Tissue-engineered microvessels: a review of current engineering strategies and applications

Abstract

Microvessels, including arterioles, capillaries, and venules, play an important role in regulating blood flow, enabling nutrient and waste exchange, and facilitating immune surveillance. Due to their important roles in maintaining normal function in human tissues, a substantial effort has been devoted to developing tissue-engineered models to study endothelium-related biology and pathology. Various engineering strategies have been developed to recapitulate the structural, cellular, and molecular hallmarks of native human microvessels in vitro. In this review, we first summarized recent progress in engineering approaches, key components, and culture platforms for tissue-engineered human microvessel models. Then, we review tissue-specific microvessel models of the human brain and other organs, followed by the major applications of tissue-engineered human microvessels in modeling development, diseases, drug screening and delivery, and vascularization in tissue engineering. Finally, we discuss the future research directions for the field.

Graphical Abstract

Advances in tissue engineering and stem cell technology have led to the rapid development of vascular models that seek to recapitulate organ level functions. Highly reductive 2D models enable high-throughput screening, whereas, more complex 3D tissue-engineered models are crucial for disease modeling. Here we summarize progress in field including engineering approaches, applications, and future research directions.

Introduction

The cardiovascular system includes the heart, blood, and the network of blood vessels composed of endothelial cells (ECs), pericytes (PCs), smooth muscle cells (SMCs), fibroblasts, and vascular basement membrane (BM), a specialized extracellular matrix (ECM). The vascular system plays an important role in the transport of nutrients and waste products, maintenance of homeostasis, and immune surveillance in humans as well as other vertebrate organisms. Blood from the heart is pumped through the aorta, and into major arteries, arterioles, capillaries, venules, veins, and lastly the vena cava. Arterioles (lumen diameter ~10 - ~300 μm)^[1], capillaries (~6 - ~14 μm)^{[1b, 2]}, and venules (~10 - ~200 μm)^[3] are often referred to as microvessels (microvasculature or microcirculation) (Fig. 1). The exact dimensions of microvessels varies significantly across different tissues and species: large arterioles branch into a network of small arterioles, which feed into capillaries and connect to venules which have a similar network structure to arterioles. Microvessels regulate blood flow resistance (arterioles), enable substance exchange (capillaries), and facilitate immune surveillance (venules)^[4]. Microvascular dysfunction is associated with many lethal human diseases such as stroke, heart attack, cancer metastasis, and neurodegenerative diseases. Their pivotal role in human health and disease has motivated efforts to engineer in vitro 3D microvessels to address questions associated with development, disease mechanisms, screening therapeutics, and regenerative medicine. Compared with traditional 2D models and animal models, 3D microvessels enable incorporation of human specific vascular cells to mimic key physiological or pathological features of native human microvessels.

Fig. 1.

Schematic illustration of a microvessel system in healthy adult human tissue. EC, endothelial cell; PC, pericyte; SMC, smooth muscle cell; BM, basement membrane. Note that pulmonary and umbilical microcirculation have opposite oxygen tension in venules and arterioles. Capillary oxygen tension in the lung can reach 100 mmHg. Created with Biorender.

Here we review recent advances in tissue-engineering of human-specific microvessels. Although, there are many different definitions of microvessels^[5], in this review, we define tissue-engineered microvessels as a single vessel or a network of vessels with a perfusable lumen smaller than 1 mm diameter formed by ECs with or without supporting cells in vitro. Compared with large vessels (diameter over 1 mm), tissue-engineering of microvessels requires overcoming challenges associated with control of structure and function, mainly due to the small dimensions. Despite these challenges, significant progress has been made, specifically in achieving confluent human endothelial monolayers across different microvessel length scales and diameters, and using these models to advance the understanding of vascular physiology and pathology. In this review, we first summarize the approaches, key components, and the platforms for engineering and maintaining human-specific microvessels. Then, we review tissue-specific microvessel models of the brain and other organs, followed by applications of microvessels in modeling development, disease, drug screening and delivery, and vascularization in tissue engineering. Finally, we discuss future research directions for the field.

1. Engineering strategies

Engineering strategies for fabricating microvessels include templating (direct, sacrificial, sandwich templating), viscous finger patterning, laser lithography (photodegradation), 3D printing, and self-assembly (Fig. 2a). The detailed process, advantages, and disadvantages of these techniques have been extensively discussed in several recent reviews^{[5a, 5b, 6]}. The structures formed using these techniques can be categorized as: single EC vessels, networks of EC microvessels, co-cultured single vessels, and co-cultured networks (Fig. 2b). Single microvessels are defined as microvessels with a perfusable lumen, and network microvessels are defined as a collection of perfusable microvessels with multiple branches and connections. Single microvessels with well-defined diameters can be produced by templating^[7], viscous finger patterning^[8], laser lithography^[9], and 3D printing^[10]. Network microvessels have been created using self-assembly (vasculogenesis, angiogenesis), 3D printing^[10], sandwich templating^[11], laser lithography (photoablation)^[9], sacrificial templating^{[10d, 12]}, and hybrid approaches^[13]. Depending on the cell seeding strategy, ECs can be mixed with a hydrogel matrix (matrix seeding) or perfused into pre-formed hydrogel channels (perfusion seeding). The key steps, advantages, and disadvantages of each engineering strategy are summarized in Fig. 2c.

Fig. 2. Engineering microvessels using different techniques.

(a) Key steps involved in different engineering approaches to fabricate microvessels. (b) Schematic illustration of single and network microvessels and the corresponding tissue-engineering methods. (c) Summary of key advantages and disadvantages of different engineering approaches. Cost is estimated based on the materials and equipment. Fabrication time is estimated based on the time from device assembly to the formation of perfusable microvessels.

Templating is a simple and cost-efficient way to engineer microvessels, and can be classified as direct, sacrificial, and sandwich templating depending on the template material. In direct templating, a cylindrical channel in a hydrogel matrix is created by removing a template rod^[7b]. Metallic and polymeric rods have been used to generate microvessels with diameters from ten to several hundred microns. The two major challenges in this method are achieving capillary dimensions and integrating capillary-scale microvessels into hierarchical structures. Reproducibility in achieving capillary-scale dimensions is limited by the fact that ECs are typically larger than capillary diameters making it difficult to use conventional perfusion seeding approaches. A double-templating system using polymer fibers was developed to address both of these issues^[14]. In this approach, ECs are first seeded into a large microvessel channel and then directed EC migration is exploited to seed downstream capillary channels. This directed migration method avoids EC occlusion during perfusion seeding in capillary-scale channels. Sandwich templating and sacrificial templating (e.g. gelatin or carbohydrate) have also been used to achieve complex microvessel architectures^{[11–12, 15]}, including branching and tortuosity which are difficult to achieve using direct templating. However, these microvessel networks typically have relatively large diameters (> 50 μm) and non-physiological features (e.g. right angle turns). Co-cultured single microvessels, such as venules with diameters over 100 μm, have been produced using direct templating^[16]. Using a templating strategy, mural cells can be seeded into the hydrogel matrix or perfusion seeded into the channel followed by EC perfusion seeding^[16–17]. Seeding mural cells in the hydrogel matrix creates more surface area in the channel lumen for EC adhesion but increases the number of mural cells not in direct contact with ECs. On the other hand, perfusion seeding mural cells into the channel requires seeding ECs directly onto mural cells under conditions that promote adhesion and spreading.

Another common strategy is microvessel self-assembly, which relies on the native ability of ECs to form tubular networks in the presence of angiogenic factors, mimicking the physiological processes of vasculogenesis and angiogenesis. In vasculogenesis, endothelial cells with or without mural cells are mixed in a hydrogel followed by stimulation with angiogenic factors. The endothelial cells grow and merge into tubular structures inside the hydrogel and eventually form a network of microvessels. In angiogenesis, vascular sprouts grow from pre-formed microvessels. Self-assembly via vasculogenesis^[18] or angiogenesis^{[18b, 19]} can produce hierarchical microvessels with mixed diameters from 20 to ~200 μm in over a week. These structures capture the branching and tortuosity in native blood vessels, but there remain challenges in consistently achieving physiological microvessel architectures. Compared with other engineering methods, the self-assembly process is slower and hence fabrication is often significantly longer. In addition, the microvessels typically have a range of diameters and are more difficult to reproduce. On the other hand, self-assembly allows creation of microvessels incorporating mural cells without multiple seeding steps. Co-cultured PCs and ECs in microfluidic devices can also self-assemble into microvessel networks with an EC monolayer and abluminal PCs^[20]. A triple cell co-culture model with ECs, PCs, and astrocytes seeded into a fibrin hydrogel was used to create a perfusable microvessel network with diameters from 20 – 300 μm^[21].

Viscous finger patterning is not as common as other strategies but has been reported to successfully generate perfusable microvessels^{[8, 22]}. Microvessel channels are created by dispensing a droplet of medium through a pre-polymerized viscous hydrogel solution using passive pumping^[23]. The degree of pre-polymerization of the hydrogel is dependent on the target microvessel diameter, hydrogel composition, and microfluidic channel geometry. Typical diameters of microvessels created using this method are greater than 50 μm. Nevertheless, it is a cost-effective way to generate microvessels with branching and curvature. Laser lithography (photoablation) utilizes high energy photons to ablate the scaffold, which creates small channels with diameters down to ~ 10 μm^{[9b, 24]}. Microvessel channels in poly(lactic-co-glycolic acid), polyethylene (PEG), and collagen I hydrogels have been achieved using this approach. For example, laser ablated capillary channels connecting two parallel large microvessels were created in collagen hydrogels^[9b]. The migration of ECs from the parallel microvessels into the capillary channels facilitated the formation of hierarchical microvessel networks with capillaries. The advantage of laser lithography is the ability to create hierarchical microvessels with high precision, however, it requires a specialized optical system.

3D printing, also known as additive manufacturing, has been widely used to fabricate large vascular grafts (on the order of millimeters in diameters) because of its potential for expandability. 3D printing of a sacrificial templating ink loaded with ECs allows the assembly of an interconnected microvessel network^[29c]. 3D printing can be used to pattern sacrificial materials for creating hydrogel channels followed by perfusion seeding into the channel, or to directly print a hydrogel matrix mixed with ECs^{[29a, 29c]}. Straight, tapered, curved and spiral microvessels can be achieved using 3D printing^[30]. Co-axial 3D printing has successfully generated co-cultured EC/SMC models with a wall thickness of around 200 μm and inner diameter ~700 μm^[29b]. The major limitation of current 3D printing techniques is the resolution, which typically yields diameters over 50 μm. In addition, high-precision 3D printers often have a limited range of resin materials.

These engineering approaches have their advantages and disadvantages (Fig. 2c). The type of microvessel (single or network, monoculture, or co-culture), relative cost, turnover time, scalability, requirements for specialized equipment, etc. should be considered when selecting a strategy. In general, templating is the simplest and least expensive engineering strategy with moderate scalability. Self-assembly allows the creation of microvessel networks with mural cells. Laser lithography and 3D printing have better scalability but require specialized equipment.

Re-seeding ECs (re-endothelialization) into decellularized tissues is another strategy to create in vitro microvessels. Typical examples include decellularized heart^[31], lung^[32], liver^[33], and kidney^[34]. With appropriate protocols for decellularization, these tissues can retain the vascular BM structure. Infusion of ECs into the vascular BM lumen can be achieved by perfusion seeding via arteries or veins, which can result in the re-endothelialization. Perfusion conditions such as pressure, flow rate, perfusion time, cell density, source of ECs, incubation time, and perfusion medium all contribute to the efficiency of EC adhesion and spreading. The final microvessel structure is pre-defined by the vascular BM, which facilitates the building of hierarchical structures. Microvessels with diameters over 50 μm show more successful EC adhesion and spreading. Seeding capillary beds is very difficult, often resulting in occlusion or emission of ECs, which is detrimental for re-cellularized tissues. Strategies for re-endothelialization of capillaries in decellularized tissue remain to be established. Re-seeding is mainly performed with ECs resulting in the formation of microvessels without mural cells.

2. Major components of tissue-engineered microvessels

2.1. Vascular cells

Three major sources of human vascular cells (ECs and supporting cells) include: primary, immortalized, and stem cell-derived cells. The advantages and disadvantages of these cell sources are summarized in Table 1.

Table 1.

Cell sources for tissue-engineered microvessels.

Cell type	Primary	Immortalized	iPSC-derived
Origin	fresh tissues	genetically reprogramed to proliferate	derived from adult somatic cells
Advantages	native cell morphology and phenotype genetically consistent high relevance to in vivo conditions	unlimited cell availability easy maintenance high reproducibility and consistent cell phenotype	unlimited cell availability high reproducibility genetically matched cell types relevance to in vivo conditions
Disadvantages	limited cell availability high donor-to-donor variation limited regenerative capacity and loss of phenotype	changes in cell morphology and phenotype reprogramed genome	specialized differentiation and maintenance variable transcriptomic accuracy

Primary vascular cells have been isolated and purified from a wide range of human tissues including skin, liver, kidney, brain, lung, umbilical vein, and different arteries. PCs have been isolated from placenta, lung, and brain tissue, and SMCs from aorta, lung, and brain. Primary cells from many different tissues and organs are commercially available, along with specialized culture medium. The advantages of primary cells include: high morphological and biochemical similarity to native vascular cells^[35]. However, primary cells require donor tissues, often show phenotypic drift, have limited regenerative capacity in vitro, and often suffer from low purity, all of which significantly limit their applications.

Immortalized cells have nearly unlimited potential for expansion and hence provide a scalable and reproducible source. It has been reported that immortalized human microvascular ECs showed no sign of senescence after 95 passages^[36]. However, immortalized cells often exhibit phenotypical and morphological changes. In addition, the use of lentiviruses in the creation of immortalized cell lines limits potential human applications due to safety concerns.

iPSC-derived vascular cells offer a new approach to avoid some of the limitations with primary and immortalized vascular cells. Using patient-derived vascular cells also enables patient-specific disease modeling, drug testing, and toxicity screening. iPSCs have the potential to be a scalable and renewable source of human vascular cells. Many protocols for differentiation ECs^[37], PCs^{[37e, 38]}, and SMCs^[39] have been reported. Validation of iPSC-derived ECs and supporting cells is usually based on expression of common markers such as CD31, vWF, VE-Cadherin (ECs), NG2 and PDGFRβ (PCs), and ACTA2, CNN1, TAGLN, MYH11, and SMTN (SMCs). These protocols have demonstrated that iPSC-derived vascular cells not only express many key markers, but also recapitulate many functional characteristics of their native counterparts. Numerous studies have shown that iPSC-derived ECs exhibit many of the hallmarks of human ECs while enabling highly scalable and patient-specific studies^{[25c, 40]}, however, the fidelity of transcriptomic similarity to human ECs remains an issue. Unlike primary ECs, which are generally sub-cultured on uncoated surfaces, iPSC-derived ECs are usually seeded and expanded on fibronectin- or gelatin-coated surfaces^[41]. iPSC-derived ECs generally have limited capacity for regeneration and, depending on the local microenvironment, may exhibit phenotypic drift over long culture periods.

The total number of cells needed for fabrication is dependent on the microvessel diameter, microvessel density, and the engineering strategy. For perfusion seeding ECs into a microvessel channel, densities from 10 to 80 million cells per mL are needed to achieve a confluent endothelium^{[7b, 16b]}. Higher cell densities are often required for iPSC-derived ECs, which often show lower adhesion efficiency in hydrogel channels. For self-assembled microvessel networks, the density of ECs in the hydrogel matrix ranges from 5 to 10 million cells per mL^{[18a, 42]}.

2.2. Biomaterials

In native microvessels, ECs adhere to vascular BM, which is a thin layer of specialized ECM. Vascular BM is mainly composed of collagen IV, laminin, and proteoglycans, although the exact composition and structure is not well understood and is thought to vary across different tissues^[43]. Some studies have shown that the mechanical strength of vascular and retinal BMs is over 500 kPa^[44], compared with the ECM materials used in microvessel engineering (~0.5 to ~10 kPa)^[45]. Vascular BM molecules are generated by ECs and supporting cells, which in turn provides the structural matrix for adhesion^{[43a, 46]}. ECs, PCs, and SMCs cultured in vitro also generate vascular BM molecules^{[18b, 46]}, but it is unknown whether the structure and composition are comparable to in vivo vascular BM. To support the adhesion and spreading of ECs in microvessels, matrix materials should present appropriate cell binding sites (e.g. binding partners for integrins), be able to withstand the mechanical deformation from cells and flow, and be stable over the lifespan of the device. Other design criteria may be dependent on the engineering strategy, for example for some 3D printing approaches, the matrix must enable photopolymerization or photodegradation.

Native biomaterials extracted from animals such as rat tail collagen I, bovine fibrinogen, and mouse tumor ECM (Matrigel) have been widely used in tissue-engineered microvessel models. Rat tail collagen I at concentrations from 2 to 8 mg mL⁻¹ polymerizes under neutral pH forming a fibrous hydrogel network. Fibrinogen forms fibrin hydrogels on reaction with thrombin. 3 to 10 mg mL⁻¹ fibrin hydrogel is commonly used in self-assembly models with thrombin concentrations around 2 – 3 U mL⁻¹. Matrigel is a mixture of 60% laminin, 30% collagen IV, ~8% entactin (nidogen-1), and ~2–3% perlecan^[47]. These native biomaterials all have binding sites that promote EC adhesion and monolayer formation. The mechanical strength and degradation rate of native ECM hydrogels are mainly dependent on the monomer concentration and cell culture medium. Mixtures of hydrogels such as collagen I/fibrin^[18b], collagen I/alginate^[28b], collagen I/chitosan^[48], collagen I/Matrigel^{[8a, 49]}, and others^[50] have also been used to improve the mechanical strength, stability, cell adhesion, or adjust the degradation rate of the matrix. These composite hydrogels have been used in engineering microvessels with different types of ECs (Fig. 3).

Fig. 3. Biomaterials used in tissue-engineered microvessels.

hECs, human endothelial cells; HUVECs, human umbilical vein endothelial cells; iECs, induced endothelial cells; HEKs, human embryonic kidney cells; hBMECs, human brain microvascular endothelial cells; phECs, primary human endothelial cells; phMVECs, primary human microvascular endothelial cells; iBMECs, induced brain microvascular endothelial cells; hMVECs, human microvasculature endothelial cells, pmECs; primary mouse endothelial cells; hCMEC/D3, human cerebral microvascular endothelial cells; porAoECs, porcine aortic endothelial cells; LAP, lithium phenyl-2,4,6-trimethylbenzoylphosphinate.

Synthetic or semi-synthetic biomaterials such as polydimethylsiloxane (PDMS)^[51], methacrylate-modified gelatin^{[10a, 26a, 29d]}, polyethylene (PEG)^{[28a, 29d]}, polypropylene^[52], polyurethane^[53], and others have also been explored to engineer microvessels^{[29d, 54]}. Compared with native biomaterials, these synthetic biomaterials have little batch-to-batch variation, a wide range of tunable mechanical properties, and tunable degradation rates. The major disadvantage of synthetic biomaterials is the lack of cell binding sites for EC adhesion. Therefore, many synthetic biomaterials used in supporting microvessel growth are modified, supplemented, or coated with fibronectin, collagen I or other native biomolecules that promote EC adhesion and spreading.

2.3. Culture platform

Microfluidic chips with microchannels that allow the manipulation of flow at small volumes have been used in a variety of biomedical applications, such as biomolecular synthesis/purification/analysis, tissue engineering, and medical diagnostics^[58]. Advances in microfluidic hydrogel chip technology have contributed to the development of a class of hybrid microvessel models. Many materials have been used to fabricate microfluidic chips, including glass/silicon, PDMS, and polymethylmethacrylate (PMMA)^[58b]. PDMS is a transparent, hydrophobic, and highly porous material with high biocompatibility and high O₂ permeability but does not promote cell adhesion^[59]. PDMS chips are usually fabricated by casting into molds that are created by lithography, 3D printing, or machining. Glass is a transparent, highly stable, hydrophilic material with low O₂ permeability. Glass is usually processed using photolithography. Due to the low cost and fast prototyping, PDMS microfluidic chips have been widely used in microvessel engineering.

Many approaches, such as gravity-driven systems, syringe pumps, and peristaltic pumps have been used to control microvessel perfusion. Gravity-driven systems are low-cost and usually home-made devices, that allow the creation of flow based on a pressure difference between inlet and outlet reservoirs. The pressure difference between the reservoirs may change over time and is determined by the dimensions of the reservoirs and the flow resistance. As a result, the pressure and shear stress decrease over time, which is not physiological. Syringe pumps can create a steady flow, where flow rate and shear stress do not change over time, which allows the simulation of relatively constant flow. The flow rate can be adjusted from a few microliters to milliliters per minute in a typical device. Lastly, peristaltic pumps generally result in pulsatile flow due to the motor geometry, tubing, and physical properties of the medium.

Cell culture media used for perfusion of microvessels are mainly designed to maintain EC health. Commercially available media for ECs typically includes vascular endothelial growth factor (VEGF), basic fibroblast growth factor (bFGF), epidermal growth factor, heparin, ascorbic acid, hydrocortisone, antibiotics, and fetal bovine albumin (FBS). However, the exact concentration of each component typically varies between vendors. Commercial medium used for primary ECs have also been used to culture and perfuse iPSC-derived ECs. Additional factors such as retinoic acid have been added to media for culture of iPSC-derived ECs to improve the barrier function^[60]. Increased amounts of VEGF or bFGF are used in self-assembly models during the initial formation of microvessel networks. The viscosity of cell culture medium with 0 to 10% FBS ranges from 0.7 – 1.0 cP^[61], significantly lower than whole blood which has a viscosity of 2 – 5 cP at 37 °C depending on hematocrit and vessel diameter^[62]. Molecules such as 70 kDa dextran can be added to increase the viscosity of perfusion medium^{[7b, 63]}. For example, 4% dextran supplementation increases the medium viscosity to 1.63 ± 0.07 cP at 37 °C^[7b]. For co-culture models, the medium should be able to support the normal function of all cells. Despite its importance in maturation and longevity, medium optimization is largely overlooked in tissue-engineered models, and many reports of co-cultured microvessel models do not include details of the medium^{[8b, 20, 50, 55a]}. Co-culture models typically use EC medium^{[17, 64]}, or a mixture of different cell culture media with or without supplements^{[16a, 21, 65]}.

3. Tissue-specific microvessels

Microvessels in different organs show distinct differences in cellular structure and architecture to meet the local needs in tissue function. Many microfluidic-based models with 2D confluent monolayers of ECs have been developed to mimic tissue-specific vessels^[66]. Here we summarize tissue-specific microvessels (Fig. 4).

Fig. 4. Tissue-specific microvessel models.

a. Fluorescence images of a tissue-engineered brain-specific microvessel venule model. Recreated from ^[16a] with permission. iPCs, iPSC-derived pericytes; iBMECs, iPSC-derived brain microvascular-like endothelial cells; z pro., z projection; xy proj., xy projection. b. Fluorescence images of tissue-engineered human dermal microvessels stained with F-actin. Bb, Borrelia burgdorferi. Recreated from ^[25d] with permission. c. Confocal images of a PCV microvessel on day 2 showing the growth of angiogenic sprout. Recreated from ^[16a] with permission. d. Tumor organoid–microvessel interactions. MMTV-PyMT, mammary specific polyomavirus middle T antigen overexpression mouse tumor organoids; HUVEC-GFP, human umbilical vein endothelial cells labeled with green fluorescence protein. Recreated from ^[75] with permission. e. Phase/fluorescence and confocal images of THP1 immune cell adhesion and transmigration. Recreated from ^[14] with permission.

3.1. Brain microvessels

The brain vasculature is part of the blood-brain barrier (BBB) which regulates transport into and out of the brain. Microvessels in the brain cortex have high barrier function with brain microvascular ECs (BMECs) expressing tight junction proteins and low levels of endocytic vesicles. Key benchmarks for tissue-engineered brain microvessels have been reported previously^[67]. Direct templating, viscous finger printing, and self-assembly techniques have all been applied to engineer BBB models. Most BBB models have been engineered from brain microvascular endothelial cells (BMECs) without any supporting cells. More recently, reports have described co-culture of microvessels similar to post-capillary venules with BMECs surrounded by PCs^{[8b, 16]}, and microvessels with BMECs, PCs, and astrocytes^{[8b, 21, 64a]}. The fabrication of brain-specific arterioles with BMECs surrounded by SMCs has been more challenging. Brain-specific capillary-scale networks have been formed by self-assembly in microfluidic devices and using templating^{[14, 21]}.

Engineering brain-specific microvessels has largely focused on recapitulating barrier function and transport. Microvessels with diameters of ~200 μm engineered from rat brain ECs showed expression of ZO-1 and low permeability to 40 kDa dextran^[68]. Brain microvessels created from induced brain microvascular-like ECs (iBMECs) showed significantly higher barrier function to Lucifer yellow (P = 3.5×10⁻⁷ cm s⁻¹) and rhodamine123 (P = 1×10⁻⁷ cm s⁻¹), compared with the HUVECs^[49b]. Several groups have reported incorporation of PCs and astrocytes into BBB models^{[8b, 16, 21, 64a]}. Co-cultured iBMECs with iPSC-derived PCs (iPCs) have recapitulated EC/PC interactions (Fig. 4a). Whether the presence of supporting cells such as PCs and astrocytes significantly affect the barrier function of in vitro brain microvessel is still debatable. Co-cultured microvessel models using iBMECs and iPCs demonstrated similar barrier function to Lucifer yellow (P =~ 4×10⁻⁷ cm s⁻¹) compared with iBMEC microvessels. It has been suggested that the presence of pericytes does not influence barrier function in healthy microvessels, but can rescue barrier function in stressed microvessels^[16b]. Induced non tissue-specific induced ECs (iECs) co-cultured with primary human PCs and astrocytes showed improved barrier function to 10 kDa dextran (~2 ×10⁻⁷ cm s⁻¹) compared with only iECs (~1.2×10⁻⁶ cm s⁻¹)^[21]. Co-cultured human BMECs with astrocytes showed decreased permeability while co-cultured BMECs and PCs showed no difference to BMEC microvessels to 3 kDa dextran^[8b]. These contradictory results may be due to the different cell types and molecular weight of the solutes. A self-assembled BBB model including iPSC-derived brain ECs, PCs, and astrocytes was formed in Matrigel^[69], however, the microvessels were cultured under static conditions.

3.2. Other organs

In general, there have been relatively few reports of 3D microvessels in non-brain tissues. These have mainly been generic microvessels without supporting cells. A review of co-cultured models of ECs with other parenchymal cells under static conditions can be found elsewhere^[70].

The arterioles and venules in dermal microvessels form two horizontal plexuses connected by ascending arterioles and descending venules. Most of the microvessels in skin have a continuous endothelium except for the capillaries close to sweat glands and dermal papilla of the hair^[71]. Tissue-engineered human dermal microvessel models have been developed using direct templating and perfusion seeding with human dermal microvascular ECs into collagen I hydrogel (Fig. 4b)^{[7b, 25d]}. The dermal microvessel model displayed a strong barrier function over a few days and showed low adhesion of immune cells under quiescent conditions.

Microvessels in glomeruli include afferent arterioles, capillaries, and efferent arterioles^[72]. The proximal portion of afferent arterioles and the glomerular capillary network in the kidney have a fenestrated endothelial layer, which improves filtration of blood waste products. The size of the fenestrations is typically around ~60 nm^[73] and constitute 20% - 50% of the vessel surface area^[73]. Human kidney microvascular ECs (HKMECs) have been used to engineer renal-specific microvessels using sandwich templating^[74]. 3D renal microvessels engineered from HKMECs demonstrated well-defined fenestrae of ~ 60 nm and expression of plasmalemma protein, a marker of fenestrae diaphragms. The presence of diaphragms increased the overall convective flux of 40 kDa dextran.

Capillary ECs in lung alveoli share vascular BM with air sac epithelial cells. The BM in alveola capillaries varies from 20 nm to 100 – 200 nm depending on the region^[76]. Lung-specific microvessels were fabricated via self-assembly using human lung microvascular ECs and lung PCs^[77]. The network of human lung microvessels formed after 14 days showed a permeability of 2 × 10⁻⁶ cm s⁻¹ to 70 kDa dextran.

4. Applications of tissue-engineered microvessels

Animal models capture the biological complexity of living organisms, but differences at the genetic, cellular, and anatomical levels often hinder translation of results from animal studies into humans. 2D models based on confluent monolayers of ECs have been widely used to study the physiology and pathology of endothelium. While these models allow the use of healthy or patient-derived cells but represent a very reductive system. Nevertheless, previous studies using patient-derived vascular cells in 2D models have paved the way for applications in 3D models^[40b-g]. The biological complexity of 3D tissue-engineered microvessel models continues to increase, significantly expanding the range of physiological relevance for mechanistic and translational research. Microvessels have been used to model vascular development, disease, screening for therapeutic delivery, and for engineering vascularized human tissues (Fig. 5).

Fig. 5. Major applications of tissue-engineered microvessel models.

Models have become increasingly complex and tissue specific, enabling applications that span human development, disease, vascularization, and drug screening and delivery. NDD, neurodegenerative disease.

4.1. Modeling development

Vasculogenesis and angiogenesis play an important role in embryonic development. Tissue-engineered microvessel models have provided insight into these processes by allowing precise control of the microenvironment and enabling visualization of cell migration in real time. These models typically use EC matrix seeding or perfusion seeding into hydrogel channels. For example, HUVECs co-cultured with fibroblasts in fibrin hydrogels can self-assemble and form perfusable microvessel networks via vasculogenesis^[18b]. In another example, the presence of a VEGF gradient has been shown to induce the formation of angiogenic sprouts, and the addition of an angiopoitin-1 gradient further promoted the stability of HMVEC angiogenic sprouts^[28c]. Growth factors released from fibroblasts have been shown to promote angiogenic sprouting into a fibrin gel^[86b]. In other work, HUVECs have been shown to invade collagen I gels and migrate towards the high concentration of angiogenic growth factor cocktails (VEGF, hepatocyte growth factor, bFGF, monocyte chemoattractant protein-1, sphingosine-1-phosphate)^[19b]. Other studies using the perfusion of an angiogenic growth factor inside iBMEC and HUVEC microvessels also showed the formation of angiogenic sprouts and migration of tip cells into the collagen hydrogel without the presence of a growth factor gradient^{[11, 16a, 86b]}. In addition to angiogenic factors, interstitial flow also promoted the formation of angiogenic sprouts in the opposite direction of interstitial flow in fibrin and collagen I matrix^{[42, 86a, 86e]}. Vasculogenesis enhanced by interstitial flow is dependent on upregulation of matrix metalloproteinase-2^[87]. These studies report angiogenic growth rates from 30 to 170 μm per day depending on the type of ECs, ECMs, growth factors, and flow conditions.

Tissue-engineered co-culture microvessels have advanced the understanding of the role of mural cells in angiogenesis processes. A co-cultured brain microvessel model incorporating iBMECs and iPCs showed that PCs can lead angiogenic sprouts as tip cells, co-migrate with the endothelial stalk cells, or maintain a quiescent state (Fig. 4c)^[16a]. The presence of PCs also significantly enhanced sprout formation and the growth rate of sprouts^{[16a, 86d]}. The role of SMCs in angiogenesis is largely unknown.

4.2. Modeling vascular diseases

Blood vessel dysfunction plays a key role in almost all human diseases, including stroke, ischemic heart disease, cancer, and neurodegenerative diseases. Tissue-engineered microvessel models have provided new tools to probe risk factors for disease, mechanisms of disease progression, and the screening of therapeutic strategies.

The development of tissue-engineered vascular models has been important for cancer research since cancer cell/vascular interactions are a key step in dissemination and metastasis^{[75, 81a, 81e, 81f, 81h]}. In cancer research, tissue-engineered microvessel models have been developed to provide mechanistic insight into tumor growth and proliferation, migration, invasion, matrix remodeling, dormancy, intravasation, extravasation, angiogenesis, and therapeutic delivery (Fig. 4d). Model variables include cell sources (patient-derived cells, commercially available cell lines, stem cells, stromal cells, immune cells, etc.), biophysical properties (oxygen partial pressure, pH, interstitial flow, etc.), extracellular matrix (stiffness, architecture, etc.), and biochemical cues (chemoattractants, angiogenic factors, inflammatory cues, etc.). Changes in microvessel function are largely dependent on the nature of the tumor cell/microvessel interactions. For example, EC microvessels with breast or colorectal cancer cells seeded in the matrix showed increased permeability^{[81c, 81i]}, possibly due to elevated expression of VEGF^[81i]. A vascularized glioblastoma multiforme model showed that tumor cells did not significantly alter the permeability of brain microvessels, which recapitulates brain microvessel phenotype in human glioblastoma^[81b]. Imaging of tumor cells embedded in the matrix surrounding HUVEC and HMVEC microvessels revealed a multistep intravasation process including, matrix degradation, vessel interaction, insertion into the matrix-microvessel interface, deformation of the endothelium, penetration of the endothelium, and flow into circulation^{[81f, 81h]}. A HUVEC microvessel model co-cultured with breast cancer tumor organoids recapitulated vessel interactions observed in vivo, including mosaic vessel formation, vessel co-option, and vessel pinch^[75]. In another study, human glioblastoma cells were found preferentially localized around 50 μm from the HUVEC microvessel, specifically around the bifurcation and about one third of the tumor cells fully integrated with the microvessel^[81d].

Microvessel models have also recapitulated aspects of the immune response and immune cell interactions with endothelium. Activation of the endothelium by inflammatory cytokines, and adhesion and transmigration of immune cells have been confirmed using tissue engineered microvessels. Tumor necrosis factor alpha (TNF-α) perfusion in BMEC and HUVEC microvessels upregulated the secretion of inflammatory cytokines, intercellular adhesion molecule-1 (ICAM-1), and vascular cell adhesion molecule-1 (VCAM-1)^{[8b, 14, 49b]}, and increased immune cell (THP1 or HL-60 cells) adhesion (Fig. 4e)^{[7b, 14, 17]}. In a model of hematologic diseases, HUVEC microvessels seeded on fibronectin-coated PDMS. Perfusion of whole blood into microvessels following TNFα activation showed adhesion of leukocytes and platelets to the endothelium resulting in reduced flow velocity^[83]. Perfusion of neutrophil-like THP-1 cells in tissue-engineered iBMEC capillaries treated with TNFα^[14] resulted in occlusion similar to neutrophil clogging reported in capillaries in the mouse brain^[88].

Infectious diseases such as malaria and meningitis have been studied in microvessel models. Red blood cells infected with plasmodium falciparum, the microorganism that causes malaria, showed no deformation when perfused in a tissue-engineered HUVEC capillary model, which leads to rapid occlusion of microvessels^[9b]. A tissue-engineered dermal microvessel model of a tick bite was used to study the migration and intravasation into circulation of Borrelia burgdorferi (Bb), the bacterium that causes Lyme disease^[25d]. It was shown that migration did not involve chemoattraction and was independent of adhesin-mediated interactions between Bb and ECM components. Perfusion of cryptococcus neoformans, the most common pathogen leading to fungal meningitis, into a multicellular BBB model incorporating human neural stem cells, BMECs and brain pericytes showed the formation of clusters of the fungus that penetrated the microvessel without altering tight junctions, suggesting a transcytosis-mediated mechanism^[82].

Increasing evidence suggests that brain microvessel dysfunction occurs prior to the onset of symptoms of neurodegenerative disease^[89]. iBMECs differentiated from a juvenile Huntington’s disease (HD) patient were used to model the influence of HD mutations on microvessel function. The tissue-engineered HD microvessel model showed increased cell turnover and immune cell adhesion compared to control^[25c]. Microvessel models have also been used to understand how different risk factors contribute to the development of vascular dysfunction in neurodegenerative diseases. Perfusion of human BMEC microvessels with neuronal secretomes from iPSC derived neurons harboring the APP Swedish mutation showed upregulation of genes associated with endothelial inflammation and immune-adhesion ^[27a]. Finally, an iBMEC microvessel model demonstrated that chronic oxidative stress led to structural defects, increased permeability, and up-regulated inflammatory pathways^[84a].

4.3. Drug screening and delivery

Over 90% of drugs fail in human trails due to either lack of efficacy or severe side effects, despite showing promising results in animal models^[90]. This motivates the development of alternative models with improved human specificity. Tissue-engineered microvessel models allow the incorporation of human cells to mimic the architecture of human tissues for screening lead compounds and developing strategies for drug/cell delivery.

3D microvessel models co-cultured with tumor cells recapitulate aspects of the disease microenvironment and enable screening of potential drug candidates. For example, tumor cells showed a significant difference in gene expression, tumor growth, and responsiveness to drug treatment in a 3D human endothelial colony forming cell-derived ECs (ECFC-EC) microvessel model, compared with a corresponding 2D model^[81c]. In another ECFC-EC microvessel model co-cultured with human colorectal tumor spheroids, tumors showed strong metabolic heterogeneity consistent with previously reported in vivo studies^[81g]. A self-assembled microvessel model with iPSCs derived ECs, primary PCs, ACs and glioblastoma cells seeded into the fibrin matrix was used to develop glioblastoma-targeting nanoparticles^[81b]. The microvessels maintained high barrier function to 40 kDa dextran (P < 4 × 10⁻⁸ cm s⁻¹) with and without glioblastoma cells in the matrix. Perfusion of angiopep-2 peptide-conjugated nanoparticles resulted in increased accumulation of nanoparticles in tumors. Similar results were obtained in a mouse glioblastoma model suggesting that the in vitro model accurately predicted in vivo response.

Microvessels have also been used to develop strategies for drug and gene delivery. Brain microvessel models have been used to study methods for transient opening of the BBB with melittin and mannitol ^{[85a, 85b]}. A 3D BBB model with iBMECs in a collagen I hydrogel was used to optimize the dosing of the membrane active peptide melittin to reversibly open brain microvessels^[85a]. Time-lapse images showed reversible BBB disruption and increased permeability to 500 kDa dextran. The in vitro BBB model revealed that melittin dosing increased endothelial paracellular permeability by disruption of cell-cell junctions. Transient BBB opening was also validated in a mouse model. In vitro microvessel models have also facilitated the development of localized drug delivery using ultrasound ^{[27c, 85d]}, showing that enhanced drug delivery and cell death is dependent on the acoustic pressure and microbubbles.

4.4. Vascularized tissue constructs and organoids.

Tissue engineering has the potential to generate functional tissues or organs to treat many severe human diseases. However, vascularization of de-cellularized tissues or organoids remains a major challenge. The main issue in vascularization of decellularized tissues is in seeding ECs into capillary beds since the cells in suspension are of similar or larger size than capillary diameters. The two main issues in vascularization of organoids are creating a vascular network within the organoid to supply nutrients and making connection to the network to enable continuous perfusion^[91]. Vascularization using templated, self-assembled, or 3D printed microvessels have shown promising results in promoting the longevity, maturation, and function of engineered tissues.

A perfusable microvessel network is key to maintaining a supply of nutrients (oxygen and glucose) and removing waste products in engineered tissue constructs. In a 3D printed microvessel model, organoids were used as building blocks for biomanufacturing thick human tissue constructs^[10b]. Microvessels were introduced into organoid aggregates using a 3D printed gelatin sacrificial ink. After removing the sacrificial ink, HUVECs were seeded to the channels with diameters from 400 to 1000 μm. Using this system, a perfusable cardiac tissue construct with a density of 10⁸ cells mL⁻¹ were maintained for 1 week. In another approach, a sacrificial vascular ink network of pluronic and thrombin was 3D printed inside a gelatin/fibrin hydrogel with fibroblasts or mesenchymal stem cells^[10c]. After removal of the sacrificial ink, HUVECs were seeded into the channels to form microvessels. Using this approach, a dense osteogenic tissue with thickness over 5 mm was successfully created and maintained over at least 30 days. Vascularized human heart and liver tissues have been created using a chip model^[78a], which utilizes 3D stamping and layer-by-layer deposition to form complex internal channel networks and external cavities. HUVECs were seeded into the internal channels with diameters over 100 μm and maintained with a gravity-driven perfusion system. Seeding human hepatocytes and supporting cells into the external cavities of the vascularized chip resulted in a functional liver tissue which metabolizes terfenadine. Using the same model, a vascularized cardiac tissue with ~2 mm thickness and high density of cells was also created by seeding cardiomyocytes into the external cavities.

The presence of microvessels in organoids promotes perfusion and enhances maturation and longevity in vitro due to enhanced nutrient transport, waste removal, and paracrine signaling from vascular cells. HUVECs seeded with fibroblasts in a fibrin gel were used to vascularize human kidney organoids seeded on top of a fibrin gel^[79b]. Under perfusion, HUVECs self-assembled into microvessel networks infiltrating kidney organoids. The vascularized kidney organoids had more mature podocytes and tubular compartments with enhanced adult gene expression^[79b]. In an “organ-on-VasculaNet” model, microvessel networks vascularizing islets were formed via self-assembly of ETV2 transfected HUVECs. Vascularized islets in microfluidic chips showed functional response to glucose within 10 minutes^[79a].

Pre-vascularized tissue constructs can promote the recovery of ischemic tissue and facilitate the integration of tissue constructs into host tissues^{[27b, 78b, 92]}. Human microvascular networks with a diameter of 125 μm were fabricated using sandwich templating with embryonic stem cell-derived ECs in a collagen matrix^[27b]. The tissue construct with a thickness of 1 mm and a diameter of 8 mm was sutured to the epicardial surface of rat ventricle. After 5 days, constructs with microvessels showed more visible perfused vessels compared with collagen constructs without microvessels^[27b]. Similar results were obtained in 3D-printed fibrin hydrogel tissue constructs containing microvessels^[92]. A parallel microvessel array was created by perfusion seeding of HUVECs into carbohydrate glass templated channels. Tissue constructs with the microvessel array were sutured to rat myocardial tissues and showed improved collateral circulation in ischemic heart. It is still unknown whether the improved recovery of ischemic tissues was due to anastomosis of engineered microvessels with the host tissues or paracrine signaling from the implanted ECs. In another example, HUVEC microvessels with diameters ~ 300 μm were created in 3D printed alginate followed by seeding of keratinocytes on top to form a vascularized dermal tissue construct^[78b]. In vivo grafting of the dermal construct into skin of immunodeficient mice showed integration with host blood vessels, as evidenced by the presence of blood in microvessels inside the dermal tissue construct.

5. Future studies

5.1. Increasing the structural integrity of microvessels in engineered platforms

Microvessels have a BM layer between the luminal EC layer and the abluminal mural cell layer. A current challenge is to improve the structural complexity of engineered microvessels by incorporating BM and mural cells. Mural cells play an important role in tissue homeostasis and maintenance of the structural integrity of microvessels^[93]. Many previously published studies have successfully achieved a confluent monolayer of ECs in microvessels with typical endothelial cell-cell junctions. Strategies for incorporation of different mural cells are still in the early stages of development. Recent research has focused on co-culture of mural cells, such as PCs or SMCs, into tissue-engineered microvessel models. The density, morphology, orientation, and stability of mural cells in current models do not fully match native microvessels and needs further refinement. The key components of native vascular BM include collagen IV, laminin, and proteoglycans, although the exact composition and structure is still not well defined and may vary across different tissues^[43]. Immunostaining studies have shown the presence of collagen IV and laminin on the abluminal side of tissue-engineered microvessels^{[8b, 18b]}. However, it is unknown whether the structure and composition of in vitro vascular BM is comparable to that in native microvessels. Introducing these structural components into in vitro models will further improve the biological similarities to native human microvessels.

Microvessels in native tissue transition from arterioles to capillaries and venules. Growing evidence indicates coupling between arterioles and downstream capillaries and venules^[94]. Therefore, hierarchical models recapitulating the complexity of the arterio-venous axis as a unit will significantly increase the physiological relevance of the in vitro microvessel models.

5.2. Modeling microvascular interactions with tissues

The primary function of microvessels is to supply nutrients to and remove metabolic waste from tissues. Unlike the main arteries and veins, microvessels are embedded into specific tissues in each organ, with unique function and metabolism. Cells in the connective tissues around microvessels release different biochemical cues such as growth factors, cytokines, and metabolic wastes, all of which modulate microvessel function. Studies have shown that cell culture medium supplemented with additional biochemical factors can influence the diameter, morphology, barrier function, and stability of microvessels^{[19b, 25b, 55b, 85c, 86b, 86c, 86f]}. In some self-assembly models, the presence of fibroblast-secreted factors promoted angiogenesis and vasculogenesis ^{[55b, 86f]}, however, the role of fibroblasts in subsequent microvessel maintenance is not fully understood. One study showed that the initial presence of fibroblasts within the first 1–3 days of co-culture is crucial for EC vasculogenesis, however, their removal does not structurally impact the resulting vasculature^[95]. Most current tissue-engineered microvessel models are isolated from parenchymal cells, which means they have limited crosstalk. In a few of the vascularized tissue models, microvessels were surrounded by parenchymal cells seeded in the matrix. However, whether parenchymal cells in the in vitro conditions (e.g. low cell density, presence of only one type of cell) can induce proper paracrine signaling with microvessel is not known. Future studies are needed to address the question of how lack of tissue interaction affects the behavior and function of microvessels.

5.3. Perfusing blood through engineered microvessels

Although many in vitro microvessels have been used to model the hemodynamics in human microcirculation, two major issues need to be addressed: the composition and rheological properties of cell culture medium. Traditional 2D cell culture media (such as MEM, DMEM, and RPMI), containing glucose, amino acids, vitamins, and inorganic salts, have been used to perfuse 3D microvessels. These media are usually supplemented with fetal bovine serum (FBS) or heat inactivated FBS. The full composition profile of complete cell culture medium is significantly different from human plasma^[96]. The fluid properties of cell culture medium are also very different from blood. Furthermore, blood is a non-Newtonian, shear thinning fluid with 35 – 50% red blood cells. More research is needed to develop substitutes to better mimic the composition and physical properties of human blood.

5.4. Extending the culture time of engineered microvessels

Experiments with tissue-engineered microvessels commonly last from a few days to a few weeks. The longest reported lifespan of a tissue-engineered microvessel is 7 weeks ^[25a]. Improving the longevity of microvessels will not only expand the capability to model chronic vascular diseases, but also has the potential to further extend the lifespan of vascularized human tissue models. Many factors, including mechanical factors (shear stress, transmural pressure, modulus of the matrix, etc.) and biological factors (growth factors and small molecules), are known to impact the longevity of microvessels. Optimization of each factor and the combination of these factors will potentially prolong the duration of microvessel models in vitro.

Conclusions

We reviewed the approaches, key components, and the cell culture platforms for perfusable tissue-engineered microvessel models along with tissue-specific microvessel models and their major applications. Single microvessels and network microvessels from 10 to 500 μm in diameter have been successfully created using different techniques such as templating, self-assembly, 3D printing, and laser lithography, while more work needs to be done to include other components in native microvessels. We anticipate that future studies will further improve the complexity of microvessel models and the physiological relevance for specific applications.

Biographies

Bios

Nan Zhao, PhD

Dr. Nan Zhao is an Associate Research Scientist in the Searson lab at Johns Hopkins University. His current research is focused on developing tissue-engineered blood vessels for modeling and treating human diseases. He has a broad background in biomaterials and tissue-engineering with specific training and expertise in hydrogel materials and vascular engineering.

Alexander Pessell

Alexander Pessell is currently pursuing a Ph.D. in Biomedical Engineering at Johns Hopkins University. He earned his bachelor’s degree in biomedical engineering from Trine University in Angola, Indiana. His research in the Searson lab involves developing tissue-engineered models to study the effects of pathological and therapeutic perturbations on the blood-brain barrier.

Ninghao Zhu, PhD

Ninghao Zhu received his Ph.D. in 2022 from the Department of Biomedical Engineering, Pennsylvania State University under the supervision of Dr. Pak Kin Wong. He is now a postdoctoral fellow at the Institute for Nanobiotechnology, Johns Hopkins University under the guidance of Dr. Peter C. Searson. His research interest includes developing tissue-engineered 3D microvessel models to study the vasculature in health and disease.

Peter C. Searson, PhD

Searson is the Reynolds Professor of Engineering at Johns Hopkins University. He holds appointments in the Department of Materials Science and Engineering, the Department of Biomedical Engineering, the Department of Physics and Astronomy, and the Department of Oncology. He is a fellow of the American Physical Society, the American Association for the Advancement of Science, and the Electrochemical Society.