Vascular tissue engineering by computer-aided laser micromachining

Anand Doraiswamy, Roger J. Narayan

Abstract

Many conventional technologies for fabricating tissue engineering scaffolds are not suitable for fabricating scaffolds with patient-specific attributes. For example, many conventional technologies for fabricating tissue engineering scaffolds do not provide control over overall scaffold geometry or over cell position within the scaffold. In this study, the use of computer-aided laser micromachining to create scaffolds for vascular tissue networks was investigated. Computer-aided laser micromachining was used to construct patterned surfaces in agarose or in silicon, which were used for differential adherence and growth of cells into vascular tissue networks. Concentric three-ring structures were fabricated on agarose hydrogel substrates, in which the inner ring contained human aortic endothelial cells, the middle ring contained HA587 human elastin and the outer ring contained human aortic vascular smooth muscle cells. Basement membrane matrix containing vascular endothelial growth factor and heparin was to promote proliferation of human aortic endothelial cells within the vascular tissue networks. Computer-aided laser micromachining provides a unique approach to fabricate small-diameter blood vessels for bypass surgery as well as other artificial tissues with complex geometries.

1. Introduction

Coronary grafting is a common treatment for ischaemic coronary disease, which results from occlusion of the coronary arteries. Coronary arteries are muscular arteries that contain three concentric layers. The innermost layer, the tunica intima, contains endothelial cells. The middle layer, the tunica media, contains smooth muscle. The outermost layer, the tunica adventitia, contains connective tissue. Ribatti et al. (2006) noted that extracellular matrix comprises more than 50 per cent of blood vessel mass. It is thought that fibronectin, microfibrils, proteoglycans, glycoproteins and other extracellular matrix components play important roles in blood vessel function. In the tunica media, smooth muscle cells produce collagen, elastic fibres as well as other extracellular material. The elastic fibres are believed to form circumferential net-like structures.

Kolessov (1967) described the first successful anastomosis of the coronary artery with the internal mammary artery. Favaloro et al. (1971) subsequently bypassed an occlusion to the proximal third of a patient’s right coronary artery with a portion of the patient’s own saphenous vein; angiography demonstrated that the bypass remained patent 8 days after the surgery. Bailey & Hirose (1968) later demonstrated the use of internal mammary artery to bypass the coronary artery in a beating heart. According to Rosamond et al. (2007) and King et al. (2009), approximately 427 000 coronary artery bypass graft surgical procedures are performed in the USA each year; one-third of these procedures are performed on women. Ayanian (2006) noted that the use of coronary artery bypass grafting has dropped since 1997; however, it remains a common procedure, particularly for conditions that cannot be managed by means of coronary angioplasty or stenting.

According to Eagle et al. (1999), the internal mammary artery is most commonly used as an autologous graft material; other arteries, including the greater saphenous vein, the lesser saphenous vein, the right gastroepiploic artery, the inferior epigastric artery and the radial artery have also been used as sources of autologous graft material in the past. Garrett et al. (1973) described complications associated with the use of autologous tissue for coronary grafting. Autologous arterial tissue suffers from constrained dimensions as well as limited supply. Autologous venous tissue may undergo hyperplastic occlusion, thrombosis and infection. Edelman (1999) noted that trauma and infection may occur at the site at which graft tissue is harvested. According to Ribatti et al. (2006), up to 39 per cent of patients do not have venous tissue that is appropriate for use in autologous vascular grafts as a result of poor vein quality, vein abnormality or absence of venous tissue as a result of previous surgical procedures.

Synthetic materials, including polyethylene terephthalate and expanded polytetrafluoroethylene, have been examined as an alternative to autologous materials for use in coronary grafting; however, small-diameter (less than 4 mm) vascular prostheses fabricated out of these biomaterials exhibit unacceptable rates of thrombosis. Several approaches have been investigated for improving interactions between blood and conventional cardiovascular biomaterials, including the use of cell, polymer and protein coatings. For example, Rumisek et al. (1986) examined the preparation of albumin-coated knitted polyethylene terephthalate grafts; these materials were prepared by soaking porous polyethylene terephthalate grafts in albumin solutions. In vitro studies in a canine model revealed that albumin-coated grafts and blood-preclotted control grafts exhibited no differences in healing or patency. Herring et al. (1979) demonstrated seeding of polyethylene terephthalate grafts with endothelial cells. An in vivo study involving a canine model showed that the seeded cells formed an inner lining; these cells expressed endothelial cell-specific markers. Williams et al. (1994) examined the incorporation of cellular linings containing microvascular endothelial cells, which were derived from canine falciform ligament fat, within small-diameter expanded polytetrafluoroethylene vascular grafts. In vitro studies involving a canine model revealed that cell-seeded grafts exhibited higher patency rates than unmodified expanded polytetrafluoroethylene grafts. In addition, cellular linings were shown to exhibit non-thrombogenic properties. According to Ribatti et al. (2006), the layer of endothelial cells must be both confluent and adherent in order to provide resistance to inflammation, neointimal proliferation and thrombus formation. Mann et al. (1995) demonstrated functional genetically engineered grafts in a lapine model. The grafts incorporated gene therapy agents, specifically antisense oligodeoxynucleotide blockage of smooth muscle cell proliferation. The treated rabbits demonstrated resistance to diet-related atherosclerosis. Pharmacologic agents have also been incorporated within vascular prostheses; for example, Aldenhoff et al. (2000) demonstrated immobilization of dipyridamole, an inhibitor of vascular smooth muscle cell proliferation as well as an inhibitor of platelet activation, on the surface of a porous polyurethane vascular prosthesis. The polyurethane graft material exhibited deterioration and was not considered to be suitable for use in permanent small-bore vascular grafts. On the other hand, an in vivo study involving a caprine model demonstrated patency of the dipyridamole-coated polyurethane grafts in three of eight cases; this finding suggested that dipyridamole promotes growth of an endothelial cell lining and positively influences the patency rate. Tiwari et al. (2002) covalently modified a poly(carbonate-urea)urethane graft with arginine–glycine–aspartate or/and heparin using spacer arm chemistry. These RGD-modified materials exhibited improved cell retention and provided antithrombogenic functionality. Although somewhat successful in limiting thrombosis and hyperplasia, these surface-modified synthetic grafts do not provide vascular responsiveness or other biochemical secretory functions seen in normal blood vessels. Like all synthetic grafts, these modified synthetic grafts are susceptible to colonization by Staphylococcus epidermidis or other micro-organisms; as described by O’Brien & Collin (1992), graft infection may result in inflammation, degradation of the artery–graft anastomosis or haemorrhage. In addition, synthetic grafts do not exhibit biochemical and mechanical properties similar to those of their natural counterparts; tissue ingrowth or calcification are common issues. Synthetic grafts also possess limited capabilities for responding to patient growth or other changes in patient anatomy over time; for example, replacement of grafts is commonly required in paediatric patients due to patient growth.

Owing to limitations of autologous and synthetic graft materials, several investigators have examined the fabrication of artificial blood vessels by means of tissue engineering. In tissue engineering, artificial tissues are fabricated by placing living cells within three-dimensional polymeric, ceramic or naturally derived materials known as scaffolds, which help guide cell proliferation. The cell-seeded scaffolds are placed in bioreactors, which enable cells to multiply within the scaffolds. The artificial tissues are subsequently implanted in the body, so that it can resume normal function. For example, Weinberg & Bell (1986) fabricated artificial blood vessels using cultured bovine endothelial cells, fibroblasts as well as smooth muscle cells. Fibroblasts were located on the outermost portion of the structure, smooth muscle cells were located in the wall of the structure and endothelial cells lined the lumen of the structure; strength was provided by collagen and polyethylene terephthalate scaffold materials.

Shinoka et al. (2001) fabricated a scaffold out of polycaprolactone–polylactic acid copolymer, which was reinforced using woven polyglycolic acid. Cells from the peripheral vein were cultured and expanded on this scaffold for 10 days. Radiographic examination showed that the vessel remained completely patent seven months after implantation. Shinoka et al. (1998) also described the seeding of cells derived from ovine arteries and veins onto polyglactin/polyglycolic acid tubular scaffolds. In vivo studies on these grafts were performed using an ovine model. Acellular polymer tubes sealed with fibrin glue developed progressive obstruction; on the other hand, the tissue-engineered grafts remained patent. The tissue-engineered grafts also exhibited an increase in diameter, which was attributed to the development of the endothelium and media layers. Niklason et al. (1999) developed vascular grafts with arbitrary lengths out of bovine endothelial cells and smooth muscle cells. The vascular grafts were prepared under pulsatile conditions and remained patent after implantation in a porcine model. Remy-Zolghadri et al. (2004) used an in vitro self-assembly method to create artificial small-diameter blood vessels. In this study, human fibroblasts were used to create decellularized extracellular matrix structure; endothelial cells were subsequently seeded on the interior of this structure. Xu et al. (2004) fabricated a nanofibrous scaffold by electrospinning aligned poly(l-lactid-co-ε-caprolactone) (75 : 25) copolymer. The scaffold contained aligned approximately 500 nm fibres; these structures promoted migration and alignment of smooth muscle cells.

Conventional techniques for producing artificial tissues involve cell seeding on pre-formed scaffolds; these techniques may not provide sufficient control over overall scaffold geometry or cell position within the scaffold. Control of overall scaffold geometry as well as scaffold macro-architecture would facilitate the development of patient-specific scaffolds, which would be especially useful for paediatric patients. According to Tan et al. (2003), scaffolds produced using conventional methods exhibit poor scaffold-to-scaffold consistency since conventional methods are sensitive to the manual skills and experiences of the operator. Since conventional scaffold fabrication methods demonstrate poor consistency and poor repeatability, they may not be suitable for commercial use. Novel scaffold fabrication approaches are required to address these issues. Naing et al. (2005) described the use of computer-aided design (CAD) software to design and fabricate customized scaffolds with well-controlled shapes and spatial arrangements. This approach, which Sun et al. (2004) refer to as computer-aided tissue engineering, would enable the fabrication of scaffolds that are not only readily reproducible but also consistent in microstructure. Several investigators have described the use of computer-aided tissue engineering for the fabrication of vascular tissues. For example, Sodian et al. (2002) used X-ray computed tomography data in order to prepare stereolithographic models of a human aortic root scaffold and a human pulmonary heart valve scaffold. These models were subsequently used to prepare biodegradable scaffolds out of poly-4-hydroxybutyrate and polyhydroxyoctanoate using a thermal process. Sodian et al. (2005) also used CAD software to design a replacement for a coarcted aortic segment. A biodegradable vascular scaffold for the descending aorta was subsequently fabricated using stereolithography. Schaefermeier et al. (2009) described the use of X-ray computed tomography data of a human aortic homograft for the fabrication of a silicone model. The silicone model was subsequently used to create a scaffold out of polyglycolic acid; the scaffold showed a deviation of only 3–4% in height, inner diameter and length when compared with human aortic homograft. Xu et al. (2008) used CAD software and a solid free form fabrication process to create three-dimensional micro-tunnels similar to human liver architecture. Structures were fabricated out of a biodegradation polyurethane based on polycaprolactone and poly(ethylene glycol). It should be noted that many of these computer-based tissue engineering processes are not suitable for fabricating scaffolds containing biological molecules, since the biological molecules may be irreversibly damaged by high temperatures or variations in pH during processing.

Laser-based processes exhibit several advantages over other computer-aided tissue engineering processes. Most importantly, laser-based processes may be used for modifying a wide variety of materials since they do not require harsh chemicals or heating of material. In addition, scaffolds with complex geometries that conform to a patient’s medical condition and anatomy can be fabricated. Laser-based processes can be performed under ambient conditions. It should be noted that lasers are a ubiquitous feature in a wide variety of clinical environments. Laser-based processes may readily be scaled up for commercial activities; for example, laser-based processes are commonly used for the fabrication of stents, prostheses and other medical devices. Laser machining has been used to fabricate stents out of tantalum, stainless steel, platinum alloys, niobium alloys and cobalt alloys; this process involves minimal heating of the stent or alteration of overall stent geometry. For example, Stoeckel et al. (2004) described the fabrication of self-expanding stents out of nickel–titanium shape memory alloy using laser machining.

Computer-aided laser micromachining may be used to create scaffolds for tissue engineering. For example, Miller et al. (2009) used ArF laser micromachining to pattern channels on silicon surfaces; human aortic vascular smooth muscle cells were subsequently seeded on these surfaces. The human aortic vascular smooth muscle cells grew on the non-ablated silicon substrate and formed contacts with the grooves. Cell patterning on silicon surfaces has potential applications in biosensors and other electronic devices. Patz et al. (2005) used an ArF laser to micromachine channels with 60–400 μm widths onto 2 per cent agarose surfaces. The surfaces of the channels were subsequently lined with basement membrane matrix solution. Matrigel created a hydrophobic environment within the channels. C2C12 murine myoblast cell suspensions were then placed within these micromachined channels. The murine myoblast cells were aligned parallel to the 60–150 μm channels; vertical stacking was also observed. Some of the myoblasts fused and differentiated into multinucleated myotubes. Live/dead assays showed that cell number, cell size and number of nuclei per cell increased over 72 h. Doraiswamy et al. (2006) performed a similar study involving growth of B35 murine neuronal cells within laser-micromachined channels. Channels were micromachined in 2 per cent agarose hydrogel using an ArF laser; these channels were filled with basement membrane matrix. Neuronal cells were placed in the micromachined channels; vertical stacking of daughter cells as well as differentiation of cells into multinucleated structures were observed. Between 72 and 96 h after cell seeding, increasing cell density caused the neuronal cells and basement membrane matrix to delaminate from the agarose substrate and form a free-standing structure. The dimensions of the free-standing structure corresponded with those of the laser-micromachined channel that was used to generate it. In a more recent study, Harris et al. (2008) demonstrated differential adherence and growth of A431 human epithelial carcinoma cells within micromachined agarose channels. The results from these studies indicate that laser micromachining is a versatile technique for processing artificial tissues with arbitrary geometries.

This study investigates the use of computer-aided laser micromachining to fabricate scaffolds for vascular tissue networks. Computer-aided laser micromachining was used to construct patterned surfaces in 1 per cent agarose hydrogel, which were used for differential adherence and growth of cells into vascular tissue networks. The results of this study indicate that computer-aided laser micromachining provides a unique approach for fabricating vascular tissues and other artificial tissues with complex geometries.

2. Material and methods

Liquid agarose was created by combining electrophoresis-grade agarose powder (Sigma Aldrich, St Louis, MO, USA) with water; the resulting solution was heated to 80°C. Agarose hydrogel was formed by pouring the liquid agarose into a Petri dish and allowing the liquid to solidify at room temperature. The agarose hydrogel-filled Petri dish was mounted on a motor-driven linear stage (Aerotech, Pittsburgh, PA, USA). The agarose hydrogel was micromachined with an ArF excimer laser (λ=193 nm, repetition rate = 10 Hz) using several laser energy densities and aperture sizes; a schematic of the computer-aided laser micromachining process is shown in figure 1. Fused silica biconvex lenses (L1 and L2) located in a telescopic arrangement were used to reduce the primary spot size of the laser. Apertures (A1 and A2) were used to further reduce laser spot size and enable laser alignment. Ultraviolet dielectric mirrors (M1 and M2) were used to reflect laser energy at a 45° incident angle. A 10× ultraviolet objective (O) was used to focus the laser beam to spot sizes as small as 5 μm. A charged-coupled device (CCD) camera was aligned with the mirrors and objective to monitor the computer-aided laser micromachining process. The micropositioning system provided X–Y–Z stage movement corresponding to CAD geometry; stage movement was synchronized with laser firing. Ablation of the agarose hydrogel was examined as a function of aperture and laser energy density. Channels with widths between 60 and 400 μm were prepared. A reservoir was created at the end of the channel in order to facilitate transport of material into the channel. Non-diluted Matrigel basement membrane matrix solution (American Type Culture Collection, Manassas, VA, USA) was micro-pipetted into the reservoirs; this material polymerized within the channels at room temperature.

Figure 1.

Schematic of the computer-aided laser micromachining process.

HAAE-1 human aortic endothelial cells and HA-VSMC human aortic smooth muscle cells were obtained from a commercial source (American Type Culture Collection). Cells were sub-cultured and incubated at 37°C, 5 per cent CO2 prior to seeding on the test surfaces. Matrigel basement membrane matrix (Becton Dickinson, Franklin Lakes, NJ, USA), HA587 human soluble elastin (Elastin Products, Owensville, MO, USA), vascular endothelial growth factor (Sigma Aldrich) and heparin (Sigma Aldrich) were obtained from commercial sources. In order to assess the effects of vascular endothelial growth factor and heparin on cell activity, HAAE-1 human aortic endothelial cells and HA-VSMC human aortic smooth muscle cells were seeded in 24-well culture plates (approx. 5000 cells per well) containing vascular endothelial growth factor (1, 10, 100 and 1000 ng ml−1) and heparin (1, 10, 100 and 1000 units ml−1) separately using unmodified media as a control. Four samples of unmodified media and modified media were examined in this study. At 24 h, the cells were exposed to media and MTT (3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl tetrazolium bromide) assays were performed. The cells were incubated in MTT medium (0.5 mg ml−1) for 4 h. The MTT assay, which was originally developed by Mossman (1983), is based on the reduction of a yellow tetrazolium salt to a purple formazan dye by succinnic dehydrogenase within mitochondria. Absorbance was determined using a UV/visible spectrophotometer at a wavelength of 550 nm (Multiskan RC Labsystems, Helsinki, Finland).

Cell proliferation on surfaces prepared using computer-aided laser micro- machining was observed with a Meiji 5500 inverted optical microscope (Meiji Techno, Santa Clara, CA, USA), which contained a digital still camera and an epifluorescence attachment. A live/dead cell staining kit (Invitrogen, Carlsbad, CA, USA) was used to examine cell viability. This two-colour fluorescence cell assay used two probes in order to simultaneously provide information on live and dead cells. Intracellular esterase activity of live cells led to enzymatic conversion of calcein AM to intensely fluorescent calcein, which was retained within the cells. On the other hand, ethidium homodimer entered cells with damaged membranes; this dye exhibited bright red fluorescence upon binding to nucleic acids.

3. Results and discussion

Agarose hydrogels were selected as substrates for computer-assisted laser micromachining of vascular tissue networks. As shown by Normand et al. (2000), agarose is a thermoreversible hydrogel that solidifies upon cooling below the ordering temperature of approximately 35°C. Agarose is commonly used in tissue engineering; Gruber et al. (2006) and Jain et al. (2006) have noted that it exhibits mechanical properties similar to those of soft tissue and exhibits good cell compatibility. In addition, agarose can be readily patterned into a desired shape. Gruber et al. (1997) indicated that agarose exhibits low rates of cell adhesion since it is hydrophilic and does not contain functional groups that interact with cell adhesion molecules. Computer-assisted laser micromachining of agarose hydrogel may be used to create differentially adherent surfaces for the formation of vascular networks. In a recent study, Harris et al. (2008) evaluated the laser ablation characteristics of 1 per cent agarose gel as a function of laser spot size and as a function of laser energy density. One per cent agarose hydrogel was micromachined using a laser spot size (D) of approximately 250 μm, a repetition rate (R) of 10 Hz and a feed rate of 500 μm s−1; these parameters allowed for approximately 20 per cent overlap of spots. The laser spot size may be varied by altering the aperture on the optical path between the laser and the objective. For each spot size, a feed rate that allowed for 20 per cent overlap was selected. The feed rate of the translation axes determined the extent of overlap between laser pulses for a given laser spot size and a given repetition rate. The overlap defined the roughness of the undercut. Greater than 50 per cent overlap resulted in deep trenches near the central portion of the channel; less than 10 per cent overlap resulted in small ripple structures along the channel. Figure 2a,b contains cross-sectional and top-view images, respectively, of 1 per cent agarose gel that was micromachined using several energy densities. Laser energy densities of 1.0–2.2 J cm−2 in 0.2 J cm−2 increments were examined. Figure 2c contains a graph that shows the mean ablation width and the mean ablation depth of agarose hydrogel at several laser energy densities. These data suggested that ablation depth increased linearly with laser energy density. The ablation width did not show a significant increase at energy densities between 1 and 2 J cm−2. An abrupt nonlinear increase was observed at higher laser energy densities; this behaviour was attributed to nonlinear laser–agarose interaction. Figure 3a contains an optical micrograph of laser-micromachined agarose channels that were prepared using several apertures. In this figure, apertures 1, 2 and 3 correspond to laser spot sizes of approximately 150, 600 and 1000 μm, respectively. Figure 3b shows the mean ablation width and the mean ablation depth for these apertures. Ablation width and ablation depth were shown to increase with aperture size. Structures with dimensions between approximately 200 and 1100 μm were formed by varying the aperture size and the laser intensity. Additional studies such as the one performed by Harris et al. (2008) are needed in order to better understand the laser ablation behaviour of commonly used scaffold materials.

Figure 2.

(a) Top-view optical micrograph of channels fabricated in 1% agarose hydrogel. (b) Cross-sectional optical micrograph of channels fabricated in 1% agarose hydrogel. The labels A (2.2 J cm−2), B (2 J cm−2), C (1.8 J cm−2), D (1.6 J cm−2), E (1.4 J cm−2), F (1.2 J cm−2) and G (1 J cm−2) represent the fluence values used for micromachining. (c) Plots showing width of agarose channel versus laser energy density and depth of agarose channel versus laser energy density. Data are expressed as mean ± standard deviation (dotted line with triangle, width; solid line with triangle, depth). Scale bar (a,b), 500 μm. Adapted from Harris et al. (2008) with permission from Elsevier.

Figure 3.

(a) Cross-sectional optical micrograph of channels that were prepared using several apertures (scale bar, 500 μm). (b) Plots showing width of agarose channel versus aperture size and depth of agarose channel versus aperture size (dotted line with triangle, width; solid line with triangle, depth). The labels 1 (150 μm), 2 (600 μm) and 3 (1000 μm) represent the aperture sizes that were used for micromachining. Data are expressed as mean ± standard deviation. Adapted from Harris et al. (2008) with permission from Elsevier.

In order to examine the fabrication of multilayered vascular networks, a concentric ring structure in which each ring was connected to an independent reservoir was fabricated. Each ring in the concentric ring structure represented a single layer within a layered vascular network. Such structures may enable placement of dissimilar growth factors within each ring of the concentric ring structure. Figure 4 contains an optical micrograph and a scanning electron micrograph of a two-ring structure that is connected to two 2×2 mm reservoirs; this structure was micromachined on the surface of a Si (111) wafer. The outer ring was 4 mm in diameter and the inner ring was 3.5 mm in diameter. Both rings exhibited thicknesses of approximately 250 μm. Figure 5 contains images from several stages of computer-aided laser micromachining of a single-ring structure. Figure 5a shows the design of the structure that was prepared using CAD software. Figure 5b shows an image of the structure processed on Si (111) using computer-aided laser micromachining. Figure 5c shows an image of the structure processed on agarose hydrogel using computer-aided laser micromachining. Figure 5d shows confluent HAAE-1 human aortic endothelial cell growth on an agarose hydrogel structure that was filled with Matrigel.

Figure 4.

(a) Optical micrograph of a two-circle ring. (b) Scanning electron micrograph of a two-circle ring at the channel/ring interface. Scale bar, 400 μm.

Figure 5.

Computer-aided laser micromachining process from the CAD to the cell-seeded structure. (a) Design of structure prepared using CAD software. (b) Image of structure processed on Si (111) using computer-aided laser micromachining. (c) Image of structure processed on agarose hydrogel using computer-aided laser micromachining. (d) Image of confluent HAAE-1 human aortic endothelial cell growth on structure that was processed on agarose hydrogel. The structure was modified using Matrigel prior to cell seeding. Scale bar, approximately 1 mm.

In previous work by Patz et al. (2005) and Doraiswamy et al. (2006), channels prepared using computer-aided laser micromachining were lined with Matrigel basement membrane matrix solution. Matrigel is derived from Engelbreth-Holm-Swarm sarcoma, which is a transplantable tumour in C57BL mice. Kleinman et al. (1982) noted that Matrigel contains a high concentration of extracellular matrix components, including collagen IV, heparan sulphate, laminin and proteoglycans. Matrigel also contains basic fibroblast growth factor, epidermal growth factor, insulin-like growth factor-1, platelet-derived growth factor and transforming growth factor-beta. Kleinman & Martin (2005) noted that Matrigel promotes outgrowth of differentiated cells as well as differentiation of many cell types; several cell types have been shown to undergo differentiation after being plated on Matrigel surfaces. For example, primary endothelial cells as well as endothelial cell lines exposed to Matrigel form tube-like structures. Kubota et al. (1988) demonstrated that human umbilical vein endothelial cells and human dermal microvascular endothelial cells form tubes within 1–2 h after exposure to Matrigel. Laminin and to a lesser extent collagen IV were shown to play roles in rapid differentiation and tube formation. Maley et al. (1995) demonstrated that primary skeletal muscle cultures from adult mice formed myotubes on Matrigel; Matrigel was shown to support large myotube formation better than collagen IV, entactin-free laminin or fibronectin alone. Previous work by Patz et al. (2005) and Doraiswamy et al. (2006) indicated that incorporating basement membrane matrix within laser-micromachined channels may enable controlled alignment of cells.

Wilkinson et al. (2002) noted that microscale patterns give physical cues to proliferating cells that guide cell position and alignment. For example, Harrison (1911) noted that embryonic ranine (frog) cells grown on spider web fibres demonstrated alignment along the fibres. More recently, several groups have previously investigated the use of microscale channels for guiding cell proliferation and cell elongation. For example, Lu et al. (2005) examined 5–15 μm deep and 10 μm wide grooves, which were fabricated out of poly(l-lactic acid) and polyethylene terephthalate. Rat skin fibroblasts and rat aortic smooth muscle cells proliferated within the channels and demonstrated alignment parallel to the channels. Shen et al. (2006) demonstrated that ultraviolet micro-embossing could be used to produce 40–160 μm wide microchannels. C2C12 cells in the 40 μm channel were shown to merge together and create myofiber structures.

Biological molecules may be incorporated within the channels in order to either promote or inhibit cell proliferation. Pieper et al. (2002) described the use of growth factors to increase in vivo angiogenesis. Ribatti et al. (2006) noted that growth factor activity involves two steps, namely: (i) dissociation of the growth factors from matrix binding sites and (ii) subsequent activation through either conformational changes or dissociation from latency proteins. For example, Koch et al. (1994) and Conway et al. (2001) noted that vascular endothelial growth factor induces migration of endothelial cells, proliferation of endothelial cells as well as tube formation. According to Nillesen et al. (2007), vascular endothelial growth factor stimulates angiogenesis by increasing cell production of matrix metalloproteinases. Asahara et al. (1995) showed that in addition to stimulation of endothelial cell proliferation, vascular endothelial growth factor also indirectly inhibits smooth muscle cell proliferation. Peirce et al. (2004) demonstrated focal delivery of vascular endothelial growth factor-164 and angiopoietin-1 using alginate microbead delivery devices in a murine model. Their work indicated that the formation of vascular network patterns may be correlated with spatially and temporally controlled delivery of vascular endothelial growth factor-164 and angiopoietin-1; for example, increases in the length and number of blood vessels smaller than 25 μm were observed within a 1 mm radial distance of vascular endothelial growth factor-164 delivery devices. Kelm et al. (2007) described the activity of vascular endothelial growth factor as being extremely dependent on dose.

Work by several investigators has indicated that heparin provides non-selective inhibition of aortic smooth muscle cell proliferation. Clowes & Karnovsky (1977) described diminished smooth muscle cell proliferation in response to heparin. Heparin was shown to suppress hyperplasia of smooth muscle cells in a murine model. Several investigators, including Regelson (1968) as well as Heilbrunn & Wilson (1949), have suggested that heparin directly acts on smooth muscle cells. More recent work by Weatherford et al. (1996) demonstrated by means of in vitro assays that heparin and vascular endothelial growth factor together inhibited proliferation of human aortic smooth muscle cells as well as stimulated proliferation of human aortic endothelial cells; fibrin glue or another biological glue containing both biological molecules may improve patency of vascular grafts by selectively promoting endothelial cell proliferation while inhibiting smooth muscle cell proliferation.

Vascular endothelial growth factor and heparin were selected to develop differential adherent interfaces for growth of HAAE-1 human aortic endothelial cells and HA-VSMC human aortic smooth muscle cells. Proliferation of HAAE-1 human aortic endothelial cells and HA-VSMC human aortic smooth muscle cells was examined using vascular endothelial growth factor in concentrations of 1, 10, 100 and 1000 ng ml−1 and heparin in concentrations of 1, 10, 100 and 1000 units ml−1. The MTT cell viability data were initially analysed between several concentrations of vascular endothelial growth factor or heparin for each cell type. The MTT cell viability data were subsequently analysed between the HAAE-1 human aortic endothelial cells and the HA-VSMC human aortic smooth muscle cells at each concentration. Figure 6 contains viability data for HAAE-1 human aortic endothelial cell exposure and HA-VSMC human aortic smooth muscle cell exposure to vascular endothelial growth factor in concentrations of 1, 10, 100 and 1000 ng ml−1. These data suggested that higher concentrations of vascular endothelial growth factor reduced viability of HAAE-1 human aortic endothelial cells and HA-VSMC human aortic smooth muscle cells in comparison with unmodified media. Viability of HAAE-1 human aortic endothelial cells and HA-VSMC human aortic smooth muscle cells was significantly higher for cells exposed to media containing 1 ng ml−1 of vascular endothelial growth factor than for cells exposed to media containing 10, 100 or 1000 ng ml−1 of vascular endothelial growth factor (p<0.05). There was no significant difference observed in the viability of cells exposed to 10, 100 or 1000 ng ml−1 of vascular endothelial growth factor (p<0.05). When comparing the two cell types, viability of HAAE-1 human aortic endothelial cells was significantly higher than that of HA-VSMC human aortic smooth muscle cells for vascular endothelial growth factor concentrations of 1, 10, 100 and 1000 ng ml−1. These results suggest that vascular endothelial growth factor induced proliferation of HAAE-1 human aortic endothelial cells and inhibited proliferation of HA-VSMC human aortic smooth muscle cells. The difference in mean viability between HAAE-1 human aortic endothelial cells and HA-VSMC human aortic smooth muscle cells was found to be greatest at a vascular endothelial growth factor concentration of 1 ng ml−1. Figure 7 contains viability data for HAAE-1 human aortic endothelial cell and HA-VSMC human aortic smooth muscle cell exposure to heparin at concentrations of 1, 10, 100 and 1000 units ml−1.

Figure 6.

MTT cell viability of HA-VSMC human aortic smooth muscle cells and HAAE-1 human aortic endothelial cells for various amounts of vascular endothelial growth factor (1–1000 ng ml−1); these data were normalized with cell viability on unmodified media. Data represented as mean ± standard deviation. Black bars, HA-VSMC; grey bars, HAAE-1.

Figure 7.

MTT cell viability of HA-VSMC human aortic smooth muscle cells and HAAE-1 human aortic endothelial cells for various amounts of heparin (1–1000 units ml−1); these data were normalized with cell viability on unmodified media. Data represented as mean ± standard deviation. Black bars, HA-VSMC; grey bars, HAAE-1.

MTT viability assays suggested that media containing heparin reduces viability of HAAE-1 human aortic endothelial cells and HA-VSMC human aortic smooth muscle cells in comparison with unmodified media. Media containing higher concentrations of heparin exhibited reduced HAAE-1 human aortic endothelial cell viability and HA-VSMC human aortic smooth muscle cell viability. Viability of HAAE-1 human aortic endothelial cells and HA-VSMC human aortic smooth muscle cells was significantly higher for cells exposed to media containing 1 unit ml−1 of heparin than for cells exposed to media containing 10, 100 and 1000 units ml−1 of heparin. There was no significant difference between viability of cells exposed to media containing 10 units ml−1 of heparin and viability of cells exposed to media containing 100 units ml−1 of heparin. Viability of HAAE-1 human aortic endothelial cells and HA-VSMC human aortic smooth muscle cells exposed to 10 and 100 units ml−1 of heparin was significantly higher than that of cells exposed to 1000 units ml−1 of heparin. When comparing the two cell types, the viability of HAAE-1 human aortic endothelial cells was significantly higher than the viability of HA-VSMC human aortic smooth muscle cells for media containing 1, 10 and 100 units ml−1 of heparin. There was no significant difference between viability of HAAE-1 human aortic endothelial cells exposed to media containing 1000 units ml−1 of heparin and viability of HA-VSMC human aortic smooth muscle cells exposed to media containing 1000 units ml−1 of heparin. The difference in viability between HAAE-1 human aortic endothelial cells and HA-VSMC human aortic smooth muscle cells was found to be greatest for media containing 100 units ml−1 of heparin. These results indicate that heparin may be used to increase proliferation of HAAE-1 human aortic endothelial cells and inhibit proliferation of HA-VSMC human aortic smooth muscle cells within artificial vascular tissues. Figure 8 shows MTT viability of HA-VSMC human aortic smooth muscle cells and HAAE-1 human aortic endothelial cells in Matrigel as well as in Matrigel containing 10 ng ml−1 of vascular endothelial growth factor and 100 units ml−1 of heparin; data for the test materials were normalized with data for the unmodified media. Viability of HA-VSMC human aortic smooth muscle cells reduced significantly and viability of HAAE-1 human aortic endothelial cells increased significantly in Matrigel containing 10 ng ml−1 of vascular endothelial growth factor and 100 units ml−1 of heparin as compared with unmodified Matrigel. Figure 9 contains optical micrographs of HA-VSMC human aortic smooth muscle cells and HAAE-1 human aortic endothelial cells in Matrigel containing 10 ng ml−1 of vascular endothelial growth factor and 100 units ml−1 of heparin. These results confirmed work by Weatherford et al. (1996), which indicated that the combination of vascular endothelial growth factor and heparin stimulated proliferation of aortic endothelial cells and inhibited proliferation of aortic smooth muscle cells. Matrigel containing 10 ng ml−1 of vascular endothelial growth factor as well as 100 units ml−1 of heparin was used to promote proliferation of HAAE-1 human aortic endothelial cells and inhibit proliferation of HA-VSMC human aortic smooth muscle cells within the vascular network.

Figure 8.

MTT cell viability of HA-VSMC human aortic smooth muscle cells and HAAE-1 human aortic endothelial cells for Matrigel as well as Matrigel containing 10 ng ml−1 vascular endothelial growth factor (VEGF) and 100 units ml−1 heparin; these data were normalized with cell viability on unmodified media. Data represented as mean ± standard deviation. Black bars, HA-VSMC; grey bars, HAAE-1.

Figure 9.

(a) Optical micrograph of HA-VSMC human aortic smooth muscle cells grown in Matrigel containing 10 ng ml−1 vascular endothelial growth factor and 100 units ml−1 heparin. (b) Optical micrograph of HAAE-1 human aortic endothelial cells grown in Matrigel containing 10 ng ml−1 vascular endothelial growth factor and 100 units ml−1 heparin. Image was obtained 24 h after cell seeding. Scale bar, 100 μm.

HAAE-1 human aortic endothelial cells and HA-VSMC human aortic smooth muscle cells were seeded on concentric three-ring structures that were fabricated on agarose hydrogel substrates. In these structures, the inner ring contained human aortic endothelial cells, the middle ring contained HA587 human elastin and the outer ring contained human aortic vascular smooth muscle cells. Figure 10 contains an optical micrograph of an agarose channel containing human elastin; in this figure, the elastin appears with a dark purple colour. Between 6 and 9 h after cell seeding, the cells remained attached to the filled channels and did not adhere to other regions of the agarose hydrogel surface. Between 12 and 24 h after cell seeding, the cells continued to proliferate within the filled channels; increased cell coverage within the channels was observed. Confluence of cells within the channels was observed 48 h after cell seeding. Figure 11a contains an optical micrograph of an HA-VSMC human aortic smooth muscle cell–Matrigel network that was obtained 72 h after cell seeding. Confluence of cells is observed in this figure. Figure 11b contains an image of a partially delaminated HA-VSMC human aortic smooth muscle cell–Matrigel network. The cells delaminated and formed a free-standing vascular network 72 h after cell seeding. The channel created by means of computer-aided laser micromachining is visible in this figure. The cells were also examined in 50 : 50 co-culture with unmodified media.

Figure 10.

Optical micrograph of a stained agarose channel that was filled with HA587 human elastin. The elastin was stained with a purple colour. Scale bar, 250 μm.

Figure 11.

(a) Optical micrograph of a fully grown HA-VSMC human aortic smooth muscle cell–Matrigel network. Image was obtained 72 h after cell seeding. Scale bar, approximately 2 mm. (b) Optical micrograph of a delaminated HA-VSMC human aortic smooth muscle cell–Matrigel network. Image was obtained 72 h after cell seeding. Scale bar, approximately 250 μm.

Figure 12a contains an optical micrograph of a network containing HAAE-1 human aortic endothelial cells, HA-VSMC human aortic smooth muscle cells, human elastin, as well as Matrigel containing 10 ng ml−1 of vascular endothelial growth factor and 100 units ml−1 of heparin. Figure 12b contains a live/dead stained image of an HAAE-1 human aortic endothelial cell network that was obtained 48 h after cell seeding. Figure 13 contains a live/dead image of HA-VSMC human aortic smooth muscle cells and HAAE-1 human aortic endothelial cells within a vascular network. HAAE-1 human aortic endothelial cells in the inner ring, which were grown on Matrigel containing 10 ng ml−1 vascular endothelial growth factor and 100 units ml−1 heparin, demonstrated confluence 48 h after cell seeding. Proliferation of HA-VSMC human aortic smooth muscle cells was observed on the outer layer. Figure 13c shows a live/dead image of HA-VSMC human aortic smooth muscle cells and HAAE-1 human aortic endothelial cells within the vascular network. The green colour observed in the live/dead stained cells indicates plasma membrane integrity as well as intracellular esterase activity. A few red-coloured cells indicating cell membrane damage were observed in vascular networks that were prepared in 50 : 50 co-culture with unmodified media. Live/dead analysis of the vascular networks showed greater than 99 per cent cell viability 72 h after cell seeding. The vascular network delaminated from the agarose hydrogel template and formed a free-standing structure at that time. It should be noted that the aspect ratio of the delaminated free-standing structure corresponded with that of the agarose hydrogel template. The free-standing vascular network was sustained in media for 96 additional hours.

Figure 12.

(a) Optical micrograph of vascular network containing HAAE-1 human aortic endothelial cells, HA-VSMC human aortic smooth muscle cells, human elastin as well as Matrigel containing 10 ng ml−1 of vascular endothelial growth factor and 100 units ml−1 of heparin. Image was obtained 72 h after cell seeding. (b) Live/dead stained optical micrograph of HAAE-1 human aortic endothelial cell network. Image was obtained 48 h after cell seeding. Scale bar, approximately 250 μm.

Figure 13.

(a) Live/dead stained fluorescent micrograph of vascular network containing HA-VSMC human aortic smooth muscle cells and Matrigel. (b) Live/dead stained fluorescent micrograph of vascular network containing HAAE-1 human aortic endothelial cells and Matrigel. (c) Live/dead stained fluorescent micrograph of vascular network containing HA-VSMC human aortic smooth muscle cells, HAAE-1 human aortic endothelial cells, Matrigel and human elastin. Images were obtained 48 h after cell seeding. Scale bar, approximately 200 μm.

4. Conclusions

Computer-aided laser micromachining provides a unique approach for fabricating tissue engineering scaffolds with complex geometries. Computer-aided laser micromachining enables small amounts of material to be removed from bulk material in a controlled manner. In addition, it enables a structure with a desired shape to be produced; scaffold-to-scaffold variation is minimized in this automated process. In this work, computer-aided laser micromachining was used to construct patterned surfaces in agarose or in silicon, which were used for differential adherence and growth of cells into vascular tissue networks. Vascular endothelial growth factor and heparin were incorporated within basement membrane matrix in order to promote growth of aortic cells into vascular tissue networks. Computer-aided laser micromachining is suitable for translation into clinical use since it is a readily scalable process that is commonly used for modifying stents as well as other medical devices. Computer-aided laser micromachining may benefit from decreases in minimum feature size as well as improvements in materials selection. Reductions in the minimum feature sizes offered by computer-aided laser micromachining will broaden the use of this technique. In addition, it should be noted that laser machining using nanosecond excimer lasers is limited to ‘line-of-sight’ processing of material; ablation of material cannot take place at locations where the laser path is obstructed. Some geometries (e.g. structures with overhangs) cannot be fabricated by means of nanosecond laser micromachining. Femtosecond lasers may be used when processing structures with smaller feature sizes, as well as structures with interior features. Farson et al. (2006) described the use of femtosecond laser micromachining in combination with gas-assisted material removal in order to prepare internal channels with diameters between 2 μm and 20 μm. In addition, the effects of laser–scaffold material interaction must be more fully evaluated. For example, studies are needed to compare the biological, chemical and mechanical properties of laser-micromachined scaffold materials with those of as-prepared scaffold materials. It is anticipated that computer-aided laser micromachining and other computer-aided processes will enable the development of tissue engineering scaffolds with unique geometries and other patient-specific attributes.

Footnotes

References

View Abstract