Vascularized organ bioprinting: From strategy to paradigm

Abstract Over the past two decades, 3D bioprinting has become a popular research topic worldwide, as it is the most promising approach for manufacturing vascularized organs in vitro. However, transitioning from bioprinting of simple tissue models to real biomedical applications is still a challenge due to incomplete interdisciplinary theoretical knowledge and imperfect multi‐technology integration. This review examines the goals of vasculature manufacturing and proposes new strategic objectives in three stages. We then outline a bidirectional manufacturing strategy consisting of top‐down reconstruction (bioprinting) and bottom‐up regeneration (cellular behaviour). We also provide an in‐depth analysis of the four aspects of design, ink, printing and culture. Furthermore, we present the ‘construction‐comprehension cycle’ research paradigm and the ‘math‐model‐based batch insights generator’ research paradigm for the future, which may have the potential to revolutionize the biomedical field.


| EXCITING GOAL: MANUFACTURING ORGANS IN VITRO
The in vitro manufacturing of human organs is anticipated to bring about a revolution in biomedical fields such as organ transplantation, drug development and pathophysiology emulation. Bioprinting has emerged as one of the most promising biofabrication 1 strategies for tissue engineering and regenerative medicine,2,3 owing to its capacity to precisely arrange cells and biomaterials in three-dimensional space.
Despite the remarkable advances in science and technology, the intricate nature of human organs still constitutes a major impediment to in vitro organ manufacturing. Heran Wang and Xin Liu contributed equally to this work.

| Three-step objectives
We propose three objectives for constructing human organs in vitro: (1) Short-term goal (in 5 years): creating vascularized, implantable, volumetric organs; (2) Long-term goal (in 15 years): producing functional, transplantable, full-size organs; (3) Ultimate goal (in 30 years): achieving clinical, patient-matched, autogenous organs. Early bioprinting efforts with regards to microvasculature have focused on constructing a hollow lumen and forming an endothelial monolayer. 4 Currently, scientists are further investigating the creation of vascular networks, that can supply nutrients and oxygen to volumetric The solid organs are volumetric and multi-scale in structure with multi-tissue composition, multi-cellular interactions and multi-level tubular networks. From a reverse engineering standpoint, the biological organism is the most difficult to replicate due to their spontaneous emergence as a complex systems at many levels, such as tissues, multi-cellular structural units, cells, organelles and biomolecular structures. 10

| Microcirculation: a central objective in bionics
The key to sustaining volumetric tissue activity is to reconstruct the microcirculatory system, which is a network of arterioles (<0.3 mm), capillaries and venules (<0.2 mm). Vasculature reconstruction can be decomposed into three stages: vasculogenesis, angiogenesis and vascular remodelling. To facilitate this multicellular self-organization process, biophysical and biochemical parameters, such as extracellular matrix (ECM) viscoplasticity, vascular endothelial growth factor (VEGF) gradients, 6

| Vasculature from functional perspectives
Vascularization is ultimately necessary for achieving the functions of tissue oxygenation, nutrient delivery and waste disposal. 7 Natural selection has tended to maximize both metabolic capacity (by maximizing surface area for exchange) and mechanical efficiency (by minimizing transport distances and time). 8 In this case, organisms have evolved fractal hierarchical branching vascular networks that terminate in capillaries, which must eventually be located within 200 μm of their target cells, depending on the maximum distance of diffusion of critical substances in vivo. 8 Mimicking vasculature based on its functional goals, rather than blindly copying its hierarchical structure, is essential for our success in reconstructing it.

| Artificial vascularization
The reconstruction of vascularized tissues in vitro should aim to replicate natural conditions to the greatest extent possible, such as nutrient supply kinetics, blood flow mechanics and developmental dynamics. Nevertheless, natural vasculatures develop in a stage-wise manner during embryogenesis, whereas artificial vasculatures must be able to provide nutrient supply immediately upon bioprinting. This challenge is akin to a Mars mission for bioengineering, 9 possibly even more complex due to its intricacy at the micro-scale. We suggest a 4-level capability for vasculature morphology fabrication with increasing precision: (1) coarse simple planar branches (1 mm); (2) fine, complex three-dimensional networks (0.5 mm); (3) dense finegrained endothelial networks (250 μm); (4) volumetric microvascular network anastomosis with capillaries (50 μm).

| Dual deficit in scientific knowledge and bionic technologies
Although molecular and cell biology have seen rapid advances in recent decades, the complexity of multi-biomolecular and multicellular interactions at the tissue and organ levels, as well as the complex time-dependent dynamics of development, still leave us with an inadequate understanding of the tissues and organs to be mimicked in bionics. 10 Moreover, simply mimicking mature organ morphology may not be sufficient, and researchers are beginning to recognize the potential need to target earlier stages of organ development. 11 Therefore, questions on how to reconstruct multiscale microenvironments and macrostructures have been challenging to answer. 12 Figure 2) does not aim to completely mirror a mature organ, but instead to craft a regenerative environment with biophysical and biochemical cues to direct cell behaviour. Certain biological elements can be incorporated into the design of bioinks, such as cell sources, growth factors, cell adhesion ligands and mechanical properties. Geometric elements such as matrix fibre and signal distribution, and vascular topology, can also be printed by design.
Interestingly, the lack of ability to print at capillary-scale resolution is often viewed as a crucial hurdle. Nevertheless, capillaries cannot be constructed through direct printing since even if a 10-micron tube could be printed, ECs with a similar diameter would be unable to perfuse it to effect vessel wall endothelialization. What then is the minimum bioprinting precision needed to reconstruct a well-functioning, phenotypically accurate and reproducible organ system? We posit that the smallest printable duct should be at least several times the diameter of the ECs, thus the minimum 'sufficient resolution' for bioprinting is approximated to be 50 μm. Additionally, it is noteworthy that cells suspended in bioinks are nearly spherical in shape, while F I G U R E 1 Schematic diagram of the RR framework for in vitro organ biofabrication, with a focus on information correspondence.
F I G U R E 2 Schematic of a 3D bioprinting configuration.
after bioprinting and growth, they gradually differentiate and flatten into the 'ultimate state'.

| Bottom-up regeneration: cellular selforganization
From as early as the design stage, we should consider the dynamic conditions needed for regeneration, such as growth factor sustained-release, oxygen gradients, morphogenesis, blood flow, etc.
Nevertheless, many biological issues remain to be explored and understood; thus, the current rule-of-thumb is to remain as close as feasible to the in vivo environment. Throughout perfusion culture, cells autonomously respond to the surrounding mechanical and chemical environment to generate tissue-level morphogenesis, such as vasculogenesis and angiogenesis ( Figure 3). For complex biosystems, we must employ devices equipped with quantitative detection tools to monitor and control all relevant parameters. An analogy may be used to comprehend the dynamical control relationships between the in vitro culture device and the cultured tissue: the in vitro culture device is analogous to the pregnant mother, the cultured tissue is analogous to the foetus, and when the tissue is thoroughly developed and ready to be used in vivo, it is comparable to the birth of the foetus.

| Design for vascularized tissues/organs
Design processes function as a compass for organ manufacturing, deciding the tissue's ultimate biological function. Nevertheless, current design research predominantly concentrates on simple geometric morphological sketching. We posit that bioprinting is entering a function-oriented and model-based designable phase, which could have a great impact on biofabrication.

| Biophysical models
1. Substance diffusion model. 17 The nutrient exchange functionality of vasculature relies on diffusive and convective solute transport ( Figure 4). Recently, a parametric characterization based on the metabolically active (Krogh) radius has been unearthed, 18 which is a comprehensive index combining the impacts of cellular matrix permeability, cell density and metabolic intensity. Literature frequently references capillaries with a maximum distance, 5 which is practically twice the Krogh radius; however, in vitro organs possess superior matrix permeability, lower cell density and lower cellular metabolic intensity compared to in vivo, resulting in an enlarged Krogh radius, which theoretically denotes the design basis for the vascular network density of in vitro tissues. Given that nutrient exchange is situated at the vessel surface, we propose that the 'vascular surface area' coupled with the 'vascular surface area per parenchymal tissue unit' quantifies the functional-oriented geometric traits of the vasculature. Moreover, the design of concentration gradient fields of biochemical molecules can be computed and simulated based on the reaction-diffusion model, which can be referred to as Turing pattern related studies. 19 2. Hemodynamic model. 20 Vascular networks that are not hemodynamically compatible are susceptible to thrombotic issues since blood clotting is sensitive to the mechanical state of the vasculature. 21 For example, rough vessel wall surfaces and nonstreamlined ducts can lead to turbulent flow, creating high local shear stresses and prompting a platelet clotting reaction. Murray's law, derived from the principle of minimum action in mechanics, is a beneficial guide for the structural design of branches, and has yielded the vessel wall shear stress (WSS) set point theory (SPT, Figure 5). The forces that blood flow exerts on the vasculature affect cellular behaviour, such as EC sensitivity to WSS, and SMC sensitivity to circumferential tensile stress, resulting in transformations in the vasculature's short-and long-term morphology. 22 3. Vascular development model. Microvascular remodelling adheres to the WSS SPT, and ECs typically behave as WSS sensors (sensor-pathway model and tensegrity model), 22,23 which tend to adjust vessel diameter to maintain a stable level of pressure and WSS. Simultaneously, upstream and downstream responses must also be considered in order to finish a computable vascular development model (Figure 6). 20 Naturally, the vascular development process can be computationally simulated through building mathematical models to comprehend these biological mechanisms and form an automated vasculature design algorithm.
F I G U R E 3 Schematic of bottom-up cellular self-organization. Reproduced with permission. 16 Copyright 2021, WILEY.
F I G U R E 4 Basic principles of solute transport to tissue.

| Design methodology
Organ design is markedly different from traditional industrial design due to its information-richness in three-dimensional space. To accommodate the 3D printing process as well as the dynamic computable specifications, we propose that the model foundation for organ design should be a voxelized multidimensional information digital model ( Figure 7). 24 We predict that the philosophy of organ design will gradually transition from simple to complex systems. 28 Thus we should formulate biophysical equations and cellular behaviour models based on biological principles, by using straightforward algorithmic rules to simulate and calculate tissue patterns. 25 In this way, the design methodology will evolve from principle-based to model-based, from static analysis to dynamic simulation, and from 'structure-oriented' to 'structure-function integration oriented'. It is foreseeable that modelbased computable digital designs will propel the field of in vitro organ manufacturing to become more scientific and inspiring.

| Bioprinting inks for vascularized tissue
Bioprinting inks encompass bioinks and biomaterial inks (mainly assistive materials, including sacrificial inks and support baths 26 ); the distinction lies in whether they contain cells. 27 Much literature has been generated regarding the development of bioprinting inks; nevertheless, the material properties that are essential for fabricating cellmaterial constructs that accurately imitate biological function, remain indeterminate. 28 In addition, the R&D on biomaterials is largely search-based rather than function-oriented, drawing inspiration from ECM, food additives, cosmetics, or even industrial products to attain novel properties through blending and modifying.

| Bioink material
Bioink materials should not only possess fundamental properties such as printability, crosslinkability, structural stability, cytocompatibility and cell blendability, but also properties that stimulate cellular behavior. 5 Cellular behaviours (e.g., migration) can be contingent on ECM viscoplasticity (i.e., viscosity, 29 elasticity and plasticity), which is a near-universal mechanical feature that requires an understanding of porosity, degradation, 30 and dynamics, and which is indispensable for the replication of human tissue properties. 31 Moreover, the local properties rather than the global properties of the material are pertinent to the cell-material interaction behaviour, which necessitates increased focus on the sophisticated structure of natural tissue ECM.
Bioinks are primarily natural materials (Table 1) or even decellularized extracellular matrix (dECM), 32 complete with cell adhesion ligands, 33 natural signalling capabilities and mechanical characteristics similar to those in vivo ( Figure 8). Nevertheless, synthetic materials, such as PEG (polyethylene glycol), also have immense potential owing to their stability, programmability, 34 and medical availability, especially if we can fully uncover the target properties of matrix materials through reductionism. 4 Furthermore, after printing, the bioink can be crosslinked physically or chemically, 35 to obtain a microscopic network, which can significantly and programmably influence cellular behaviour.
F I G U R E 5 Classical WSS set point theory according to the concepts of control theory. Reproduced with permission. 22 Copyright 2020, Frontiers.
F I G U R E 6 Responses to stimuli on microvascular diameters.
F I G U R E 7 Voxelized multidimensional information digital model for organ (in vitro) design. Reproduced with permission. 24 Copyright 2021, IOP.

| Cells in bioink
The majority of tissues comprise a variety of functional and supporting cells. In addition to the requisite functional cells, tissues also contain cell types that provide support, perform structural or barrier functions, form blood vessels, or support the maintenance or differentiation of stem cells. 45  The cell types employed in bioinks must be able to replicate the target cell types various functions and be expanded in vitro to largescale organ printing quantities. 48 For long-term applications, the printed cells must adapt to all physiological conditions, such as shear stress, enzymes, etc., in culture or during use. 49 They must also be resilient enough to endure the printing process or have sufficient proliferative capacity to preserve cell numbers through self-renewal.
It is essential to recognize that the incorporation of cells modifies

| Sacrificial inks
Sacrificial or fugitive inks were introduced in the 2010s to sustain vessels throughout printing process and subsequently removed. The sacrifice mechanism typically includes aqueous dissolution, thermal gelation and melting and physical crosslinking disruption (Table 2). In bioprinting, the essential requirement for sacrificial inks is excellent printability and cell-compatible removability. There is also another classification for the use of sacrificial inks: (1) support materials, that is, as external auxiliary supports for non-regular structures with very low removal requirements or even manual peeling; (2) soluble core materials, which are sacrificial inks specifically designed to print microvascular networks thus require excellent removability, preferably in a phase change to liquid; (3) reinforcing materials, as components to temporarily improve the printability of bioinks; and (4) porosifier materials, which serve as phase separation components that produce cell-friendly pores for bioinks after removal. 53

| Mechanical process of printing
Printing is the process of assembling ink in 3D space as designed. This involves two key mechanical processes: material 'transport' and 'assembly'. 'Transport' is the regulated movement of materials under the influence of forces, while 'assembly' is the combining of discrete materials. The accuracy and precision of these processes decide how closely the print outcomes coincide with the design.
Mass transport is the consequence of a combination of factors associated with energy sources and flow channels. This mechanical perspective can enhance our comprehension of various printing approaches. For instance, the pneumatic printing type cannot be volumetrically dosed, and the nozzle tends to experience permanent blockage with poorly homogenized materials or agglomerated cells.
Conversely, the electric piston type is volume-controlled, and the pressure rises when obstruction occurs, thus automatically declogging the nozzle. The term 'transport precision' alludes to the volume discrepancy between actual output and intended output, which is the primary concern in transport; especially when faced with the vast amount of starts and stops caused by the geometric complexity of a 3D hierarchical vascular network. We propose that this dynamic process should be viewed as a relaxation phenomenon (Figure 9), which can be quantitatively characterized by the 'transport relaxation time τ'.
Assembly is the merging process of discrete materials, homogeneous or heterogeneous, when the old surface vanishes and a fresh surface appears. The ultimate morphology layered on the pre-process structure depends on (1) the material self-supportability, which is jointly determined by yield stress, viscosity and surface tension, and (2) the bonding and infiltration between materials and the pre-process material properties, which is determined by interfacial tension, physical diffusion and chemical reaction; and the material relaxation time determines the dynamic processes. The assembly requirements of the materials vary depending on their purpose, for example, the vascular soluble core materials, should be sufficiently supported but not excessively infiltrated with native materials.

| Bioprinting approaches
Currently, bioprinting is classified into nozzle-based and light-based categories,64 where we can divide nozzle-based into transportfeatured and assembly-featured categories according to mechanical characteristics (Table 4) printing is undoubtedly high-speed; yet, like DLP methods, the multi-material distribution is still challenging. Some methods, such as multi-vat-photopolymerization, sequential injection, and sequential deposition, enable multi-material printing to a certain extent.
Nevertheless, frequent switching and cleaning limit efficiency and precision severely.

| Issues and developments of printing
Under the premise of multi-material distribution,93 bioprinting engineering today confronts a triple paradox: precision, speed and cytocompatibility. Precision is the most concerning issue for users because actual tissue heterogeneity often takes place at a scale lower than printers can achieve. It should be noted that the actual precision is dissimilar from the machine's declared precision. Printing result fidelity and minimum feature size should be taken into account as co-criteria.
In addition, high precision often leads to slow speed, which poses a challenge for large-volume printing and cell activity assurance. We use 'ink volume flow rate' to characterize the printing speed, but note that the auxiliary action time must be accounted for, as this is a long-

| Culture: regeneration and application
In the RR strategy, bioprinting only constitutes half of the work.
Following bioprinting, long-term nutrient solution perfusion should be employed to guarantee tissue activity and encourage cell selfassembly, while culture effects should be quantitatively tested before final implantation into animal models to evaluate tissue function in a real-world environment ( Figure 10).

| Perfusion
The
We can search for suitable physical parameters in well-established areas of cell biology or physiology. Nevertheless, bioprinting-based organ manufacturing also has its own peculiarities, such as it differs from simple cell culture and evaluation as it includes interactions between matrix materials and cells in 3D space. Effective transport and assembly processes are vital for the proper functioning of cells.
However, even with in vitro constructed tissue components and theoretical models, the physiological functions and morphological structures remain relatively basic and fall short when compared to those found in natural tissues and organs. As a result, the detectable physiological indicators might not match actual physiology, which calls for further research and development.

| Implantation
The requirements and methods for implantation have yet to be systematically studied, limiting future applications for pathophysiological models and organ transplantation. Implantation strategies will vary marginally for different tissue types and volumes, but there are typically four aspects to consider. (1) Anastomosis: seamless connection with the blood vessels in the body is essential, especially considering F I G U R E 1 0 Process to realize the Regeneration aspect of the RR framework. Reproduced with permission. 40 Copyright 2021, WILEY.
the contradiction between the pressure-bearing nature of vessels and the need for porosity to enable nutrient penetration. 40  Science' research paradigm, which fosters an upward spiralling progression ( Figure 11).

| New paradigm, new hope
The diminishing returns on investments into biomanufacturing have become a prevalent theme in recent years. For example, the past decade has yielded countless studies on pioneering bioink material development. Nonetheless, many of these studies offer so few groundbreaking insights that it remains unfeasible to design materials in a target-oriented manner, which has sparked discussions about necessary shifts in the scientific research paradigm ( Figure 11). Thanks to advancements in bioprinting technology, batch ordering and experimental mentality can now be employed at the same time. Several printers (e.g., the 'SIA bioprinter PRO' we designed) can accomplish extensive batches that cover multi-factor variables in a single experiment through concentration gradient printing technology, to enable automatic analysis and mathematical modelling. However, despite the fact that big data and artificial intelligence (AI) have become popular research topics in recent years, paradigmatic shifts must be adopted