New Frontiers for Biofabrication and Bioreactor Design in Microphysiological System Development

Abstract

Microphysiological systems (MPS) have been proposed as an improved tool to recreate the complex biological features of the native niche with the goal of improving in vitro-in vivo extrapolation. In just over a decade, MPS technologies have progressed from single-tissue chips to multi-tissue plates with integrated pumps for perfusion. Concurrently, techniques for biofabrication of complex 3D constructs for regenerative medicine and 3D in vitro models have evolved into a diverse toolbox for micrometer-scale deposition of cells and cell-laden bioinks. However, as the complexity of biological models increases, experimental throughput is often compromised. This review discusses the existing disparity between MPS complexity and throughput, then examines an MPS-terminated biofabrication line to identify the hurdles and potential approaches to overcoming this disparity.

The Rise of Microphysiological Systems

In tissue engineering, assays reliant on 2D cellular monolayers are utilized for deciphering the mechanisms of action for drugs or other external stimuli such as atmospheric composition[1–3], however, these assays cannot be extrapolated to predict native behavior in response to the same stimuli[4,5], e.g. screening the efficacy of a drug in a monolayer metabolic activity assay will not reflect the response of native 3D tissue to those same drug doses; monolayers simply lack too many of the properties of the native niche. Not only do they lack biochemical and atmospheric gradients[6], but stiff tissue culture plastic causes cells to lose their original phenotype.[7] The monolayer organization prevents cells from experiencing the appropriate cell-cell and cell-extracellular matrix (ECM) interactions.[8] However, despite these limitations, 2D cultures are used routinely as they require extremely low reagent volumes[9], and are compatible with existing ‘omics’ and high-content analysis platforms.[2,8,10]

To overcome these limitations of 2D cultures, a series of 3D model platforms have been developed. These include, but are not limited to, cell sheets, microtissues (see Glossary), matrix encapsulation, and biofabricated constructs.[11]

Cell sheets are confluent 2D monolayers recovered with the ECM intact as a single sheet from temperature-responsive[12], or fibrin-coated dishes.[13], whereby thick constructs can be formed by stacking multiple sheets. However, the challenge of fabricating thick constructs of clinically-relevant size often results in development of necrotic regions due to a lack of nutrient transport and oxygen diffusion.[14] One possible approach to avoid necrotic regions was to vascularize the cell-sheet constructs, as demonstrated for cardiac tissue.[15] By coculturing endothelial cells (EC) and cardiac cells, vascular networks formed within each sheet. After recovery and stacking, the multi-layered constructs resolved nutrient and oxygen transport within the construct, as well as provided the support of an ex vivo vascular bed to adjacent tissue.

The development of microtissue models relies on self-assembly of cells and their native ECM to create dense aggregates (also termed spheroids or organoids) smaller than 300 μm. These microtissues can be formed from one or more cell types (cocultures) in large quantities using ultra-low attachment or hanging-drop microplates.[16] When compared with monolayers, microtissues allow cell-cell and cell-ECM interaction in a 3D environment and permit self-organization in response to cues from biochemical and atmospheric gradients. This self-organization, however, leads to heterogeneity, and the presence of the necrotic core, which may be detrimental to assay robustness.[17] Microtissues offer the ability to observe tissue-level behavior, importantly, in a format compatible with imaging-based high-content screening and analysis with fluorochromes or genetically-modified cells.[8]

An additional commonly adopted 3D model platform employs an exogenous ECM to encapsulate single or multiple cell suspensions in a synthetic or naturally-derived matrix (or hybrid thereof), typical in hydrogel form.[18] Research into exogenous ECM, or hydrogel, development started in the 1960s[19] and is still rapidly advancing.[20] Briefly, this ECM composition may be synthetic (e.g. polyethylene glycol[21]) or naturally-derived[22] from plant (e.g. alginate[23]), animal (e.g. silk[24,25], gelatin[26], collagen[27]), or recombinant (tobacco collagen[28], bacterial collagen[29]) sources as necessary to satisfy particular tissue requirements. Furthermore, the ability to tailor the electrical/mechanical properties[30,31] and functional properties[32,33] has permitted researchers to investigate a variety of applications, including musculoskeletal[34–36] tissue engineering and cancer research.[37–39] However, despite the broad selection and customization of hydrogels, they cannot be assumed to recreate all of the native ECM aspects, such as the presence of appropriate adhesion ligands and proteins as well as the resulting structural organization.[5]

Biofabrication[40] of complex 3D tissue constructs in which combinations of cells[41], ECM components (i.e. bioink[32,35,42,43]), and support structures (e.g. scaffolds for bone[44]) are spatially organized (e.g. for neural[45], vascular[46], or tumor[47–49] targets) in a predetermined manner[50,51] has rapidly advanced and is now an established field.[52–56] These multi-material hybrid constructs[57] are part of a trend to control the spatial organization over multiple cell-types to achieve a more realistic recapitulation of the native tissue organization.[58] This degree of automation and control has enabled 3D cell-based assays to be executed with the reliability and robustness required to advance the technology towards validated replacements for animal testing (https://ntp.niehs.nih.gov/pubhealth/evalatm/accept-methods/index.html).[48,59]

At peak complexity, are microphysiological systems (MPS). [60] MPS combine the biological complexity of spatially-organized tissue models, through biofabrication, with the technical complexity of a supporting environment tailored to replicate aspects of the tissue model’s native niche. MPS are uniquely positioned to examine systemic responses by combining multiple tissue or organ compartments in bioreactors providing native-like stimuli (e.g. mechanical compression in an articular joint model or fluid shear stress in a vascular tissue model) under monitored conditions.[60] A number of articles and reports from academia, the pharmaceutical industry, and government agencies suggest MPS will provide the best opportunity to predict in vivo responses using an in vitro assay[1,60–63], but a multitude of obstacles are hindering development.

In the technical domain of the MPS, i.e. the bioreactor, the obstacles are associated with bioreactor material selection to avoid compound absorption, maintaining sterility, incorporating bubble traps and in-line sensors, and providing each tissue module with the correct flow rate despite being connected in series.[64] Also, the tissue modules must permit the appropriate allometric considerations in terms of proportional tissue and media volumes.[65,66] These technical challenges may be summarized as a need to increase system complexity in order to accommodate target tissues and monitor media conditions for temporal assays, all while maintaining compatibility with external analysis platforms and imaging modalities.

In the biological domain, i.e. the tissue construct, advanced biofabrication techniques[58] may be employed to incorporate vasculature[67], innervation[68], and other local (e.g. mural[69]) and systemic cell types (e.g. immune[70]). The target niche must be fully characterized with the biochemical, electrophysiological, and atmospheric gradients refined to match accordingly.[71] Until the biological and technical challenges are resolved, systems seeking to increase sample numbers must sacrifice capabilities (e.g. perfusion or mechanical stimuli), resulting in an apparent complexity-throughput dichotomy when examining the MPS in the field. This review will discuss these challenges to MPS development and propose collaborations to expedite resolution of these challenges in context of a complete MPS experimental workflow; an MPS-terminated biofabrication line.

Balance of Biological Complexity and Throughput in MPS Development

Current MPS in Terms of Chips and Plates, Tissues and Bodies

In the context of MPS, experimental throughput is a function of the number of independent model replicates residing on a single platform and the speed of tissue production and data acquisition. On the other hand, model complexity is a function of the number of interconnected tissues within each model, the inclusion of biomechanical stimuli, the presence of high-level tissue architecture, and the integration of on-chip sensors. However, to better understand the relationship between throughput and complexity, we quantified the two factors by limiting our consideration to the number of independent models on a single platform and the number of interconnected tissues within each model, respectively (Figure 1). Based on these criteria, we have examined and summarized 10 different representative state-of-the-art technologies in the field and review potential trade-offs between captured complexity and throughput. These technologies with supported sample dimensions are provided in Table 1.

Figure 1.

Interplay between complexity and throughput in representative MPS bioreactors. A body is a fluid circuit containing one or more tissue models. Tissue chips with 1 tissue or tissue interface are represented by the black diamond on the plot with a representative chip.[85] Each point represents a permissible configuration that has been assigned for the purposes of this review. For those systems with interconnected tissue/organ modules, lines connect these points to represent the entire complexity-throughput spectrum of that MPS. Systems with single points, such as those from InSphero, 4Design, and MIMETAS, implemented on-plate replicates (tissue plates) which typically cannot be connected in series to form more complex systems, thus, have been assigned a point representing the right-most end of a horizontal line starting at 1. Tissue/organ-on-a-chip modules representing a single body consisting of a single tissue are at the lower left of the plot (black diamond). At the bottom left are commercial multi-tissue body platforms. At the bottom right of the plot are microplates designed for arrays of independent tissues (one-tissue bodies). The plot reveals a trend in the perfusion mechanisms where high-throughput MPS use gravity-driven perfusion, and high-complexity MPS use pump-driven perfusion. Platform data from: 4Design[81], CNBio[75,77], Draper HOS[87,88], EU BOC[84], MIMETAS[89], TARA IV (InVade) [90], TissUse [73,91], UOtago[80], and UPitt OC[78,92]. Images adapted from respective articles with permission.

Table 1.

Sample geometries utilized in current MPS bioreactors.

MPS Bioreactor	Classification	Body Complexity (max. # tissues/circuit)	Bioreactor Throughput (max. # replicates)	Perfusion Method	Largest Sample Size	Ref
4Design	Tissue Plate	1	16	Gravity- driven	1 mm × 2 mm × 0.12 mm	[93]
EU BoC	Tissue Plate	6	11	Gravity- driven	∅200 μm to ∅300 μm	[83]
MIMETAS	Tissue Plate	1	96	Gravity- driven	4.5 mm × 0.2 mm × 0.12 mm	[82]
TARA InVade	Tissue Plate	2	10	Gravity- driven	4 × 3 × 2 mm	[90]
CNBio	Body Chip	12	10	Internal pump	∅15 mm tissue scaffolds, ∅12 mm Transwell® inserts	[77]
Draper HOS, EVATAR	Body Chip	12	10	Internal pump	∅9 mm Transwell® insert	[73]
TissUse	Body Chip	4	2	Internal pump	∅6.5 mm × 5.6 mm mass, ∅9 mm Transwell® insert	[72]
UOtago	Body Plate	192	192	External pump	∅5.5 mm × 10 mm	[80]
UPitt	Body Plate	96	10	External pump	∅3 mm × 6 mm	[78]

A step up in complexity from tissue chips are body chips, which are multiple tissue chips linked in a continuous fluid circuit. The TissUse/ProBioGen body-on-a-chip systems (TissUse) employed on-chip micropumps to transfer media to 4 tissues either as a 2-tissue circuit in duplicate as demonstrated for discs of cast cell-laden hydrogel[72], or in series as a 4-tissue circuit populated with an intestinal tissue model, an infant skin biopsy, 20 liver microtissues, and proximal tubule cells seeded onto a membrane.[73] Whereas the TissUse system implemented on-chip micropumps, the 2-tissue Homunculus implemented electrodes and real-time fluorescent sensors.[74] A larger variant targeted a 6-tissue absorption, distribution, metabolism, and excretion (ADME) pathway.

The CNBio Single/Multi Organ-on-a-Chip™ plates (CNBio) operated according to a similar principle, adding docking-station capabilities to maintain a microplate footprint while supporting up to 12 tissue scaffolds as single-tissue body replicates[75,76], or interconnected in multi-tissue circuits as demonstrated with tissue scaffolds and Transwell® inserts.[77] The Draper Human Organ System™ (Draper HOS) permitted cultivation of up to 12 tissue modules. Under the US National Institutes of Health (NIH) Tissue Chip for Drug Screening program, the Woodruff lab utilized the system as a base for a female reproductive platform (EVATAR) supporting tissues in Transwell® inserts.[73]

Moving on to body plates, or body chips with on-plate replicates, the University of Pittsburg (UPitt) osteochondral platform, supported by the NIH Tissue Chip program, was designed to recapitulate the bone-cartilage region of the articular joint, as such it was based on a mechanically-stimulated dually-perfused tissue (osteogenic media and chondrogenic media). The total platform capacity as specified in the introductory article was 96 tissue constructs divided into 12 8-tissue bodies.[78,79] Theoretically, these 8-tissue bodies could be placed in series to form a larger multi-tissue body as the system utilized external pump and tubing to connect the sample housing to reservoirs.

A recently-described body plate bioreactor platform for culture of parenchymal and barrier tissue constructs under active perfusion was designed to marry high complexity with high throughput (UOtago).[80] The external pump and the ability to individually-address two fluid circuits in each of the 96 wells, permitted arbitrary tissue combinations between 192 one-tissue bodies and a single 192-tissue body. In contrast to common MPS utilizing external peristaltic pumps, systemic priming volume was minimized by implementing 24-channel peristaltic pumpheads in a microplate footprint between the sample housing and the microplate media reservoirs. The platform was demonstrated for use with hydrogel-based vascular constructs and bioassembled ovarian cancer models.

The last classification, tissue plates, are arrays of tissue chips for on-plate replicates. The 4Design Biosciences Vascularized Micro-Organ Platform™ (4Design) is and example of a microplate designed for arrays of independent tissues (one-tissue bodies), as described in the literature supported 12 modules with 3 tissues in series as replicates[81], while the commercialized version supported 16 modules. The MIMETAS OrganoPlate™ (MIMETAS) supported up to 96 tissues.[82] Perfusion in the 4Design and MIMETAS MPS was achieved via gravity-assisted media transfer as in the EU/FP7 BOC MPS. The European Union (FP7-ICT) Programme “Body-on-a-Chip” collaborative project (BOC, Grant agreement ID: 296257) developed a platform to perform hepatotoxicity testing on liver microtissues. The resulting platform arranged 66 microtissues into 11 6-tissue bodies.[83] Further, more complex, configurations were not possible as perfusion was achieved via gravity-assisted media transfer between on-chip reservoirs on either side of the tissue series.[84]

Trends in MPS Development

Closer examination of the above MPS platforms revealed the technical trends driving the complexity-throughput dichotomy. The systems focused solely on sample numbers (e.g. EU/FP7 BOC, 4Design, MIMETAS) utilized the ANSI/SLAS microplate standards (ANSI/SLAS 1–2004 through ANSI/SLAS 4–2004), and mainly implemented gravity-assisted perfusion in place of an active pump. Additionally, these MPS relied on small, thin tissue volumes with thicknesses between 0.1 mm and 0.12 mm, thus remaining within the focal depth of high-content imaging (HCI) modalities.[94] These ensured compatibility with laboratory robotic and imaging systems for rapid handling and analysis.

The complexity focused systems, i.e. those that permitted interconnections between tissue modules (e.g. TissUse, Homunculus, CNBio, Draper, UPitt), used active pumping. All systems (with the exception of UPitt and UOtago) implemented on-chip pneumatically-driven pumps to minimize the priming volume of the media circuit. The introductory articles to these MPS performed detailed allometric or functional scaling to compare the achieved media-to-tissue volume ratios and perfusion rates with respect to their in vivo application targets, e.g. the Draper-based EVATAR system.[95] MPS not capable of achieving in vivo media-to-tissue volume ratios, would have a potential drawback given the dilution of soluble factors for cell signaling to the exterior of the same cell (autocrine), between cells locally (paracrine), and between tissues (endocrine).[96] With respect to the supported tissue models, the complexity-focused MPS were designed to accommodate tissues in well inserts, either as part of the tissue construct (e.g. with hydrogel in UPitt[79] or with cells in CNBio[77]), as a carrier (e.g. for microtissues in TissUse[73], or for primary tissue in EVATAR[88]), or as a supplied model (e.g. MatTek EpiIntestinal™ in TissUse[73]).

In general, the advantages of gravity driven flow include simplicity of operation and ease of removing bubbles in open well plate configurations, targeting users such as scientists in pharmaceutical companies who could use the platform for drug discovery and disease modelling. A regular work flow in pharma includes screening in well plates, high content imaging in machines that accommodate well plates as well as fluid handling by pipetting robots and ultrasonic fluid dispensers. It is important to consider that pumps have not been included in a regular pharma workflow. The drawbacks of gravity driven flow include a limited range of flow-rates available in any single device, inability to achieve pulsatile flow and transient application of perfusion. Flow driven by syringe or peristaltic pumps provide a solution for these limitations, albeit at the expense of requiring specialized equipment and incorporation of bubble traps.

Lastly, the MPS considered in Figure 1 do not include on-chip sensors, with the exception of the Homunculus.[97] However, given developments with single-tissue chips, future devices may monitor cellular activity across a spectrum of technologies.[98] A recent full-featured collection of modules implemented a suite of electrochemical, physiological, and biochemical sensors to monitor and assay conditions.[85] Other systems have implemented functionalized electrodes to quantify glucose and lactate concentrations in microtissue cultures[99], an in-line enzyme-linked immunosorbent assay (ELISA) to quantify hepatocyte albumin secretion[100], and mass-spectrometry compatibility was implemented to measure soluble factors secreted by adipocytes[101] and Jurkat T-cells.[102]

The UOtago platform focused on balancing complexity and throughput by adopting the ANSI/SLAS microplate format for both the sample housing and the media reservoir to maintain compatibility with standard microscopy and liquid handling platforms, while retaining region-specific independence of media in tissue models carried by well inserts (similar to those utilized by UPitt).[80] In addition, the platform expanded non-invasive sample assessment to high-resolution computed tomography (e.g. μCT) via modular microplates and leveraged a docking station to permit in situ imaging during perfusion. For perfusion, reliance on an external peristaltic pump caused the priming volume of each circuit to remain 10× higher than the near-physiological values achieved by the EVATAR system[95], however, the docking station also permitted the tissue module interconnections similar to the UPitt system in external tubing, but at an expanded degree of flexibility by permitting an arbitrary number of tissues in series (1 to 192) with varying degrees of modulated crosstalk.

Over the past decade, government initiatives have supported the development of MPS technology, largely through the US NIH Tissue Chip[71] and EU/FP7 BOC programs (with AstraZeneca).[83] The systems have reached a state of maturity to permit comprehensive validation initiatives. Future developments, such as implementing onboard sensor technologies, may facilitate translation of these body chips into body plates for high-throughput studies.

Opportunities for MPS-terminated Biofabrication Lines at the Convergence of Biofabrication Techniques and Bioreactor Technologies

Biomechatronic design methodology correlates existing technologies and techniques to fulfill the end biological and technical requirements[103]; when applied specifically to MPS development, one may identify the rate limiting steps of the experimental workflow and pinpoint the causes to the complexity-throughput dichotomy discussed in the previous section. In Figure 2, the MPS experimental workflow is portrayed as a biofabrication line to clarify the movement of cells and constructs (blue arrows) through laboratory equipment (rectangles). Through this representation, the degree of integration in tissue construct and bioreactor production (in the box labelled Bioprinter) may be weighed against the overall workflow. For example, in the fully-integrated situation where the bioprinter concurrently fabricates the bioreactor components and tissue construct to eliminate intermediate loading steps, as in a recently described liver chip[104], the labor savings in the overall workflow alone may not warrant the imposed constraints on bioreactor materials and fabrication method. This will be discussed in more detail at the Build stage.

Figure 2.

Stages of a biofabrication line. Arrow key: black is workflow direction, grey is movement of bioreactor, blue denotes movement of all other physical entities such as cells or constructs, orange denotes flow of information, and yellow encompasses all forms of biochemical and energetic stimulation. The Bioprinter details three possible degrees of integration of bioreactor and tissue construct (bio)fabrication; decoupled, partially-integrated, and integrated.

Stages 1 and 2: Design and Prepare

Key steps in the Design stage could involve conversion of data from in silico studies investigating basic biology and elucidating systemic interactions (Figure 2, in silico), prior data from in vitro/vivo studies, and known aspects of human and animal physiology (combined in Figure 2 as Prior Data) into a desired spatial organization of cell-laden (e.g. microtissues) and acellular (e.g. polymer) constituents of a tissue-engineered construct. At this point, the manufacturing methods would be decided amongst one or more common techniques described in to achieve the desired approximations of the composition, organization, and niche of the target tissue or organ achievable with the available fabrication modalities, materials, and cell sources. The construct setpoint would then be converted into a set of instructions for the bioprinter using computer-aided drafting/manufacturing (CAD/CAM) software. Here execution speed would be dependent on algorithmic efficiency and IT infrastructure.

Key steps for the subsequent Prepare stage include purification of cell populations, expansion of the purified populations, differentiation to the desired lineages, and if necessary, combination of cells with additional, sterile, material such as extracellular matrix (ECM) components and/or hydrogel. The resulting feedstock for the next stage may include cell-laden bioink [57] or 3D cellular modules such as microtissues (dense collections of cells held together with autologous ECM), microspheres (cells encapsulated in exogenous ECM), or even microspheres encapsulating microtissues.[109] For tissues utilizing primary cells rather than cell lines, a population of donor cells, either primary or differentiated progeny of induced pluripotent stem cells (iPSC), could be purified to isolate the populations with the phenotypes for the target tissue model (Cell Selector). For both primary and immortalized sources, the cells may be expanded on microcarriers, tissue culture plastic, or hollow fiber reactors to achieve the yield necessary to create a tissue with the desired volume and cell density (Expansion System). Depending on the construct geometry, the cell populations would then be mixed with a macromer suspension in any cell culture vessel to form bioink (Bioink Mixer), seeded into microtissue-forming microplates and matured (Microtissue Generator), and/or encapsulated in hydrogel and passed through a spray nozzle or multiphase microfluidic chamber and crosslinked to form microspheres (Microsphere Generator).

Previously described devices and technologies exist for each of the proposed stages above.[110] For example, the StemCell RoboSep™ for cell separation[111] and the Sartorius CompacT SelecT™[112] or bespoke systems[113] to expand cell lines in a variety of culture vessels. Fluid handling stations paired with hanging-drop microplates can generate microtissues (e.g. PerkinElmer Zephyr® G3 with InSphero GravityPLUS™[114]), microfluidic systems can generate microspheres (e.g. Dolomite μEncapsulator 1™[115]), and generic fluid handling stations can mix bioinks. When placed into context of the convergence of the biofabrication and MPS fields, a high-throughput MPS supporting biologically-complex tissue models will require an extensive Prepare stage. The cell quantities to achieve the throughput will lengthen the expansion step, while the increasing number of different cell types or 3D cellular modules will require researchers to spend longer periods of time interacting with, and transferring material, between more pieces of equipment. Fortunately, the burden on the researcher may be alleviated by automating and integrating the separate preparation steps in a single robotic laboratory cabinet such as the BioNex Hive™.[116]

Stages 3 and 4: Build and Culture

The Build stage can utilize additive manufacturing technologies to transform construct blueprints into a construct using a series of dots and fibers of possibly cell-laden bioinks, adopting the (bio)fabrication techniques as specified in the blueprint (). Execution speed of the build stage would be a function of production speed, and entity transference into and out of the bioprinter. These key factors would be influenced by the degree to which tissue construct biofabrication and bioreactor fabrication are integrated. This degree of integration would strike a balance between expediency and limitations of the available production techniques and resolution, as well as bioink and (bio)material properties (Figure 4).

Figure 4.

Degrees of integration for tissue construct and bioreactor production with examples of decoupled[78], partially-integrated[55], and integrated[104] methods. Graphics adapted from the respective articles with permission.

In a decoupled approach, the tissue constructs and bioreactor are manufactured separately. This approach offers the greatest flexibility with regards to manufacturing techniques keeping both open to future advances: the bioreactor can leverage proven methods for mass production (e.g. injection molding, computer-controlled machining) no matter the incompatibility of that method to biologic entities and without constraining the choice of biofabrication technique, that is, as long as the final construct is structurally-sound and compatible with pick-and-place (PnP) robotics. Likewise, construct fabrication itself may be decoupled, as in bioassembly[40], to combine techniques reliant on tightly-controlled build conditions (e.g. melt electrospinning to create stiff scaffolds with high spatial resolution[117]). Additionally, considerable attention must be paid to the design of the bioreactor and construct temporal geometries to guarantee robust anastomosis.

In terms of biofabrication speed, material loading and construct transfer to the bioreactor may be performed manually by a technician. Few bioprinters (e.g. Brown University Bio-P3[118], Cyfuse Regenova, Advanced Solutions BioAssemblyBot, and nScrypt 3Dn) specifically include PnP end effectors to automate the latter, while at this moment only the BioAssemblyBot has the movement envelope to handle both. In addition to the lack of automation, the biofabrication process itself is extremely slow. The fastest known techniques (extrusion of bioink and stereolithography printing of bioresin[32]) have only achieved volumetric deposition rates of 0.5 cm³ min⁻¹.[119] The complexity-focused MPS reviewed in the previous section utilized this decoupled approach. For example, the TissUse system[73] relied on prefabricated MatTek tissue models, and ex vivo tissue located in transwells.

The flexibility of the decoupled approach is represented visually in Figure 5. This broad range of currently permissible combinations of bioreactor fabrication and construct biofabrication methods (black border cells) offers a promising route to achieve high-throughput MPS with highly-complex tissue constructs. In the future, the decoupled approach may become even more flexible through adaptation of powder-bed technologies (3DP and SLS) to biofabrication (Figure 5, red border cells), through the use of biopowder. Given bioresin is a cell-laden SLA resin and bioink is cell-laden ink for 3D plotting, biopowder is cell-laden 3DP or SLS feedstock. Biopowder could be microspheres encapsulating cells in hydrogel, as currently produced by the Dolomite μEncapsulator 1™.[115] In 3DP, biopowder would be suspended in density-matched carrier fluid at a high powder-carrier ratio (a highly-viscous emulsion) to ensure the powder particles remained in contact yet restricted from fusing spontaneously. The printer head would deposit nanoscale volumes of high-viscosity exogenous ECM at particle interfaces, simultaneously forcing the carrier fluid out of the interface and bonding the particles. Similarly, in SLS, the biopowder would be suspended in acellular photocrosslinkable resin. The laser would fuse neighboring particles by selectively crosslinking the resin between said particles. While this is not strictly sintering due to the additional material, neighboring particles of a powder bed are still fused through the controlled application of light. In short, future 3DP or SLS bioprinters may be applied to create the cellular components of tissue constructs.

Figure 5.

Degrees of integration as related to (bio)fabrication technologies. terms in alphabetical order as defined in [119] where appropriate: 2PP is two-photo polymerization, 3DP is 3D printing, 3D Plot is 3D plotting of extruded material and includes PAM, Cast is any technique reliant on intermediate components such as a disposable mold, includes injection-molding, DLP is digital light processing, ES is electrospinning, FDM is fused deposition modeling, also known as 3DF, Jet represents all jet printing including inkjet and valve-based, Jig is a build platform contacting the component on more than one surface which includes a machine chuck or removable mold, LAB is laser-assisted bioassembly also known as LIFT and MAPLE, MA is general modular assembly, includes bioassembly with spheroid-based approaches such as Kenzan, PAM is pressure-assisted microsyringe dispensing and wet-spun technologies, PnP is pick-and-place robotics, SLA is stereolithography, SLS is selective laser sintering, Sub is the broad field of subtractive machining technologies, including laser ablation.

On the opposite end of the spectrum is the integrated approach; a specialized example of lean manufacturing in which the construct and bioreactor are produced simultaneously. While this approach is logistically riskier due to the reduction of permissible pause points (e.g. the inability to manufacture bioreactors en masse via injection molding and store for later use), the construct and bioreactor geometries may be highly entangled. This high degree of entanglement, or integration, reveals two important advantages: the complexity of the construct-bioreactor interface may be increased, and the material required for structural integrity of soft constructs may be reduced possibly increasing the sample density, hence throughput, in the same bioreactor footprint.

Emerging examples of this approach (Figure 5, grey cells) rely on nozzle dispensing print technologies (e.g. 3D Plot, ES, FDM) and/or component manipulation via PnP (MA). Specifically, recent studies created individual perfused tissue models targeting liver[104,120], cancer[121], and vascularized kidney.[122] It has also been applied to skin[123] and neural[124] constructs for static (non-perfused) wellplate cultures. Current research into novel biomaterials suitable for biofabrication feedstock[57] ensures researchers have an expanding pallet with which to satisfy MPS structural design requirements while mitigating exposures to cytotoxic volatiles Conversely, utilization of volumetric technologies to increase print speed (e.g. DLP), or metals to increase mechanical strength (e.g. sintered titanium), is unlikely given the relative incompatibility of fabrication conditions. The future for this approach (Figure 5, red cells) may be to leverage the increased printing resolution of existing planar dispensing biofabrication technologies (e.g. Jet or LAB) to further increase the complexity of the construct-bioreactor interface, or to create smaller features in either component.

Between decoupled and integrated, a partially-integrated approach allows for the fabrication of the construct directly inside the bioreactor. It permits soft or amorphous constructs and facilitates a mechanical coupling between the construct and mechanical stimuli (e.g. seeding cells onto a membrane to undergo a tension regime to mimic a lung[125]). From an execution speed perspective, this technique has the advantage of permitting mass production methods for bioreactor fabrication and also removes the construct transfer step, however it still imposes constraints on the bioprinter print heads (e.g. an extrusion nozzle or jetting head must reach the bottom of the sample chamber in the bioreactor). Thus, while the bioreactor may be fabricated with any technique, construct biofabrication is limited to nozzle dispensing and PnP-based techniques (Figure 5, black patterned cells). With the advent of bioresins[32,126], modified SLA, DLP, or printers may facilitate the inclusion of these techniques with open-top bioreactor geometries clear of internal opaque obstructions (Figure 5, red patterned cells). Likewise, the introduction of new tomographic DLP printing technology[127] may accelerate development of a light-based partially-integrated approach, provided the bioreactor materials are transparent to the relevant wavelengths[128,129]. Therefore, in the future, mass produced bioreactors may be combined with the most rapid high-resolution biofabrication methods resulting in a streamlined MPS biofabrication line centered around constructs with increased physiological relevance.

Lastly, an MPS is formed once a construct is located inside the bioreactor and ancillary control and monitoring components are connected. Final MPS complexity is a function of the complexities in the technical and biological domains. An increase in biological complexity (i.e. vertical movement toward Future Directions in Figure 1) may be considered as the drive to create physiologically-relevant tissue constructs. Similarly, the technical domain (i.e. lateral movement toward Future Directions in Figure 1) encompasses the efforts to perform the following for an ever-increasing number of samples: generate and analyze data for both maintaining construct health (e.g. pH, dissolved oxygen and metabolite levels), observing the impact of induced stimuli of interest (e.g. paracrine levels, localization of damage-sensing fluorescent markers), and the degree to which automation is used to perform routine tasks (e.g. to apply stimuli and refresh culture media).

Concluding Remarks and Future Directions

Microphysiological systems are emerging platforms to predict in vivo behaviors by means of in vitro experimentation. However, further developments are necessary before the systems can achieve both high model complexity representative of human physiology and high experimental throughputs, resulting in the current complexity-throughput dichotomy. In the short-term, the context of the biofabrication line may be used to guide the design of tissue models and bioreactors to match the tissue requirements (e.g. mechanical coupling, intact removal for histological assessment) to the implemented construct-bioreactor production integration approach. A decoupled approach offers an opportunity to utilize the widest range of biofabrication methods to both maximize production speed and construct complexity, with the notable requirement that the resulting construct must survive a transfer into a receptive MPS. The integrated approach overcomes this need for structural integrity thereby permitting constructs closer to their target in vivo tissue. However, in this approach, fewer techniques for construct and bioreactor production are interoperable.

One potential future direction in biological modeling is the creation of vascularized functional organoids where dynamic perfusion can be applied through an internal vasculature integrated within a sophisticated stem cell derived organoid. Current solutions so far all adopt a decoupled approach. Because of the specific requirement on the bio-matrices and the complex differentiation protocol, organoids are grown separately and then transferred to an organ-on-a-chip device with an open-well design.[130,131] Although we have seen impressive success in this area, to establish perfusable vasculature that will fully integrate with organoids, a partially-integrated approach might be necessary to seamless integrate organ-on-chip bioreactors with the organoid differentiation and maturation process. This strategy could avoid the need to transfer fragile organoids tissues which introduce arbitrary interruption in the tissue development process. Future advancements in bioresin formulations that are more suitable for stem cell culture and bioprinter hardware may position the partially-integrated approach as an attractive compromise between decoupled and integrated (see Outstanding Questions). The corresponding MPS-terminated biofabrication line could be automated, employ rapid high-resolution biofabrication technologies, and utilize mass-produced bioreactors to stimulate and monitor an array of near-physiological tissue constructs – thus resolving the complexity-throughput dichotomy. It’s also important to note that the application of producing large arrays of complex functional tissues and maturing them on-chip is not restricted to drug testing or fundamental biological studies. If a platform allows tissue extraction for downstream analysis via a partially integrated approach, the same capability will also enable tissue extraction for implantation with a modular tissue engineering approach. Tissue engineering in drug discovery and regenerative medicine are just two sides of the same coin.

Figure 3.

Classes of Additive Manufacturing technologies commonly adopted for MPS production utilizing acellular (blue) (bio)material feedstock (a, b, d, g, h, i, l[105] and c[106]), which include subsets of Biofabrication technologies for cell-laden (orange) bioinks (e, f, j, k, m, n)[107], or hybrid combinations thereof (o[108], p[109]). Images adapted from respective articles with permission. To give an insight into the various layer-by-layer fabrication approaches, each Additive Manufacturing (blue) and Biofabrication (orange) technology can generally be assigned to 4 technology classes: (i) Laser- or Light-based Technologies: These Additive Manufacturing technologies consist of: a) Selective Laser Sintering (SLS) of powders, b) laser Stereolithography (SLA) onto liquid resins, (c) Digital Light Processing (DLP) of a projected mask of light onto liquid resins, d) high-resolution Two-photon Polymerization (2PP) of laser pulses into photosensitive resins. The subset of Biofabrication technologies include: e) Lithography-based Bioprinting that adopts SLA/DLP approaches combined with cell-laden photosensitive resins, f) Laser-assisted bioprinting or laser-induced forward transfer (LIFT) that uses pulsed laser to selectively project cell-laden droplets onto a substrate.

(ii) Extrusion-based Technologies: These Additive Manufacturing technologies consist of: g) Fused Deposition Modeling (FDM) of a (bio)material filament spool extruded from a heated nozzle, h) Extrusion 3D Plotting (3DF) of a molten (bio)material under pressure or displacement from a heated nozzle, i) Solution-electrospinning of randomly arranged (bio)material/solvent nano-filaments from high voltage spinneret onto a collector, or alternative Melt-electrowriting (MEW) of ordered molten (bio)material micro-filaments from high voltage heated spinneret onto a computer controlled x-y-z collector. The subset of Biofabrication technologies include: j) Extrusion 3D bioprinting of cell-laden (shear thinning) bioink under pressure or displacement from a temperature controlled nozzle often followed by subsequent photo-crosslinking, k) Granule-based medium-assisted bioprinting cell-laden bioink into a granule-based or buoyant bath or support medium.

(iii) Jetting- or Powder-based Technologies: These Additive Manufacturing technologies consist of: l) 3D Printing (3DP) combines inkjetting droplets of a binder material onto a (bio)material powder bed. The subset of Biofabrication technologies include: m) Inkjet bioprinting using thermal or piezoelectric jetting head to jet cell-laden bioink droplets onto a surface, n) Multi-jet bioprinting is similar to Inkjet bioprinting with multiple jetting heads and multiple cell-laden bioink droplets.

(iv) Hybrid Biofabrication Technologies: These rapidly emerging technologies utilize combinations of the above Additive Manufacturing and/or Biofabrication technologies to create more complex hybrid constructs or 3D tissue models. For example, o) Hybrid 3D Bioprinting of molten (bio)material filament (or other light- or powder-based technologies above) with multiple cell-laden bioink filaments via extrusion, typically using multiple combinations of temperature controlled extrusion nozzles as well as (shear thinning) bioinks with different cells, p) 3D Bioassembly of tissue spheroids can involve combinations of extrusion 3D Bioprinting molten (bio)material filament (or other light- or powder-based technologies above) with automated (bio)assembly of tissue spheroids/organoids or cell-laden bioink spheroids typically fabricated using high throughput methods, whereby each spheroid can contain different cells and/or bioink formulations.

Outstanding Questions Box

• Can we overcome the trade-offs between the complexity of 3D biological models and the experimental throughputs and solve the complexity-throughput dichotomy?

• Can we yield greater flexibility to the biological model in accordance to experimental needs by integrating or partially integrating the fabrication of tissue constructs with the design of bioreactors and MPS?

• Can we further improve the biological relevance of our engineered tissue models with the routine use of primary patient-specific cells or induced pluripotent stem cells?

Highlights Box

• A range of advanced biofabrication techniques have been developed in recent years to enhance our ability to position cells and extracellular matrices in 3D, leading to the construction of increasingly complex tissues.

• Complex organ-on-a-chip devices with modular design and advanced fluid circulation control has been developed to model complex cellular communication in tissues as well as inter-organ interaction.

• The need for more advanced tissue models with greater structural complexity and dynamic external stimulation motivated the convergence of biofabricated tissues and organ-on-a-chip devices in recent years.

Glossary

Bioassembly

‘The fabrication of hierarchical constructs with a prescribed 2D or 3D organization through automated assembly of pre-formed cell-containing fabrication units generated via cell-driven self-organization or through preparation of hybrid cell-material building blocks, typically by applying enabling technologies, including microfabricated molds or microfluidics.’

Biofabrication

‘The automated generation of biologically functional products with structural organization from living cells, bioactive molecules, biomaterials, cell aggregates such as micro-tissues, or hybrid cell-material constructs, through Bioprinting or Bioassembly and subsequent tissue maturation processes.

Biofabrication line

A collection of laboratory equipment employed to convert computer models, cellular material, and acellular material, into one or all components of a microphysiological system in a maximally-automated manner.

Bioink

‘A formulation of cells suitable for processing by an automated biofabrication technology that may also contain biologically active components and biomaterials’.

Bioprinting

A subset of Biofabrication technologies using computer-aided transfer processes for patterning and assembling living and non-living materials with a prescribed 2D or 3D organization in order to produce bioengineered structures serving in regenerative medicine, pharmacokinetic and basic cell biology studies.

Body chip, Body plate

A body chip is a bioreactor housing a collection of tissues (or tissue/organ constructs) connected by one or more fluid circuits. A body plate is a device with multiple onboard replicates of a body chip.

Hybrid construct

A tissue construct composed of multiple bioinks or biomaterial-inks fabricated using one or more techniques.

Microsphere

In the field of biofabrication, a microsphere is a collection of cells encapsulated in spheroid of exogenous extracellular matrix (typically a hydrogel), either natural or synthetic, typically used as building blocks in bioassembly. In this article, the same spheroid without cells is a cell-free microsphere.

Microtissue

A collection or aggregate of cells (typically spheroid shape) with a high degree of cell-cell contact, bound by secreted autologous extracellular matrix. Prior to unification of terminology by the International Society for Biofabrication, microtissues were also known as spheroids or cell aggregates.

Microphysiological system (MPS)

‘a microfluidic device capable of emulating animal biology in vitro at the smallest biologically acceptable scale, defined by purpose.’

Tissue chip, Tissue plate

A tissue chip is a bioreactor housing a single tissue/organ construct. A tissue plate is a device with multiple onboard replicates of a tissue chip.