Cellular Patterning Alone Using Bioprinting Regenerates Articular Cartilage through Native-Like Cartilagenesis

Few studies have proved that bioprinting itself helps recapitulate native tissue functions mainly because the bioprinted macro shape can rarely, if ever, in�uence cell function. This can be more problematic in bioprinting cartilage, generally considered more challenging to engineer. Here we show a new method to micro-pattern chondrocytes within bioprinted sub-millimeter articular micro-cartilage tissues (PA-MCTs) under the sole in�uence of bioprinted cellular patterns. A pattern scoring system is developed after over 600 bioprinted cellular patterns are analyzed. The top-scored pattern mimics that of the isogenous group in native articular cartilage. Under the sole in�uence of this pattern during PA-MCTs bio-assembling into macro-cartilage and repairing cartilage defects, chondrogenic cell phenotype is preserved, and cartilagenesis is initiated and maintained. Neocartilage tissues from individual and assembled PA-MCTs are comparable to native articular cartilage and superior to cartilage bioprinted with homogeneously distributed cells in morphology, biochemical components, cartilage-specic protein and gene expression, mechanical properties, integration with host tissues, zonation forming and stem cell chondrogenesis. PA-MCTs can also be used as osteoarthritic and healthy cartilage models for therapeutic drug screening and cartilage development studies. This cellular patterning technique could pave a new way for bioprinting to recapitulate native tissue functions via tissue genesis.


Introduction
Live-cell 3D bioprinting has recently progressed considerably. [1]This holds promise in solving organ shortages and in providing better models to predict drug treatment outcomes prior to clinical trials.However, few studies have proved that bioprinting itself has helped recapitulate native tissue functions.[2] [3] Unlike a 3D printed construct using only polymers, a bioprinted construct using live cells must go through a post-bioprinting process to mature.To positively in uence a bioprinted construct to mature towards its native analog tissue, bioprinting itself needs to provide a cue to initiate and maintain tissue genesis during the post-bioprinting culture.However, existing bioprinting techniques largely fail to accomplish this goal.
Numerous reported successes have focused on generating anatomical tissue shapes at macro-scales.However, tissue functions are determined by cells, which are largely unable to sense the in uence of a macro-scale shape.In contrast, as proved by other bioengineering techniques, a micro-scale cellular pattern is an important biophysical signal in the extracellular microenvironment.nfortunately, existing bioprinting techniques are not capable of patterning cells at a micro-scale to provide this favorable cue.Even with the highest cellular resolution, single-cell droplet bioprinting is not suitable for this patterning purpose because it cannot print any hydrogel. [7] This interface separation limits matrix remodeling and inhibits the bioprinted tissue from maturing (Figure S5C).
Using DVDOD, we can pattern cells into multiple clusters within a droplet of bioink.Using optimized bioprinting parameters, we can control the distribution patterns of cell clusters and the number and area percentages of the clusters that support chondrogenesis and cartilagenesis.Here, we report the success of generating native-comparable sub-millimeter patterned micro-articular-cartilages tissues (PA-MCTs).
Articular cartilage has a unique pattern of isogenous groups. [13]Articular cartilage defects are prevalent, and osteoarthritis (OA) is the leading cause of disability.Because of donor shortage, native-compatible engineered cartilage is highly desired for clinical, pharmaceutical and research applications.Articular cartilage is composed of only one cell type without any vasculature or lymphatic vessels and was predicted to be one of the rst tissue to be successfully engineered.However, numerous studies over 30 years have shown that the simplicity of the cellular component and the complexity of the structural organization of cartilage makes it more di cult to engineer than other cellularly complex tissues. [14]The main reason for this di culty is that chondrocytes lose their native phenotype and cartilagenesis rarely happens during the in vitro process.Because the pattern in articular cartilage is formed from condensed chondrocyte clusters during prenatal cartilagenesis and persists throughout the lifespan of healthy cartilage, we postulate that patterning chondrocytes as clusters alone, without exogenous factors, maintains the function of native chondrocytes and initiates and maintains cartilagenesis.
Here we demonstrate that without any exogenous factors, solely under the cue of the bioprinted cellular pattern, the native phenotype of chondrocytes is preserved and the chondrogenesis of human mesenchymal stem cells is enhanced.The bioprinted pattern also initiates and maintains cartilagenesis in individual PA-MCTs and during bio-assembly (PA-MCTs integration).PA-MCTs are comparable to native articular cartilage in morphology, biochemical components, cartilage-speci c protein and gene expression, and mechanical properties.PA-MCTs can also be used as miniature models of healthy and osteoarthritic (OA) cartilage.MCTs bioprinted with homogeneously distributed cells (H-MCTs) were used as the control.

Results and Discussion
First, we demonstrate the feasibility of DVDOD patterning cells from homogenous distribution to clusters within the smallest available bioprinting unit-a droplet.The principal process of DVDOD is accomplished in two steps: (i) under linear motion, a droplet of bioink of a speci c volume is driven to the tip of the nozzle, and (ii) the droplet is blown to the substrate by pulsed driving air.Thus, unlike any existing airassisted bioprinting techniques, the volume of the dispensed droplet in DVDOD is independent of the driving air pressure (Fig. 1A-i).With this unique feature, we rst dispensed a droplet of bioink composed of chondrocytes, collagen and brin hydrogels, using the lowest pressure possible to dispense the bioink onto a substrate.Immediately after dispensing the rst droplet, using a second dispenser, we dispensed the second layer of multiple droplets of bioink containing thrombin but in the absence of cells.The second layer of droplets collided with the rst layer changing the homogeneous chondrocyte distribution to clusters.Almost immediately after dispensing the second layer droplets, the thrombin solidi ed the brin in the entire droplet of bioink.Thus, the newly formed chondrocyte patterns were xed in place without generating interfaces between clusters.(Fig. 1A-ii, 1E-i) We used the distributions of the equivalent circular diameter (ECD) of cell clusters to represent the patterns of cell clusters within a bioprinted micro-construct (deliberately controlling the ID > 2mm to include more cell clusters for analysis, Fig. 1B-ii, 1E-i, ii).By varying the volumes of droplets, driving pressures, the number of droplets and the location of droplet-dispensing, we generated different patterns of cell clusters (Fig. 1B-i).Using Empirical Cumulative Distribution Function (ECDF, Fig. 1C) plotting and Kolmogorov-Smirnov (KS, Fig. 1D) distribution analysis, we selected 20 unique patterns from an array composed of over 600 patterns as the standard patterns-every two patterns are statistically different (Fig. 1B-iii and D).Replicates of each standard pattern were bioprinted using the same corresponding bioprinter control parameters.Only the replicates with the statistically same distribution as the corresponding standard distribution were included for further pattern analysis (Fig. 1B-iv, Figure S6).
Within the same assay, using the same lot of cells and hydrogels, 20 groups of PA-MCTs of 20 same unique patterns were also bioprinted using the corresponding bioprinter control parameters.The PA-MCTs were further cultured for 3 weeks before qPCR analysis of the expression of the cartilagenesis positively related genes Col-II, Sox-9 and ACAN and the negatively related gene COL-I (Fig. 1B-v and 1F).
One of our goals was to develop a scoring system to quickly select the pattern that best supports cartilagenesis during each bioprinting assay's initial bioprinter calibration stage.Using the bioprinter control parameters among the 20 unique patterns, we selected the optimized pattern supporting chondrogenesis based on qPCR scores (Fig. 1F).Because KS distribution analysis is nonparametric and it does not rank multiple distributions, we chose three parameters to represent the cluster patterns: (i) Area percentage: the summed area of clusters in a speci c ECD range over the total cluster area, which re ects the percentage of cells under the in uence the pattern (bar plots in Fig. 2A), (ii) Number percentage: the summed number of clusters in a speci c ECD range over the total cluster numbers, which re ects the e ciency of patterning (raincloud plots in Fig. 2A), (iii) Distances between clusters (Fig. 1E-iii).First, we observed that once the distances between clusters were greater than 15 µm, which was observed in most of the selected patterns, the cartilagenesis was not affected.Therefore, this parameter was excluded from the scoring system.We introduced a pattern score (Fig. 1B-iv), the average of the number percentage and area percentage of the clusters in a speci c ECD range.We chose 20 µm as the lower limit of the ECD range because clusters with ECD ≤ 20 µm can form spontaneously, and 220 µm as the upper limit to exclude samples containing large homogenous clusters.We evaluated pattern scores calculated from 66 different ECD ranges (Table S1), and ranked each pattern and qPCR score in descending order (Fig. 2B-i).According to Spearman's correlation analysis of the ranking (Fig. 1B-vi), the pattern scores using the ECD range of 20 to 80 µm have the highest correlation with the corresponding qPCR scores (Fig. 2B-ii).This means that when evaluated using the 20 to 80 µm ECD range, the pattern of the highest score ranking best supports cartilagensis.For each assay, after an array of microconstructs is bioprinted using different printer control parameters, pattern scores were calculated.The control parameters with the highest score were selected for each bioprinting batch.H-MCTs were bioprinted without generating cell clusters by omitting the droplet colliding.The nal concentrations of cells and hydrogels in PA-MCTs and H-MCTs and the volumes of corresponding PA-MCTs and H-MCTs were controlled to be the same in each assay.
The conceptual design of assays used to analyze the PA-MCTs is illustrated in Fig. 3. First, clusterpatterning alone in PA-MCTs preserved the native phenotype of chondrocytes as evidenced by morphology, gene expression and mechanical properties.Compared with clustered cellular patterns in PA-MCTs, H-MCTs showed homogeneous chondrocyte distribution after bioprinting (Fig. 4B-i).The overall sizes of the individual PA-MCTs are uniform (Fig. 4B-ii).Our bioprinting process does not affect cell viability (Fig. 4B-iii).Chondrocytes in PA-MCTs also expressed signi cantly higher levels of native phenotype (chondrogenic) genes, namely, collagen type II and aggrecan, than those in H-MCTs, and a signi cantly lower level of the brotic phenotype gene of collagen type I at week 5 (Fig. 4D).
At the tissue level, cluster patterning also initiates and maintains the cartilagenesis (Fig. 4A) in PA-MCTs as evidenced by morphology and biochemical components, biomechanical enhancement and overall scores.PA-MCTs show smooth white surfaces, which is a typical macroscopic appearance of native cartilage (Fig. 4E-i).PA-MCTs are rigid and small enough to be delivered through injection (Fig. 4E-ii).Microscopically, lacunae and isogenous groups, the structural morphology of native cartilage, formed at week 3 and became extensive by week 5 (Fig. 4C-i, Figure S7).In contrast, lacunae or isogenous groups did not form in H-MCTs, similar to engineered cartilage reported previously using homogeneous cells (Fig. 4C-i).For biochemical components, (i) collagen type II content was signi cantly higher in PA-MCTs than in H-MCTs and showed no difference compared with native cartilage (Fig. 4C-ii, Fig. 4G); (ii) sulfated glycosaminoglycans (sGAG) content was higher in PA-MCTs than in native cartilage and signi cantly higher than in H-MCTs (Fig. 4C-i, Fig. 4F).For biomechanical properties, as characterized by a capillary aspiration assay (Fig. 4H-i), PA-MCTs also demonstrated signi cantly greater stiffness than H-MCTs (Fig. 4H-ii).PA-MCTs also demonstrated a signi cantly higher Modi ed Bern Score-an overall evaluation of the quality of tissue-engineered cartilage-than H-MCTs (Figure S8, Table S3).
Previously, modular cartilage tissues, including chondrocyte spheroids or miniature cubes of minced native cartilage, were expected to assemble into a macro-cartilage-like tissue through a bottom-up approach. [15]However, because chondrocytes change to brotic phenotype in the spheroids and cartilagenesis is absent, spheroids are only able to fuse into a small and thin (< 1.5 mm in long axis, < 0.75 mm in height) cartilage-like tissue.A larger fusion usually assembles incompletely in vitro: boundaries of individual modules remain; necrotic cores form; and mechanical properties of fused tissues are poor. [16] [17] [18]For minced cartilage, cartilage-to-cartilage integration is extremely di cult to accomplish because low metabolism and dense anti-adhesive matrix exist in native cartilage and cartilagenesis cannot be initiated. [19]n contrast, our approach of bioprinted cellular patterning solved both problems mentioned above.
Without any exogenous factors, the biophysical cue of cellular patterning maintains the native phenotype of chondrocytes.After an initial short period of cartilagenesis (2 weeks) in individual PA-MCTs, the cartilagenesis continues during the entire bio-assembly process.Thus, we successfully generated macro-articular-cartilage tissues.
First, using less than 20 PA-MCTs, we studied the dynamics of cartilagenesis during bio-assembly using the process illustrated in Fig. 5A.Bio-assembly, as we are using the term, refers to the integration of micro-tissues (building blocks or modular tissues) to form macro-cartilage as opposed to self-assembly, where cells attach to form a spheroid or a sheet (a single building block). [20]We found three major changes occurred and the activity ratios between any two changes are dynamic over the time course (Fig. 5C, S9): (i) Cell migration.Chondrocytes migrated out of PA-MCTs into the surrounding hydrogel.This occurred as early as the rst 24 h when a single PA-MCT was tracked.Chondrocyte phenotype was well preserved, as indicated by positive Alcian blue staining (Fig. 5B).(ii) New matrix deposition.The chondrocytes that migrated to the spaces between PA-MCTs deposited sGAG starting from week 1 (Fig. 5C; Figure S10, semi-quanti cation of the sGAG content distribution).The newly deposited matrix interconnected individual PA-MCTs.(iii) Structure reorganization (Fig. 5C).Each PA-MCT began reorganizing internal structures as the boundary of each PA-MCT gradually diminished starting from week 5.When reorganization was completed at week 7, individual PA-MCTs disappeared, and the macrotissue showed a global native-articular cartilage morphology.
We further demonstrate that the biophysical cue of bioprinted cellular patterns can maintain cartilagenesis when ~ 800 PA-MCTs assembled into large macro-cartilage at a clinically-applicable scale (9 mm in diameter).The cartilagenesis was evidenced by native-comparable morphology and matrix components as well as enhanced mechanical properties.Macroscopically (Fig. 5D), when PA-MCTs were loaded into a transwell and covered with hydrogels, individual PA-MCTs were clearly visualized.At week 5, PA-MCTs bio-assembled into a macro-tissue and its surface became at and only blurry outlines of the individual PA-MCTs were visualized.At week 7, the macro-tissue showed articular cartilage-comparable appearances: smooth white surfaces; and intense sGAG deposition as evidenced by dark blue throughout the tissues after whole-tissue Alcian blue staining (Fig. 5E).At the microscopic level, wholemount Safranin O staining (Fig. 5F-i; Figure S11) shows that sGAG was intensely expressed throughout the macro-tissue.All the PA-MCTs were interconnected and the assembly was structurally seamless.No necrosis was observed in any space within the assembled cartilage.Collagen type II was also intensely expressed in assembled macro-articular-cartilage tissues (Fig. 5F-i).In comparison, similar to the results previously reported by others [21] , H-MCTs did not fully assemble and gaps were visualized both macroscopically and microscopically (Fig. 5F-ii) at week 7. H-MCTs did not show native-cartilage-like morphology, and the assembled macro-tissues showed signi cantly less intense sGAG staining and fewer cells.The chondrocytes in the H-MCTs showed elongated brotic morphology and also behaved like broblasts by compacting individual H-MCTs into irregular shapes (Fig. 5F-ii).The assembled macroarticular-cartilage tissues from PA-MCTs demonstrated superior mechanical properties to those from H-MCTs (Fig. 5G): (i) signi cantly higher Young's modulus, (ii) signi cantly higher aggregate modulus, and (iii) signi cantly lower permeability constant indicating the PA-MCTs formed more mature native-likemore solid and less porous-internal structures.
Next, we demonstrate that PA-MCT-assemblies are able to integrate with host cartilage and subchondral bone through cartilagenesis.As mentioned above, native cartilage-to-cartilage integration rarely happens.Therefore, clinically, the capability of an engineered implant to integrate with host cartilage and bone determines its success.In the integration assay, macroscopically, PA-MCT-assemblies grew into smooth white macro-cartilage on top of the native cartilage and bone (Fig. 6-i).Microscopically, the PA-MCT-assemblies formed articular cartilage and showed seamless histological integration with the host cartilage and bone tissues (Fig. 6-iii).In contrast, at week 7 of in vitro culture, H-MCT-assemblies failed to form intact macro-cartilage as individual H-MCTs were still visible macroscopically (Fig. 6-ii), and void spaces were visualized throughout the samples microscopically (Fig. 6-iv).The H-MCT-assemblies failed to attach to the native cartilage as a sizable gap was observed between individual H-MCTs (Fig. 6-ii top).
Most H-MCT-assemblies fell off while a few loosely attached to the underlying bone (Fig. 6-ii bottom).Intense matrix expression of sGAG (Fig. 6-iii) was observed in the PA-MCT-assemblies.The sGAG content of PA-MCT-assemblies in the cartilage and bone integration assays were 5.4-fold (P < 0.01) and 6.9-fold (P < 0.05) of those in H-MCT-assemblies, respectively (Figure S12).We further demonstrate the capability of PA-MCTs to repair cartilage defects through cartilagenesis in a more clinically-relevant model (Fig. 7A).Two types of articular cartilage defects exist clinically: (i) a partial-thickness defect, which is di cult to treat because existing implants do not adhere to it, and (ii) a full-thickness defect, which is a more advanced stage of the disease. [22]We demonstrate repairing both per bovine osteochondral allograft.Mimicking a clinical minimally invasive approach, PA-MCTs were injected into the defect.The outer diameter of the delivery device is small enough (< 4 mm) to potentially operate in a minimally invasive manner during surgery (Fig. 4E-ii, Figure S13).Macroscopically (Fig. 7B): (i) at weeks 2 and 5, individual PA-MCTs had partially assembled, and their outlines became blurry; (ii) at week 7, PA-MCTs assembled into macro-articular-cartilage with a smooth white surface.The macro-articular-cartilage repaired the two types of defects completely by lling the full-thickness defect and by adding a layer of cartilage tissue to the partial-thickness defect, increasing its height to normal.Microscopically, PA-MCT-assembly seamlessly integrated with the irregularlyshaped cartilage walls (Fig. 7C-i), the host subchondral bone (Fig. 7C-ii) and the partial-thickness defect (Fig. 7C-iii), demonstrating the native-comparable morphology.Intense sGAG was deposited throughout the PA-MCT-assembly, including the deep zone (Fig. 7C-iv), as demonstrated by the color intensity quanti cation of Safranin O staining (Fig. 7F).As the depth of the defect that PA-MCTs repaired (3 mm) is greater than the average depth of cartilage in the human knee (2.14 mm) [23] , PA-MCTs show promise for clinical application.In comparison, H-MCTs failed to repair defects, which is similar to the result in our integration assay and those reported elsewhere using spheroids composed of homogeneously distributed chondrocytes. [24] [25]A large proportion of the defect was void due to unassembled H-MCTs that fell off the explants during culture and a large degree of H-MCT compaction (Fig. 7D).Histologically, necrosis, compaction of H-MCTs, void space underneath the neo-tissue surface, and unconnected and unintegrated H-MCTs were observed (Fig. 7E).The average adhesion strength of PA-MCT-assemblies to the host cartilage reached 26% of the force needed to push to tear an intact healthy native cartilage tissue.On average, PA-MCT-assemblies showed a 9.6-fold higher adhesion strength than H-MCTassemblies (Fig. 7H, P < 0.05).Explants repaired by PA-MCTs also demonstrated a signi cantly higher Modi ed O'Driscoll Score-an overall evaluation of the quality of cartilage defect repair-than those by H-MCTs (Figure S14, Table S4).
The unique zonal structure developed during postnatal cartilagenesis largely improves the capability of native articular cartilage to withstand shear and compressive forces.However, previous attempts, including 3D bioprinting to create articular cartilage zonation, have not been successful.These attempts mostly construct a layered structure using different materials or concentrations of cells.However, the initially layered setup does not lead to growing a zonal structure at the end of the assay.Instead, we rely on the capability of cartilagenesis to create the zonation.We observed encouraging zonal structure formed in the macro-articular-cartilage assembled from PA-MCTs, but not from H-MCTs (Fig. 7G).Three structure zones can be clearly visualized: (i) the top zone is composed of at and elongated chondrocytes being parallel to the cartilage surface, which almost recapitulates the characteristics of the super cial zone in native articular cartilage; (ii) some chondrocyte-stacks, perpendicular to the cartilage surface, formed in the bottom zone.This represents some characteristics of the deep zone in native articular cartilage; (iii) the medial zone, corresponding to the native intermediate zone, is distinct from the top and bottom zones by its morphology but resembles the native characteristics less than the other two zones do.Interestingly, when PA-MCTs were assembled without integrating to a native tissue (Fig. 5F), the zonal structure did not form.We think the possible factors determining this might be some cues from the native tissue and/or the asymmetry between the native tissue and the culture medium.Future study is needed to elucidate this and other articular cartilage zonal properties, such as biomechanical and biochemical zonations.The formation of zonal structure is another evidence that cartilagenesis occurs during the assembly process.
We also demonstrate the feasibility of several potential therapeutic applications using PA-MCTs and analyzed their cartilagenesis capability in a proof-of-concept manner.(i) Repairing large (long axis > 15 mm) irregularly-shaped cartilage defects (Fig. 7I): it was modeled in clinically relevant sizes for a translational purpose; in addition, PA-MCTs can self-t concave and convex surfaces, demonstrating the potential to be applied in complex-shaped defects (Figure S15).(ii) Resurfacing entire femoral condyles (Fig. 7J): this demonstrates the potential of cartilage regeneration for the whole joint, (iii) Assembling and generating a personalized osteochondral graft from the data of a patient's computed tomography: it can be potentially used to treat a more complex articular disease of an osteochondral defect where both cartilage and bone losses have occurred.(Figure S16 and other information detailed in the Supplementary data).
In addition to the characteristics for therapeutic applications that we have demonstrated above, PA-MCTs can also be used as miniature models in drug screening and developmental biology studies.
Miniature models are largely preferable because of their higher throughput. [26] [27] [28]However, existing spheroid models using homogeneously distributed chondrocytes poorly represent native cartilage. [28]A native cartilage explant (typically 3 mm in diameter, 1 mm in height) is too large (more than 60 times the volume of a PA-MCT) to be used for drug screening.Additionally, explants have topography-related variations even when harvested from the same joint. [29]In contrast, as evaluated by the coe cient of variation (CV) of sGAG content, PA-MCTs demonstrated a smaller variation than all groups of explants, and signi cantly smaller variation than explants from the medial tibial plateau, lateral femoral condyle, and the entire joint (Fig. 8A).Aiming at disease-modifying osteoarthritis drugs (DMOADs), we developed a micro-osteoarthritis model (OA-PA-MCTs) using PA-MCTs, and it demonstrated native arthritis-like characteristics (Fig. 8B).Using OA-PA-MCTs, we tested several biomolecules with known antiin ammatory effects. [30] [31] [32]Compared to native articular cartilage reported previously, OA-PA-MCTs responded to these drugs similarly (Fig. 8C): Dexamethasone, IGF-1 and TGF-β demonstrated robust effects in reducing sGAG loss, inhibited expression of in ammatory genes (MMP-3 and MMP-13) and upregulated expression of cartilage positively related genes (Sox9, collagen type II and ACAN) than Celecoxib.The overall anti-in ammatory ranking of drugs is in the order: Dexamethasone > IGF-1 > TGF-β > Celecoxib (Table S2).Collectively, OA-PA-MCTs show native-OA-like characteristics with high uniformity and miniature size, which make them suitable for DMOADs screening.PA-MCTs also show an advantage in studies of cartilage development biology.Microinjecting genes into a cartilage explant is superior to transfecting genes into individual chondrocytes because it models at a tissue or organ level to study related signal pathways in normal cartilage development. [33] [34]However, this method suffers the same limitation of explant variations.We demonstrate that PA-MCTs can overcome this limitation by successfully microinjecting dextran, which mimics a gene carrier such as adenovirus, into PA-MCTs as a proof-of-concept study (Fig. 8D).Similar to our drug screening model, PA-MCTs can increase throughput and minimize sample variations compared with native cartilage explants.

Conclusion
In summary, to the best of our knowledge, this is the rst time that patterning cells within a bioprinted unit has been accomplished and its practical signi cance demonstrated.We report the success of recapitulating native functions and characteristics of articular cartilage in PA-MCTs and their assemblies solely by bioprinted cellular patterns.The patterned structure provides a biophysical cue that positively in uences cells toward the chondrogenic phenotype and initiates and maintains cartilagenesis.Our technique represents a potential solution to arthritis.In the future, we anticipate implanting PA-MCTs in vivo, screening new candidate DMOADs, and generating other modular tissues through bioprinted cellular patterns to activate the corresponding tissue genesis.This technique could pave a new avenue for bioprinting to recapitulate native tissue functions.Figure 5 Characterizing the cartilagenesis during the bio-assembly of PA-MCTs vs. H-MCTs.A) Schematic of the process: individual PA-MCTs were cultured for 2 weeks, and multiple PA-MCTs were placed together to assemble for a total of 7 weeks.B) A representative image demonstrating chondrocytes migrating from a PA-MCT into the surrounding hydrogel after 24 h.C) Representative histological images demonstrating the progress of cartilagenesis during the bio-assembly ofPA-MCTs, and darker blue indicates a higher level of sGAG expression.The width of the color bar approximates the ratio of the indicated activity.D) Representative images of the gross appearance of the assembled macro-articular-cartilage over time.E) A representative image of a completed macro-articular-cartilage demonstrating robust sGAG expression.the PA-MCT-assembly forming neocartilage with a smooth white surface (B).Microscopically the PA-MCT-assembly demonstrates seamless structural integration at all locations (C): an irregular cartilage wall interface (i), the bone interface (ii), the partial-thickness defect interface (iii), the deep region in the neocartilage (iv).D, E) H-MCTs failed to repair the defect.A large proportion of the defect was void due to unassembled H-MCTs fell off the explant during culture and a large degree of H-MCT compaction (D-i and sliced views/D-ii).Necrosis, compaction of H-MCTs and void spaces underneath the neo-tissue surface were observed histologically (E).(F) Semi-quanti cation of sGAG content: yellow, green, and cyan dotted regions correspond to native cartilage, the result of full-thickness defect repair and partialthickness defect repair, respectively.The intensity of every pixel in the red channel, ranging from 0 to 255 (color bar), represents the level of sGAG expression.The intensity-distribution-map demonstrates intense sGAG expression throughout the neocartilage, including the deep zone.G) Preliminary zonal structure formed in the PA-MCT-assembly, which is similar to that in the native articular cartilage.H) PA-MCT-assemblies demonstrated signi cantly higher adhesion strength to the host cartilage than H-MCTassemblies in the push-out test.n ≥ 3, data are means ± SD, * P < 0.05 by ANOVA.I, J) Proof-of-concept repairing a large irregularly-shaped cartilage defect in a sheep femoral condyle (I-i) and resurfacing an entire bare bone rabbit femoral condyle (J-i) by the PA-MCT-bio-assembly.All PA-MCTs were cultured for cartilagenesis individually and implanted at week 2. Both demonstrate similar progress: individual PA-MCTs were clearly visible at implantation (I-ii, ~800 PA-MCTs); boundaries between PA-MCTs gradually became blurry (J-iii), and smooth white cartilage surfaces were regenerated at week 7 that defects were repaired (I-iii) and the femoral condyle bone was resurfaced (J-ii).

Figure 6 .
Figure6.Cartilagenesis during integration with native articular cartilage and bone using PA-MCTs vs. H-MCTs.All PA-MCTs and H-MCTs were cultured individually for 2 weeks before integration assays.A) PA-MCTs demonstrate the capability of cartilagenesis.Macroscopy: (i) When integrating to native cartilage and bone, PA-MCTs were individually visible at week 5 and assembled to intact macro-articular-cartilage with smooth and white surfaces at week 7. (ii) H-MCTs only partially assembled and adhered to native cartilage, and adhered loosely to or fell off native bone (arrow).Microscopy: (iii) PA-MCTs seamlessly and histologically integrated with native cartilage and bone with intense sGAG expression indicated by dark blue staining.(iv) Cartilagenesis did not occur in H-MCTs assembly, and weak sGAG was expressed as indicated by light blue staining.

Figure 1 Cell
Figure 1

Figure 4 A
Figure 4

F
Figure 6