A Method for High-Throughput Robotic Assembly of Three-Dimensional Vascular Tissue

Abstract

An essential step toward commercializing engineered tissues is to scale-up and automate their production. This presents a challenge for self-assembled tissues that are fragile at early time points and difficult to handle using robotic systems. The goal of this study was to automate tissue-engineered blood vessel (TEBV) fabrication by creating a custom cell seeding and self-assembly system that is conducive to robotic manipulation, coupled with a robotic system to assemble smooth muscle cell ring units into tissue tubes. To generate self-assembled tissue ring units manually, cells are seeded at a high density into custom agarose wells that have center posts (2 mm inner diameter), around which cells aggregate and contract to form rings. Agarose is well-suited for cell seeding and ring self-assembly because it is noncell adhesive, can be autoclaved, and reproducibly cast to form wells using silicone templates. However, agarose gel is soft and makes reliable robotic manipulation challenging. To solve this problem, we designed a custom ring self-assembly plate utilizing polyetherimide (PEI) wells with MED610 three-dimensional-printed center posts and an MED610 well negative that allows the casting of individual, annular agarose cell-seeding troughs within the PEI plate. Rings cultured in the new plate system were morphologically similar to rings cultured in control agarose wells and had a slightly higher failure load. To automate tube assembly, we created a unique robotic punch system to push tissue rings out of the PEI-MED610 plate onto a stainless steel mandrel to enable tube fusion. Tubes fabricated by manual or robotic ring removal and placement demonstrated similar morphology after tube fusion, and the automated system substantially reduced the time required to assemble tubes. In summary, we developed a novel robotic assembly system to precisely manipulate self-assembled tissue rings and enable scale-up and automation of TEBV biofabrication.

Impact Statement

Self-assembled tissues have potential to serve both as implantable grafts and as tools for disease modeling and drug screening. For these applications, tissue production must ultimately be scaled-up and automated. Limited technologies exist for precisely manipulating self-assembled tissues, which are fragile early in culture. Here, we presented a method for automatically stacking self-assembled smooth muscle cell rings onto mandrels, using a custom-designed well plate and robotic punch system. Rings then fuse into tissue-engineered blood vessels (TEBVs). This is a critical step toward automating TEBV production that may be applied to other tubular tissues as well.

Introduction

Tissue-engineered constructs have enormous potential as implantable grafts, or as tools for disease modeling and high-throughput drug screening. A key step for translating these technologies into marketable products is automating and scaling-up their production. Automation can not only significantly reduce production time and increase output, but also reduce the risk of human error and improve product consistency. Technology has been developed for automated cell seeding, passaging, and medium changes for two-dimensional cell cultures, which are the first steps to creating engineered tissue.¹ However, there are limited technologies available for scaling-up and automating the fabrication of three-dimensional (3D)-engineered tissues.

Bioprinting is commonly used as an automated approach to create 3D tissues.^2–5 With bioprinting, a wide range of scaffold-based, cell-laden bio-inks^5,6 or scaffold-free cell spheroids⁴ can be precisely patterned directly onto a surface. Layer-by-layer printing approaches may allow for even more complex shapes, or multi-tissue structures.^4,7 Cellular self-assembly may have advantages over scaffold-based approaches to tissue fabrication because of the biocompatibility, enhanced cell–cell and cell–matrix interactions, and physiologically relevant mechanical properties of scaffold-free tissues.^8–11 These characteristics make cellular self-assembly ideal for fabricating many different tissue types, including tissue-engineered blood vessels (TEBVs).^11–14 Additive manufacturing is frequently used when fabricating engineered tissues by scaffold-free approaches. Smaller tissue subunits are first formed through cellular self-assembly, and those subunits are then fused into larger tissues.

Spheroids are the most commonly used tissue subunits for both bioprinting and other additive manufacturing techniques because they are relatively simple to fabricate.^4,15,16 However, constructs fabricated from even tightly packed spheroids often have remaining gaps after fusion, as spheroidal units cannot be pushed into complete contact with one another.^4,17,18 Alternatively, toroidal units can be used as building blocks when fabricating 3D tubular structures.^19,20 Rings may fuse more seamlessly than spheroids, as greater portions of ring surface area are in contact with one another, making them more conducive to TEBV fabrication.

Another challenge with additive manufacturing of self-assembled tissue-engineered constructs is the ability to lift individual tissue units and precisely place them together to build a complex composite structure. This may be especially difficult at early time points in culture, when self-assembled tissue units are fragile, as they may not yet have secreted sufficient extracellular matrix proteins to form a robust structure. However, tissue units harvested at earlier time points have improved tissue fusion compared with more mature tissue units^21–23; thus, it is essential that tissue building blocks can be manipulated early in culture.

A handful of technologies currently exist for manipulating tissues in culture.²⁴ For example, the “Kenzan” method for fabricating scaffold-free tissues utilizes microneedles that are inserted through the center of self-assembled spheroids. This holds the spheroids in a patterned shape, and has successfully been used to make tubular constructs.²⁵ However, a major disadvantage of this method is that it requires the use of relatively large spheroid building blocks. Each spheroid is punctured by a needle with a 170 μm diameter,²⁵ which may be damaging to smaller constructs. Robotic systems have also been developed that can precisely pick up and place self-assembled tissues.^26,27 However, this system is not completely automated, and the user must still manually place the tissue using a series of positioning knobs. In addition, it was difficult to stack toroidal tissue units precisely enough for seamless tissue fusion, and individual tissue unit borders were still visible after 3 days of culture.²⁶

Our laboratory has developed a system for creating self-assembled TEBVs from individual tissue-ring subunits with manual assembly of rings into tissue tubes.^19,23 In brief, smooth muscle cells are seeded into an agarose well with a center post, where they aggregate to form vascular smooth muscle rings.²⁸ Vascular tissue ring units are then manually removed from wells, threaded onto a mandrel, and pushed together, where they fuse into tubes,²³ as shown schematically in Figure 1.

FIG. 1.

Manual method for fabrication of self-assembled vascular tissue rings and tubes. Cells are seeded into individual ring-shaped agarose wells, where they aggregate to form self-assembled tissue rings. Tissue rings are manually harvested and threaded onto silicone tubing mandrels, where they remodel and fuse together to form a tissue tube. Color images are available online.

Our overall goal was to develop a completely automated robotic assembly system to scale-up the fabrication of TEBVs for commercial purposes (Fig. 2). In this study, we focused on developing a means to robotically remove rings from agarose wells and thread them onto mandrels for fusion. There are currently no other systems available that are suitable for this specific purpose. To do this, we redesigned the agarose wells used for cell seeding, tissue ring self-assembly, and ring culture. Agarose is well-suited for enabling cellular aggregation because of its noncell-adhesive properties, but the softness of agarose makes it challenging to grip and manipulate consistently using conventional robotic systems. The first goal of this study was thus to develop new cell-seeding wells that would enable both self-assembled ring formation and robotic culture plate manipulation. We then evaluated the effect of the custom plate on tissue ring morphology and mechanical properties compared with rings cultured in typical agarose gel wells. Finally, we tested the feasibility of using a custom robotic punch to harvest tissue rings directly from the plates and stack them onto a stainless steel mandrel, a labor- and time-intensive process when performed by hand. Overall, this novel tissue culture plate and robotic punch system enabled robotic fabrication of tissue tubes from individual tissue ring subunits.

FIG. 2.

Overview of proposed method of robotic assembly of TEBVs fabricated from self-assembled vascular tissue rings. Color images are available online.

Methods

Tissue ring culture plate design

A new plate system was designed and prototyped, consisting of a custom 96-well plate, a center-post plate inserted from the bottom, and a removable well negative for casting agarose cell-seeding troughs (shown schematically in Fig. 3). The 96-well plate is machined from polyetherimide (PEI, trade name Ultem^™ 1000; SABIC) plastic, with dimensions based on ANSI SLAS 4-2004 (R2012) standards for a 96-well microplate. PEI plates contain bottomless wells with a 6 mm diameter to match the dimension of our control agarose gel system^19,28 that flare out to 9 mm at the top of the well to match the 96-well microplate footprint. An array of 2 mm center posts that fit within the wells of the PEI plate was 3D printed from MED610 (Stratasys Ltd.) material using a photopolymer printer (Object Connex 260; Stratasys Ltd.). The well negative was also printed from MED610 material, which, when inserted in the PEI plate, allows agarose troughs to be cast within the PEI plate wells. For the prototype testing presented in this study, the 96-well plate format was scaled down to 16 wells (two rows) to reduce the amount of reagents and cells required. Two 16-well plates were prototyped for cell seeding, ring culture, and robotic ring harvesting and assembly studies.

FIG. 3.

PEI-MED610 plate system. A MED610 center-post plate fits under the open-bottomed PEI 96-well plate. A MED610 negative sits on top of the 96-well plate, to allow casting of rounded-bottom agarose troughs in each well. Color images are available online.

Cell culture

Rat aortic smooth muscle cells (WKY-3M22²⁹) were cultured in medium comprising Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal bovine serum, 1% nonessential amino acids, 1% l-glutamine, 1% sodium pyruvate, and 1% penicillin–streptomycin.

Agarose gel preparation

For the control configuration in this study, agarose gels were fabricated as described previously.²⁸ Two percent agarose (Lonza #50000) in DMEM was autoclaved and poured into polydimethylsiloxane (PDMS) templates. When cooled, each agarose gel was removed from the PDMS, placed into a well of a six-well plate, and allowed to equilibrate overnight in culture medium before use. Each agarose gel contained wells for five individual rings (2 mm inner diameter; Fig. 1).²⁸

To prepare PEI-MED610 plates for the proposed automated approach, the assembled plate was ethylene oxide sterilized and allowed to de-gas 48 h before use. A solution of 3% low-melting-temperature agarose (Lonza #50101) in DMEM was autoclaved and allowed to equilibrate for 30 min in a 45°C water bath. This allowed the agarose to cool to a temperature below 52°C to avoid damaging the 3D-printed MED610 pieces. Agarose was then carefully pipetted into the assembled PEI-MED610 plate system and allowed to solidify. The well negative was removed, and a hydrophobic 1% Pluronic F-127 (Thermo Fisher) solution was pipetted into wells to reduce cellular adhesion to the PEI components in the plate wells and to the MED610 posts. After incubating 15 min at room temperature, the Pluronic was aspirated and the well plate was equilibrated overnight in culture medium before use.

Ring fabrication

Rings were seeded as described previously.²⁸ In brief, cells were trypsinized and resuspended at a final concentration of 10 million cells per milliliter. A total of 500,000 cells per ring were seeded into each well of either control agarose wells or PEI-MED610 plates. On day 1, wells were flooded with fresh culture medium. Both agarose wells and PEI-MED610 plates were submerged in medium during ring culture. Culture medium was changed after 2 days.

Mechanical testing

After 4 days of culture, rings were removed from agarose wells or PEI plates and placed in a dish filled with phosphate-buffered saline (PBS). Rings were placed under a machine vision system (DVT series 600-Model 630), and thickness measurements were taken in four locations around the ring circumference with edge detection software (Framework 2.4.6.; DVT Corp.).¹³ Average ring thickness was used to calculate ring cross-sectional area. Rings were then loaded onto a custom grip setup submerged in PBS, subjected to eight precycles from 1 to 5 mN, and then pulled to failure at a rate of 10 mm/min using an ElectroPuls E1000 (Instron). Failure load and ultimate tensile stress (UTS; failure load/ring cross sectional area) were recorded.

Histology

After 4 days of culture, rings were fixed for 1 h in 10% neutral-buffered formalin, processed, and paraffin embedded. Five-micrometer sections were prepared and adhered to charged glass slides. A hematoxylin and eosin stain was used to visualize tissue morphology, and a Picrosirius Red/Fast Green stain was used to visualize collagen deposition. Images of stained tissue sections were acquired with an upright microscope (Leica DMLB2 with DFC 480 digital camera).

Robotic punch design

To extract self-assembled tissue rings from the PEI plates and assemble them into a tubular construct, we developed an automation platform controlled by a simple graphical user interface. The system indexes to each well of the PEI-MED610 plate, lowers a 316 stainless steel mandrel over the well's center post, and punches four titanium rods up through the agarose troughs to thread the ring onto the mandrel. Steel mandrels have a thin coating of USP class VI epoxy (EP42HT-2MED; Master Bond, Inc.) applied to the tip to help prevent rings from sliding off after automated stacking. A diagrammatic representation of the ring extraction process is given in Figure 4. The platform uses two high-resolution linear ball screw stages with ±3 μm repeatability (404100XRMS; Parker Hannifin Corporation) to position a bottomless tray holding the PEI-MED610 plate between the punch and mandrel. The punch and mandrel are mounted to two pneumatic linear slides (13-MXS16–30; SMC Corporation) allowing motion in the vertical direction. To ensure rigidity, the frames supporting the linear slides and PEI-MED610 plate were machined from aluminum with system components mounted to a thick polycarbonate base. All components fit inside a standard tissue culture hood. Images of the robotic assembly system are given in Figure 5. All components that come in direct contact with the PEI-MED610 plate system are removable and autoclavable. After autoclaving, parts are reattached inside the cell culture hood by an operator wearing sterile surgical gloves.

FIG. 4.

Schematic of robotic process to remove tissue rings from a PEI-MED610 plate. A 4-prong punch pushes through a cast agarose trough and moves self-assembled tissue rings onto a stainless steel mandrel. As the mandrel is lifted away from the well, an aluminum agarose remover pushes any remaining agarose away from the mandrel while the tissue ring stays in place (See Supplementary Video S1 for real-time demonstration). Color images are available online.

FIG. 5.

Robotic assembly system not photographed in a cell culture hood. A positional stage moves the PEI-MED610 plate over a 4-prong punch. The punch pushes rings out of the well and up onto a stainless steel mandrel. For testing, a shortened 16-well plate format was used to reduce the amount of reagents and cells required to evaluate iterative prototype changes. Color images are available online.

To control the system, a graphical user interface created in Matlab2016a (MathWorks, Inc.) outputs tray positions and punch commands over RS-232 serial communication to a 2-axis motion controller (Compumotor 6K2; Parker Hannifin Corp.). Magnetic limit switches are used for the homing calibration process of the positional stages and for checking the current positions of the pneumatic slides. Absolute position commands are given to the controller based on a measured offset between the home position and the first well. Pneumatic slide positions are used for error checking, and locking out the system from indexing when the punch or mandrel are not clear of the PEI-MED610 plate.

Tube fabrication

Smooth muscle cell rings were cultured for 4 days in either control agarose wells or PEI-MED610 plates before harvesting for tube fabrication. PEI-MED610 plates were loaded onto the robotic platform inside a cell culture hood, and automatically punched out of the plate and onto a stainless steel mandrel. A brief 10 s dwell period was included between the removal of each ring from its well and the punch retraction. This allowed rings to contract, enabling them to remain on the mandrel when the punch was retracted. The epoxy coating on the mandrel tip also helped prevent rings from sliding off the mandrel. Rings in the PEI-MED610 plate and on the mandrels were periodically hydrated with culture medium using a pipette. Mandrels with stacked rings were then loaded into custom polycarbonate holders for fusion culture.¹³ For control tubes, rings are manually threaded with forceps onto either silicone mandrels as previously described¹³ or onto the stainless steel mandrels in the robotic system. Mandrels were secured in custom polycarbonate holders (polycarbonate holders shown in Gwyther et al.¹³), to evaluate the effect of mandrel material on ring fusion. Tubes were submerged in culture medium described in “Cell culture” section, with medium changed on day 3. All tubes were cultured for 4 days, and then fixed 1.5 h in 10% neutral-buffered formalin. Tubes were manually removed from their respective mandrels for processing, sectioning, and histological staining.

Statistics

Statistical analysis was performed using SigmaPlot Software (Systat, version 12.5) on ring thickness, failure load, and UTS data. A Student's t-test was performed to analyze statistical differences between mechanically tested rings on data with a normal distribution. A Mann–Whitney test was used to compare data that failed a normality test.

Results

PEI-MED610 plate design and prototyping

Agarose gels are soft and not easily manipulated with robotic systems. Therefore, our first goal was to design and prototype a new plastic plate system as an alternative to the custom agarose wells we used for cell seeding and tissue ring culture in our previous study.²⁸ The new three-part plate design utilizes MED610 3D-printed posts and well negative, and a machined PEI 96-well plate (Fig. 3).

The PEI-MED610 plate system is a result of several design iterations. Initially, all parts were fabricated from MED610 photopolymer, because of the one-step 3D-printing process, and its status as a USP plastic class VI, indicating it is safe to be in contact with tissues.³⁰ After seeding cells in the MED610-only plate, we observed cells failing to self-assemble into tissue rings, possibly because they adhered to the MED610 material. Thus, we developed a hybrid system, and used a PEI thermoplastic for the bulk of the plate system. PEI was chosen because of its availability, machinability, and previous use in culture of other cell types.³¹ We continued to use 3D printing to generate the MED610 post insert plate, as the shape of this piece could not be made easily with other prototyping methods, and the small amount of MED610 in the posts did not seem to affect ring formation. A 3D-printed MED610 “well negative” that fits over the plate assembly enabled us to cast agarose troughs in the bottom of each well in the assembled plate for cell seeding and self-assembly. Although rings successfully formed with this method, they still adhered slightly to the MED610 posts, which made it challenging to remove them from the wells. This issue was addressed by adding a Pluronic coating in our current prototype that prevents cell adhesion to materials.³²

Ring fabrication in PEI-MED610 plate system

To ensure the change in well material and the plate format did not adversely affect cellular self-assembly and ring formation, or tissue ring morphology and mechanical strength, we compared rings cultured in control agarose wells to rings cultured in the PEI-MED610 plates (Fig. 6A). We observed no significant differences in ring thickness (0.42 ± 0.02 and 0.44 ± 0.04 mm, rings cultured in agarose vs. PEI-MED610 plates, respectively; Fig. 6B). There was a slight increase in UTS (30.5 ± 7.9 and 45.7 ± 14.9 kPa for rings cultured in agarose vs. PEI-MED610 plates, respectively; Fig. 6C), although this was not statistically significant. There was significant increase in maximum load at failure (failure load) when rings were cultured in the PEI-MED610 plates (8.6 ± 2.1 and 14.2 ± 4.6 mN for rings cultured in agarose vs. PEI-MED610 plates, respectively; Fig. 6D).

FIG. 6.

Structure and strength of tissue rings grown in PEI-MED610 plates compared with control agarose gels. Images of 4-day-old rings cultured in agarose gels (left) or PEI-MED610 plates (right) shown in (A). Comparisons of ring thickness (B), ultimate tensile stress (C), and maximum load at failure (D). *p < 0.05. A nonparametric t-test was used for (B) and (C), and a Mann–Whitney test in (D). Scale bar = 1 mm; n = 6 for rings cultured in agarose controls, n = 8 for rings cultured in PEI-MED610 plates.

No visible differences were observed between rings cultured in control agarose wells or PEI-MED610 plates, either macroscopically (Fig. 7A, B) or by hematoxylin and eosin staining (Fig. 7C, D). Collagen deposition was visible in both groups, with no apparent differences (Fig. 7E, F).

FIG. 7.

Morphology of self-assembled tissue rings cultured in agarose gels or PEI-MED610 plates. Photographs of tissue rings in cell-seeding wells (A) and (B), PEI plate has been removed from (B), so only rings around MED610 posts are shown. Ring sections were stained with hematoxylin and eosin (C, D) or Picrosirius Red/Fast Green (E, F; red = collagen, green = counterstain). Rings in (A) and (B) are 2 mm inner diameter. Scale = 100 μm (C–F). Sections representative of n = 3 rings. Arrows point to rings (A, B). Color images are available online.

Robotic stacking of vascular tissue rings

Next, we evaluated the ability of the robotic punch to remove rings from the PEI-MED610 plates and stack them onto a stainless steel mandrel. Rings were cultured in two PEI-MED610 plates, scaled down to 16 wells per plate. When preparing the PEI-MED610 plates, 10 of 16 agarose troughs successfully formed on the first plate, and 14 of 16 on the second plate. Rings were seeded in all plate wells with agarose troughs, although one ring on each plate broke within 24 h of seeding. From the first plate, 6 of 9 self-assembled tissue rings were successfully pushed onto the mandrel; 5 of 13 were successfully pushed onto the mandrel from plate 2. During pilot testing before this experiment, we determined that a 10 s pause after pushing the ring onto the mandrel, before retracting the punch, helped prevent rings from sliding off the mandrel when the remnant agarose gels are removed. Still, some rings did slide off the mandrel, which contributed to this reduced yield. In addition, some rings broke during removal from the PEI-MED610 plate. This may be because rings had only been cultured for 4 days and may not have secreted sufficient extracellular matrix to withstand the forces from stacking. Four of 28 control rings cultured in agarose-only wells failed during manual stacking and could not be used. Ring failure rates are given in Table 1.

Table 1.

Summary of Failure Rates of Tissues Rings Stacked Manually or Robotically

	Total rings seeded	Ring failures during culture	Ring failures during stacking	Time required to stack 96 rings
Manual	28	0	4	3.49 h
Robotic	24	2	11	0.45 h

During this experiment, the time required to stack rings manually compared with the robotic system was recorded. We found that after the initial setup, the robotic assembly system reduced the time required to remove rings from their cell-seeding plates and thread them onto a mandrel to fabricate tissue tubes. To manually stack the 28 rings onto two silicone mandrels (including 4 rings that broke during manual placement) took ∼61 min, which averages to ∼2.2 min per ring. This is compared with 17 s per ring to remove and stack each ring using the automated system (Supplemental Video S1). For reference, on a 96-well plate this translates to 0.45 h for the automated approach and 3.49 h for the manual approach. Because of this increased speed, even when accounting for the observed 54% ring failure rate, the PEI-MED610 plates can still stack ∼3.5 times as many rings as an experienced human operator in the same amount of time.

Tube fusion after robotic ring stacking

To evaluate if the robotic ring-stacking procedure, or the stainless steel mandrels, adversely affected ring fusion, we stacked five to six rings per tube (n = 2 tubes per group) by robotic or manual stacking onto stainless steel mandrels, or manual stacking on silicone tubing mandrels as previously reported.²⁸ Robotic ring stacking was performed in a biosafety cabinet to maintain sterility of rings and tubes, and stacked tubes were collected and cultured for an additional 4 days for fusion. Photographs of tubes after fusion are given in Figure 8A–C. Longitudinal sections of fixed tubes (Fig. 8D–I) show that rings in all three groups have fused, although ring boundaries were still distinct. Rings and tubes appeared slightly longer in the PEI-MED610 group. A slight growth of cells from the rings onto the MED610 posts was observed, causing a slight increase in ring “height,” which may have contributed to the increased tube length.

FIG. 8.

Fusion of robotically or manually fabricated tissue tubes. Tissue rings were either manually stacked onto silicone (A, D, G) or stainless steel (B, E, H) mandrels, or stacked onto stainless steel mandrels using the robotic assembly system (C, F, I). Photographs of tubes after 4 days of fusion (A–C). Hematoxylin and eosin stain of fused tube tissue sections (D–F), with magnified images shown in (G–I). Scale = 0.5 mm (D–F) or 0.1 mm (G–I); n = 2. Color images are available online.

Discussion

Automation is key to scaling-up the production of human tissues for commercial use, either as implantable grafts or as tools for high-throughput drug screening. Here, we developed a unique tissue ring self-assembly plate and robotic tissue ring harvesting and tube assembly system, and demonstrated the feasibility of robotic assembly of TEBVs fabricated from individual ring units.

The first goal of our study was to develop a new well system for seeding smooth muscle cell rings that had both a rigid frame conducive to robotic manipulation and enabled cellular self-assembly. We developed a three-part plate system using a PEI plate with individual open-bottom wells, 3D printed MED610 center posts, and a MED610 well negative that enabled the casting of agarose troughs in the PEI plate wells. In addition to providing a nonadhesive cell seeding and ring self-assembly surface, the agarose troughs also enabled ring harvesting by punching through the agarose well bottoms to push the rings out of the plate and onto the stainless steel mandrel in the robotic assembly system.

After implementing the PEI-MED610 plate system, we evaluated the effects of the new plate materials on ring formation, morphology, and mechanical strength to determine whether the new system could produce tissue rings with similar properties to the agarose wells. Overall, rings cultured in the PEI-MED610 plates had a low failure load and UTS, which is likely because of the short 4-day ring culture duration used in this study (compared with previous mechanical testing studies of 2 mm rat smooth muscle cell tissue rings, typically cultured for at least 14 days^13,19). Strength was evaluated at day 4 because this is the time point when rings are stacked into tubes. We chose this early time point for ring stacking because we have shown that rings fuse more cohesively when fused at earlier time points than later ones.²³ In our initial tests (Fig. 7), we observed a significant increase in maximum load at failure with rings cultured in the PEI-MED610 plates compared with rings cultured in agarose wells. It is possible that the larger quantity of culture medium required for the new culture system contributed to this improvement. The PEI-MED610 plates are placed in a glass dish for culture, which requires ∼70 mL of medium to cover a plate containing 16 rings. This is substantially more than our agarose gel system,³³ which uses 4.5 mL medium in each well of a six-well plate, which covers five rings in an agarose gel. Most importantly, this indicates that the PEI-MED610 plates did not adversely affect ring formation or mechanical properties and can be used as an alternative to pure agarose wells for culturing tissue rings for automated assembly.

In previous studies, we showed that tissue ring strength increases with culture time (7 vs. 14 days¹⁹). To assess the feasibility of culturing rings for longer times in the PEI-MED610 wells, and to increase ring strength, we performed an additional experiment comparing rings cultured for 4 and 7 days in both PEI-MED610 plates and agarose wells. Consistent with our previous studies with agarose wells, we observed significant increases in failure load with increased culture time in both agarose wells and PEI-MED610 plates (Supplemental Results and Supplementary Fig. S1).

The presented robotic assembly system is advantageous because of its simple design and compatibility with existing standards for laboratory automation. The device requires only a commercially available two-axis positioning stage and two commercially available linear slides with no additional moving linkages, degrees of freedom, grippers, or extruders. Other systems for manipulating engineered tissue are limited and may be significantly more expensive and complex to build. Its compact design fits in a standard biosafety cabinet, allowing setup and operation in an aseptic environment.

This design was well-suited for the task of tissue ring extraction and stacking, and substantially reduced the amount of time required to fabricate tubes compared with stacking by hand. However, some future optimization is still required to improve tissue ring yield and stacking efficiency. For example, despite the thin epoxy layer on the mandrel tip (to slightly increase friction), failures sometimes resulted when rings slid off the mandrel, and followed the punch back down into its original well. This could be addressed by optimizing the mandrel shape and surface finish to prevent slipping. The current system relies on a 10 s dwell time for rings to contract around the mandrel and prevent them from sliding off. Optimizing mandrel parameters to further prevent sliding may allow us to eliminate this dwell period and increase production speed. Automating tube hydration may also reduce the total time required for ring stacking.

A small number of rings also failed during transfer out of the wells by the robotic punch. This may be improved by optimizing culture conditions to increase ring strength. In previous studies, we showed that media supplementation to promote collagen synthesis increased rat smooth muscle ring UTS by 266 kPa.⁹ Finally, when setting up the mold system, the casting of agarose troughs was sometimes not successful. In these instances, the agarose adhered slightly to the well negative and was pulled out of the PEI well when the negative was removed. In subsequent experiments, we observed improved casting of agarose wells by removing the well negative with sterile surgical gloves, instead of forceps to improve user dexterity. In the future, a coating may also be used to reduce agarose trough adhesion to the well negative. Despite these limitations, we were still able to precisely move 3.5 times as many rings onto mandrels as a human can manually stack in the same amount of time.

After automated or manual stacking of rings, rings were allowed to fuse for 4 days in culture. After 4 days, we observed ring fusion regardless of mandrel type or stacking method. Ring boundaries were still visible in all groups, which was consistent with our previously published work with rat aortic smooth muscle cells.¹⁹ We have previously tested the longitudinal strength of fused tubes as a means to evaluate fusion strength.³⁴ We observed that tubes tore at similar loads to unfused tissue rings, indicating that visible ring boundaries may not adversely affect tissue failure load.³⁴ Fusion may continue to improve with longer culture times. Tubes in this study were only allowed to fuse for 4 days (after 4 days of ring culture, for 8 days total culture) compared with 1–3 weeks of total culture time with fusion in previous studies.^19,23,35 Boundaries are primarily visible on the luminal side of the fused tubes (Fig. 8D–F). We have previously shown that ring fusion improves and boundaries between rings become less evident with the addition of luminal flow, which will be applied in longer term tube culture in future studies.

In longitudinal tube sections, rings that had been cultured in the PEI-MED610 plate appeared longer than rings cultured in agarose wells. Although we did not observe differences in ring thickness when viewed from above the plate, cells did appear to grow out from the rings along the length of the MED610 posts, resulting in slightly increased ring height compared with agarose wells with agarose center posts. However, this increase in ring height does not seem to have negatively affected ring fusion, mechanical properties, or our ability to remove rings from the PEI-MED610 plate system. Such rings may even enable us to make longer tubes without the need for additional cells and reagents.

Another challenge to automated tissue fabrication is maintaining sterility. After 4 days of fusion culture postautomated stacking, there was no visible contamination, indicating that autoclaving machined parts that came in direct contact with the tissue rings and culture wells was sufficient to maintain sterility.

In this study, we demonstrated in proof-of-concept experiments our ability to automate tissue ring stacking. We determined that tubes fabricated with our new method and mandrels fuse comparably with rings stacked by hand on silicone mandrels. This provides an advantage over existing systems, such as “Pick, Place, and Perfuse,” where the human operator could not align tissue subunits as precisely as our robotic punch and mandrel system.²⁶ It also does not require tissues to be punctured and damaged, unlike the “Kenzan” method.²⁵ Future experiments will include an evaluation of smooth muscle function and the inclusion of endothelial cells to create a more biomimetic TEBV. Endothelial cells will be added after smooth muscle rings are fully fused, and thus will not be impacted by the automation procedure.

Future work will also involve automating other components of the tube-fabrication process, such as cell seeding and medium dispensing. PEI-MED610 plates are designed to mimic the dimensions of a standard 96-well plate, which will enable the use of these existing automation technologies for this purpose.¹ The stacking procedure we developed is perhaps the most challenging step in fully automating production of engineered blood vessels from ring modules. Here, we demonstrated the feasibility of using a robotic punch to push self-assembled tissue rings up out of a custom-designed hybrid PEI-MED610 plate and onto a steel mandrel, where they fuse to form a tissue tube. This automatic system substantially reduced time required to fabricate tubes and is a step toward complete automation and scale-up of TEBV fabrication. This may enable us to scale-up the production of vascular disease models fabricated for high-throughput drug screening experiments.²³ It may also be applied more broadly to biofabrication of other tubular tissues for disease modeling or implantation purposes. For example, tissue-engineered tracheas can also be constructed from ring segments.^35,36 Scaling-up the tube fabrication process allows for increased production and improved product consistency, which is an essential step toward commercializing any tissue-engineered product.

Disclosure Statement

No competing financial interests exist.

Supplementary Material

Supplementary Results

Supplementary Figure S1

Supplementary Video S1