Dynamic microfactories co-encapsulating osteoblastic and adipose-derived stromal cells for the biofabrication of bone units

Cells with differentiation potential into mesodermal types are the focus of emerging bone tissue engineering (TE) strategies as an alternative autologous source. When the source of cells is extremely limited or not readily accessible, such as in severe injuries, a tissue biopsy may not yield the required number of viable cells. In line, adipose-derived stromal cells (ASCs) quickly became attractive for bone TE, since they can be easily and repeatably harvested using minimally invasive techniques with low morbidity. Inspired by the multiphenotypic cellular environment of bone, we propose the co-encapsulation of ASCs and osteoblasts (OBs) in self-regulated liquefied and multilayered microcapsules. We explore the unique architecture of such hybrid units to provide a dynamic environment using a simple culture in spinner flasks. Results show that microtissues were successfully obtained inside the proposed microcapsules with an appropriate diffusion of essential molecules for cell survival and signaling. Remarkably, microcapsules cultured in the absence of supplemental osteogenic differentiation factors presented osteopontin immunofluorescence, evidencing that the combined effect of the dynamic environment, and the paracrine signaling between ASCs and OBs may prompt the development of bone-like microtissues. Furthermore, microcapsules cultured under dynamic environment presented an enhanced mineralized matrix and a more organized extracellular matrix ultrastructure compared to static cultures used as control. Altogether, data in this study unveil an effective engineered bioencapsulation strategy for the in vitro production of bone-like microtissues in a more realistic and cost-effective manner. Accordingly, we intend to use the proposed system as hybrid devices implantable by minimally invasive procedures for bone TE applications.


Introduction
Since Tissue Engineering (TE) principles were defined [1], significant progress has been performed in orthopedic research. The combination of cells and biomaterial-based scaffolds, in order to construct living tissues, has been viewed as a potential alternative to the current gold standard treatment, namely the autogenous bone grafting. Whereas classical approaches have not been entirely effective in the restoration of functional bone tissue, a novel generation of devices should instruct nearby cells to promote endogenous repair mechanisms while allowing an orchestrated spatiotemporal delivery of single or multiple factors [2,3]. Therefore, prior to the design of novel regenerative approaches, a fundamental understanding of the inter-cellular network of native bone tissue is of utmost importance [4]. Moreover, the combination of cells with multilineage differentiation potential with biomimetic devices significantly broads their application for TE strategies. Among the different tissue sources, adipose-derived stromal cells (ASCs) quickly became attractive for bone TE, owing to a number of appealing features, such as being available in large quantities with diminutive donor site morbidity or patient discomfort [5,6]. However, controlling cell multipotency and engineering bone in vivo remains a significant challenge, as it often leads to heterotypic and mechanically inferior osseous tissues. Inspired by the multiphenotypic cellular environment of bone, we hypothesized that self-regulated liquefied and multilayered microcapsules [7][8][9][10], already tested in vivo [11], loaded with ASCs and osteoblasts (OBs) could be a promising attempt. For that, microcapsules generated by electrohydrodynamic atomization (EHDA) were composed by (i) a multilayered membrane obtained through the layer-by-layer assembly of three polyelectrolytes, namely poly(L-lysine), alginate, and chitosan, (ii) a liquefied alginate core, (iii) surface functionalized poly(ε-caprolactone) microparticles (μPCL), and (iv) cells. While the multilayered membrane wraps all the cargo contents and ensures permeability to essential molecules for cell survival, the liquefied core maximizes their diffusion through the entire 3D construct, and thus the oxygen limitation size of 150-200 μm of tissue engineered constructs is eliminated. Within this unique liquefied environment, cells can move freely and construct 3D cell-mediated systems by recruiting surface functionalized microparticles according to their needs. Moreover, taking advantage of the unique confined liquefied core microenvironment, the proposed microcapsules were tested using a spinner flask. The production method of liquefied and multilayered microcapsules is represented in scheme 1. In vivo, the paracrine signaling between osteoblastic cells are imperative to the development, maintenance and adaptation of the skeleton. The biophysical stimulation provided by in vitro dynamic cultures were already shown to upregulate osteogenic markers, and stimulate bone mineralization [12,13]. Therefore, the dynamic environment is expected to maximize the interaction between the different multiphenotypic cells and microparticles, while also mimicking the dynamic environment of native tissues. In this study, microcapsules encapsulating only ASCs (MONO microcapsules) or a co-culture with OBs (CO microcapsules) were cultured up to 21 d in culture medium with (OSTEO) or without (BASAL) osteogenic differentiation factors. Our hypothesis is that by recreating the specific microenvironment of the bone regenerative process inside microcapsules, new microtissues with superior quality and without requiring any osteogenic medium supplementation could be engineered.

PCL microparticles production and functionalization
Polycaprolactone microparticles (μPCL) were produced by emulsion solvent evaporation technique. Briefly, a 5% w/v PCL (molecular weight (Mw)∼ 80 000, Merck) solution was prepared in methylene chloride (Honeywell). Then, the PCL solution was slowly added to a stirring 0.5% w/v polyvinyl alcohol (PVA, Merck) solution. After 2 d under agitation at room temperature (RT), the μPCL were collected, washed several times with distilled water, and sieved to obtain a diameter range of 40-50 μm. Afterwards, the surface of μPCL was modified by plasma treatment technique. μPCL were placed into a low pressure plasma reactor chamber (ATTO, Diener Electronic) in which air was used as gas atmosphere. A low-pressure glow discharge was generated at 30 V and 0.2-0.4 mbar for 15 min. Then, μPCL were immediately sterilized and immersed in an acetic acid solution (20 mM, Chem-Lab NV) containing collagen I (10 μg cm −2 , rat protein tail, ThermoFisher Scientific) for 4 h at RT.

Bioencapsulation setup
At 90% of confluence, ASCs (passage 6) and OBs (passage 6) were washed with phosphate buffered saline (PBS) solution, and detached using trypsin-EDTA solution (Merck) at 37°C for 5 min. The two cell suspensions were centrifuged for 5 min at 300 g to discard the supernatant, and re-suspended (5×10 6 cells ml −1 ) in 2.5% w/v of low viscosity sodium alginate from brown algae (ALG, Merck) prepared in sodium chloride solution (0.15 M, NaCl, LabChem) with MES hydrate (25 mM, Alfa Aesar) containing surface-functionalized μPCL (30 mg.ml −1 ). Alginate microgels were then produced by EHDA technique (Spraybase, Avectas) using calcium chloride (0.10 M, CaCl 2 , Merck) solution as the crosslinking bath for 10 min. The operating parameters for electrospray technique were 50 ml h −1 of flow rate, 22 G needle, tip to collector distance of 8 cm, and 10 kV of voltage. Subsequently, layer-by-layer was performed using three different polyelectrolytes (0.3 mg ml −1 ), namely poly(Llysine) (PLL, Mw∼30 000-70 000, Merck), followed by ALG, and chitosan (CHT, NovaMatrix), to produce the multilayered membrane. The process was repeated until a 10-layered membrane was created. Ultimately, the liquefied core was obtained by chelation with EDTA solution (0.02 M, Merck) for 2 min at RT. The pH of all solutions was set to 6.7, excepting for CHT (pH 6.3). Two different formulations of liquefied and multilayered microcapsules were obtained, namely a mono-culture of ASCs (MONO microcapsules), and a co-culture (1:1) of ASCs and OBs (CO microcapsules). Each obtained formulation was cultured in α-MEM medium, with (OSTEO) or without (BASAL) osteogenic supplementation. OSTEO medium was obtained by supplementing BASAL medium with ascorbic acid (50 μg ml −1 , Merck), β-glycerophosphate (10 mM, Merck) and dexamethasone (10 mM, ACROS Organics). Samples were cultured under static or a dynamic environment up to 21 d at 37°C in a humidified 5% CO 2 air atmosphere. For static culture, plastic flasks were used. For dynamic culture, microcapsules were maintained in a spinner flask (Celstir, Wheaton) with double sidearms for good gas exchange, at 50 rpm. All solutions were sterilized by filtration using a 0.22 μm filter and the entire procedure was performed under sterile conditions.

Cell viability
The survival of the encapsulated cells was evaluated by live-dead fluorescence assay according to the manufacturer's recommendation (ThermoFisher Scientific). Briefly, at 1, 7, 14, or 21 d post-encapsulation, samples were washed with PBS and then stained with the kit components at 37°C for 20 min and protected from light. Afterwards, samples were visualized by fluorescence microscopy (Axio Imager 2, Zeiss).

Cytoskeleton F-actin staining
The F-actin network of the encapsulated cells in liquefied microcapsules was visualized after fluorescence phalloidin staining. At 1, 7, 14, or 21 d postencapsulation, samples were washed with PBS and fixed in formalin (4% v/v) for 1 h at RT. Following 0.1% Triton X (Merck) permeabilization for 5 min at Scheme 1. Schematic representation of the production method of liquefied and multilayered microcapsules. (I) Under influence of electrical forces, a liquid jet of alginate containing a dispersion of cells and surface modified polycaprolactone microparticles (μPCL) breaks up into droplets. After crosslinking in calcium chloride (CaCl 2 ), spherical microgels co-encapsulating adipose-derived stromal cells (ASCs), human osteoblasts (OBs) and μPCL are obtained within an average diameter of ca. 600 μm. (II). Layer-by-layer is subsequently performed using poly(L-lysine), alginate, and chitosan as polyelectrolytes to produce a 10-layered membrane. (III). The core is liquefied after immersion in ethylenediaminetetraacetic acid solution (EDTA). Ultimately, to better mimic the dynamic environment of native tissues, microcapsules are cultured in spinner flasks. Static flaks are used as control.
2.6. Scanning electron microscopy and energy dispersive x-ray spectroscopy (SEM-EDS) At 21 d post-encapsulation, the membrane of microcapsules was disrupted to expose the core content. Then, samples were fixed in 4% v/v formalin, and dehydrated in a graded series of ethanol. Afterwards, microcapsules were fixed with a carbon tape onto a graphite stub (Ted Pella) and sputtered by a thin film of carbon (K950X Turbo-Pumped Carbon Evaporator). Morphological and compositional analysis were carried out by scanning electron microscopy (accelerating voltage 15 kV, SEM Hitachi, SU-70 instrument) coupled with an energy dispersive x-ray detector (EDS Bruker, Quantax 400 detector). Calcium (Ca) and phosphorous (P) peaks were determined by EDS spectra using Esprit software. Ca/P ratio was calculated by deconvolution of Ca and P peaks after background subtraction. EDS spectra of lyophilized BASAL and OSTEO cell culture media were also performed as controls of the Ca/P values supplied to the microcapsules.

Attenuated total reflection Fourier-transform infrared spectroscopy (ATR-FTIR)
At 21 d post-encapsulation, samples were fixed in 4% v/v formalin, and dehydrated in a graded series of ethanol. ATR-FTIR spectra were measured on a Bruker Alpha infrared spectrometer controlled by the OPUS software package (version 7.0). Background and sample measurements were performed in a range between 4000 and 500 cm −1 with a resolution of 4 cm −1 and averaging 256 scans, at RT and with controlled humidity. Amide I bands were deconvoluted with the PeakFit software using the second derivative procedure as elsewhere described [14,15]. After a linear baseline correction, the curve fitting was performed with Gaussian functions using a nonlinear least squares fitting routine (coefficient of determination r 2 >0.999).

Osteopontin immunofluorescence detection
The secretion of osteogenic marker osteopontin (OPN) was analyzed by immunofluorescence detection. After 21 d of culture, microcapsules were subsequently washed with PBS, fixed in 4% v/v formalin for 1 h at RT, and permeabilized with 0.1% Triton X for 5 min. Then, non-specific binding was blocked by immersion of the samples in FBS (5% v/v in PBS) for 1 h at RT. Afterwards, microcapsules were incubated overnight at 4°C with the primary antibody mouse anti-human osteopontin (1:200 in 5% FBS, Biolegend). Upon PBS washing, samples were incubated with the secondary antibody anti-mouse Alexa-Fluor 647 (1:500 in 5% FBS, ThermoFisher Scientific) for 1 h at RT. Samples incubated only with the secondary antibody were used as controls. Ultimately, samples were counterstained with DAPI (1:1000 in PBS, 1 mg ml −1 , ThermoFisher Scientific) for 5 min at 37°C, and visualized by fluorescence microscopy (Axio Imager 2, Zeiss).

OPN and VEGF cytokine quantification
The amount of OPN and human vascular endothelial growth factor (VEGF) released by the encapsulated cells was assessed by ELISA quantification assay. For that, the supernatants (1 ml) of cell culture media at 21 d of culture of the liquefied microcapsules were stored at −80°C until analysis. Commercially available human OPN and VEGF ELISA development kits (Abcam) were performed according to the manufacture's specifications. The measurements were read at 450 nm in a microplate reader (Gen 5, Synergy HT, Biotek).

Von Kossa and Masson's Trichrome histological staining
At 21 d of culture, the presence of phosphate deposits and collagen was assessed by Von Kossa and Masson's Trichrome staining, respectively. For that, microcapsules previously fixed in 4% of buffered formalin, were processed in an automatic tissue processor (STP120, Microm) and then embedded in paraffin. Histological section (5 μm) were obtained using a microtome (HM355E, Microm, ThermoFisher Scientific) and placed in adhesive slides (SuperFrost, ThermoFisher Scientific) for von Kossa and Masson's Trichrome staining.
2.12. Hydroxyapatite fluorescence staining At 21 d of culture, the presence of hydroxyapatite (HA) crystals was assessed using the OsteoImage™ Mineralisation Assay kit (Lonza) according to the manufacturer's instructions. Samples were counterstained with DAPI (1:1000 in PBS, 1 mg ml −1 , ThermoFisher Scientific) for 5 min at RT, and visualized by fluorescence microscopy (Axio Imager 2, Zeiss).

Diameter measurements
The size of μPCL, microcapsules, and μPCL-cell aggregates was measured by ImageJ image analysis software. Each value presented for μPCL-cell aggregates correspond to the mean of length and width, for a total of 10 samples for each condition in the different timepoints. For μPCL and microcapsules, exactly 100 and 50 diameter measurements were performed, respectively.

Morphological analysis of microcapsules
The successful production by EHDA technique of spherical microcapsules encapsulating cells and μPCL was evaluated by optical microscopy ( figure 1(A)). Microcapsules and μPCL presented diameters of 608.5±122.3 μm and 45.7±7.3 μm, respectively. Regardless the distribution of microcaspules size ( figure 1(A)), the microparticles:cells ratio is maintained since microparticles and cells are encapsulated at a ratio of 30 mg and 5×10 6 per ml of alginate, respectively. The random distribution of the cells within the microcapsules core was confirmed by microscopic analysis. Prior to the encapsulation procedure, ASCs and OBs were fluorescently marked with lipophilic dyes. Both green and red fluorescence identifying respectively ASCs and OBs could be visualized after 1 and 7 d of culture ( figure 1(B)). Livedead assay shows that, up to 21 d post-encapsulation, the majority of encapsulated cells remained viable in all formulations ( figure 1(C)), evidencing the mild conditions of the cell encapsulation technique proposed. Importantly, these results also evidence the ability of microcapsules for long-term cell survival, a major challenge for tissue engineering strategies aiming cell encapsulation.

In vitro evaluation of the microtissues
The fluorescence staining of F-actin filaments ( figure 2(A)) evidences the interaction and structural organization of the encapsulated cells with the μPCL inside the compartmentalized and controlled environment of microcapsules, after one and seven days of culture. Interestingly, it is possible to observe that the dynamic environment led to the development of significantly larger aggregates of cells and μPCL comparing with static conditions ( figure 2(B)), which evidences the maximized interaction between cells and μPCL inside of the microcapsules. These results were mainly visualized in the first days of culture, since, over time, the developed microtissues tend to achieve similar sizes.

Osteogenic potential evaluation
The late osteogenic marker OPN was visualized by immunofluorescence staining in MONO and CO microcapsules cultured in BASAL and OSTEO media under static and dynamic conditions at 21 d postencapsulation. OPN is a calcium-binding glycophosphoprotein involved in bone remodeling and its expression is linked with extracellular osseous matrix mineralization [16]. Remarkably, the OPN marker was expressed in microcapsules cultured in BASAL medium under dynamic environment, evidencing that the combined effect of the shear stress and the cells coculture may prompt the development of bone-like microtissues ( figure 3(A)). The profile of OPN and VEGF released by MONO and CO microcapsules cultured in the different conditions is shown in figure 3(B). The release of OPN was in accordance with the immunofluorescence staining. Despite no statistically relevance between static and dynamic conditions was observed, the release of OPN was higher in the microcapsules cultured under mechanical stimuli. Likewise, the release of OPN was higher for CO microcapsules, with statistical difference for microcapsules cultured in BASAL medium and under static environment. Additionally, VEGF is a vascular growth factor involved in the development of new blood vessels and, therefore, critical for bone regeneration [17]. In MONO microcapsules, the release of VEGF was significantly higher in microcapsules cultured in dynamic environment. Furthermore, for static conditions, the secretion of this growth factor was significantly enhanced in CO microcapsules over MONO microcapsules. The highest amount of released VEGF was observed for CO microcapsules, with similar values for all conditions.

Extracellular matrix (ECM) deposition evaluation
With increasing culture time, the protrusion of the core contents occurs, thus microaggregates composed by cells and PCL microparticles are released to the external environment. Consequently, such microaggregates from different microcapsules start to merge, creating macroaggregates, as schematically represented in figure 4(A). F-actin staining reveals the elongation of the cells while agglomerating various PCL microparticles. Masson's Trichrome evidences the presence of collagen in the ECM of the macroaggregates. The structural analysis of the collagen-rich ECM was carried by FTIR-ATR through the deconvolution of the amide I band in the 1700-1600 cm −1 range aiming to assess qualitatively and quantitatively the different secondary structure contributions based on the band areas (Table S1, supporting information). Collagen typically presents five types of secondary structures, namely triple helix (∼1635 cm −1 ), unordered structures (∼1645 cm −1 ), α-helix (∼1655 cm −1 ), β-turns (∼1670 cm −1 ), and various absorption bands related to β-sheet conformers [18][19][20][21][22]. Bands around 1625 cm −1 and 1680 cm −1 are attributed to intramolecular β-sheet structures, while intermolecular β-sheets present a main band around 1615 cm −1 with a minor contribution at 1690 cm −1 [22]. At first glance, deconvoluted ATR-FTIR spectra of MONO microcapsules exhibit these typical features in the amide I region, which only seem to vary in intensity, and thus, confirming the presence of collagen. However, MONO microcapsules cultured in BASAL medium and static conditions ( figure 4(B)) present an additional band at lower wavelength (ca. 19%) related to free amino acid residues [23]. This contribution is vanished under dynamic conditions and without any supplemental osteogenic differentiation factors ( figure 4(D)), while αhelix (ca. 21%) and triple helix (ca. 24%) configurations increase compared to its static counterpart (15% and 14% for α-helix and triple helix, respectively). The influence of the dynamic stimulus on the structural organization of the collagen matrix was also clearly noticed in MONO microcapsules cultured in OSTEO medium (figure 4(E)), with a decrease of disordered structure proteins (from 19% to 9%) and the prominence of triple and α-helices (overall 52%), which are typically the major secondary structures of collagen [22].
Interestingly, the amide I band of the different CO microcapsule formulations are significantly shifted to lower wavelengths (figures 4(F)-(I)). This shifting is indicative of stronger intermolecular hydrogen bonds attributable to protein aggregates and it is visualized by the additional contribution (blue shaded curve) that emerges at ∼1600 cm −1 [24,25]. The band intensity increases significantly under osteogenic differentiation factor supplementation (24%-27%) as compared to basal counterparts (6%-9%). Furthermore, these spectral features were confirmed by SEM analysis, showing elongated filaments in the nanometer scale for MONO microcapsules (figures 4(B)-(E)), characteristics

Mineralization evaluation
After analysis of the extracellular matrix of the μPCLcell aggregates, we were looking for the presence of mineralization. However, in all MONO microcapsules no evidence could be found, even in MONO microcapsules cultured in OSTEO medium under dynamic environment (figure S1, supporting information is available online at stacks.iop.org/BF/12/015005/ mmedia) despite its osteogenic differentiation ability and the presence of an organized extracellular matrix ultrastructure. On the other hand, mineralization evidences could be found in CO microcapsules. The chemical characterization of the microtissues formed inside CO microcapsules was analyzed by energy dispersive spectroscopy (EDS) after 21 d of culture. For microtissues developed in BASAL microcapsules, results show that phosphorous (P) and calcium (Ca) contents were higher in dynamic cultures ( figure 5(A)). Likewise, the same results were shown for microtissues developed in OSTEO medium ( figure 5(B)). Interestingly, nodules-like structures were observed by SEM (figures 5(A1) and (B1)), and further analyzed by EDS mapping, which evidenced a P and Ca enriched ECM (figures 5(A2) and (B2)). However, by the analysis of Ca/P ratio differences could be noticed ( figure 5(C)). In BASAL microcapsules (figure 5(C) black line), the average of Ca/P content ratio was 4.92 and 2.65 for static and dynamic conditions, respectively. For CO microcapsules in OSTEO medium (figure 5(C) grey line), the average of Ca/P content ratio decreased to 3.97 (static) and 1.50 (dynamic). These results suggest that CO microcapsules in OSTEO medium and under dynamic environment allowed to develop hydroxyapatite-like minerals (HA), with a Ca/P ratio similar to the native HA (1.67) [27]. Furthermore, the elemental analysis by chemical mapping shows the distribution of P (red) and Ca (green) in microcapsules cultured under mechanical stimulation. The overlap of both elements, mainly observed in OSTEO medium, is represented in yellow (figures 5(A2) and (B2)). Additionally, in CO microcapsules it is possible to visualize HA nodules stained in green ( figure 5(D)). Of note, higher amount of HA nodules could be found in formulations cultured in dynamic environment or in OSTEO medium. In accordance, von Kossa staining shows an enhanced mineralization in microcapsules cultured under dynamic environment by the staining of the phosphate deposits marked in black ( figure 5(E)).

Discussion
We have previously developed a cell encapsulation system under static conditions that solved issues related with oxygen and nutrients diffusion as well as ensured physical support for anchorage-dependent cells [7][8][9][10][11]. Additionally, in vitro and in vivo studies demonstrated the osteogenic potential of the developed capsules co-encapsulating ASCs and endothelial cells, even in the absence of dexamethasone and ascorbic acid, two major osteogenic differentiation factors. However, such capsules presented an average diameter of 1.8 mm and microbiomaterials are currently shown to regulate cell microenvironment in spatial and temporal aspects, crucial for regenerative medicine purposes [28]. So, in order to overcome the requirements concerning the macro size of the liquefied and multilayered capsules, electrohydrodynamic atomization (EHDA) technique was performed. As shown, microcapsules encapsulating ASCs, OBs and μPCL were successfully obtained, presenting a privileged microenvironment for the development of microtissues. In all conditions, cells adhered to the surface of the μPCL, proliferated, and deposited extracellular matrix in such a way that μPCL were assembled in 3D micro-constructs. The multilayered membrane allowed an appropriate diffusion of essential molecules for the long survival of the encapsulated cells. In addition, to take advantage of the liquefied core microenvironment, the proposed microcapsules were tested under dynamic environments using spinner flasks. We hypothesized that the fluid flow could increase the interactions between cells and μPCL while improving mass transfer inside the compartmentalized microcapsules. Usually, the flow produced by spinner flasks remains restricted to the periphery of the 3D scaffolds and the distribution of cells and essential molecules is sparse [29,30]. However, the liquefied core feature of the compartmentalized and multilayered microcapsules allowed the freely dispersion of cells and μPCL.
Therefore, the interactions of the cellular components are enhanced by the hydrodynamic shear provided by the uniform and continuous rotation of the culture medium. As demonstrated in F-actin filaments staining, the dynamic environment led to the development of significantly larger aggregates of cells and μPCL comparing to static conditions. After only 1 d, the dynamic flow allowed the recruitment of almost all cells and μPCL, demonstrating the maximum efficiency of spinner flasks in developing microtissues.
Besides the improved mass transfer, the fluid flow also provided physiologically relevant physical signals to cells. For example, the late osteogenic marker OPN could be visualized in MONO microcapsules cultured in the absence of osteogenic differentiation factors, evidencing that cell differentiation was prompted by the presence of shear stress. Several studies have been reporting that cell behavior is influenced by mechanical stimulation [31,32]. The shear stress trigger mechanotransduction pathways in cells that converts mechanical signals into biochemical signals. In particular, osteoprogenitor cells have been shown to be mechanosensitive. Mechanical stimulation has been shown to enhance the osteogenic differentiation of cells by displaying increased calcium deposition, ALP activity and bone specific genes expression [12,32,33]. Importantly, herein aggregates of cells and microparticles could be generated inside microcapsules and under dynamic stimulation due to the unique liquefied and confined environment of such cell encapsulation system.
The effect of mechanical stimulation added to the microcapsules culture was also observed in the complexity of the newly deposited ECM. Natural bone is a hierarchical organized tissue composed by an organic phase, mainly constituted by collagen fibrils from type-I collagen molecules (∼90%), and an inorganic phase, comprised predominantly by nanocrystals of carbonated hydroxyapatite (HA) that are distributed along the collagen fibrils [4,34]. After SEM image analysis, it was possible to notice the presence of thin and elongated filaments in the ECM of cell-μPCL aggregates that could correspond to collagen fibrils formation in MONO microcapsules (excepting in basal/ static condition ( figure 4(B))). These collagen fibrils, with typical diameters in the nanometer range, are formed from the building-block aggregation of triplehelical tropocollagen molecules [26,35]. From this point of view, ATR-FTIR spectra are corroborated by SEM analysis. The mechanical stimulation in MONO microcapsules prompted the enhancement of triple helix structures content, absent in static conditions, and hence, achieving the primary hierarchical level of collagen, shown by collagen fibrils formation. It is worth noting that the dynamic environment, as well as, the presence of osteogenic differentiation factors from the OSTEO medium, extinguished the presence of free amino acids in MONO microcapsules. Furthermore, for CO microcapsules cultured in OSTEO medium, the content of triple and α-helix contributions decreased (figures 4(G) and (I)). These findings suggest that triple and α-helices assembled into highly ordered structures with stronger intermolecular hydrogen bonds, resulting in collagen aggregates. As showed in figure S3, for PCL microparticles there is no absorption bands in the 1700-1500 cm −1 range (figure S3 square), indicating that the contributions observed in the spectra of the different biosystems, after 21 of culture, are unambiguously related with proteins secreted by the extracellular matrix. Additionally, SEM images analysis shows that CO microcapsules cultured in OSTEO medium (figure S2(a), supporting information) are covered by a thicker extracellular matrix with randomly distributed fibers in a micrometer scale (protein aggregates), while MONO microcapsules exhibit a homogenous and smooth ECM morphology. It is also notable that cross-sections of CO microcapsules ( figure S2(b), supporting information) showed the abundance of mineral crystals embedded in the collagen matrix. These results are more evidenced in CO microcapsules under mechanical stimulation, as shown by the presence of fluorescently marked HA crystals (figure 5(D)) and von Kossa staining ( figure 5(E)). Likewise, EDS analysis suggests that the dynamic environment allowed to develop apatite-like minerals, with a Ca/P ratio similar to the native HA of the bone matrix [27]. In fact, bone tissue possesses a precise collagen-mineral balance content. In the early stages of mineralization, the nucleation of HA crystals occurs in the gap zones of collagen molecules, while in the latter stages, the crystal deposition cover the overlap zones [35,36]. Besides the mechanical stimuli added to the culture of microcapsules, we hypothesized that by increasing the cellular complexity through the co-encapsulating of bone-forming OBs with ASCs, a well-orchestrated cell-to-cell interaction would occur. Consequently, inside the liquefied and multilayered microcapsules, microtissues could be formed in a more realistic fashion of the native bone regenerative process. Several studies have demonstrated that osteoblasts have a great ability to modulate cell differentiation without requiring any osteogenic supplementation [37][38][39]. Here, we demonstrate that within the hierarchical and compartmentalized environment of microcapsules, co-encapsulated OBs and ASCs can interact in such a way that leads to the development of microtissues presenting a ECM enriched with collagen fibrils and HA crystals. CO microcapsules also release the highest amount of VEGF, compared to MONO microcapsules, since both ASCs [40][41][42] and OBs [43][44][45] are known to be involved in blood vessels formation by secreting pro-angiogenic factors, such as VEGF. This growth factor is not only involved in cell differentiation towards the endothelial lineage, but it also seems to play an important role in the osteogenic healing capacity of ASCs [17,46]. Of note, the VEGF measured was detected in the culture medium, and thus it indicates that the VEGF released by the encapsulated cells was able to cross the multilayered membrane of microcapsules. Therefore, we believe that such microcapsules can also act as angiogenic inductors, which upon implantation may stimulate the recruitment of the host vessels, and ultimately contribute to the vascularization and integration of the formed microtissues.
Altogether, the present study showed the potential of the proposed microcapsules for bone regeneration. Here, we demonstrated that the mechanical stimulation added to the microcapsules suppressed the need of osteogenic supplementation factors for osteoblastic differentiation occur. In fact, the liquefied environment allowed to the encapsulated cells to freely move inside of the microcapsules, to proliferate, to self-organize in a 3D structure and to develop a bone-like ultrastructure. Considering the reported findings, we believe that microcapsules can be promptly implanted into a bone injury, after 24 h of in vitro mechanical stimulation. Additionally, we intend to use the proposed microcapsules as hybrid devices implantable by minimally invasive procedures due to their injectability provided by the liquefied core. These compartmentalized units might facilitate the implantation of cells, avoiding their dispersion to other regions of the body, while encapsulated μPCL act as cells adhesion sites, allowing cells to adhere, proliferate and create bonelike tissues.

Conclusion
Inspired by the multiphenotypic cellular environment of the native bone, we developed a bioencapsulation system that proved to be an effective strategy for the in vitro creation of microtissues expressing different bone biomarkers. Overall, the co-encapsulation of ASCs, OBs and μPCL led to development of differentiated bone-like microtissues, even in the absence of dexamethasone, ascorbic acid and β-glycerophosphate, the three classical supplements of osteogenic differentiation medium. Additionally, the free dispersion of co-encapsulated cells and microparticles prompted by the mechanical stimulation, recreates the dynamic environment of the native bone tissue. Here, we demonstrated the possibility of an engineered selfregulated osteogenic device that can be further envisaged as an injectable procedure for bone repair.