduced markers of the myofibroblast phenotype when cultured with structures (right, top) (scale bars = 20 μm). HA microrod injections yield increased LV wall thickness, reduced local collagen deposition, and improved cardiac performance compared to saline and HA material injection (bottom) (scale bars = 2 mm) (Le et al. 2018). (D) Scanning electron microscope (SEM) micrographs of biocellulose coatings (PMC) (top) (scale bars = 300 μm) that yield reduced fibrotic capsule formation (middle) compared to native, uncoated pacemakers (BI) at 3 months (bottom, left) and 12 months (bottom, right) (scale bars = 400 μm) (Robotti et al. 2020).

During the wound healing response, fibroblasts are actively recruited to the injury via various signaling mechanisms such as IL-1β stimulation and angiotensin II type 1 receptor and β-adrenergic receptor activation.¹¹¹ Increased presence of these fibroblasts due to migration, much like with proliferation, greatly contributes to fibrosis progression. In terms of migration capability, numerous studies have reported that biophysical cues influence the migratory capacity of fibroblasts.^123–127 Cell migration on graphene oxide micropatterns revealed that the geometry of micropatterns (i.e. triangle, square) heavily influenced cell migration behavior including distance, speed, and directionality of travel and that migration likely correlated with observed cell morphology; cell migration distance and speed were significantly higher on triangle micropatterns than square micropatterns. Cells on square micropatterns were more spread than those on triangle patterns, which could have resulted in the slower migration behaviors.¹²⁴ Another study corroborated the link between adhesive area and migration speed as nanopores that constrained adhesion areas resulted in increased migration speed as adhesive area decreased.¹²⁷ Further, it was found that there was increased migration, collagen synthesis, and proliferation on rough surfaces.¹²⁵ These studies all point to designs for topographical control of fibroblast behavior to include considerations for increasing cell adhesive area and reducing surface roughness.

2.3.2. Implications of micro- and nanotopography in CVD therapeutic strategies via fibroblast modulation

Therapeutic strategies that can successfully attenuate the fibrotic response have tremendous implications for the treatment of CVDs. After MI, onset of fibrosis can lead to stiffening of the myocardial wall and impede cardiac contractility, furthering disease progression and worsening morbidity outcomes.¹¹¹ Also, of importance, is the ability to prevent device failures and complications of medical implants due to fibrous encapsulation during FBR.^75–77

In the context of MI, a few studies have demonstrated the therapeutic impact of intramyocardial injections of discrete microstructures for augmenting cardiac performance by attenuating cardiac fibrosis.^118,119,128 Pinney et al. compared the performance of injectable polyethylene glycol (PEG) microcube and microrod structures and demonstrated that both decreased collagen and TGF-β levels, increased elastin deposition, improved vascularization, and enhanced cardiac function after 6 weeks in rodent models of ischemia-reperfusion (I/R) MI.¹¹⁹ Interestingly, while in vitro studies pointed to better performance of higher aspect ratio microrods, in vivo results demonstrated that microcube structures saw increased cardiac performance as calculated by relative change in LVEF over the 6-week period post-MI.¹¹⁹ Plausible explanations for these discrepancies between in vitro and in vivo performance may be attributed to (1) lower injection efficiency of the higher profile microrods and (2) increased surface area of the volume-controlled injection of microcubes. Further investigations into the mechanisms by which discrete microtopographical structures regulate fibroblast phenotype will inform whether higher aspect ratio structures, which can better resist cellular contractile forces, or higher surface area structures, which provide more adhesive area, are optimal for dictating therapeutic potential of microstructural strategies.

In a subsequent study, Le et al. investigated the performance of hyaluronic acid (HA) microrods compared to mass-controlled soluble HA injections in rodent models of I/R MI.¹¹⁸ Improvements in morphological characteristics in the HA microrod group as evidenced by increased LV wall thickness and moderately reduced global LV collagen deposition were apparent (Fig. 3C).¹¹⁸ However, the extent of fibrosis attenuation may have been underestimated due to the reporting of global collagen deposition as HA microrods provide a local anti-fibrotic response and only exert their mechanistic effects in their immediate area. This is corroborated histological analysis that demonstrates local reduction in collagen deposition in areas around HA microrod injections versus areas distal to the injection. Observed histological improvements correlated with functional assessments as both stroke volume and relative change in LVEF were greatly enhanced in the microrod group. Less significant improvements over saline were observed for soluble HA control, indicating that while HA itself does provide benefits post-MI, the form factor of these structures was crucial to therapeutic efficacy. In a permanent ligation MI model, PEG microrods showed modest prevention of cardiac deterioration as measured by hemodynamic analysis and pathologic remodeling compared to MI controls.¹²⁸ Even so, PEG microrods attenuated gene expression of embryonic β-myosin heavy chain isoform and atrial natriuretic factor which are associated with pathological remodeling of the ventricles and preserved contractile efficiency after 10 weeks compared to MI control.¹²⁸ Overall, the use of injectable crosslinked hydrogel structures affords numerous benefits including fine control over geometry and stiffness, ability to release therapeutic factors,¹²⁸ and ease of injectability. Additionally, graphene oxide/alginate and reduced graphene oxide/alginate spherical microgels were investigated for their potential to protect delivered human mesenchymal stem cells (hMSCs) from oxidative stress after implantation into post-MI myocardium. While some therapeutic success was observed in the hMSC-loaded microgel group, minimal effects on fibrosis were observed for the microgel group alone.¹²⁹ It is interesting to note the differences in microstructure dosing, timing of administration, and MI models used across these studies and how these parameters may all influence therapeutic efficacy of microstructure strategies for MI treatment.

The potential of biophysical cues to mitigate FBR response to implanted devices has generated much excitement in the field of implantable biomaterials. Robotti et al. investigated the use of micropatterned hexagonal pits on biosynthesized cellulose (BC) membranes to reduce fibrous encapsulation of implanted pacemakers in a chronic minipig animal model.¹⁰⁹ Explants of both native and BC membrane protected pacemakers performed at 3 and 12 months after implantation showed that BC pacemakers had no signs of degradation and had ~66% less fibrotic tissue generated around the device when compared to native device (Fig. 3D). Importantly, the generator and proximal parts of the leads were free from fibrotic tissue, highlighting the value of this anti-fibrotic membrane as complications with leads and the generator are commonly documented.^130,131 While not performed in this study, it would be interesting to parse out the contribution of BC as a material from the micropatterned features in reducing FBR to pacemakers. Results from these preclinical studies showcase the efficacy by which biophysical cues can exert anti-fibrotic effects in complex in vivo models of CVDs and how the attenuation of fibrotic mechanisms correlate to improved functional cardiovascular performance.

2.4. Endothelial Cells

The endothelium serves multiple functions in maintaining vascular homeostasis, playing crucial roles in processes including coagulation, thrombosis, platelet/leukocyte interactions, angiogenesis, and vasoconstriction/dilation.¹³² The single layer of ECs that forms the lining of blood vessels is continuously exposed to diverse mechanical cues, including blood flow shear stress and ECM topographies, which help regulate behaviors such as proliferation, gene expression, and protein synthesis.¹³³ Dysfunction of vascular endothelium is highly associated with many CVDs and can be caused by various stimuli, like the formation of free radicals upon tissue injury. These pathological cues ultimately disrupt homeostatic mechanical regulation of ECs and activate pro-inflammatory and pro-thrombotic states that can contribute to cardiovascular conditions like peripheral artery disease (PAD) or restenosis after graft or stent implantation.¹³² Further, vascular grafts and stent failure modes often include inadequate endothelialization of the implanted material, greatly increasing thrombosis and neointimal hyperplasia risk.¹³⁴ Of particular importance is understanding how to harness EC mechanotransduction in therapeutic strategies to promote select behaviors that will enhance clinical outcomes. Recently published studies have showcased the use of topographical cues to obtain EC phenotypes that have enhanced proliferative behavior, reduced inflammatory state, and increased angiogenic potential. This section will briefly discuss how these features affect adhesion, proliferation, morphology, gene and protein expression, migration, and vascularization potential of ECs. Additionally, in vivo studies that have investigated the ability of topographic cues to improve disease outcomes by augmenting reendothelialization and vascularization potential will be discussed.

2.4.1. In vitro endothelial cell responses to micro- and nanotopography

The mechanosensitive nature of ECs has been detailed previously,¹³³ and studies culturing cells on patterned micro- and nanotopographies have demonstrated that physical cues can dictate various EC behaviors (Table 4). In native endothelium, ECs align themselves along the axis of blood vessels to minimize shear stress caused by blood flow.¹³⁵ It has been elucidated that topographies can modulate preferential cell alignment, proliferation, and migration, through the regulation of endothelial adherens junctions and yes-associated protein 1 (YAP) nuclear localization.^136,137 Further, orientation, adhesion, and proliferation have been shown to be enhanced in geometries such as grooves,^138–141 wrinkles,¹⁴² and fibers (Fig. 4A).^137,143,144 Studies utilizing groove patterns have found that widths of 1 μm were optimal to promote orientation, cell spreading, and proliferation in ECs.^138,141 Synergies have also been found to exist between different length-scale topographies and their abilities to influence EC orientation, elongation, and spreading, which opens up the potential for multi-dimensional topographical modulation of these cells.¹⁴⁴ A critical observation was made Hagen et al. investigated the differential responses of carotid artery endothelial cells (CtAECs) and blood-derived endothelial cell colony forming cells (ECFCs) on micropatterned grooves and found that while alignment was robustly increased in both cell types, only CtAECs showed enhanced elongation on micropatterns.¹³⁹ Differential responses of various cell types and sources can have important implications for clinical strategies depending on the treatment site. Moreover, on low diameter fibers in 2D culture ECFC and human umbilical vein endothelial cell (HUVEC) morphology were found to be similar, but differences in cytoskeletal organization and deposited collagen patterns were observed on larger diameter fibers.¹⁴⁵

Table 4.

In vitro endothelial cell response to micro- and nanotopography

Feature Geometry	Feature Size [a]	Substrate Material [b]	Cell Type [c]	Morphology & Cell Behavior [d]	Expression & Function	Ref.
concave lens	diameter: 1.8 μm pitch: 2 μm height: 0.7 μm	PDMS	HCAECs	no change in cell area or proliferation on concave lens vs. unpatterned	increased angiogenic capacity (total tube length and total branching points), no differences in oxLDL metabolism on concave lens vs. unpatterned	(Cutiongco et al., 2018)
convex lens	diameter: 1.8 μm pitch: 2 μm depth: 0.7 μm	PDMS	HCAECs	no change in cell area or proliferation on convex lens vs. unpatterned	increased angiogenic capacity (total tube length and total branching points), no differences in oxLDL metabolism on convex lens vs. unpatterned	(Cutiongco et al., 2018)
fibers	diameter: 0.5 – 3 μm	PTFE	HAoECs	increased cell area, elongation, proliferation on fibers vs. flat	decreased PECAM1 and VCAM1 gene expression and no change in ICAM1 gene expression on fibers vs. flat	(Lamichhane et al., 2016)
	diameter: 0.8, 1.2, 2 μm	ELP	HUVECs	increased junction disruption, increased proliferation, and increased migration speed on larger fibers vs. smaller fibers and flat	increased YAP nuclear localization and pERK1/2 on larger fibers vs. smaller fibers	(Mascharak et al., 2017)
grooves	width/ridge: 0.5 – 50 μm depth: 3.5 μm	TiO/Si	HUVECs	highest adhesion and proliferation on 1 and 5 μm width/spacing, increased elongation and alignment on grooves vs. flat, feature- dependent reduction in FAs/cell on grooves vs. flat	trends towards increased PECAM1 and VWF gene expression on grooves vs. flat	(Ding et al., 2014)
	width: 3 – 14 μm depth: 1 μm	PU	baboon CtAECs	increased alignment and elongation on grooves vs. flat	no difference in KLF2, VCAM1, and SELE gene expression on 6 μm grooves vs. flat	(Hagen & Hinds, 2020)
	width: 3 – 14 μm depth: 1 μm	PU	baboon ECFS	increased alignment but no difference in elongation on grooves vs. flat	no difference in KLF2, VCAM1, and SELE gene expression on 6 μm grooves vs. flat	(Hagen & Hinds, 2020)
	width: 2 μm ridge: 2 μm depth: 2 μm	PDMS	EA.hy 926	increased alignment and migration (leader cell velocity), decreased TI on grooves vs. unpatterned	increased pYAP and pLATS1/2 cytoplasmic localization on grooves vs. unpatterned	(Gorji et al., 2019)
	width/ridge: 800 nm depth: 600 nm	PEG/GelMA	HUVECs	increased adhesion, alignment, cell area, elongation, migration speed, and persistence time on grooves vs. flat	N/A	(Kim et al., 2014)
	ridge: 70 nm depth: 70 nm	TiO2	HUVECs	increased elongation, cell area, and FA size on grooves vs. flat	N/A	(Muhammad et al., 2014)
grooves & grooves with nanogroove features	gap: 2 μm ridge: 2 μm depth: 2 μm & grooves with either PAR or PERP width/ridge/depth: 250 nm features	PDMS	HCAECs	decreased cell area and no change in proliferation on all grooves vs. unpatterned, increased monocyte adhesion on grooves vs. unpatterned, increased migration (velocity and distance) on PERP grooves vs. unpatterned	increased angiogenic capacity (total tube length and total branching points) on PAR and PERP grooves vs. unpatterned, no differences in oxLDL metabolism on all grooves vs. unpatterned, decreased expression of NOS3 and immunogenicity genes (VCAM1, ICAM1, SELE, VWF) on all grooves vs. unpatterned	(Cutiongco et al., 2018)
pillars	width/spacing: 0.5 – 50 μm height: 3.5 μm	TiO/Si	HUVECs	biphasic trends in adhesion (highest on 1 μm width/spacing), proliferation (highest on 20 and 50 μm width/spacing), and number of FAs/cell (highest on 50 μm width/spacing)	feature- dependent reduction in PECAM1 and VWF gene expression on pillars vs. flat	(Ding et al., 2014)
tubes	diameter: 110 nm	TiO2	HUVECs	increased adhesion, elongation, and migration on tubes vs. flat	increased NO secretion on tubes vs. flat	(Zhong et al., 2014)
	diameter: 70, 110 nm	nitinol	HAoECs	increased cell spreading, decreased proliferation, and increased migration on tubes vs. flat	increased collagen and elastin secretion on tubes vs. flat	(P. P. Lee & Desai, 2016)
	diameter: 30, 50, 90 nm	TiO2	HCAECs	increased adhesion and decreased cell area on 90 nm tubes vs. flat	with TNF-α- stimulation— decreased VCAM1 gene expression on 90 nm tubes vs. flat; no change in FAK or pFAK on 90nm tubes vs. flat	(Cao & Desai, 2020)
wells	width: 420 nm ridge: 70 nm height: 70 nm	TiO2	HUVECs	increased cell area and FA size on wells vs. flat	N/A	(Muhammad et al., 2014)
wrinkles	amplitude: 0.05 – 4.3 μm wavelength: 0.5 – 27 μm	PDMS	HUVECs	increased alignment, elongation, proliferation, and monocyte adhesion on 10 μm wavelength wrinkles vs. flat	increased TNF, IL1B, pAKT, and ICAM1 gene expression on 10 μm wavelength wrinkles vs. flat	(Liu et al., 2020)

^[a]PAR = parallel, PERP = perpendicular

^[b]ELP = elastin-like protein, GelMA = gelatin methacrylate, PDMS = polydimethylsiloxane, PEG = poly(ethylene glycol), PTFE = polytetrafluoroethylene, PU = polyurethane, Si = silicon

^[c]CtAECs = carotid artery endothelial cells, ECFCs = endothelial colony forming cells, HAoECs = human aortic endothelial cells, HCAECs = human coronary artery endothelial cells, HUVECs = human umbilical vein endothelial cells

^[d]FA = focal adhesion, TI = tortuosity index (zipper/linear endothelial adherens junctions ratio)

Figure 4. Biophysical cues modulate endothelial cell morphology, immunogenicity, and vascularization potential.

(A) Carotid endothelial cells and endothelial colony forming cells exhibit high alignment to micropatterns with various pitches. Groups sharing letters are not statistically different according to post-hoc Tukey test with p>0.05 (Hagen & Hinds 2020). (B) Healthy ECs cultured on topographies have reduced immunogenicity. Black data bars denote healthy EC responses, gray data bars denote diabetic EC responses (Cutiongco et al. 2018). (C) EC-seeded aligned scaffolds show higher limb perfusion compared to other groups in hindlimb ischemia models (Nakayama et al. 2015). (D) SEM micrographs of unpatterned and patterned PVA graft topographies (left). Comparisons of luminal area show that 2 μm groove topographies had higher patency compared to other topographies (right, top). Explanted 2 μm groove grafts exhibit EC attachment (red) to the luminal surface (right, bottom) (Cutiongco et al. 2016). (E) Successful large lumen formation by CD31⁺ cells was observed 3 days after implantation of aligned cord patches. Loss of pattern cord-associated clustering of vessels was observed between day 3 and 7 (Brady et al. 2020).

A major barrier to endothelization of implants is the ability to maintain a confluent EC monolayer in the presence of physiological or greater wall shear stress (WSS) levels, such as those generated within vascular assist devices. Robotti et al. demonstrated that only cells cultured on 1 μm grooves had preserved monolayer integrity under high WSS; at physiological WSS, ECs were aligned with topography which was further reinforced by co-aligned flow direction.¹⁴⁶ However, at supraphysiological WSS, most ECs aligned themselves perpendicularly to the flow and topography direction.¹⁴⁶ When studying the effect of pre-polarizing ECs by culturing them on grooves oriented perpendicular to supraphysiological flow, they found endothelial layers with high connectivity index, alignment, and cell area had formed. This demonstrated the ability for topographical surface modification to enhance endothelialization under supraphysiological flow conditions. ECs cultured on aligned scaffolds also had greater resistance to detachment compared to cells cultured on randomly-oriented and partially-aligned scaffolds.¹⁴⁷

Topography-mediated alterations in EC gene expression and secretion profiles have been documented to yield phenotypes that reduce immunogenicity of ECs. These anti-inflammatory states are evidenced by reduced expression of vascular cell adhesion molecule 1 (VCAM1)^143,148,149, intercellular adhesion molecule 1 (ICAM1),¹⁵⁰ and platelet and endothelial cell adhesion molecule 1 (PECAM1) (Fig. 4B),^{138,143,149,150} increased secretion of nitric oxide,¹⁵¹ collagen,¹⁵² and elastin,¹⁵² and reduced secretion of inflammatory cytokines IL-1β, interleukin 3 (IL-3), and MCP-1.¹⁵³ The ability to reduce pro-inflammatory activities in ECs is critical as they are implicated in the progression of atherosclerosis and restenosis.¹⁵⁴

Further, enhanced migration capacity is associated with pro-angiogenic potential. Previous studies have demonstrated increased migratory potential on patterned features with regards to speed, persistence time, and distance when compared to flat substrates.^{137,140,150,151,155} Gorji et al. observed significant increases in the migration of leader cells and, to a lesser extent, also follower cells after exposure to microgroove features.¹³⁶ Lastly, the ability for ECs to coalesce and organize into vascular tubes and networks is imperative. Some have found that anisotropic topography can encourage clustering of endothelial cells and development of tube-like structures when compared to isotropic scaffolds.^156,157 One study found a synergistic effect between 1.8 μm convex lens geometries and VEGF-enriched medium for enhancing differentiation of hMSCs into ECs and promoting higher capillary density on explanted Matrigel plugs when compared to cells on unpatterned substrates.¹⁵⁸ Another study screened the impact of directional polydimethylsiloxane (PDMS) topographical gradients on human pulmonary microvascular ECs and found that larger microscale topographies inhibited vascular network formation compared to submicron topographies, which allowed for the limited formation of unstable networks. They further noted that the adhesion characteristics of a substrate coating layer largely influenced the dynamics of vascular network formation on topographic surfaces; by utilizing an appropriate instructive layer, otherwise unstable vascular networks that formed on submicron topographies could be stabilized.¹⁵⁹

2.4.2. Implications of micro- and nanotopography in CVD therapeutic strategies via endothelial cell modulation

Critical applications of EC-modulating surface topographies in CVD therapies include stent, vascular graft, and cardiac patch implantations, where proliferative, anti-inflammatory, and pro-angiogenic behaviors are paramount to tissue repair. Major causes of stent and small graft failure, such as thrombosis and intimal hyperplasia, result from inadequate reendothelialization and increased inflammatory response, leading to a lack of vascular patency after implantation.¹³⁴ These procedures are typically performed in areas of regional ischemia, highlighting the need for appropriate pro-angiogenic stimuli to aid in the creation of new vasculature. Therefore, the use of biophysical cues to repair a denuded endothelial lining after surgical intervention and promote vascular growth represents an important strategy for expediting cardiovascular repair.

With respect to preclinical investigations of stent surface modifications, micro-nanostructured biomimetic surface patterns yielded more rapid reendothelialization of drug-eluting stents at 30- and 90-days post-implantation compared to unpatterned stents.¹⁶⁰ Further, this stent modification improved drug release kinetics, opening up the potential for dual-therapeutic approaches targeting both ECs and immune response.¹⁶⁰ A study using a rabbit model of restenosis corroborated in vitro findings¹⁴⁸ and showed titanium surfaces with ~90 nm tube features favored EC growth over that of VSMCs.¹⁶¹ An investigated of nanoridge stent patterning similarly yielded rapid reendothelization and reduced neointimal thickening compared to bare stents, affirming the reliability of topographic patterns to improve EC coverage.¹⁶²

Numerous investigations have focused on understanding the impact of topographical guidance on outcomes after vascular graft implantation.^163–166 In experimental models of PAD where nonobese diabetic/severe combined immunodeficiency (NOD SCID) mice were subjected to hindlimb ischemia, iPSC-EC-seeded aligned collagen scaffolds demonstrated enhanced perfusion recovery after 2 weeks compared to cell injection, acellular aligned scaffolds, and cell-seeded control scaffolds (Fig. 4C) and improved microvascular density after 28 days when compared to PBS and cell injections.¹⁶³ Subsequent in vivo studies delivering stromal-vascular-fraction-cell-seeded aligned collagen scaffolds also resulted in improved perfusion recovery in PAD rodent models.¹⁶⁴ Interestingly, the superiority of aligned scaffolds over isotropic scaffolds for improved vascularization may be context dependent. Wanjare et al. demonstrated in an immunocompromised rodent model of mild MI that iPSC-CM and iPSC-CM/iPSC-EC-seeded isotropic scaffolds displayed either equivalent or increased arteriole density compared to cell seeded aligned scaffolds.¹⁶⁵ Subcutaneous implantation of cell-seeded scaffolds further indicated higher survival of ECs and enhanced pro-angiogenic effects in random scaffolds, while aligned scaffolds encouraged anisotropic vessels formation.¹⁶⁵ These results stress the importance of understanding the differential responses of CVD-relevant cell types to topographical cues and the differences between in vivo models: while random scaffolds have been previously shown to have deleterious effects on transplanted cardiomyocyte function,⁴⁷ improved pro-angiogenic outcomes in ECs were observed here.¹⁶⁵

Another approach attempted to utilize both physiologically relevant tubular molding and luminal surface micropatterning to impart optimal mechanical guidance of vascular cells after graft implantation. After infrarenal aortic implantation, 2 μm grating micropatterns had improved vascular patency after 20 days compared to unpatterned, pillar, convex, and concave micropatterns, and also demonstrated successful EC attachment to the luminal surface (Fig. 4D).¹⁶⁶ Additional studies have explored the use of patterned features to promote the formation of functional vasculature.^167–170 Human embryonic stem cell-derived cell (hESC)-EC-seeded microfluidic channel networks within hESC-EC-seeded collagen gel matrices showed high integration with coronary vasculature, enhanced vascular density, and comparable perfusion velocities to that of healthy myocardium when implanted into infarcted rat hearts.¹⁶⁷ Another strategy utilized aligned cords of EC and stromal cell populations formed via culture on microgroove substrates and subsequently encapsulated in a fibrin hydrogel to form a cardiac patch.¹⁶⁸ Following supra-epicardial transplantation onto an uninjured rat heart, the patches demonstrated perfusion 3 days post-transplantation and formed capillary- to arteriole-sized microvessels at 7 days. A loss of patterning in newly formed vessels was observed between day 3 and 7 (Fig. 4E), likely due to remodeling of the patch by local inflammatory response, but successful engraftment of cardiomyocytes, stromal cells, and ECs was still observed.¹⁶⁸ The advantages of delivering endothelial cells via material scaffolds were further highlighted when induced vascular progenitor cells (iVPCs) seeded on micro-bundle scaffolds showed higher survival and engraftment on myocardium compared to free iVPCs. Moreover, improved cardiac outcomes (LVEF, end systolic volume, fractional shortening), reduced infarct size, and increased vascularization in in rat models of MI, were also observed in iVPC-seeded scaffolds when compared to untreated control.¹⁶⁹ Another study demonstrated that circumferentially oriented fibrous grafts achieved complete endothelization, with cell alignment along the direction of blood flow achieved at 4 weeks. Additionally, formation of new capillaries was also observed during all time points over the 12 weeks.¹⁷⁰

It has also been demonstrated that topography-mediated effects on tight-junction expression and cell morphology is maintained during successive passages, despite the removal of topographic cues after seeding.¹⁷¹ An interesting avenue to explore is leveraging the memory of topographic modulation in ECs to improve regeneration in-situ without continuously providing such cues. Taken together, these findings indicate unique roles of biophysical cues on substrata to modulate EC behaviors including survival, proliferation, and vascularization and directly links these effects to improved cardiovascular outcomes.

2.5. Vascular Smooth Muscle Cells

In native arteries and veins, VSMCs can be found in the medial layer of vessels where they play central roles in vascular contraction, compliance, and elastic recoil in response to varying hemodynamic states.¹⁷² In healthy vessels VSMCs express high levels of contractile proteins indicative of their differentiated “contractile” phenotype and they have been found to possess great plasticity.¹⁷² Phenotype switching of VSMCs in response to events such as tissue injury has been documented. These cells take on “synthetic” phenotypes which is characterized by increased proliferation and increased expression of ECM components and remodeling enzymes¹⁷²; these de-differentiated phenotypes can prove to be problematic as they can contribute to CVD progression and lead to complications during clinical CVD therapies. During atherosclerosis, increased transition of VSMCs to a synthetic phenotype results in increased ECM deposition and contributes to intimal thickening and retention of atherogenic lipids.¹⁷² Further, during treatments for atherosclerosis such as stent deployment, inflammatory cascades triggered by denudation of the EC layer can trigger hyperproliferation and excessive matrix deposition in VSMCs and contribute to restenosis of the vessel wall.¹⁷³ As such, there is great value in being able to selectively modulate the phenotype of VSMCs to match an intended therapeutic strategy: therapies that aim to address atherosclerotic diseases and wound repair after stent implantation should discourage VSMC synthetic phenotype and strategies aiming to recapitulate native vasculature such as in vascular graft implantation should enhance VSMC contractile phenotypes. Recent efforts have successfully modified substrates with biophysical cues to direct VSMCs to a contractile phenotype to improve clinical outcomes and strategies, which are typically hindered by hyperproliferation and ECM deposition during treatment of CVDs. This section will briefly discuss findings related to how these features affect VSMC adhesion, proliferation, gene and protein expression, and secretion profiles. Additionally, preclinical studies that have assessed the ability of topographic cues to improve disease outcomes by preventing an immunogenic VSMC response will be explored.

2.5.1. In vitro vascular smooth muscle cell responses to micro- and nanotopography

In the vasculature, VSMCs are aligned circumferentially around the vessel in a helical formation to achieve optimal contraction and dilation capabilities.¹⁷⁴ These differentiated VSMCs exhibit high expression of contractile proteins, including α-SMA, smooth muscle myosin heavy chain (SM-MHC), and calponin (CNN1).¹⁷⁵ Studies culturing cells on patterned micro- and nanotopographies have demonstrated that biophysical modulation of VSMC phenotype and behavior is possible (Table 5). VSMCs interact with geometries through cytoplasmic projections and podosome formation,^148,176 and studies have pointed towards the RhoA/ROCK pathway and caveolin-1 for playing a role in the mechanotransduction of these biophysical cues.¹⁷⁷ It was found that there are topographic regimes that facilitate VSMC alignment, focal adhesion formation, and differentiation.¹⁷⁸ Further, findings indicate that overall cell alignment in conjunction with focal adhesion expression may be key regulators of VSMC differentiation.¹⁷⁸ In addition to promoting increased alignment, some studies have demonstrated reduced proliferative capacity in cells cultured on topographic cues compared to those on unpatterned substrates.^{138,148,152,179,180} This is critical as hyperproliferation of VSMCs in atherosclerosis and after device implantation prevents repopulation of ECs and may increase neointimal thickening, thereby increasing likelihood of restenosis. Cao et al. reported that TiO₂ nanotubes dampen the response of primary VSMCs to inflammatory cytokine stimulation, resulting in decreased TNF-α induced proliferation without impeding endothelialization (Fig. 5A).¹⁴⁸ It was also found that VSMCs cultured on silk-protein microgrooves had larger populations in the G0 cell cycle phase as compared to those on flat silk-protein surfaces.¹⁷⁹

Table 5.

In vitro vascular smooth muscle cell response to micro- and nanotopography

Feature Geometry	Feature Size [a]	Substrate Material [b]	Cell Type [c]	Morphology & Cell Behavior [d]	Expression & Function	Ref.
aligned fibers with pores	diameter: 1.6 μm with pore AR: 2.7, 3, 3.9	PLLA	HUASMCs	increased adhesion, proliferation, and alignment on fibers with 3.9 AR pores vs. fibers with other pore ARs	decreased OPN gene expression, increased ACTA2 gene expression, and increased collagen secretion on fibers with 3.9 AR pores vs. fibers with other pore ARs	(Zhou et al., 2015)
fibers	diameter: 0.5 – 3 μm	PTFE	HASMCs	increased cell area, elongation, and proliferation on fibers vs. flat	increased α-SMA expression on fibers vs. flat	(Lamichhane et al., 2016)
grooves	width/ridge: 0.5 – 50 μm depth: 3.5 μm	TiO/Si	HUASMCs	increased alignment and decreased cell area coverage on grooves vs. flat	N/A	(Ding et al., 2014)
	groove/ridge: 10 μm depth: 5 μm	PDMS	hMSCs	increased alignment on grooves vs. flat	increased ACTA2 and CNN1 gene expression on grooves vs. flat	(Parandakh et al., 2019)
	width: 2.8 – 3 μm depth: ~150 – 200 nm roughness: ~60 nm	silk protein (BM, AA)	PASMCs	increased alignment and decreased proliferation on all grooves vs. respective protein flat control	increased ACTA2 and MYH11 gene expression on AA grooves vs. AA flat, increased collagen, MMP2, and MMP9 production on all grooves vs. respective protein flat control	(Gupta et al., 2019)
	ridge: 3 μm depth: 5 μm	PDMS	RASMCs	increased alignment, elongation, and migration while decreased cell area and proliferation on grooves vs. flat	increased myocardin, α-SMA, and Ncadherin expression and decreased PDGFR, PCNA, and OPN expression on grooves vs. flat	(Chang et al., 2014)
	groove/ridge: 800 nm depth: 600 nm stiffness: 11MPa, 1.1GPa	PUA	HUASMCs	increased alignment and elongation on grooves vs. flat for both stiffness	increased CNN1 and TPM1 gene expression on grooves vs. flat for both stiffness, increased ACTA2 gene expression on grooves vs. flat for 1.1GPa stiffness, increased DES gene expression on grooves vs. flat for 11MPa stiffness, decreased ICAM1, F3, and IL6 gene expression on grooves vs. flat for both stiffness, decreased CCL2 gene expression on grooves vs. flat for 11MPa stiffness	(Chaterji et al., 2014)
pillars	width/spacing: 0.5 – 50 μm height: 3.5 μm	TiO/Si	HUASMCs	decreased cell spreading and cell area coverage on pillars vs. flat	N/A	(Ding et al., 2014)
stripe patterns	width: 5 – 1000 μm height: 60 nm	PS/PVA	hMSCs	no cell adhesion on 5 and 10 μm stripes, increased alignment on 20 – 200 μm stripes vs. flat	increased ACTA2 and CNN1 gene expression on 20 – 200 μm stripes vs. flat	(Nakamoto et al., 2014)
tubes	diameter: 110 nm	TiO2	HUASMCs	decreased adhesion, cell spreading, and proliferation on tubes vs. flat	N/A	Zhong 2014
	diameter: 70, 110 nm	nitinol	HASMCs	increased alignment and decreased proliferation on tubes vs. flat	no difference in collagen secretion on tubes vs. flat	(P. P. Lee & Desai, 2016)
	diameter: 30, 50, 90 nm	TiO2	HCASMCs	with and without TNF-α- stimulation— decreased proliferation on tubes vs. flat (greatest reduction on 90nm tubes)	with and without TNF-α-stimulation— decreased MCP-1 secretion on 90 nm tubes vs. other tube diameters and flat; decreased FAK expression and increased pFAK/FAK ratio on 90 nm tubes vs. flat	(Cao & Desai, 2020)
wrinkles	wavelength: 0.46 – 11 μm amplitude: 0.05 – 3.4 μm	PDMS	ASCs	increased cell and FA alignment on micron-scale wavelength wrinkles vs. flat, biphasic trends in FA area/cell (lowest on 0.7 – 1.3 μm wavelength)	with TGF-β1stimulation— decreased SM22-α expression on all topographies except 11 μm wavelength wrinkles vs. flat	(Liguori et al., 2019)

^[a]AR = aspect ratio

^[b]AA = non-mulberry Antheraea assama, BM = mulberry Bombyx mori, PDMS = polydimethylsiloxane, PLLA = poly(L-Lactic acid), PS = polystyrene, PTFE = polytetrafluoroethylene, PUA = polyurethane acrylate, PVA = polyvinyl alcohol, Si = silicon

^[c]ASCs = adipose-derived stromal cells, HASMCs = human aortic smooth muscle cells, HCASMCs = human coronary artery smooth muscle cells, hMSCs = human mesenchymal stem cells, HUASMCs = human umbilical arterial smooth muscle cells, PASMCs = porcine aortic smooth muscle cells, RASMCs = rat aortic smooth muscle cells

^[d]FA = focal adhesion

Figure 5. Micro- and nanoscale features induce contractile phenotypes in VSMCs and improve vessel patency in vivo.

(A) Titanium nanotube surfaces decrease VSMC proliferation and MCP-1 secretion (scale bars = 1 μm) (Cao et al. 2020). (B) VSMCs cultured on microgroove substrates had higher expression of contractile phenotype markers and decreased expression of synthetic phenotype marker OPN (Chang et al. 2014). (C) SEM of a bare metal stent (Ai-Aiii) and a nanotextured stent (Bi-Biii) (top, left) (scale bars = 1, 100, 500 μm from i to iii respectively). Nanotextured stents had increased patency and reduced neointimal hyperplasia compared to bare metal stents (bottom). Staining of ECs via wheat germ agglutinin (green) indicated complete endothelialization in both stent conditions (top, right) (scale bars = 1 μm) (Cherian et al. 2020). (D) Circumferentially aligned microfibers that informed resulting increased alignment over 12 weeks (top, left). SEM explant of graft showed lumen coverage by ECs (vWF = green) (right, top) (scale bars for “a, d” 500 μm, “b” 100 μm, “c, e, f” 50 μm). H&E staining (scale bars = 500 μm) as well as antibody staining (scale bars = 50 μm) demonstrated rapid cell infiltration into grafts (right, bottom) (α-SMA, SM-MHC = green, nuclei = blue) (Zhu et al. 2015).

Further, topography-mediated influences in VSMC gene expression and secretion profiles yield phenotypes indicative of contractile phenotypes, characterized by increased expression of markers α-SMA,^{143,179–183} calponin,^177,182,183 and desmin¹⁷⁷ and reduced expression of the synthetic marker osteopontin (OPN) (Fig. 5B).^180,181 Further, reduced secretion of inflammatory cytokines MCP-1, IL-6, and ICAM-1^148,177 and altered secretion of ECM components (collagen and elastin)^179,181,184 and enzymes (MMP-2 and MMP-9)¹⁷⁹ has been documented with certain geometries, including grooves, fibers, and tubes. Secretion of elastin and collagen are critical for proper mechanical function of vascular walls; elastin provides the vessel with elastic properties and coordinates mechanical response at low strains while collagen provides tensile stiffness to prevent rupture at high pressures.¹⁸⁵ Gene expression analysis of VSMCs cultured on micropatterned grooves and on flat substrates showed that cells on grooved surfaces had preferential expression of genes related to VSMC migration, differentiation, and contraction.¹⁸⁰ Further, microRNA-145, a well-known VSMC marker that is abundantly expressed in normal vascular wall and in freshly isolated VSMCs, was upregulated in cells on grooved surfaces (Fig. 5B).¹⁸⁰ Functionally, vessel walls are continuously exposed to contraction and relaxation. VSMCs cultured on micropatterned grooves were found to have greater maximum contraction.¹⁷⁷ Researchers have also developed more representative tubular constructs for graft engineering.^186–188 Rayatpisheh et al. developed circumferentially aligned tubular constructs utilizing aligned PCL nanofiber scaffolds and micropatterned PDMS substrates to yield circumferential alignment of VSMCs with expression of contractile genes, including ACTA2, CNN1, transeglin (SM22), and smoothelin (SMTN).¹⁸⁷ Further, it was reported that VSMCs cultured on topographic substrates showed enhanced migratory potential compared to those on flat surfaces, likely due to observed increased phosphorylated FAK levels on microgroove surfaces as ERK and FAK activation are involved VSMC migration.¹⁸⁰

2.5.2. Implications of micro- and nanotopography in CVD therapeutic strategies via vascular smooth muscle cell modulation

Common modes of failure for vascular grafts include intimal hyperplasia and atherosclerosis, which often develop due to excessive VSMC proliferation, migration, and ECM deposition within the intima. Reasons as to why these pathological processes may result after graft implantation include tissue trauma post-surgery, denudation of the endothelial layer, and compliance mismatch of graft and vessel¹³⁴. However, given the devastating consequences and health complications associated with stenotic blood vessels, it is imperative that therapeutic methodologies to minimize VSMC synthetic phenotype be explored further.

An important application of VSMC phenotype modulation is after stent implantation to minimize risk of restenosis. In-stent restenosis may occur as a result of denudation of the EC layer, thereby triggering inflammation and neointimal hyperproliferation.¹⁸⁹ While advancements in drug-eluting stents have aimed to counteract excessive proliferation of VSMCs, they often also hinder reendothelization, thereby increasing the risk of thrombosis.¹⁹⁰ Therefore, mechanisms that enable cell-specific modulation are imperative for enhanced therapeutic efficacy. Various preclinical models have reported that nanotextured features on stent surfaces strongly decrease neointimal thickness, have larger lumen areas, and reduced stenosis compared to unpatterned surfaces.^161,162,191 Nuhn et al. demonstrated that 90 nm nanotube stent surfaces reduced stenosis after iliofemoral artery implantation compared to unmodified stent surfaces.¹⁶¹ Other studies investigating nanotexturing of stent surfaces yielded similarly exciting results including significant reductions in neointimal thickening and in-stent restenosis as well as complete EC coverage compared to bare metal stents (Fig. 5C)^162,191. These distinct behavioral responses by ECs and VSMCs to biophysical cues highlight the importance of understanding cell-specific interactions to topographies in order to leverage them in the design of CVD therapeutic strategies.

With respect to utilizing topographical cues for rational design of vascular grafts, numerous methodologies have been utilized in order to minimize neointimal hyperplasia, improve vascular patency, and maintain appropriate arrangement of VSCMs. One study has reported the investigation of different sized pore features and resulting development of calcification. Infrarenal aortic implantation of larger-pore PLA-based grafts (~30 μm) in mice resulted in the formation of well-organized neointima and absence of neointimal calcification compared to smaller-pore grafts (~0.7 μm) after 12 months¹⁰². Further, histology demonstrated that larger-pore grafts had greater collagen and elastin content in the neointima and higher levels of α-SMA⁺ and SM-MHC⁺ cells than smaller-pore grafts.¹⁰² In a subsequent study, the authors investigated the relative expression of ECM components in larger-pore grafts compared to that of native aorta.¹⁹² They reported that collagen type I and III gene expression levels in larger-pore graft neovessels increased over time and were much greater than native aorta after 12 months, while elastin gene expression remained lower in grafts throughout all time points.¹⁹² Interestingly, despite observing elastin deposition within 4 months of implantation, incidents of aneurysmal changes across both studies suggests that the amount of elastin secreted may be insufficient. One plausible explanation for this could be that the presence of residual, non-degraded PLA fibers interfered with pulsatile stretching of VSMCs, thereby hindering mechanically stimulated elastin deposition. Another study also utilized a composite small arterial graft comprising of a ~20 μm microporous, fast-degrading poly(glycerol sebacate) tube with an outer sheath of PCL nanofibers for mechanical strength. These grafts demonstrated integration with native rat aortas, good vessel patency, and no significant stenosis or calcification one year after transplantation. Further, these grafts remodeled into neoarteries and exhibited representative elastin levels and compliance compared to that of the native aorta.¹⁹³

Several studies have designed vascular grafts with physical cues to instruct VSMC development and circumferential alignment.¹⁷⁰ Zhu et al. used a circumferentially aligned microfibrous layer for instructing VSMC behavior. They found that found that α-SMA⁺ and SM-MHC⁺ cells readily infiltrated the graft and that the percentage of aligned VSMCs increased over time, approaching values similar to that of the native artery by 3 months (Fig. 5D).¹⁷⁰ Further, histological staining confirmed that deposited fibrillar collagen also had a distinct circumferential orientation. Explanted grafts exhibited responsiveness to both endothelium-independent and endothelium-dependent vasoactivators indicating functional signal transduction between ECs and VSMCs.¹⁷⁰ A subsequent study aimed to extend the observation period and analyze long-term graft performance over 18 months.¹⁰⁶ They found no obvious changes in luminal diameter and patency at 18 months compared to 3 months post-implantation. Further, a well-organized intima mostly comprised of α-SMA⁺ and SM-MHC⁺ was observed along with evidence of collagen, elastin, and laminin deposition, which were both similar to that of native abdominal aorta. Interestingly, at this longer time point only one out of three explanted grafts demonstrated responsiveness to vasoactivators.¹⁰⁶ The stiffness and slow degradation rates of PCL fibers may have hindered vascular regeneration and led to increased inflammation at the implantation site. In addition, hyaluronan-functionalized circumferentially-aligned nanofibrous grafts were shown to promote increased circumferential orientation of contractile α-SMA⁺ infiltrating cells and accommodate higher blood flow speeds than with non-coated aligned grafts after 6 weeks in a rabbit carotid artery replacement model.¹⁸⁴ Results from these preclinical studies showcase the ability for micro- and nanoscale physical cues to promote anti-proliferative and contractile VSMC phenotypes in vivo to yield improved tissue regeneration after graft implantation.

3. Conclusion & Perspective

It is evident that micro- and nanoscale biophysical cues can effectively modulate the phenotype and function of cell types involved in CVDs. These features draw inspiration from native tissue architectures that continuously exert diverse mechanical cues to resident cells to maintain homeostatic behavior. Considering abnormal physical cues in the microenvironment can exacerbate CVD progression, strategies that introduce features that elicit native cell behaviors to impede CVD processes represent an exciting therapeutic direction to pursue.

A critical objective of materials-based therapeutic strategies is the identification of “optimal” feature geometries and dimensions for eliciting specific phenotypes. However, there are varied findings in the literature. Factors that are likely contributing to these differences include: (1) Reports have shown distinct cellular responses to physical cues depending on the nature of cells utilized, highlighting the potential of varying cell- sensitivities to mechanical cues.^139,194 Characteristics such as cell species, primary vs. cell lines, tissue source, cell heterogeneity, and cell maturity may account for the variability observed. Heterogeneous hiPSC-CM populations that can comprise of mixed nodal-, atrial-, or ventricular-like cell populations¹⁹⁵, each with distinct expression and electrophysiological profiles, may lead to inaccurate evaluations of topography-induced cardiomyocyte maturation and function. Further, immature cardiomyocytes have unique morphological and expression profiles compared to differentiated myocytes, such as Cx43 polarization^38,39 which directly impact functional outputs, and so can also contribute to differences observed across studies. (2) Both substrate surface chemistry and mechanical properties contribute to how physical cues are perceived and interpreted by cells. Recent efforts have shown how material characteristics such as material functionalization¹⁸⁴, mechanical properties¹⁹⁶, degradation behavior¹⁹⁷, pore size¹⁰², and thickness¹⁹⁸ can affect performance of therapeutic strategies. Hyaluronan-functionalized grafts promoted higher circumferential orientation of contractile α-SMA⁺ cells than that of non-coated grafts in vivo.¹⁸⁴ Further, hydrophilicity has been found to have synergistic effects with topography on cell modulation.^85,151,179 Cell phenotypes are also heavily dictated by the elastic modulus of their microenvironment. For example, VSMCs take on more contractile phenotypes on softer topographical substrates more closely aligned with healthy medial tissue elastic modulus.^177,199 Smaller diameter fibrous scaffolds correspondingly have lower stiffness which afford cells higher contractile ability against the fibers, highlighting a need to understand how stiffness and topography individually exert their roles on cell function.^46,50 (3) Ensuring that micro- and nanofabrication techniques are robust and precise enough to yield feature uniformity are vital to maximizing reproducibility in the field of therapeutic materials; failure to ensure this may lead to inability to replicate cell phenotypes and behaviors previously observed in other studies.²⁰⁰ Since it is apparent that cells exhibit unique behaviors when exposed to specific size regimes^{84,126,138,178}, rigorous characterization, assessment, and comparison of geometry dimensions to other studies will be of great importance. (4) The environmental context of cell culture and exposure to the biophysical cues may be very different across studies. Parameters such as 2D versus 3D culture environments and whether cells were non-stimulated, pre-stimulated, or co-stimulated with soluble factors during exposure to biophysical cues all may have profound impacts on topography-induced modulation of cells. When cultured in a 2D monolayer, HUVEC and ECFC morphology and phenotypes were very similar; however, while in 3D cultures, there were stark differences in the behaviors of these cells.¹⁴⁵ Further, Dollinger et al. noticed stark differences in cytokine/chemokine secretion profiles between pre-differentiated M1/M2 macrophages and co-stimulated macrophages in M1/M2 media cultured on microgrooves.¹⁰¹ (5) The length and timescale after which the effects of physical cues are examined are crucial to note. Cell culture periods with topographies and subsequent timing of morphological, behavioral, expression, and functional analysis should accurately reflect the timescales during which these changes are expected to be taking place. Cell gene, protein, and secretion profiles are likely to vary over time and therefore attention to how topography-induced cell modulation may be temporally dependent demands attention.

While there still remains much to be explored regarding the intricacies of how cardiovascular cells respond to biophysical cues and how these responses may be leveraged for treating CVDs, this review provides a broad landscape of the efforts made in the field to date. Recent work has showcased a diverse array of geometries, materials, and size regimes for modulating cardiovascular cells both in vitro and in preclinical models of CVDs. In addition to being strong cell mediators on their own, topographies can work synergistically with additional cues such as electrical stimulation, mechanical stimulation, and stiffness to further regulate cell function. The versatility of these materials-based therapeutic strategy cannot be understated— micro- and nanofabrication techniques afford compatibility with a plethora of substrate materials, functionalization strategies, mechanical properties, feature geometries, and size regimes which may be tailored as needed. With these highly modular fabrication capabilities, a fascinating direction for this strategy could be creating biophysical features based on specific patient and disease contexts as a more “personalized medicine” approach. However, first, optimization of all of the above parameters in addition to delivery and administration will need to be determined. Despite the long road ahead to effective translation of these strategies to the clinic, recent work at the intersection of materials, regenerative medicine, and cardiovascular research has demonstrated that biophysical cues have tremendous implications for cardiovascular disease therapy.