An automated fabrication strategy to create patterned tubular architectures at cell and tissue scales

The use of materials to impose tissue-like architecture at cell resolution will be important if engineered functional replacements for damaged cardiovascular, pulmonary, renal or digestive tissues are to be authentically engineered. Here, we demonstrate a coordinated system for the fabrication and subsequent culture of tubular tissues composed of multiple layers, cell-types and materials with physiological dimensions and defined architectures at cell resolution. We developed an automated tube fabricator that rolls 2D-matrices into 3D-tubular constructs directly from cells, hydrogels and scaffold biomaterials. Coordinated use of surface modification strategies allows 2D cell sheets and cell/biomaterial composites (i.e. hydrogels or electrospun scaffolds) to be fabricated which may be transferred into a perfusion bioreactor in a rapid and standardized procedure. To exemplify our strategy we fabricated structures resembling human mammary artery and gut; these can be imaged in situ and real-time electrical resistance measurements performed of the vessel walls, allowing non-invasive assessment of viability and functionality. Our system allows patterning at cellular resolution with variable tissue thickness, length, luminal diameter, and constituent biomaterial. This inherent flexibility will allow the recapitulation of the complex hierarchical biological architectures and generate functionality found natively in vivo.


Introduction
Various obstacles hinder tissue engineered vessel development including recreating the substantial mechanical strength and biological activity in situ grafts require on transplant. Furthermore these conduits need to be immunologically inert and nonthrombotic for successful long-term post-transplantation patency and for clinical adoption (Kakisis et al 2005, Baguneid et al 2006. Fundamental construct fabrication considerations include; dimension standardization, and for biologically active architectures, factors such as seeding, positioning and patterning of cells within different layers throughout tissues. To address these, various methodologies have been applied including synthetic scaffolds, biocompatible natural scaffolds, cell sheet technologies, automated (McAllister and L'Heureux 2006) or self-assembly of vessels (Nerem and Seliktar 2001, Kakisis et al 2005, Baguneid et al 2006, Kubo et al 2007, Williams et al 2009, Peck et al 2012. Some systems have shown good patency in animal models (Niklason et al 1999, Shum-Tim et al 1999, Kim et al 2008, Wang et al 2013, with few progressing to human trials (Shin'oka et al 2001, McAllister et al 2009. This limited success highlights the necessity to improve graft fabrication processes and facilitate the use of multi-phase materials to create tissues (Baguneid et al 2006, Peck et al 2012. Multilayer tubes have been engineered using various approaches including manually rolling cell sheets around a mandrel (L'Heureux et al 1998, Seliktar et al 2000, tubular constructs requiring external structural supports to prevent tube disassembly (Papenburg et al 2009), cell sheet rolling technologies (Kubo et al 2007), or self-rolling stretched synthetic materials (Yuan et al 2012). Manual rolling with or without additional external support is technically complex and requires lengthy tissue maturation, whilst the presence of stabilizing materials prevents usefulness in downstream applications. Self-rolling materials are generally not biocompatible for longterm viability, and the mechanical forces placed on cells may directly affect cell behaviour (Engler et al 2006). The application of cell sheet patterning technologies allows precise positioning of cells within a 2D layer (Williams et al 2009), however the only previous example of an automated tube fabrication strategy required sheets to be suspended in solution, preventing controllable sheet orientation (Kubo et al 2007). Here, we focused on devising a methodology to fabricate example tissues with these attributes.

Materials
Materials were purchased from Sigma-Aldrich (Dorset, UK) unless stated otherwise. NIH-3t3 mouse embryonic fibroblasts (3T3s), human umbilical vein endothelial cells (HUVECs), BJ6 fibroblast cells and CACO-2 cells were obtained from the LGC Standards (Middlesex, UK). Primary human airway smooth muscle cells (SMCs) were isolated from bronchial biopsies at the Glenfield Hospital (Leicester, UK) as described previously. (Kaur et al 2006) The research was approved by the Leicestershire Ethics Committee, and patients gave their written informed consent. HUVEC media was supplied by Promocell (Heidelberg, Germany). DMEM media, rat tail collagen I, mouse monoclonal anti-E-cadherin, and rhodamineconjugated secondary antibody were obtained from Invitrogen Life Technologies (Paisley, UK). Lentiviral labelling of cell lines was performed as described previously (Dixon et al 2011, Paik et al 2012. Cells were infected with enhanced green fluorescent protein (GFP)-or monomeric red fluorescent protein (mRFP)-labelled lentivirus at confluence and selected with puromycin for seven days which produced >95% labelled cells confirmed by flow cytometry.

Preparation of collagen cell sheets
Collagen sheets (2 mg ml −1 ; 25 × 15 mm 2 ) loaded with cells (3t3s, 2 × 10 6 /sheet) were cross-linked within filter paper frames for 1 h at 37°C and were cultured until confluency (three days). Filter paper frames were used to transfer sheets to the automated tube fabricator (ATF) and the frame excised before fabrication.
Preparation of thermo-responsive surfaces and monolayer or micro-patterned cell sheets Non-treated tissue culture plastic was coated with 5 μl cm −2 of 5% PNIPAAm (w/v) in distilled water using an EClA500 HP-CS gravity-fed air-brush at 20 pounds per square inch (psi) (The Airbrush Company Ltd, UK). For grafting these were heated to 65°C for 12 h. 5 μl cm −2 of 100% FCS was deposited on PNIPAAm-coated dishes by air brushing coating the full surface for monolayers or through micropatterned stencils (Tannlin Ltd, UK) for patterning cells (Paik et al 2012). Cells (6 × 10 6 cells) were seeded onto surfaces or micropatterns in media without FCS or Ca 2+ . For monolayers after confluency was achieved alginate-wetted (1.2% w/v) filter paper frames were placed onto the monolayer, the alginate cross-linked with 135 mM CaCl 2 to adhere to the underlying outer portion of the monolayer and the sheet detached through reduction in temperature to 4°C and gentle agitation. For patterns alginate (1.2% w/v) was used to overlay the pattern within a filter paper frame, the alginate cross-linked and detached as above. Filter paper frames were used to transfer sheets to the ATF and the frame excised before fabrication.

Electrospun scaffold fabrication
Electrospun scaffolds were produced in a vented chemical fume hood at room temperature using 10% polyethylene terephthalate (PET) (w/v) (foods grade drinking bottle quality) dissolved in a 1:1 trifluoroacetic acid:di-chloromethane (Fisher Chemicals, Loughborough, UK). Polymer solution was loaded into a syringe attached to a blunt 23 gauge (G) needle (BD Falcon™, Oxford, UK) and placed in a pump-driver (Harvard Apparatus Ltd, Kent, UK) extruding at 0.5 ml hr −1 . A 15 kV voltage was applied and nanofibres were collected on a steel collector plate positioned 15 cm from the needle tip. Scaffolds were attached to perspex frames and sterilized by UV-irradiation for 30 min. For seeding, scaffolds were soaked in media prior to CACO-2 cell seeding (1 × 10 6 Caco-2 cells/ 25 × 15 mm 2 scaffold). Perspex frames were used to transfer sheets to the ATF and the scaffold was excised before fabrication.

Fabricating vessels
The ATF (developed and manufactured by the Medical Engineering Unit, University of Nottingham) consists of stepper-motor driven modules, a mandrel rotation module driven by two inversely rotating facing stepper-motors and a sample delivery module, and a stage to deliver the material to the mandrel rotation module. Distance between the stage and mandrel is manually (or automatically) controlled to bring the sample in contact with the mandrel. Using the programmable system the automated fabrication was initiated and the mandrel rotation module rotated (0.1-1 rotations/ second) to draw up the sample. This action occurs simultaneously as the stage advancing; the stage covering the same distance as the circumference of the mandrel at the same rate as the mandrel rotation. The fabricated tube (directly integrated into the tube holder mounts) was transferred into the jig. For fibrin hydrogel bonding; 20 μl cm −2 of 16 mg ml −1 human fibrinogen was incubated with cell sheets, hydrogels or biomaterial for 5 min at 25°C. 20 μl cm −2 of 200 U ml −1 human thrombin was either airbrushed onto the tube mounts or incubated with the previously fabricated layer for 5 min at 25°C. Bonding of the tube mounts and the first layer was allowed to take place for 5 min before fabrication.
Transepithelial electrical resistance (TEER) measurements with calcium depletion CACO-2 populated tubular constructs connected to the bioreactor system had electrodes inserted within the tube lumen and the perfusion chamber that were connected to an EVOM 2 Volt-Ohm-metre (World Precision Instruments, Hitchin, UK). Background conductance was removed by subtracting acellular scaffold readings. Media was replaced with calcium reducing buffer (CRB) (Dixon et al 2014) for 5 min at room temperature which was then replaced by growth media (which contains 2 mM CaCl 2 ). TEER readings were taken periodically over a 24 h time period and are expressed as a percentage of the initial TEER reading (Ω/Ω 0 )cm −2 .
Dextran permeability CACO-2 populated tubular constructs were incubated in serum-free media at 37°C for 1 h. A solution containing both 4 kDa FITC-tagged dextran and 70 kDa-rhodamine-tagged dextran (1 mg ml −1 ) was added to the tube lumen. Media in the perfusion chamber was replaced with CRB for 5 min at room temperature before replacing with normal media. Samples were collected from the perfusion chamber over a 2 h period post-CRB application and fluorescence quantified on a plate reader (excitation 520 nm/ emission 590 nm) (Tecan Infinite M200, Reading, UK).

Immunocytochemistry
Samples were calcium depleted for 5 min before immediate fixation and permeabilization using 100% methanol at −20°C. Non-specific antibody binding was reduced by incubation in 3% (w/v) bovine serum albumin solution. Samples were incubated with anti-E-cadherin antibody overnight at 4°C, with protein expression visualized with species-appropriate secondary antibody.

Construct visualization
Composite tubular formations were visualized using a Nikon Eclipse TS100 Microscope (Nikon Instruments, Surrey, UK). Entire surface micropatterns images were captured by a Nikon SMZ 1500 microscope with a SPOT insight camera. To view 3D tubes, and E-cadherin staining a TCS LSI super zoom confocal microscope was used (Leica Microsystems, Milton Keynes, UK).

Data and statistical analysis
Data are presented for cells cultured on at least three separate occasions and are expressed as mean ±SEM. Data were analyzed (GraphPad Prism, San Diego, CA) using unpaired T-test.

Results
Development of the ATF To produce viable, multi-layered, 3D-tubular architectures from 2D-cellular sheets, we developed the ATF to perform pre-determined automated tube rolling protocols (figures 1(A) and (B)). We have developed this technology to be unrestricted as to the material used for fabrication and to allow patterned materials to be employed in an orientated fashion. The device is integrated into a sterile culture hood and consists of three stepper-motor driven modules: (i) a 'mandrel rotation' module driven by two inversely rotating facing stepper-motors, (ii) a 'sample delivery' module; a stage to deliver the material to the mandrel rotation module, and (iii) a 'transfer module'; a system to transfer the construct to a compatible perfusion bioreactor. The fabrication process is complete within minutes and can rapidly reset to build multi-layer, multi-phase or multi-material architectures. The distance between the stage and the mandrel is automatically or manually controlled to bring the 2D-material in contact with the mandrel (figure 1(A), N°10). Fabrication is remotely initiated and the mandrel rotates at the same speed as the delivery stage (programmed from 0.1-1 rotations/second) to draw up and roll the sample. If required, thrombin/fibrinogen bonding can be employed to integrate multiplephases; the mandrel coated with thrombin and the cell sheet/material treated with fibrinogen solution (figure 1(C)). Contact between the thrombin (mandrel) and the fibrinogen (sheet/material) creates a fibrin hydrogel which securely bonds the surfaces. During fabrication, fibrin forms cross-links between adjacent layers and creates a fully adhered structure. Fabricated tubes are securely integrated into tube holder mounts ( figure 1(A), N°4) and transferred to sterile culture jigs using the rotating tube transfer Yarm ( figure 1(A), N°3). This can then be installed in the complementary bioreactor system ( figure 3(A)), ensuring no further tube manipulation is required for effective construct perfusion. (Tube fabrication demonstrated in video S1.) If necessary the fabricated vessels can be detached from the bioreactor manually which is not prevented by using fibrin within the structure.

Fabrication of vessels with varying dimensions and phases
Construct dimensions can be manipulated by varying different parameters: wall thickness is determined by the 2D-material thickness and/or number of rotations around the mandrel (5 μm->1 cm). The length can be adjusted by repositioning the mandrel rotation motors (figure 1(B); purple arrows) and/or varying cell-sheet width (<1 mm-5 cm). Here we demonstrated that short vessels of several centimetres, but clinically relevant lengths (15-60 cm long), could be fabricated by modifying the motor distance and using further cell sheets to bond the junctions between several tube sections. Luminal diameter is varied using differently sized holder mounts (100 μm-5 cm). This flexibility ensures constructs possess dimensions comparable to various human tubular tissues. As examples, using a PET electrospun scaffold and alginate composite, we fabricated single-phase tubular architectures with dimensions applicable to grafts from the aorta, gut, trachea, medium diameter arteries and urethra ( figure 1(D)).
Multi-phase constructs are fabricated by sequentially rolling cellular-sheets/materials. A tri-phasic tubular structure consisting of a single cell monolayer and two cell-seeded collagen gels was fabricated (figure 2(A)) as an exemplar. Initially a GFP-labelled NIH3T3 (GFP-3t3s) cell monolayer was rolled around the mandrel (1.05 rotations, 5 μm layer thickness), followed by an mRFP-(mRFP-3t3s) and a GFP-3t3s seeded collagen gel (both 1.05 rotations, ∼250 μm layer thickness) (figures 2(A)-(E)). Cell monolayers were created by surface modification of non-tissue culturetreated surfaces airbrushed with the temperature-sensitive polymer poly(N-isoproplacrylamide) (PNI-PAAm) and FCS prior to cell culture to confluence (Williams et al 2009). PNIPAAm grafting was conducted using temperature-dependent absorption (65°C for 12 h). This was chosen over other techniques such as UV-or microwave-radiation due to the ease of processing samples. For simplicity and before temperature-sensitive detatchment, cell monolayers were cross-linked to an outer paper frame coated in alginate. This suspends the monolayer between the support allowing it to be easily transferred to the ATF for fabrication (figure S1). A temperature reduction (4°C) allows the harvest of a complete cell monolayer sheet (figure S2A) without damaging cell-deposited extracellular matrix (Shimizu et al 2001(Shimizu et al , 2006. We analyzed the complete fabricated tubes demonstrating layer integrity was maintained, with no significant cell migration into adjacent phases (figure 2(D): hydrogel to hydrogel; figure 2(E): monolayer to hydrogel) after seven days of culture. Importantly however layers were integrated with cells at the phase surfaces in direct contact (figure 2(D); white arrow show integration, black arrows show phase interface).

Creation of patterned architectures
To demonstrate the utility of our technology interand intra-layer cell patterning (with micrometre resolution over an extended distance) within the tubular architectures was achieved by adapting our previous high-resolution cell patterning technology (figure S2A) (Paik et al 2012). We employed micropatterned stencils and airbrushing to generate 50 μm wide lines of PNIPAAm interspersed with 50 μm intervals on non-tissue culture plastic surfaces which we used to pattern GFP-3T3s (figure 2(F)). We chose such large spacing between patterned lines to help visualization; however smaller gaps could be used to create orientated sheets rather than simple micropatterns. Patterns could be transferred by overlaying and detaching alginate-cross-linked sheets. These sheets maintained cellular patterning and were fabricated into vessels using the ATF (figure 2(G)). In many tubular tissues luminal capacity is controlled by the surrounding SMCs which orientate uni-axially to ensure co-ordinated muscle contraction (Dobrin 1978). We therefore used this patterning to demonstrate effective cell alignment that can be directed into a helical orientation (evident in the larger airway bronchioles and small resistance arteries) (Jeffery 2001, Rhodin 1980 or herringbone alternating orientations (found in larger elastic arteries which maintain wall structural integrity) (Rhodin 1980). The herringbone motif was recapitulated using micropatterning and ATF sheet fabrication; GFP-3T3 and mRFP-3T3 micropatterned cell sheets at 60°(mRFP-3T3) and 120°(GFP-3T3) from horizontal respectively (figure S3; example orientation). These two layers were fabricated sequentially to produce a 3Dtubular architecture with a herringbone motif found in native blood vessels (Rhodin 1980  Fluorescently-labelled human endothelial (HUVEC-GFP), smooth muscle cells (SMC-mRFP) and fibroblasts (BJ6-GFP) were sequentially rolled to fabricate a tube mimicking muscular vessel architecture. The vessel was fabricated to create a sectioned structure to aid visualization. Bar is 2 mm.

Fabrication of an arterial biomimetic
We recapitulated a complete vascular conduit comprising of four cell layers to mimic the individual regions found within a blood vessel wall (Rhodin 1980 (Dixon et al 2011). Initially, individual 2D-sheets of GFP-HUVEC cells (3 × 10 6 cells), two aligned sheets of mRFP-SMC (1 × 10 6 cells), and a sheet of GFP-BJ6 cells (3 × 10 6 cells) were cultured as 2D-sheets for two days before sequential cell-sheet rolling and tube imaging ( figure 2(J)). The two SMC layers were rolled in the same orientation to allow visualization of the individual cell layers.
Creation of tubular tissues with gut-like barrier function Electrospun scaffolds are another biomaterial extensively used in tissue engineering that can mimic the 3D nano-topography of basement membranes on which epithelial cells reside in tubular architectures (Teo and Ramakrishna 2006, Booth et  The construct is held in position on culture jigs and is continually perfused using a peristaltic pump. Tube is shown with electrodes connected to an EVOM 2 machine to allowing measurements of real-time electrical resistance (TEER); amplitude (I) and voltage (V) are measured within the lumen (V2 and I2) and externally to the tube within the perfusion chamber (I1 and V1). (B) Photograph of construct secured in perfusion chamber with electrodes inserted within lumen. Bar is 10 mm. (C) Scanning electron microscope image of an acellular PET electrospun scaffold. Bar is 10 μm. (D) Immunostaining of E-cadherin (red) in a confluent layer of CACO-2 cells on the PET electrospun scaffold before Ca 2+ depletion, 5 and 20 min post-Ca 2+ depletion (left to right). Bar is 60 μm. (E) Electrical resistance through CACO-2 tube after Ca 2+ depletion. CACO-2-populated PET electrospun scaffold tube was fabricated and loaded into the perfusion bioreactor with electrodes placed as shown in (A). Ca 2+ was depleted for 5 min before replacement in normal media. Electrical resistance was measured periodically over a 24 h period. (F) Para-cellular permeability through a CACO-2 tube. Acellular and CACO-2 populated electrospun tubes were placed in perfusion bioreactor. Serum-free media containing both 4 kDa FITC-tagged and 70 kDa rhodamine-tagged dextran was perfused through the tube lumen. Samples from the perfusion chamber were taken over a 2 h period, and fluorescence expressed as a percentage of total initial fluorescence. (G) Para-cellular permeability through a CACO-2 tube pre-and post-Ca 2+ depletion. electrospun scaffold was created ( figure 3(C)). The epithelium's primary function is to provide a physical barrier between adjacent environments whilst allowing a controlled para-cellular flow of nutrients and ions, and maintaining cellular polarization and structural integrity (Mullin et al 2005). Inter-cellular barrier junctions consist of tight-and adheren-junctions to control barrier integrity and porosity (Gumbiner 1996, Schneeberger andLynch 2004, ). The TEER was measured as an indicator of barrier function. During initial 2D-cell sheet culture period, CACO-2 cells showed a TEER increase, with negligible increases seen after ten days indicating functional barrier formation. 3D-tubular, cell-populated scaffolds were fabricated on the ATF and transferred to the compatible bioreactor system ( figure 3(A)). Real-time TEER measurements within the bioreactor system were taken by inserting electrodes inside the lumen of the tube (figure 3(A) (V2 and I2) and (B), and externally to the construct in the bioreactor perfusion chamber (figure 3(B), V1 and I1). These electrodes were connected to an EVOM 2 machine measuring voltage (V1 and V2) and amplitude (I1 and I2) to quantify resistance across the tubular wall ( figure 3(A)).
Relative levels of extracellular and intracellular calcium can directly affect epithelial layer barrier integrity (Cereijido et al 1978, Gumbiner andSimons 1986), a phenomenon applied to examining the modulation of barrier integrity in gut and kidney models (Gumbiner and Simons 1986, Ye et al 1999, Ivanov et al 2004b, Kwak et al 2012. This short-term, reversible effect is induced through the dis-assembly of the inter-cellular junctional complexes (Ivanov et al 2004a). CACO-2 populated tubes were incubated in a CRB, known to chelate calcium (Dixon et al 2014) for 5 min before replacing with normal media (containing 2 mM Ca 2+ ) for a further 24 h. Epithelial barrier permeability during this period was measured through TEER and para-cellular permeability. The reduction of extracellular calcium produced an instantaneous drop in TEER (67 ± 5 versus 144 ± 8 Ω cm −2 ; mean ±SEM, n = 3) with the re-introduction of calcium providing an equally robust restoration in electrical resistance that returned to initial levels after 24 h (figure 3(E)). Inter-cellular junction integrity was visualized by immunostaining the adheren-junction protein E-cadherin at 5 min and 24 h post-calcium depletion. The classic 'chickenwire' E-cadherin distribution prior to calcium reduction was disrupted at 5 min and restored after 24 h ( figure 3(D)). In parallel, barrier integrity was monitored by small-molecule permeability. Using fluorescently-labelled dextran (4 kDa FITC-dextran and 70 kDa rhodamine-dextran) allowed the analysis of different sized molecule diffusion within individual samples. Dextran was applied to the lumen of either blank electrospun scaffold tube, or a CACO-2 populated tube, and samples were taken from the perfusion chamber over a 2 h time course. Diffusion through acellular tubes was greater than cellular tubes (figure 3(F)), whilst CRB treatment of cellular tubes significantly increased the rate of dextran diffusion over the time course. Lower weight dextran (4 kDa) showed an increased diffusion rate compared to larger dextran molecules (70 kDa) (figure 3(G)). These data indicate CACO-2 fabricated architectures possess functional barrier properties that restrict para-cellular diffusion in a size dependent manner. Therefore we fabricated an intestinal tubular tissue that could be employed for drug screening strategies.

Discussion
Here we have demonstrated an automated fabrication system to create multi-layer, multi-phasic, patterned tubular structures in which biomaterials, hydrogels or cells can be used as the base material. In conjunction with previously described patterning technology (Paik et al 2012) and temperature-responsive polymers (PNIPAAM) (Williams et al 2009) this technology allows precise positioning of cells in 3D architectures.
This study has demonstrated that dimension standardization can be achieved in biologically active architectures and issues such as seeding, positioning and patterning of cells within different layers throughout tissues can be addressed directly during fabrication beyond the attempts of other studies (Baguneid et al 2006, Peck et al 2012. Manual fabrication of vessels used in previous studies (L'Heureux et al 1998, 2006, Seliktar et al 2000 is not easily achievable when multi-layer-phasic patterned architectures are required which prompted the development of the ATF system. An automated system presents theoretical advantages over manual fabrication. However cost and system complexity must be offset with an improvement in vessel quality, speed of production and less need for technical ability needed for manual fabrication. Previous cell sheet rolling technologies (McAllister andL'Heureux 2006, Kubo et al 2007) have produced fabricated vessels however the technicality of manipulating delicate and hydrated hydrogels or cell sheets prompted us to develop a system in which 2D materials are fed into the fabrication mechanism yielding much more reproducible constructs and allowed patterned cellular structures to be fabricated (Williams et al 2009). Furthermore by integrating the ATF system to transfer constructs to perfusion chambers directly the inherent variability of manual manipulation to allow bioreactor mounting and perfusion is removed. We demonstrated the use of fibrin to bond sheets during fabrication. Fibrin confers the advantage of rapid adherence between fabricated sheets but does have it disadvantages; fibrin inclusion adds cost and could interfere with fusion of layers due to the biological activity of cells following fabrication during maturation. Furthermore the fibrin component was not sufficient to provide sufficient mechanical strength to create viable blood vessel mimetic constructs. Importantly, fibrin is not essential to our system and its use could be emitted. Further analyses regarding any thrombogenic activity of fibrin in these constructs will also need to be tested before in vivo translation however this is unlikely (Kheirabadi et al 2006).
Even with our technology, materials that are fully bioactive and provide the substantial mechanical strength will need to be developed if living tissue is to be engineered as vascular grafts. However as the ATF system can be employed with any 2D material this methodology could be used to fabricate vessels with such materials (Kakisis et al 2005, Baguneid et al 2006. Using the ATF system for fabrication of vasculature employing materials with the correct mechanical properties is now the focus of future work.

Conclusion
This novel automated platform allows accurate, standardized and repeatable fabrication of tubular architectures; comprised of multiple compact, stable and secure cell and/or material layers. These fabricated tissues are directly transferred to a compatible perfusion bioreactor without further manipulation. Rolling each layer takes less than a minute and because no external material is required to maintain the structure except the cells with or without a biomaterial, it can be considered as an authentic living structure. We demonstrated our systems utility in fabricating vessels with arterial-like cell patterning and gut-like barrier function. This technology is easily adaptable to fabricate multiple biomaterial-types to engineer a variety of biologically stable tissues directed for regenerative or in vitro diagnostic protocols.