Bioprinting of human dermal microtissues precursors as building blocks for endogenous in vitro connective tissue manufacturing

The advent of 3D bioprinting technologies in tissue engineering has unlocked the potential to fabricate in vitro tissue models, overcoming the constraints associated with the shape limitations of preformed scaffolds. However, achieving an accurate mimicry of complex tissue microenvironments, encompassing cellular and biochemical components, and orchestrating their supramolecular assembly to form hierarchical structures while maintaining control over tissue formation, is crucial for gaining deeper insights into tissue repair and regeneration. Building upon our expertise in developing competent three-dimensional tissue equivalents (e.g. skin, gut, cervix), we established a two-step bottom-up approach involving the dynamic assembly of microtissue precursors (μTPs) to generate macroscopic functional tissue composed of cell-secreted extracellular matrix (ECM). To enhance precision and scalability, we integrated extrusion-based bioprinting technology into our established paradigm to automate, control and guide the coherent assembly of μTPs into predefined shapes. Compared to cell-aggregated bioink, our μTPs represent a functional unit where cells are embedded in their specific ECM. μTPs were derived from human dermal fibroblasts dynamically seeded onto gelatin-based microbeads. After 9 days, μTPs were suspended (50% v/v) in Pluronic-F127 (30% w/v) (µTP:P30), and the obtained formulation was loaded as bioink into the syringe of the Dr.INVIVO-4D6 extrusion based bioprinter. µTP:P30 bioink showed shear-thinning behavior and temperature-dependent viscosity (gel at T > 30 °C), ensuring µTPs homogenous dispersion within the gel and optimal printability. The bioprinting involved extruding several geometries (line, circle, and square) into Pluronic-F127 (40% w/v) (P40) support bath, leveraging its shear-recovery property. P40 effectively held the bioink throughout and after the bioprinting procedure, until µTPs fused into a continuous connective tissue. µTPs fusion dynamics was studied over 8 days of culture, while the resulting endogenous construct underwent 28 days culture. Histological, immunofluorescence analysis, and second harmonic generation reconstruction revealed an increase in endogenous collagen and fibronectin production within the bioprinted construct, closely resembling the composition of the native connective tissues.


Introduction
Bioprinting is well-known for having revolutionized the fabrication of complex 3D tissue constructs with precise control over cellular organization and biomaterial composition [1].This technology has paved the way for the development of in vitro tissue equivalents that faithfully replicate native microenvironments, facilitating a deeper understanding of tissue development, disease progression, and therapeutic strategies [2].
To date, two primary bioprinting strategies have emerged: scaffold-based and scaffold-free [3,4].Scaffold-based bioprinting employs biomaterials to sustain cells during the printing process and the tissue maturation; however, this methodology relies mainly on the use of exogenous matrices, such as hydrogels mimicking the extracellular matrix (ECM) features, pre-mixed with cells alone or together with biochemical compounds [3,5].Despite being widely exploited, this approach lacks reliability in achieving the precise assembly of complex and functional tissue equivalents.On the other hand, scaffold-free bioprinting, relies on the assembly of building blocks (cell sheets, spheroids, microtissues) for the tissue fabrication, without the need for scaffold support [6].
Microtissues, which are micrometric pieces of tissue precursor constituted by cells surrounded by selfsecreted neo-ECM, are particularly interesting for connective tissue fabrication, since unlike spheroids and cell sheets, microtissues possess the ideal combination of cells and extracellular materials [7].These building units not only contain cells and biochemical factors, but they also contain the information necessary for their organization in supramolecular structures.The significance of µTPs relies in their pivotal role in achieving structurally competent in vitro tissues, as they serve as the foundation for organized development and integrity of the final tissue structure.Upon contact between each other's, µTPs undergo fusion and self-assemble to form larger tissues, in a faster way, compared to individual cells [8].Thus, with this approach it is possible to obtain a construct made only of cells and self-secreted ECM, better recapitulating the formation of native tissues or organs, compared to the use of exogenous matrices.
However, the lack of a vehicle which holds together the building blocks during the bioprinting process is the main disadvantage of the scaffoldfree approach [9].To address the limitations of both approaches, tunable 'sacrificial' bioinks were introduced, with the goal of supporting the embedded building blocks during the bioprinting process and during their post-printing assembly, promoting inter-and intra-cellular crosstalk, but being able to be removed on-demand, by leaving a scaffold-free endogenous construct [10].
Several state-of-art methodologies were exploited for manipulating and positioning microtissues.A widely exploited technique was the Kenzan method, which involves the precise microtissues arrangement by using an array of fine microneedles [11].Kenzan approach possess some limitation, such as fixed interneedle distances, restricting blocks size and difficulties in fabricating structures with curves and geometrical complexity in z-axis or even structures with overhangs [12][13][14].Other approaches are: (i) biogripper, limited to positioning building blocks of 600-1000 µm [15,16]; (ii) aspiration-assisted bioprinting, which harness the power of aspiration forces for picking and placing building blocks, often causing low cell cohesion and microtissues rupture [17]; (iii) magnetic bioprinting, relying on the use of magnetic field to manipulate and pattern magnetic nanoparticles-labelled building blocks, but the incorporation of magnetic nanoparticles is the drawback [18].These techniques are usually not automated, very time consuming and demand extreme attention to avoid damaging the microtissues during positioning [8,19].Additionally, microtissues should be in direct contact to fuse after deposition and this is not easily achievable with current technologies [20].In contrast, extrusion-based bioprinting technique, by embedding the building blocks inside a matrix (exogenous or endogenous) allows to dispense microtissues of multiple sizes, by using nozzles available in several diameters, without altering the viability of encapsulated cells at optimized shear stress.This approach, firstly exploited by De Moor et al to bioprint spheroids and investigate their vascularization potential, was successfully used over the years to obtain tissue models [21].Common challenges encountered during the printing of microtissues are their sedimentation within the syringe and the clogging of the nozzle [8].Microtissues sedimentation can be prevented by increasing the viscosity of the hydrogel while keeping a good printability of the bioink; whereas, to avoid the nozzle clogging, monodisperse microtissues of smaller size compared to the nozzle diameter should be used and their concentration should be carefully optimized [8].These challenges can be overcome by tuning the bioink properties (mechanical and compositional) and bioprinting parameters (nozzle diameter, printing speed and pressure).
In the context of skin tissue engineering, human dermal microtissues have gained significant attention due to their ability to recapitulate essential features of the native connective counterpart [7,22].Skin is an anisotropic tissue and its manufacturing by using microtissues could provide solution to long-standing problem of anisotropic properties [6,23].We previously proposed a two-step bottom-up approach, based on: (i) the manufacturing of micrometric tissue precursors (µTPs) by means of dynamic seeding of human dermal fibroblasts (HDFs) onto porous gelatine microcarriers within a spinner flask bioreactor, and (ii) the µTPs assembly in appropriate chambers to allow their fusion through cellcell and cell-ECM interactions, in optimized culture conditions, to obtain a completely endogenous dermis equivalent [7,24,25].Nevertheless, to unlock µTPs full potential, their accurate positioning to instruct the formation of a specific pattern is crucial, since the complex hierarchical architecture of native tissues defines their specific function and should be recreated as faithfully as possible.As a result, incorporating these cellular building blocks into a bioprinted construct would represent a promising approach to create a more physiologically relevant in vitro endogenous connective tissue model.
Herein, we applied the extrusion-based bioprinting to obtain dermal µTPs precise dispensing in proximity between each other's, to enable their fusion post-printing in an organized manner, facilitating their assembly into higher-order tissue constructs with better control over their architecture.To achieve µTPs desired packing and fusion throughout the bioprinting process, we took advantage of PF127 both in the bioink and as a support printing bath.PF127 concentration as µTPs embedding-matrix was optimized to obtain an extrudable bioink, whose printing parameters were finely tuned.On the other side, PF127 as bioprinting support bath was used to aid the µTPs packing during the bioprinting process and facilitating their fusion over 4 d of culture to establish functional tissue structures.This optimized approach holds promise for manufacturing several mature endogenous connective tissues.

General workflow
The entire procedure involves three steps: (i) manufacturing of µTPs in dynamic conditions (figure 1(A)); (ii) extrusion-based bioprinting of a bioink composed of µTPs and PF127, in optimized condition to obtain µTPs homogenous dispersion inside the syringe without clogging the nozzle (figure 1(B)), (iii) post-printing phase, relying on the µTPs fusion at optimized microtissues packing factor (MPF), according to JHP theory (figure 1(C)).
Rheological analyses were performed to evaluate the mechanical properties of P20, P25 and P30 formulations and their suitability for extrusion-based bioprinting, by using a rotational rheometer (Anton Paar, MCR 302) with a 25 mm-diameters plate geometry, equipped with a Peltier element.The distance between the two plates was set to 0.958 mm and 600 µl of ice-cold solution was poured on the bottom plate for each test.
Rotational measurements were performed to assess the formulations viscosity and shear-thinning behavior.Steady shear sweep test was performed at 37 • C (600 µl of solution was poured on the bottom plate and pre-heated for 120 s at 37 • C), within the range 1-100 s −1 .Information about the gelation kinetics were obtained from the temperature ramp test at a constant shear rate of 100 s −1 and a heating rate of 1 • C min −1 within a temperature range from 4 • C to 39 • C.
An oscillatory strain sweep test was performed at 37 • C to evaluate the linear viscoelastic region (LVER) of each formulation.Oscillatory measurements of dynamic moduli (storage (G ′ ) and loss (G ′′ )) vs. temperature (4 • C-39 • C) were performed at a frequency of 1 rad s −1 and 0.1% strain.The crossover of G ′ and G ′′ was considered as the gelation temperature.
Gelatin porous microbeads were produced with an optimized double emulsion (O/W/O) approach and crosslinked with 4% w/w of glyceraldehyde (Sigma-G5001).µTPs were obtained by using a previously described two stages bottom-up tissue engineering approach, based on seeding HDFs onto gelatin microbeads and growing them dynamically in a spinner flask bioreactor (Integra) at 30 rpm [27].Cell medium, supplemented with 0.5 mM ascorbic acid (2-0-α-D-Glucopyranosyl-L-Ascorbic-Acid TCI) was replaced every 2 d. µTPs were dynamically cultured for 9 d before their bioprinting, to guarantee the initial collagen synthesis (figure 1(A)).

Sacrificial bioink formulation and characterization
µTPs suspension was transferred to a 50 ml falcon tube and all the medium was removed to have the µTPs dry.An equal volume (50% v/v) of cold (stored at 4 • C) P30 was added to the µTPs and gently mixed with a spatula, in order to have a solution made of µTP:P30 at a ratio 1:1, which was the highest µTPs concentration enabling their homogenous dispersion within the P30 at 4 • C. At this concentration, the bioink formulation underwent sol/gel transition at 37 • C in less than a minute, allowing the maintenance of µTPs homogenous dispersion inside P30 hydrogel.Rheological analysis were performed on µTP:P30 formulation, according to the protocol reported in the previous section: G ′ and G ′′ vs. T (4 • C-39 • C), G ′ and G ′′ vs. shear stress (1-100 Pa), η vs. T (4 • C-39 • C) and η vs. shear rate (1-100 s −1 ).Also, oscillatory stress sweep test (G ′ and G ′′ vs. shear stress) was performed to evaluate the critical stress, within the range 1-100 kPa, at a fixed temperature of 37 • C, frequency of 1 rad s −1 and 0.1% strain.

Supporting bath formulation and characterization
P40 hydrogel was selected as a support bath.G ′ and G ′′ vs. T (4 • C-39 • C), as well as η vs. T (4 • C-39 • C) were performed, as reported in section 2.2.Shear recovery was assessed for P40 formulation with rotational thixotropy by applying low (0.1 s −1 ) and high (100 s −1 ) shear rate at 37 • C and measuring the viscosity.The value of high shear rate was calculated from the shear rate in capillary of power law, by assuming pneumatic extrusion dispenser system is similar to the capillary rheometer and that the geometry of the bioink strand is constant from the start point to the end point and is extruded in a continuous manner (supplementary data, S1) [28].Two cycles of low and high shear rate were performed, each of them of 200 s.

Bioprinting process optimization
For the bioprinting process, AutoCAD 2023 3D modeling software (Autodesk software, San Rafael, CA, USA) was used to model several shapes.The organ regenerator slicer (ROKIT Healthcare Inc., Seoul, Korea) allowed to obtain a G-code-based script to produce the bioprinted structure via an extrusion-based 3D bioprinter (Dr.Invivo 4D6; ROKIT Healthcare Inc., Seoul, Republic of Korea).For bioprinting, 2 ml of the formulation made of µTPs:P30 (50:50) was loaded in the Dr. Invivo 4D6 cartridge.The biodispenser, containing the bioink solution, and the bed, underneath the support bath, underwent pre-warming at 38 • C for 20 min prior to bioprinting.Both temperatures were set and maintained at 38 • C during and after the extrusion process.The bioink was then extruded and printed inside the support bath within a 6-well plate.A 18G nozzle (I.D. 0.84 mm) was selected to ensure the extrusion of µTPs without the nozzle clogging.The slicer was configured with the following settings: a layer height of 1.5 mm, 100% infill density, single-line pattern.Several printing pressures were tested, in the range of 60-78 kPa and the pressure allowing the optimal extrusion of a continuous filament was selected.Multiple printing speed, in the range of 3-6 mm s −1 were analyzed and the speed which allowed to deposit a continuous filament, without showing discontinuities between the printed µTPs, therefore having a high ratio (>50%) between µTPs area and the filament area was selected.Analysis were performed on samples after 24 h, following culture media addition.The optimal parameters were used to bioprint multiple shapes (square, circle and a line) and to assess the bioprinted constructs evolution and µTPs fusion over time.The multi-well plates and the petri dishes containing the bioprinted constructs were placed in the incubator at 37 • C and 5% CO 2 on an Orbital shaker (Neo Biotech, Switzerland) at 200 rpm, and cultured up to 28 d with a change of medium, enriched with ascorbic acid, every 2 d.

Cells viability assessment
The Live/Dead assay (LIVE/DEAD ® Cell Imaging Kit, Thermo Fisher Scientific) was employed following the guidelines provided by the manufacturer.This fluorescence-based kit utilizes a combination of calcein AM and ethidium bromide, enabling a twocolor discrimination of live cells (in green) and dead cells (in red).Prior to incubation with the staining solution, each cell culture condition underwent two washes with phosphate buffered saline (PBS).The Live/Dead solution, comprising 4 µM ethidium homodimer-1 and 10 µM calcein diluted in PBS, was prepared and incubated in the dark for 30 min at 37 • C. Samples were imaged after the incubation with the Live/Dead by using confocal laser scanning microscope (Confocal Leica TCS SP5 II).Z-stacks were captured in several points of the samples with 12-bit resolution, 512 × 512 pixels and 10× or 25× waterimmersion lens.Z-projections were analyzed with Image-J software (Fiji).Cellular metabolic activity was assessed utilizing the Thiazolyl Blue Tetrazolium Bromide (MTT) assay, employing a standard kit procured from Merck, Italy.A 5 mg ml −1 MTT solution was prepared by dissolving MTT powder in PBS, and this stock solution was then combined with serum-free MEM without phenol red at a ratio of 1:10.Following a 4 h incubation at 37 • C in darkness, the MTT solution was replaced with 400 µl of dimethyl sulfoxide (DMSO, Merck, Italy) per well, and the plates were agitated for 30 min on a Plate Shaker to dissolve the formazan crystals, a product of MTT digestion by the cells.Subsequently, 200 µl of each well solution (in duplicate) was transferred to a clear-bottom 96-well plate, and absorbance at 570 nm was measured using a Victor 3 spectrophotometer (Perkin Elmer, Waltham, MA).Cell numbers were estimated based on a standard curve generated by seeding HDF cells at various densities (0, 1000, 5000, 10 000, 50 000, 100 000, and subsequently up to 500 000 with a 50 000 increase).

Post printing analysis
2.8.1.µTPs fusion assessment µTPs fusion was tracked using an optical microscope over 8 d of culture.Three different configurations were analyzed: two microtissues (µTP = 2), three microtissues (µTP = 3) and four microtissues (µTP = 4).µTPs fusion was visually observed at various time points (1 h, 8 h, 1 d, 2 d, 4 d, 5 d, 6 d, 7 d, 8 d) and the images were analyzed using Image-J software.µTPs fusion was evaluated in the three configurations by calculating the ratio between the µTPs area over the whole field area at each time point.
The configuration µTP = 2 was selected to analyze in-depth the fusion between two adjacent bioprinted µTPs over time (fusion index (FI), fusion angle, neck contact region).µTPs FI was calculated as the ratio between two µTPs and the area between them: FI = µTPs AREA Total AREA .
Image-J angle tool was used to measure the included angles between merged µTP at the different fusion time points.Additionally, the length measuring tool of ImageJ was utilized to measure the lengths of the merged region ('neck') between two spheroids [29][30][31].This measurement process was repeated for 10 different merged spheroids using the same method.

Immunofluorescence analysis
For immunofluorescence, samples were fixed in 4% Paraformaldehyde for 30 min at room temperature (RT) and then washed with PBS 1X.Subsequently, they were permeabilized by using 0.1% Triton (Merck, Italy) in PBS 1X for 5 min at RT, followed by another wash with PBS 1X.
For cytoskeleton staining, permeabilized samples were incubated for 1 h with Phalloidin-iFluor 647 Reagent (Abcam, UK) to stain actin filaments, followed by nuclei staining with DAPI.
Collagen and Fibronectin quantification was performed on 20 random acquisitions.The obtained images were subjected to semi-quantitative analysis using Image-J, by measuring the fraction of signal per cell.The area corresponding to collagen and fibronectin was thresholded and quantified: this value was divided by the total number of cells, which were obtained by analyzing the DAPI channel, using the Analyse particles function of the software.

Second harmonic generation (SHG) assessment
To assess collagen production and assembly, the samples were examined using a Confocal Leica TCS SP5 II in combination with a Multiphoton Microscope (Leica Microsystems, Italy).The microscope setup utilized a near-infrared femtosecond laser beam generated by a tunable compact mode-locked titanium: sapphire laser (Chameleon Compact OPO-Vis, Coherent).Collagen in the samples was observed using SHG imaging, with an excitation wavelength (λ EX ) of 840 nm and an emission wavelength (λ EM ) of 420 ± 5 nm.Z-stacks were performed as reported above.Image-J software was used to obtain the collagen fraction, by measuring the amount of collagen present in the ECM space, represented by bright pixels (number of pixels from the collagen-Nc), as compared to the non-collagenous portion, represented by black pixels (number of pixels in the non-collagenous portion-Nb).Collagen fraction (CF) for each time point was calculated by using the following equation in the selected Region of Interest (ROI): CF = Nc Nc + Nb .

Histological staining
For histological analyses, samples were fixed at day 7, day 21 and day 28 with 10% neutral buffered formalin for 30 min at RT.Following that, samples were dehydrated in graded ethanol of 75%, 85%, 95% and 100% (twice), each time for 20 min at RT and treated with 100% xylene (A9982 ROMIL, Italy) for 30 min, prior to the embedding in paraffin.Samples sectioning at 7 µm-thickness slices was performed with a microtome and the obtained sections were stained with Hematoxylin and Eosin (H&E) (Bio Optica, Italy) and Masson's Trichrome (Merck) and mounted with Histomount Mounting solution (Invitrogen, Italy).Images were acquired with Olympus BX53 light microscope with 20x and 40x objective lenses.

Statistical analysis
All the experiments that were performed in triplicate.For image analysis, at least 10 images were analyzed for each sample.Graphics show data expressed as mean ± SD.Differences between groups were determined using the statistic test ANOVA Tukey HSD test for all the comparisons.MTT assay was analysis with non-parametric t-test.Significance between groups was established with a * p value < 0.05, * * p value < 0.01, p value < 0.001 and * * * p value < 0.0001.

Assessment of the bioink formulation and rheology
To assess the effect of different PF127 concentration on rheological properties, rotational and oscillatory tests were performed (supplementary data S2).
All the formulations displayed a decrease in viscosity with respect to the increasing shear rate (shear thinning behavior) (figure S1(A)), which makes PF127 an ideal biomaterial candidate for extrusion bioprinting.Once recognized the LVER, via oscillatory strain sweep test (figure S1(B)), temperature-dependent properties were analyzed.P20, P25 and P30 displayed differences in the dynamic moduli (G ′ and G ′′ ) vs. temperature analysis and in the viscosity ramp.Specifically, higher polymeric concentration showed an increase in viscosity and dynamic moduli at a lower temperature, compared to lower concentrations (figures S1(C) and (D)).P30 showed an increase in viscosity starting at approximately 15 • C, when the PEO and PPO monomers dispersed in solution start to be in contact triggering an increase in the activation energy and the nucleation of micelles: PPO becomes less soluble and forms the core of micelles with the PEO, which forms the shell [32].The viscosity reached its highest value and then remained stable at 22 • C. With the same principle, P25 viscosity started to increase at around 20 • C and reached the equilibrium at 25 • C, whereas P20 at around 25 • C, reaching the equilibrium at 33 • C. P30 was selected as bioink sacrificial material, since it was the most stable formulation in gel state at RT.For the bioprinting process, P30 was mixed with µTPs (1:1) (figure 2(A)) and the rheological properties of P30 formulation changed.P30:µTP showed a shift of the viscosity ramp to higher temperatures: the nucleation of micelles started at 23 • C and stabilized at around 29 • C (Sol/Gel transition temperature), as demonstrated by the oscillatory temperature sweep analysis (figure 2(B)).This shift in gelation temperature indicates that the addition of µTPs hinders the interaction between the micelles, which is needed to form the cubic liquid crystalline phase for the gelation [33].The values of the elastic modulus at equilibrium also showed a decrease by the addition of µTPs, from 26.1 ± 1.2 kPa to 8.9 ± 0.9 kPa (figure 2(C)).P30:µTP displayed a decrease of viscosity at increasing shear rates (figure 2(D)), confirming the shear thinning behavior.Also, amplitude sweep test, represented as a function of the shear stress, showed the yield stress point (Y P ) as the limit of the viscoelastic range and the flow point (F P ) at around 90 kPa, where the loss modulus G ′′ was above the storage modulus G ′ and therefore, the flow could start (figure 2(E)).

Rheological analysis of the support bath
P40 was selected as a support bath.As depicted in figure 2(F), the gradient of concentration (∆C) between the bioink (P30) and the support bath (P40) during the extrusion, was the driving force enabling the flow of water at their interface.The loss of 10% of water between the two phases allowed µTPs packing inside the bioink: from 50% (before and during the extrusion) to 60%, after the water diffusion at the interface between the bioink and the support bath, during the filament deposition.At this MPF, µTPs are in contact between each other's.P40 rheological analysis results are reported in figures 2(G)-(J).P40 showed a temperature-dependent viscosity increase (figures 2(G) and (H)), such as the other PF127 formulation analyzed.However, its gelation temperature was 13 • C, which allowed its stability in gel state in a wide range of temperatures.At rest, P40 exhibited solid-like properties; upon the application of high shear rate, such as the movement of the needle, it becomes fluid-like and can be displaced.This behavior is reported in the graph of thixotropic analysis (figure 2(I)): when a low shear rate was applied (0.1 s −1 ), the viscosity was high (∼6.2kPa•s), whereas at higher shear rate (100 s −1 ), the viscosity became lower (∼10 Pa•s) (figure 2(J)) (supplementary data S1).This means that following the removal of the applied stress, the support bath very quickly recovers its solid-like properties in a self-healing manner, entrapping and supporting the deposited bioink (µTPs).Specifically, the P40 transition between low viscosity (during the printing phase, figure 2(I-I, III)) and high viscosity (after the nozzle removal, figure 2(I-II, IV)) happens in less than 10 s, which is a reasonable time to avoid the µTPs dispersion and disorganization within the support bath, and to allow their confinement within a channel.

Printability assessment and parameters optimization
The bioink printability was evaluated after establishing the process's main parameters.Bed and dispenser temperatures were fixed at 38 • C, to ensure bioink and support bath stability in gel state and cells viability, whereas the nozzle used was 18G, having an inner diameter suitable for µTPs extrusion.Two parameters were investigated: printing pressure and printing speed (figure 3).At pressures lower than 70 kPa, it was not possible to extrude the bioink, which left the nozzle in forms of drops; whereas starting from 78 kPa the bioink was defined printable, meaning that it was able to be extruded in a form of continuous filament [34].At increasing pressures, it was observed a loss of printing accuracy, due to an increase in the filament diameter.Consequently, 78 kPa was defined as the minimal pressure initiating the printing process (figure 3(A)).On the other side, speeds in the range of 3-6 mm s −1 were analyzed.In the graph in figure 3(B) it is reported the ratio between the area occupied by the µTPs over the filament area: at 3 mm s −1 speed, the filament was full of µTPs, indeed their percentage of occupancy within the extruded filament was twice the one at 6 mm s −1 .Thus, at 3 mm s −1 , it was possible to extrude a continuous filament, without leaving any gap between the µTPs, whereas at increasing speed the µTPs deposition was discontinuous (figure 3(B)).In addition, lower speeds were excluded to fasten the process, while obtaining optimal results.The printing parameters that had the most accurate results in terms of filament extrudability, resolution and continuity were a printing pressure of 78 kPa and a printing speed of 3 mm s −1 .Apart from the square, even a line and a circle were printed to demonstrate the printability of the bioink with the optimized parameters after 24 h (figures 3(C) and (D)).

Evaluation of bioprinting process effect on cells viability
The viability of cells post-printing remained consistently high at both 3 h and 48 h, emphasizing the robustness and enduring biological sustainability of the bioprinting process (figures 3(E) and (F)).Remarkably, the extrusion procedure, coupled with the inherent associated shear stress, exhibited no detrimental impact on cellular viability: only a minimal number of dead cells were observed at 25× magnification, at both time points.Furthermore, MTT assay revealed a continued proliferation of cells following the bioprinting process from 3 h to 48 h, emphasizing the sustained metabolic activity of the cellular population.These results demonstrated the biocompatibility of the developed biofabrication technique developed and of the materials used in the process.

Evaluation of the µTPs fusion
After successfully demonstrating our capability of bioprinting µTPs within self-healing support hydrogels at high resolution, we aimed to investigate the potential of µTPs fusion for creating controlled structures.This analysis was performed on 2,3 or 4 µTPs bioprinted in proximity.We observed that, when µTPs were bioprinted in direct contact, they underwent adhesion between each other's over a 4 d culture period, mediated by cell-cell and cell-ECM interaction, which is fundamental to prevent them from scattering within the supporting medium during erosion.Notably, the support hydrogel (P40) used to sustain the bioprinted structure, was gradually eroded and washed away over time by the culture medium and the dynamic conditions.After 4 d of culture, PF127 totally disappeared.Throughout an extended culture time (up to 8 d), the adjacent µTPs underwent cohesion between each other's, until forming a single construct, where individual units were no longer distinguishable and completely assembled, at day 8.The fusion process involved not only cell-cell and cell-ECM adhesion and interaction, but also cells and ECM molecules migration at the interface between the adjacent micromodules, together with the synthesis of self-secreted neo-ECM [7].Interestingly, the fusion occurred even when starting from µTPs in loose contact just over a small portion of the external surface, as previously observed by Daly et al [20].The graph in figure 4(B) shows that in the three conditions (µTPs = 2, µTPs = 3, µTPs = 4) the area occupied by the µTPs, with respect to the total area of the acquisition field, decreased over time: from ∼36.1% at 8 h to ∼29.7% at day 8 for µTPs = 2; from ∼46.2% at 8 h to ∼35.1% at day 8 for µTPs = 3; from ∼47.0% at 8 h to ∼37.2% at day 8 for µTPs = 4.With the gradual µTPs fusion, bioprinted constructs underwent shrinkage and compaction, with a decrease of the total area (figure 4(A)) [35,36].
Figure 4(C) reports scanning confocal laser microscopy multichannel images at day 8 of fused µTPs in the three conditions (µTP = 2, µTP = 3 and µTP = 4), highlighting cell's nuclei (green), cell's cytoskeleton (magenta) and unstained collagen (gray).Cell-ECM interaction and actin cytoskeleton are known to play an essential role in µTPs fusion and matrix production, which are strongly regulated by their surrounding microenvironment [37].These images confirmed fusion between adjacent µTPs, and regions of high fibroblast density at their interface, together with neo-ECM deposition.
Building on these findings, we analyzed the sample µTP = 2 at the interface between the two adjacent µTPs (figure 4(D)).To explain the process of fusion, we conducted measurements of the FI, revealing that the area occupied by the µTPs increased over time, as the fusion occurred, from ∼80% to ∼90% (figure 4(E)).In addition, we measured the tangent angles at the fusion points between two µTPs.Eight hours after merging, the original shapes of the µTPs remained clearly distinguishable, and the angles between them varied from ∼49 • to ∼160 • and from ∼25 • to ∼155 • over 8 d of culture, indicating almost the complete fusion (figure 4(F)).The region between the two merged µTPs, referred to as the 'neck' , measured ∼102 µm 8 h after printing and gradually expanded to ∼1 mm after 8 d, due to the increase of the points of contact and fusion between the two adjacent µTPs (figure 4(G)).

ECM formation
To analyze the production of endogenous ECM by HDFs, the bioprinted constructs were cultured for 28 d.As abovementioned, after 4 d of culture PF127 had completely liquefied due to superficial erosion, allowing the construct made of fused µTPs to mature under dynamic conditions.Collagen and fibronectin, two essential components of connective ECM, exhibited increased production over the 28 d of culture (figures 5(A) and (B)).Additionally, SHG signal showed a weaker signal at day 7, compared to day 21 and day 28, suggesting that collagen production was still not organized in fibers (figure 5(C)).Quantitative analysis confirmed the qualitative assessment, revealing a progressive increase in the ECM components over time.Histological analysis showed that the µTPs fused over time, creating a bigger-size construct rich in ECM.Masson Trichrome staining of collagen highlighted an increase in collagen production over time (evidenced by increased blue staining), along with microbeads degradation, leaving place to neo-endogenous ECM.Beads were mostly degraded after 28 d of culture, as clearly observable from their weaker staining in Masson trichrome images (figure 5(D)).

Discussion
The emergence of 3D bioprinting technologies in the field of tissue engineering has introduced the potential to fabricate in vitro tissue models with precise control over their geometry and architectural composition.Indeed, the replication of the intricate characteristics of tissue microenvironments is essential for gaining deeper insights into tissue regeneration processes, disease progression, and the effectiveness of treatment options [38].
Leveraging our extensive expertise in generating 3D tissue equivalents that exhibit histological and functional competence, such as skin, gut, and cervix, we had previously developed a two-step bottomup approach [24,27,39,40].This methodology involved producing µTPs and dynamically assembling them in a maturation chamber to form macroscopic tissue equivalents.However, the main drawback of this approach lays in the absence of automated and scalable technology capable of precisely and coherently assembling µTPs into intricated functional in vitro tissues.To address this limitation, herein, we integrated bioprinting technology into this well-established bottom-up paradigm to replicate the native connective tissue, which provided the ability to organize and guide the assembly of µTPs into predefined shapes.
To the best of our knowledge, we are reporting for the first time the synergic integration of manufacturing endogenous in vitro constructs with modular bioprinting technology.The goal is to couple a highly physiologically relevant tissue construct with the automation, standardization, and enhancement of the production of in vitro functional tissues.Our primary inquiry in this study was 'how is it possible to start from a suspension of µTPs, existing as single units, and to obtain their packing in a close enough manner to guarantee their fusion following the bioprinting process?' .To address this question, we assumed our µTPs can resemble hard spheres in a cubic unit, as outlined in the JHP theory, born with Kepler.According to this theory, the loosest random way to pack these spheres (referred to as microtissues packing factor-MPF) results in a density of about 55%, while the most compact randomized packing yields a of about 64% [41][42][43].To accomplish this, we manufactured HDF µTPs, as previously reported [27], in optimized dynamic culture conditions, to enable massive production of viable and functional micrometric building blocks.We explored several PF127 concentrations, starting from 20%, which represents the minimum PF127 concentration required for the formation of PEO-PPO micelles.We extended the investigation to 30%, which is the optimal for having a cohesive gel capable of being stable at RT, as demonstrated by our rheological characterization, and able of being extruded [44,45].Amongst P20, P25 and P30, the latter was selected as the most suitable for being combined with µTPs to obtain an extrudable bioink.
The presence of P30 in the bioink had a double function: (i) to sustain µTPs homogenous dispersion within the syringe: as demonstrated by the temperature-dependent rheological analysis, the bioink exhibited liquid characteristics at temperatures below 30 • C, allowing µTPs dispersion, and underwent sol/gel transition at physiological temperature within 1 min; (ii) to enable the µTPs printability, by providing shear-thinning properties to the bioink, as demonstrated by the rotational rheological analysis.The optimal µTPs concentration which allowed their homogenous dispersion, without causing clusters clogging the nozzles during P30 sol/gel transition, was found to be 50% v/v.However, according to the JHP theory, MPF should ideally fall between 55% and 64% to achieve their effective packing [42].In fact, at 50% µTPs are not in direct contact with each other's, a crucial factor in ensuring their post-printing fusion and preserving the shape of the bioprinted filament.
Hence, to achieve an optimal MPF of 60%, it was necessary to reduce the water content from the bioink by 10%.To accomplish this, the bioink was extruded into a support bath composed of P40.Indeed, the interface between the support bath and the bioink created a concentration gradient of water, serving as the driving force for the removal of 10% of water from the bioink through osmosis during the extrusion phase when viscosity is lower.As a result, the postprinting MPF reached approximately 60%, causing the µTPs to come closer together through packing.The thixotropic properties of the P40 support bath, particularly its ability to recover its shape after shear stress application (as shown by rheological analysis), played a key role in this process.This property allowed the support bath to locally decrease in viscosity upon stress application, facilitating bioink deposition (and the water suction of 10%), while recovering to its original state after the shear stress was removed (after the nozzle movement).The shear stress demonstrated to not affect cells viability, as demonstrated by Live/dead and MTT assays at 3 h and up to 48 h post-printing.
Once optimized the bioprinting parameters, the support bath allowed µTPs physical confinement and the maintenance of their spatial patterning following extrusion, sustaining their fusion and the formation of a continuous filament and various differently shaped structures.In this way, it was possible to preserve the shape fidelity of the bioprinted filament during the extrusion.The support bath could be removed by placing the petri dish onto an ice bath or at temperatures lower than 14 • C, as showed in the rheological temperature sweep test.However, during the constructs culture and the change of the medium, P40 underwent gradual surface erosion.Notably, after 4 d of culture the support bath was barely observable in the petri dish.Interestingly, the study of fusion revealed that HDFs µTPs were capable of adhere and start their fusion within 4 d in different configurations (2, 3 or 4 µTPs).In doublets configuration, we observed that the FI overpassed the 80% at day 4, with a length of neck of approximately 600 µm (µTPs adhesion phase).FI reached a value of almost 100% by day 8 (µTPs cohesion and fusion), with a contact of 180 • and a neck length of 1 mm.The fusion of µTPs was facilitated by the immature nature of the ECM produced within the µTPs during their maturation stage, which enabled not only cell/cell and cell/ECM adhesion, but also cells and ECM molecules migration when in contact between each other's, until completely being assembled forming a continuum of connective tissue [46].Compared to previous work, whose focus was often confined to the study of short time cell/cell adhesion between microunits (cell sheets or spheroids), the herein reported cohesion and fusion process between µTPs, over a longer-term culture time, resembled more faithfully the native tissue repair and regeneration process [3,20,47].Also, this result was fundamental in ensuring the mechanical stability of the printed construct, a concern often influenced by the extraction process of non-mechanically stable construct during PF127 liquefaction [48].With our approach, it was possible to obtain bioprinted in vitro endogenous connective tissues composed solely of cells and their produced ECM.In fact, as observed from the immunofluorescence and histological analysis, most of the gelatin microbeads were degraded at 28 d, while the ECM was rich in collagen and fibronectin.The gradual production of ECM over the 28 d of bioprinted construct culture, was fundamental to establish a matrix-dependent communication network, providing cell adhesion sites, biochemical and biomechanical signals regulating cell survival, migration and differentiation.

Conclusion
This article delves into the cutting-edge research involving the bioprinting of human dermal µTPs as fundamental building blocks to manufacture an endogenous and functional in vitro connective tissue.Compared to non-bioprinting methods, the use of bioprinting technology provides automation and precise control over cell arrangement, ECM composition, and their spatial organization, mimicking the native connective tissue.
Implementation of successful tissue guidance holds significant potential for the engineering of spatially complex tissues, where it is important controlling the architecture of the ECM and specifically the collagen network, which remains a key challenge in the field.Our next step in this research field involves refining and standardizing our approach to fabricate more complex 3D structures, including tubular shapes.We aim to implement this technique with microtissues of varying dimensions to expand its versatility.Additionally, the herein developed strategy, could also be potentially used to arrange µTPs composed of diverse cell types.In this way it will be possible to address the challenge posed by the histologically heterogeneous nature of most organs, where multiple cell types coexist.This pioneering approach holds significant potential for manufacturing complex heterotypic organs and tissues, marking a distinctive advancement in the field of biofabrication.
Finally, this approach, owing to its ability to fabricate functional constructs in a short and automated manner, holds potential for in vivo microtissues bioprinting.This can be achieved by selecting suitable µTPs as bioink building blocks and printing customized forms, based on the 3D scan of the defect.

Figure 1 .
Figure 1.Schematic illustration of the whole process.(A) Fabrication of µTPs by seeding human dermal fibroblasts (HDFs) onto gelatin microbeads, in dynamic conditions, within a spinner flask bioreactor, followed by the (B) µTPs homogenous suspension inside the PF127 for extrusion-based bioprinting.(C) Finally, µTPs are allowed to fuse post-extrusion, by ensuring their close proximity at a concentration (MPF) between 55% and 64%, according to Jammed Hard-Particle Packings theory (JHP).MPF is calculated as the ratio between the volume of the µTPs in a cubic unit (Vµ TPs ) and the volume of the cubic unit (VCUBE).

Figure 2 .
Figure 2. Formulation and rheological analysis of bioink and support bath.(A) Sacrificial bioink was composed of P30 and µTPs at 1:1 concentration, which allowed their homogenous dispersion inside the syringe for bioprinting.(B) Temperature ramp: viscosity vs. temperature analysis (4 • C-39 • C). (C) Oscillatory measurement of G ′ and G ′′ vs. temperature (4 • C-39 • C). (D) Shear sweep test: viscosity vs. shear rate analysis (1-100 s −1 ).(E) Oscillatory stress sweep test: G ′ and G ′′ vs. shear stress (1-100 kPa).(F) Sacrificial support bath was composed of P40, to have a gradient of water concentration of 10% between the supporting bath and the bioink, allowing the water flow at their interface during the extrusion and the µTPs packing factor (MPF) increased from 50% to 60%, according to JHP theory.(G) Temperature ramp: viscosity vs. temperature analysis (4 • C-39 • C). (H) Oscillatory measurement of G ′ and G ′′ vs. temperature (4 • C-39 • C). (I) Thixotropic analysis of support bath, by applying low (0.1 s −1 ) and high (100 s −1 ) shear rate for two cycles at 37 • C. (J) Table of the values of viscosity for each thixotropic analysis phase.

Figure 3 .
Figure 3. Bioprinting parameters optimization and assessment of the bioprinting process on cells viability.(A) Pressure optimization: at pressures lower or equal to 70 kPa the bioink was extruded in forms of drops, while at 78 kPa it was possible to obtain a continuous filament.(B) Speed optimization: at a speed of 3 mm s −1 it was possible to deposit a filament full of adjacent µTPs, whereas at 6 mm s −1 these were distant and the filament discontinuous.Graph showing the ratio between the area occupied by the µTPs over the filament area (%).(C)-(D) Bioprinting of different shapes (square, line and circle), with the optimized parameters, reported in the table.Scale bars =10 mm.(E) Live/Dead assay results, showing at 3 h and 48 h calcein (green, live cells) and ethidium bromide (Eth.Br., red, dead cells) and their merge, at 10× and 25× magnification.Scale bars =200 µm.(F) Analysis of live cells number at 3 h and 48 h post-printing, performed with MTT assay.Statistics: * p < 0.05 (non-parametric t-test).

Figure 4 .
Figure 4. Analysis of the µTPs fusion.(A) Snapshots of the fusion of spheroids obtained from experiments at different time point post-printing (8 h, day 1, day 2, day 4, day 5, day 6, day 7, day 8).Scale bar = 1 mm.(B) Analysis of the decrease in the area occupied by the µTPs over the capture area, over time.(C) Scanning confocal laser microscopy of fused µTPs at day 8 in the three conditions (µTP = 2, µTP = 3, µTPs = 4): cell's nuclei (green), cell's cytoskeleton (magenta) and unstained collagen (gray).(D) Snapshots of the fusion of two µTPs at their interface.Scale bar = 250 µm.(E) µTPs fusion index measurements: Fusion index % = µTPs area/total area.(F) Analysis of the fusion angles, and of the (G) length of the neck between two adjacent µTPs.

Figure 5 .
Figure 5. Composition of constructs.(A) Immunofluorescence images showing the collagen I (green) and nuclei (blue) over time, and quantification of the collagen I signal (µm 2 /number of nuclei).(B) Immunofluorescence images showing the fibronectin (green) and nuclei (blue) over time, and quantification of the fibronectin signal (µm 2 /number of nuclei) (C) Images of SHG (gray) signal and cells nuclei (blue) and quantification of the collagen fraction.(D) Histological analysis: Hematoxylin/Eosin show the increase in cell-synthetized ECM over time and the fusion points between adjacent µTPs are indicated by arrows; Masson Trichrome showing the increase in the deposition of collagen in the construct over time and the degradation of the gelatin microbeads (indicated by asterisk * ).Scale bars: 200 µm.Statistics: plots report mean values ± SD. * p value < 0.05; * * p < 0.01; * * * p < 0.001; p < 0.0001.