*2.2.* μ*Ps as Building Blocks for In Vitro and In Vivo Modular Tissue Engineering (TE) Sca*ff*olds*

The use of μPs for engineering biological tissues may follow two main approaches. In the first approach, named cell-free, μPs are used as building blocks and assembled together to form a sintered porous scaffold. Therefore, the scaffold can be used for in vitro cell culture studies before in vivo implantation. Alternatively, the scaffold is directly implanted in vivo to deliver bioactive molecules and to promptly restore tissue anatomy and functions. In the second approach, named cell-laden, μPs are used as μ-scaffolds for in vitro cell expansion and proliferation. The as-obtained cell-laden μ-scaffolds are subsequently assembled in vitro inside bioreactors to stimulate cell biosynthesis and material degradation, finally leaving a biological tissue replicating native tissues' composition and structure. Both approaches require building blocks assembly into 3D large (centimeter scale) structures by two main ways: random and ordered assembly [61]. The following sections will describe techniques of μPs' assembly for cell-free and cell-laden TE strategies, bringing to light some of the most relevant results achieved to date.

## 2.2.1. Porous Scaffolds Prepared by the Random/Ordered Assembly of μPs

The literature review has evidenced a plethora of works reporting the design and fabrication of scaffolds by using biodegradable and biocompatible μPs, demonstrating the possibility to achieve tailored porous structure, full interconnected porosity, high mechanical stiffness and, ultimately, drug loading and controlled release features. In a typical process, researchers prepared bioactive and biodegradable μPs using traditional or advanced methods, such as those described in the previous section. The μPs were then poured into appropriate molds and sintered together to form a continuous matrix. As shown in Figure 2a, the resultant scaffolds have a particles-aggregated structure while their size and shape replicated the mold (cylinder) geometry [4].

A scaffold's morphology as well as its pore structures were correlated to the size and shape of the μPs and the sintering process. Sintering depended on the motion of polymeric chains from the μPs surface to contact points that leads to chain inter-diffusion and the subsequent formation of connecting necks between μPs. This mechanism depends on polymer plasticization and can be promoted by heat, organic solvents, or high-pressure fluids [62–67]. For instance, PCL scaffolds were fabricated using thermal sintering of spherical μPs with two different size ranges, smaller (300–500 μm) and larger (500–630 μm) at 60 ◦C for 1 h. A double emulsion process was also implemented for bovine serum albumin (BSA) encapsulation inside the depots of smaller (50–180 μm) PCL particles for drug delivery purposes. The authors reported the decrease of scaffolds' porosity and pore size as well as the increase of compression moduli with the decrease of μPs' size. This effect is ascribable to an enhanced μPs' compaction and a concomitant higher number of fusion points between smaller μPs [4]. However, low porosity and pore size may result in decreased cell adhesion and colonization. The use of porous μPs enable to overcome this limitation and achieve higher scaffolds' porosity. This aspect was studied by

Qutachi and co-workers, who fabricated highly porous PLGA μPs by the double emulsion technique, where phosphate buffered saline (PBS) was used as the internal aqueous phase. Hydrolysis treatment on μPs using 30% 0.25 M NaOH:70% absolute ethanol enabled the formation of a double-scale sintered matrix at body temperature that can therefore be used as a minimally invasive injectable scaffold (Figure 2b) [62].

The optimization of the sintering step is a critical aspect for scaffolds prepared by μPs' assembly. Indeed, sintering not only affects the integrity of the scaffold structure, but also influences some key properties, such as porosity and mechanical stiffness. Borden et al. addressed this aspect for melt-sintered scaffolds [68], while Brown et al. [69] and Hyeong Jeon et al. [63] addressed it for solvent-sintered scaffolds and for a high-pressure CO2 sintering, respectively. As shown in Figure 2c, the mechanical properties of PLGA scaffolds increased with the increase of fusion time from 2 to 4 h, while higher treatment times produced the complete collapse of the pore structure due to the extensive polymer melting. Overall, these scaffolds, with a range in modulus from 137.44 to 296.87 MPa, appeared to be capable of sustaining loads in the mid-range of cancellous bone [68]. The solvent/non-solvent chemical sintering, is an alternative strategy for sintering a wide range of polymeric μPs at a low temperature for developing TE scaffolds and drug delivery vehicles [69]. Polymers such as polyphosphazenes, exhibiting glass transition temperatures from −8 to 41 ◦C, and PLGA were tested to optimize solvent/non-solvent mixtures and the treatment time based on the affinity between polymer and solvent mixtures. The authors reported that the solvent/non-solvent sintering technique produced scaffolds with median pore size and porosity similar to the heat-sintered microspheres [69]. Nevertheless, the use of potentially toxic organic solvents is a critical issue for this approach. A low-temperature organic solvent-free approach was proposed for μPs' sintered scaffolds fabrication [62,63]. This approach used high-pressurized CO2 to produce scaffolds from a large variety of polymeric materials, such as PCL, PLGA, and PLA [63]. For instance, it was reported that the optimal CO2 pressure for PLGA scaffolds was in the 15–25 MPa range, and that sintering increased with pressure due to the enhanced polymer plasticization, representing a useful way to tune scaffolds' porosity and mechanical properties [63].

As pointed out in the Introduction section, engineering tissues and organs requires combinations of biomaterials, cells, and bioactive signaling cues. The design of bioactive molecules releasing scaffolds has to consider that the spatial patterning of bioactive signals is vital to some of the most fundamental aspects of life, from embryogenesis to wound healing, all involving concentration gradients of signaling molecules that have to be replicated by scaffolds. μPs have been long studied as drug delivery systems for a variety of molecules as they enable an easy control of the release kinetics of loaded therapeutics. Alendronate (AL)- and dexamethasone (Dex)-loaded PLGA-based scaffolds were proposed by Shi and co-workers for bone regeneration [70]. These molecules were chosen as AL is a bisphosphonate able to promote the activity and maturation of osteoblasts and mesenchymal stem cells (MSCs) differentiation, while Dex is a glucocorticoid with osteogenic properties. Scaffolds' capability to release AL and Dex up to two months in a sustained fashion resulted in a marked osteogenic differentiation of MSCs in vitro and in vivo, as evidenced by significantly higher expression of bone-related proteins and genes, such as alkaline phosphatase activity (ALP), type-I collagen, osteocalcin, and bone morphogenic protein (BMP)-2, if compared to unloaded scaffold. Additionally, drug-loaded scaffolds showed significantly higher new bone formation at eight weeks implantation into rabbit femurs bone defects [70]. Jaklenec et al. used PLGA μPs loaded with dyes to demonstrate the feasibility of creating spatially controlled particles' distribution inside porous scaffolds (Figure 2d) [71]. In another study, Singh and co-workers developed a two-syringe pumping device for the controlled deposition of functional microspheres to create gradients of releasing molecules for interfacial tissues' regeneration [72]. By controlling suspensions composition and flow rates during pumping, it was possible to engineer multiple gradient configurations, such as the bi-layered and multi-layered concentration profiles. The authors further used this technique to prepare a PLGA microspheres scaffold containing opposing gradients of BMP-2 and transforming growth factor (TGFb1) for osteochondral interface TE (Figure 2e) [73]. After six weeks

of in vitro culture, MSCs-seeded scaffolds evidence regionalized gene expression of major osteogenic and chondrogenic markers.

Overall, scaffolds based on the assembly of μPs are versatile for a wide range of TE applications, from soft to hard tissues. For instance, the use of synthetic polymers resulted in high mechanical stiffness and a slow degradation rate for in vivo load-bearing implantation, such as bone [74] and osteochondral tissue [75]. Conversely, soft biopolymeric chitosan μPs scaffolds were proposed as a 3D, functional neuronal networks' regeneration platform [76]. Even if all of these studies clearly evidenced the potential of μPs-based scaffolds in TE applications, some key issues are still to be addressed for their successful clinical translation. As previously discussed, biological tissues are characterized by hierarchical-ordered architectures at both nano- and micro-metric size scales, that can be replicated only in part by μPs' random assembly. Furthermore, μPs-based scaffolds require multiple steps of fabrication, from μPs' fabrication up to assembly and sintering. Therefore, the possibility to reduce scaffolds' fabrication time by automated processes will be a great step towards clinical implementation.

One of the most investigated methods to obtain ordered scaffolds from sintered μPs is selective laser sintering (SLS). As shown in Figure 2f–h, this powder-based AM technique enabled patient-specific implantable scaffolds with interconnected multi-scaled porosity [77]. SLS employs a CO2 laser beam to selectively sinter a powder bead, based on a computer-aided design (CAD) scaffold model. Du and co-workers fabricated PCL and PCL-hydroxyapatite scaffolds for bone TE [78]. Both in vitro and in vivo evaluations demonstrated that these scaffolds not only promoted cell adhesion, supported cell proliferation, and induced cell differentiation in vitro, but also evidenced in vivo bone formation and vascularization. This effect was higher for composite scaffold as the hydroxyapatite increased surface roughness and positively charged the PCL surface. The same authors also explored the fabrication of bioinspired multilayer scaffolds mimicking the complex hierarchical architecture of the osteochondral tissue that was used to repair osteochondral defects of a rabbit model [78]. It is, however, important to point out that SLS techniques create ordered structures (Figure 2g) inside randomly assembled μPs (Figure 2h), while they cannot manipulate μPs and allow their precise positioning inside the scaffold structure. One of the first attempts to solve this aspect and to fabricate porous μPs'-sintered scaffolds with highly ordered pore structure at the μPs-scale was recently proposed by Rossi and co-workers [79]. The authors prepared PCL μPs with size in the 425–500 μm range and used alignment PDMS molds for precise particle positioning and sintering. Final scaffolds were achieved by the stacking of three μPs layers followed by a solvent bonding step (Figure 2i). If compared to randomly assembled scaffolds, the ordered scaffolds showed a better vascularization in the inner core, as evidenced by the deeper blood vessel penetration and the larger diameter of the infiltrating vessels [79]. This approach was tested with large μPs (500 μm), as smaller μPs require the implementation of new advanced automated manufacturing. Nevertheless, these recent results pave the way on the importance on μPs' scaffold design features and provide the basis for the future development in this extremely promising scaffold design research field.

#### 2.2.2. Porous μPs as μ-Scaffolds for In Vitro Tissue Building

In the past decade, researchers in TE have focused the attention on the possibility of recreating large implantable living and functional tissues in vitro by assembling cell-laden μ-scaffolds. The advantage of this approach relies on the fact that porous micro-sized scaffolds can be designed and modularly assembled to guide the correct spatial composition and organization of the de-novo synthesized cell/ECM construct. Furthermore, by this approach, it is possible to overcome limitations related to cells culturing in 3D thick scaffolds, such as cells' seeding efficiency and oxygen and nutrients' transport inside the scaffold core. The capability of recreating in vitro fully biologic centimeter-sized tissues was validated for a large variety of applications. For instance, Urciuolo and co-workers [80] have studied the fabrication of a dermis-equivalent tissue by culturing human dermal fibroblasts (HDFs) onto gelatin μ-scaffolds. The developed process involved two main steps: (Step 1) dynamic cell-seeding of fibroblasts on porous μ-scaffolds using a spinner flask bioreactor for up to nine days

to obtain micro-tissue precursors (μTPs). (Step 2) assembly of μTPs and maturation in a specifically designed chamber for up to 28 days [80]. Following this strategy, a 3D functional dermal tissue has been created and used as a base platform to study natural and pathologic tissue morphogenesis mechanisms, such as follicle-like structure formation [81] and tissue vascularization [82], as well as to study dermis remodeling and epidermis senescence after UV radiation exposure [83]. The feasibility of using cell-laden μ-scaffolds to fabricate highly complex biomimetic tissues was also explored in the case of bone [84], cardiac tissue [85], and liver tissues [86]. For example, Chen et al. cultured human amniotic MSCs onto gelatin μ-scaffolds for up to eight days, after which, cells were induced to undergo osteogenic differentiation in the same culture flask and cultured for up to 28 days. These bone-like μ-tissues were finally used as building blocks to fabricate a macroscopic cylindrical bone construct (2 cm in diameter, 1 cm height) evidencing good cell viability and homogenous distribution of cellular content (Figure 2j) [87]. The modularity of this approach was explored by Scott et al., who combined human liver cancer cell line, HepG2, and different types of PEG μPs to study the effect of porosity and drug delivery on cells' behavior (Figure 2k) [88]. In particular, the authors considered three types of PEG microspheres: the first type provided μ-scaffolds mechanical support, the second type provided controlled delivery of the sphingosine 1-phosphate (S1P), an angiogenesis-promoting molecule, and the third type served as a slowly dissolving non-cytotoxic porogen. After components' centrifugation into a mold and incubation at 37 ◦C overnight, μPs fused together and, within two days of culture, macropores formed thanks to the dissolution of the porogenic particles. The S1P delivery combined with the structural properties allowed HepG2 cells' migration through the scaffolds' macropores (Figure 2j).

AM processes are successfully used to obtain cell-laden μ-scaffolds with ordered structures for biomedical applications. Among these techniques, bioprinting is the most popular as it allows the fabrication of living constructs with custom-made architectures by the controlled deposition of cell-laden μ-scaffolds bioinks [40,89]. In a recent study, Levato and co-workers seeded MSCs onto PLA μ-scaffolds via static culture or spinner flask expansion and loaded these samples in gelatin methacrylamide-gellan gum bioinks [40]. The optimization of the composite material formulation and printing conditions enabled the fabrication of highly ordered constructs with enhanced mechanical properties and high cell-seeded densities (Figure 2l). Process flexibility was also validated by designing and fabricating bi-layered osteochondral scaffolds (Figure 2l). Tan et al. presented a similar approach for the recreation of vascular tubular tissues, based on the micropipette extrusion bioprinting method (Figure 2m) [89]. The selected bioink was made of cell-laden PLGA porous microspheres encapsulated within agarose-collagen hydrogels. Furthermore, the authors demonstrated the possibility to use concomitantly C2C12 and Rat2 cell-laden μ-scaffolds.

Manipulation of cell-laden μ-scaffolds at the micro-scale was also investigated to obtain precisely designed 3D structures for TE. The μ-scaffolds were soaked in an inert medium (mineral oil) while their assembly was obtained by geometrical constraints, specifically by the use of guiding structures or by more complex mechanisms, such as magnetic actuation. The picture in Figure 2n highlights 3D structures obtained by assembling cell-laden μ-scaffolds fabricated by soft-lithography (Figure 1) and starting from a UV-photo-cross-linkable metacrylated gelatin solution [90]. The assembly process was controlled by geometrical constraint or by using a syringe needle swiped uniaxially against the linear array of ring-shaped μ-scaffolds [90]. Liu and co-workers combined μ-scaffolds shape and magnetic field for the construction of artificial bioarchitectures [91]. Magnetite-alginate-chitosan composite microcapsule robots characterized by magnetization along the central axis were magnetically actuated to grab the building components during the transportation and assembly processes. Position and orientation remote control of the cell-laden μ-scaffolds offered a non-invasive and dynamical manipulation system for the creation of complex 3D structures for TE.

**Figure 2.** Overview of μPs applications in tissue engineering (TE) scaffold-based strategies classified by random (left column) and ordered (right column) assemblies, cell-free (first row) and cell-laden (second row) approaches. (**a**) morphology of μPs' sintered polycaprolactone (PCL) scaffold obtained by thermal sintering. (**b**) Morphology of porous μPs' sintered PLGA scaffold obtained by chemical sintering. (**c**) Effect of μPs' diameter and thermal sintering time on mean pore size and compressive modulus of PLGA-sintered scaffolds. (**d**) Optical images of sintered scaffolds with homogeneous and heterogeneous spatial distribution of loaded μPs. (**e**) Release profiles of bone morphogenic protein (BMP)-2 and transforming growth factor (TGF) b1 from μPs'-sintered scaffolds for osteochondral interface TE. (**f**) Optical image of ordered scaffold obtained by selective laser sintering (SLS) and made of PCL μPs. (**g**,**h**) morphology of SLS scaffold evidencing the order and random structures, respectively. (**i**) Comparison of random and ordered PCL scaffolds on degree of vascularization in vivo. Results proved that the internal vascularization of the ordered scaffolds has significantly better vascularization in the inner core if compared to the random scaffold. (**j**) Culture device used to generate three-dimensional (3D) bone in vitro by cell-laden μPs' assembly and morphological and optical visualization of corresponding tissue. (**k**) Assembly of cells and multifunctional poly-ethylene glycol (PEG) μPs to study cells migration in vitro as a function of scaffolds porosity and sphingosine 1-phosphate (S1P) release. (**l**,**m**) Porous scaffolds obtained by μPs' printing for osteochondral and vascular tissues repair, respectively. (**n**) Schematic of assembly processes of cell-laden μ-scaffolds obtained by soft-lithography process and resulting cell-laden constructs. **a** Reproduced with permission from Reference [4] (Luciani, Biomaterials; published by Elsevier Ltd., 2008); **b** Reproduced with permission from Reference [62] (Qutachi, Acta

Biomaterialia; published by Elsevier Ltd., 2014); **c** Reproduced with permission from Reference [68] (Borden, Biomaterials; published by Elsevier Science Ltd., 2002); **d** Reproduced with permission from Reference [71] (Jaklenec, Biomaterials; published by Elsevier Ltd., 2008); **e** Reproduced with permission from Reference [73] (Dormer, Annals of Biomedical Engineering; published by Springer Nature, 2010); **f**–**h** Reproduced with permission from Reference [77] (Du, Colloids and Surfaces B: Biointerfaces; published by Elsevier B.V, 2015); **i** Reproduced with permission from Reference [79] (Rossi, Journal of Materials Science Materials in Medicine; published by Springer Nature, 2016); **j** Reproduced with permission from Reference [87] (Chen, Biomaterials; published by Elsevier Ltd., 2011); **k** Reproduced with permission from Reference [88] (Scott, Acta Biomaterialia; published by Elsevier Ltd., 2009); **l** Reproduced with permission from Reference [40] (Levato, Biofabrication; published by Institute of Physics Publishing, 2014); **m** Reproduced with permission from Reference [89] (Tan, Scientific Reports; published by Springer Nature, 2016); **n** Reproduced with permission from Reference [90] (Xiao, Materials Letters; published by Elsevier Ltd., 2018).
