The strategy for engineering LVD tissues
To generate functional cardiac and vascular tissues, the critical point is to recapitulate the physiologically relevant, highly aligned and densely packed cellular arrangement, eliciting the tight intercellular connections which govern their concerted biological activity (1, 30). To achieve this, we propose an LVD strategy using leaf venation microchannels as geometric confinement to guide the morphological evolution of high-density cells in ECM hydrogel. As described in our previous works, a polydimethylsiloxane (PDMS) substrate with biomimetic interconnected channels, consisting of the primary channel and branched channels, can be produced from the skeleton of leaf venation networks (Fig. 1Ai) (31). The amphiphilic treatment is conducted to prevent the adhesion between PDMS and cell/hydrogel matrices (Fig. 1Aii). Cell-laden fibrin hydrogel precursor solution is then introduced into the microchannels in the PDMS substrate and gelled (Fig. 1Aiii). The uniformly-encapsulated cells begin to spread and exert contractile forces on the surrounding matrix, initiating LVD morphological evolution process (Fig. 1Aiv,Av). Subsequently, the cell-contraction-induced forces can induce the shrinkage of the cell/hydrogel, which detach from the amphiphilic-treated PDMS surfaces and self-organize into densely packed tissues (Fig. 1Avi). This process eventually leads to the formation of interconnected tissue bundles aligned along the microchannels (Fig. 1Avii). As a control, the cells cultured in non-amphiphilic-treated PDMS spread in the matrix but could not detach from the microchannels, leading to their random distribution (Fig. S1). As a demonstration, we successfully employed the LVD strategy to engineer a large-scale vascular tissue in high viability, with interconnected hierarchical networks confined within the leaf venation microchannels (Fig. 1B). It was found that the mixture of human umbilical vein endothelial cells (HUVECs) and fibrin hydrogel precursor solution detached from the channels immediately after seeding, and gradually contracted in the microchannels. The randomly-distributed round cells eventually became a stable, interconnected cellular bundle with smooth borders after 48 hours of culture (Fig. 1C). On the contrary, HUVEC/fibrin hydrogel cultured in the control group remained random and anchored to the channel surface, since the cell-contraction-induced force was not enough to break down the adhesion between the non-amphiphilic-treated PDMS channels and cell-laden fibrin hydrogel (Fig. 1D). Scanning electron microscopic (SEM) images also confirmed that the highly-elongated and compacted morphology of cell bundles along the microchannels in LVD tissues (Fig. 1E). Due to significant hydrogel compaction and tissue remodeling, the average width of the HUVEC/fibrin hydrogel was reduced ~ 2-fold in the primary channel and ~ 5-fold in branch channel (Fig. S2).
Immunofluorescence staining was performed to characterize the cellular arrangement and functional protein expression within LVD vascular tissues. Cytoskeletal staining demonstrated that HUVECs in LVD tissues reached the confluence with a highly elongated morphology (Fig. 1F). The quantitative results revealed that most cells were highly aligned along the longitudinal direction of the microchannels (Fig. 1H). In contrast, the cells in control groups anchored to all boundaries of the channels were randomly oriented in an irregular morphology (Fig. 1G,H). A high level of CD31, a specific endothelial marker of platelet-endothelial cell adhesion molecule, was expressed in regions of EC-EC contact throughout the entire LVD vascular tissues, with no noticeable differences in the varying channel regions; this expression and localization indicated that the cells self-assembled into densely-packed LVD tissues, with the formation of tight intercellular junctions between neighboring cells and the maintenance of the functional endothelial phenotype (Fig. 1I). The intercellular junctions and pinocytotic vesicles found in the ultrastructural analysis further indicated the normal endothelial function in transporting biological molecules (Fig. 1K). In contrast, the cells in the control groups expressed a lower level of CD31 in an irregular pattern (Fig. 1J). Another essential characteristic of LVD tissues is their interconnected tubular structures in all the primary and branch channels as well as the bifurcation regions (Fig. 1L), while the cells in control were distributed in the bottom of channels (Fig. 1M). These self-assembled tubular structures are in accordance with previous studies, which produced endothelial tubes with simple geometry and proved their value as templates to trigger the formation of new capillaries in vivo in the prescribed pattern (32, 33). Our LVD vascular tissues possess much more hierarchical networks, which might benefit the formation of biomimetic vasculatures in vivo.
The Dynamic Formation Process Of Lvd Tissues
We next sought to characterize the dynamic self-assembly process of LVD tissues from uniformly-distributed cells/hydrogel into the highly aligned and densely packed tubular structures (Fig. 2A). The cells and fibrin hydrogel composites cultured at different time points were visualized with a laser scanning confocal microscope (Fig. 2B). After seeding into the microchannel, all the cells were settled in the bottom of the crosslinked fibrin hydrogel. At the beginning 6 hours of culture, a decrease in the thickness of fibrin hydrogel was found, indicating that the cells had begun to spread and exerted contractile forces on the surrounding ECM hydrogel. In the next 6 hours, the fibrin hydrogel experienced a sudden massive shrinkage in the cross section, especially in the thickness direction, and was compacted into a thin rod-like structure. While most cells were still in the bottom of the hydrogel, some ones had begun to line along the out surface of the condensed fibrin rod. After 18 hours, the cells migrated from the bottom onto the top surface, forming a confluent, tubular endothelial structure filled with densely-packed fibrin. After 48 hours, the tubular structure was further compacted into a smaller dimension (Fig. S3). Eventually, in all the multiscale channels, including the primary channel and bifurcation regions, the randomly-distributed cells/fibrin transformed into continuous and interconnected tubular structures with densely-packed fibrin encircled by cellular layer (Fig. 2C, Fig. S4). Indeed, the interconnecting LVD cellular bundles were suspended between the microchannels rather than settling at the bottom. In contrast, no apparent morphological change of the cells and fibrin hydrogel was found in the control group after 48 hours of culture (Fig. S5). The quantitative results indicated that the average cross-sectional area fraction of the cells/fibrin hydrogel was reduced ~ 13-fold in the primary channel and ~ 33-fold in-branch channels, during the progression of LVD tissue development (Fig. 2D). This significant decrease in volume is expected to induce the improved mechanical property of the cells/fibrin hydrogel composites. Therefore, atomic force microscopy (AFM) nanoindentation was conducted to measure the local stiffness of the cells/fibrin hydrogel mixture and the final resultant LVD tissues (34). After 48 hours of culture, the mean local Young’s modulus of the constructs was increased ~ 45-fold, from roughly 0.8 kPa to ~ 37.3 kPa (Fig. 2E). Meanwhile, similar to the aligned cells, the fibrin hydrogel fibers in LVD tissue of both the primary-channel and branch-channel regions were also densely-packed and oriented along the microchannels (Fig. 2F) after 48 hours of culture, while the fibrin in the control group remained in porous structures and randomly-distributed (Fig. 2G). The quantitative results demonstrated that the fibrin hydrogel fiber alignment in the LVD and control tissues (Fig. 2H) approximately followed the pattern of cellular alignment along the longitudinal channel directions (Fig. 1H).
Taken together, these data reveal a self-organization process of LVD tissues through six consecutive stages (Fig. 2I): Stage 0, cells and fibrin precursor solution fill the microchannels (oriented along y-axis) uniformly. Stage 1, the round cells settle in the bottom of the crosslinked fibrin hydrogel, consisting of with randomly-distributed fibers. Stage 2, cells spread and exert contractile force on the surrounding ECM and cells, which induce them to gather towards the central region gradually; during this process, a strain field is formed within the hydrogel along the longitudinal direction of the microchannel (y-axis), which guides the local alignment of cells and fibrin fibers in the same direction. Stage 3, Under the combined action of the cellular contractile force and the longitudinal strain force, the fibrin fibers are compressed into a highly aligned rod-like structure; cells gradually migrate from the bottom to cover the surface of the compressed fibrin fibers. Stage 4, The fibrin fibers are further compressed; the cells reach confluence encircling the tight fibrin fibers and orient along the longitudinal direction in a highly aligned arrangement.
Structural Maturation And Molecular Maturation Of Lvd Neonatal Rat Cardiac Tissues
We further investigate the potential of the LVD strategy to engineer highly aligned and densely packed cardiac tissues and the contribution of this biomimetic tissue anisotropy to myocardial tissue-like maturation and functionality. After 5 days of culture, the seeded mixture of neonatal rat CMs and fibrin hydrogel self-assembled into the meshed constructs defined by the leaf-venation pattern (Fig. 3A). Higher magnification of the cells revealed that the CMs of LVD tissues were densely and uniformly packed, and strongly elongated within the interconnected tissue bundles in all scales of the channels (Fig. 3B); the control tissue showed random cellular orientation (Fig. 3C). The cross-section of the confocal file indicated that the CMs also compacted to form a tubular structure (Fig. 3D). The local stiffness of the 5-day-old LVD cardiac tissues, measured by AFM nanoindentation, was ~ 40 kPa, which was quite close to that of the native rat myocardium (~ 63 kPa). This biomimicking stiffness might be an essential contributor to the differentiation, hypertrophy, and electromechanical function of the CMs in the aligned LVD cardiac tissues (35, 36).
The maturation of LVD cardiac tissues was assessed using immunostaining staining of CM-specific markers, including contractile protein sarcomeric α-actinin and the gap junctional protein connexin 43 (CX43). The 5-day-old LVD cardiac tissue construct exhibited highly-aligned and interconnected striated sarcomeric structures, as well as higher expression and uniform distribution of CX43 compared to the control tissues with the random and loose cellular distribution (Fig. 3E,F). The quantified data based on the immunofluorescence staining, indicated longer length of the sarcomere and higher level of CX43 in LVD cardiac tissue (no evident difference between the primary-channel and branch-channel regions) than that in the control tissues (Fig. 3G,H). In addition, maturation-related cardiac marker gene expression was investigated. After 5 days of culture, higher expression of electrical coupling-related (Kcnj2F, Pdk4F) and metabolic-related genes (Cpt1bF, PpargclaF) was found compared to the control tissue (Fig. 3I), which results were consistent with the morphometric analysis (Fig. 3E-H). All these results demonstrated that the LVD cardiac tissues with highly aligned and densely packed structures benefited CM maturation with improved cell-cell coupling and contractile phenotype.
Electrophysiological Functions Of Lvd Neonatal Rat Cardiac Tissues
Calcium transients related to excitation-contraction coupling were investigated to evaluate the synchronous contraction functions of the LVD rat cardiac tissues. Within each recorded tissue area, the fluorescent intensity of calcium spikes for five locations of interest was plotted over 8 seconds. The resultant calcium transients in both the primary-channel and branch-channel regions displayed strong and synchronous calcium mobilization, indicating the synchronous beating of the LVD tissues, which was visible under the naked eye on day 5 of culture (Fig. 4A,C, Movie S1,2). On the other hand, disorderly and asynchronous Ca2+ puffs without rhythmic patterns took place in the control tissues (Fig. 4B,D, Movie S3,4). Furthermore, the quantitative parameters based on calcium transients demonstrated higher beating frequency and less time to reach the calcium peak in the LVD tissue (no evident difference between the primary-channel and branch-channel regions) compared with the control tissue (Fig. 4E,F).
The LVD rat cardiac tissues with highly aligned and densely packed structures enabled the production of high-amplitude extracellular field potentials. To demonstrate it, we integrated conductive micropillar electrodes into the LVD system for in-situ biologically relevant electrophysiological monitoring of CMs. Specifically, thin platinum wires with an average diameter of 30 µm were embedded into the bottom PDMS layer, and the end was exposed to the seeded CMs/hydrogel in the microchannels. During the formation of LVD cardiac tissues, the thin platinum wires could be enclosed by the highly aligned and densely packed structure (Fig. 4G). Through this system, we detected the extracellular field potentials of 2-day-old LVD rat cardiac tissues in 20 mV spaced with a frequency of ∼0.5 Hz, whose amplitude increased to 30 mV with an improved frequency of ∼1–2 Hz after 4 days of culture (Fig. 4H,I). In contrast, no electrical signal could be detected in the control tissue. This implied that our system of LVD tissues with built-in micropillar electrodes enables long-term noninvasive monitoring of the natural maturation and other rhythmic cellular phenomena of the cardiac tissues, which might be used for long-term online cardiotoxicity testing or functional cardiac tissue engineering. Then, the LVD system’s ability to monitor changes in the frequency of electrical signals in response to the supplement of the drug was investigated. For example, the field potential recording revealed that the signal frequency reached 1.78 Hz from an initial value of 1.13 Hz without a significant change or degradation in signal quality following the addition of 5 µM isoproterenol to the culture (Fig. 4J). Besides, electrical stimulation could also be applied through the two built-in micropillar platinum electrodes to pace the CMs. The interfering by applying acute electrical stimulation (10 V, 50 ms) at different frequencies (1 and 2 Hz) activated the cells throughout the LVD cardiac tissues and modulated their beating activities (Fig. 4K,L).
Therefore, these results indicate that the presented LVD strategy enables engineering functional cardiac tissues with interconnected, highly aligned, and densely packed structures, thus benefiting their in-vivo-like cellular phenotype, synchronous beating activities, and electrophysiological functions. In addition to a therapeutic approach, the LVD system can be exploited for in-vitro studies, such as drug screening assays in a 3D microenvironment, as well as long-term culture and maturation of cardiac cells.
Engineering Functional Lvd Human Cardiac Tissue From Hipsc-cms
While the above results have comprehensively shown the advancement of the LVD strategy for producing rat cardiac tissues with hallmarks of the native myocardium, neonatal rat CMs cannot be utilized for clinical therapeutics. Engineering human cardiac tissues based on hiPSC-CMs, avoiding issues of controversial ethical value and limited expansion and regeneration capacity of primary CMs, have been endowed with great expectations to provide solutions (3). One crucial concern is to drive their structural and functional maturation, evidenced by elongated morphology, stable contractile machinery, gene expression profiles, and electrophysiology (2, 3).
Therefore, we continued investigating the generation of matured human cardiac tissues by culturing hiPSC-CMs using the LVD strategy over a 17-day culture. A small amount of individually-beating CMs were observed within 24 hours after seeding. During the following 2 days, hiPSC-CMs/fibrin hydrogel remodeled and decreased in width, resulting in a large-scale LVD hiPSC-derived cardiac tissue with interconnected cellular bundles directed by the microchannels (Fig. 5A). Meanwhile, the initial asynchronous beating converted to a coherent, synchronous macroscopically visible contractions (Movie S5). From day 2 to day 5, these synchronous contractions gradually decreased from 50–60 to 35–40 beats per minute and then stayed at a consistent and stable beating rate in the following days, indicating long-term electrophysiological coupling of the CMs (Fig. 5B). After 17 days of culture, the LVD human cardiac tissue remained synchronous macroscopically visible contractions and relatively stable configuration with interconnected bundles confined in the microchannels (Fig. 5C). The video-based analysis of contractile motion in LVD tissues demonstrated their longitudinal contraction direction along the microchannels (Fig. 5D), relatively uniform contraction velocity between different regions (Fig. 5E), and stable contraction velocities over 17 days of culture (Fig. 5F).
Immunofluorescent staining of the cardiac tissues was performed to assess hiPSC-CM microscale organization of myofibrils and intercellular coupling. The initially formed LVD cardiac tissues after 3 days of culture had shown a relatively densely-packed arrangement with significant tendencies in alignment, but some cells are still close to round morphology (Fig. 5G). After 17 days of culture, LVD cardiac tissues demonstrated a more regular sarcomeric α- actinin and cardiac troponin T (cTnT) organization with clear contours, and nearly all the cells were elongated in a high aspect ratio, resulting in highly aligned and densely packed tissue structures (Fig. 5H). In contrast, the 17-day-old hiPSC-CMs cultured in non-amphiphilic-treated PDMS remained in round morphology (Fig. 5I). The sarcomere length of the 17-day-old CMs in LVD cardiac tissues displayed a statistically significant increase compared with that of day 3, extended from ~ 1.74 µm to ~ 2.02 µm, whose value was quite close to that in adult CMs (Fig. 5J). In addition, the 17-day-old LVD cardiac tissues expressed a higher level of CX43. These results indicated that long-term culture could benefit the structural and electrophysiological maturation of LVD cardiac tissues.
In addition to the structural proteins of sarcomere and cTnT, we further sought to reveal the molecular signatures underlying the advanced functional maturation of LVD tissues. 6 cardiac marker genes were chosen, including two groups: excitation-contraction coupling (S100A1, PLN, SCN5A) and metabolic (CKM, PDK4, COX6A2), thus reflecting key maturation processes in developing hiPSC-derived CMs. 5 of the 6 genes progressively increased at day 17 in LVD tissues compared to day 2, suggesting enhanced LVD hiPSC-derived cardiac tissue maturation with improved electrical and metabolic function over time in culture (Fig. 5K).
Engineering Of 3d Pre-vascularized Cardiac Tissues With Programmed Mechanical Property
Engineering 3D functional cardiac tissues with a pre-vasculature and proper mechanical property are pivotal to promoting their long-term survival and serving as temporary mechanical support to prevent the progression of postinfarction left ventricular remodeling, thus benefiting the restoration of the heart’s normal contraction behavior and function in vivo (9, 37). For this purpose, we develop a scaffold-assisted method to achieve the transfer of LVD cardiac and vascular tissues and assembly of them into 3D tissue composites with the programmed mechanical property. The elastic scaffold with precisely-defined serpentine microarchitecture was applied in the follow-up study, which we hypothesized was able to reproduce the anisotropic and viscoelastic behavior of the native myocardium
Figure 6A shows the transfer and assembly process of the multiple LVD tissues. Specifically, when the culture medium was removed, an elastic scaffold was placed onto the well-formed LVD vascular tissue in the PDMS substrate, and 3mg/mL bovine fibrin precursor solution was added and polymerized to encapsulate them together (Fig. 6Ai,ii). With the mechanical support of the scaffold, the LVD tissue in fibrin hydrogel could be easily transferred out of the PDMS substrate (Fig. 6Aiii). This transferred LVD tissue was then placed onto the other well-formed LVD cardiac tissue, with 100uL bovine fibrin precursor solution added between them as glue for bonding (Fig. 6Aiv). The 2-layers LVD tissue could then be obtained (Fig. 6Av). By repeating this elastic-scaffold-based transferring and fibrin-glue-based bonding process, the multiple LVD cardiac and vascular tissues could be assembled into 3D pre-vascularized cardiac tissue constructs (Fig. 6Avi). The finite element analyses (FEA) calculations were then conducted for the rational structural design of the elastic scaffolds, which could endow 3D pre-vascularized cardiac tissues with directionally dependent mechanical properties along the circumferential (CIRC) and longitudinal (LONG) axe that conform to the nonlinear deformation of the native myocardium in plane (Fig. 6Avii). The detailed FEA calculation can be found in Fig. S6.
EHD printing was employed to produce the elastic polycaprolactone (PCL) scaffold since it enabled the high-resolution fabrication of microscale serpentine structures comparable to the design, thus granting us freedom in tailoring the mechanical property of assembled tissue constructs (38, 39). The experimental results demonstrated that the dimensions of the EHD-printed scaffolds agreed with those designed dimensions (Fig. S7,8). Figure 6B,C and Fig. S9 provide the resultant FEA simulations and macroscopic images of the 3D tissue composite upon uniaxial stretching (from top to bottom: 0%, 10%, and 20%) along the LONG direction, and Fig. S10 provides that along the CIRC direction. Figure 6D demonstrates good consistency in stress-strain curves along the CIRC and LONG direction among the corresponding FEA calculations, experimental measurement and native heart tissue, indicating that the combined theoretical calculations and EHD printing can provide an accurate and effective pathway for the generation of the scaffold-reinforced 3D assembled LVD tissues with cardiac-tissue-like nonlinear mechanical behaviors.
Figure 6E shows the transferred monolayer LVD cardiac tissue, demonstrating the structural integrity and stability of the meshed tissue integrated with the serpentine PCL scaffold. Figure 6F,G show the as-prepared 3D cardiac construct, with LVD cardiac and vascular layers vasculature labeled red and green, respectively.
Injectable Delivery Of 3d Lvd Cardiac Constructs
To avoid open chest surgery, we further explore the feasibility of minimally invasive implantation of the assembled LVD cardiac constructs. With the mechanical aid of the EHD-printed elastic scaffolds, the cardiac constructs are expected to be injected by a tubing passing through the chest wall and placed onto the diseased epicardium. Before injection, the assembled cardiac constructs were rolled up around a rod with a diameter of 2.5 mm and then transferred into a tubing with an inner diameter of 6 mm (Fig. 7A). The rolled 3D tissues could stay in the tubing stably (Fig. 7B). Under the flowing water, the cardiac constructs could be rushed out of the tubing and recovered to their original shape (Fig. 7C).
As a proof of concept, the assembled cardiac construct was injected onto the ventricle of the ex vivo porcine heart, which shared a similar structure and volume comparable to the human heart. The injected cardiac construct was capable of covering a clinically-relevant size of the porcine heart with a conformable and tight attachment between them (Fig. 7D). The EHD-printed fiber scaffold and interconnecting meshed LVD tissues could preserve their structural integrity after ejection (Fig. 7E). No lamination in the LVD tissues was observed during this whole assembly and injection implantation process, indicating that the assembly strategy based on fibrin hydrogel bonding enables the generation of stable thick tissue structures.
During injection, the applied force is mainly stored in the scaffold, thus facilitating the tissues’ return to their original shape; relatively little force in the LVD tissue-laden hydrogel is beneficial for their viability. To investigate this, the FEA calculations were conducted to study the distribution of strain force within the LVD tissue consisting of the scaffold and fibrin hydrogel, subjected to out-of-plane bending during the rolling process (Fig. 7F). The calculated results confirmed that the maximum strain force in the scaffold was ten thousand higher than the hydrogel with different wrapping angles (Fig. 7G). The empirical results of live-dead staining show no significant differences between the percentage of viable cells before and after the injection of the assembled cardiac tissues in vitro, suggesting that the rolling-induced large compressive deformation and flow-induced recovery process did not negatively influence the assembled 3D LVD tissues (Fig. 7H,I).