Preconditioning Methods to Improve Mesenchymal Stromal Cell-Derived Extracellular Vesicles in Bone Regeneration—A Systematic Review

Simple Summary The evidence of the therapeutic effects of mesenchymal stromal cells (MSCs), so-called stem cells, in several diseases relies mostly on the substances they secrete, including their extracellular vesicles (EVs). EVs are an important component of cell communication and they carry a cargo that is similar to their parent cell. Cells respond differently based on their microenvironment, and so it is expected that the therapeutic potential of these vesicles can be modulated by the enrichment of their parent cell microenvironment. With this in mind, we conducted a systematic search for papers that preconditioned MSCs and collected their EVs to assess their potential to favor bone formation. The results showed different methods for MSC preconditioning, including chemical induction, culture conditions, and genetic modifications. All methods were able to improve the therapeutic effects of the derived EVs for bone formation. However, the heterogeneity among studies—regarding the type of cell, EV concentration, and scaffolds—made it difficult to compare fairly the types of preconditioning methods. In summary, the microenvironment greatly influences MSCs, and using preconditioning methods can potentially improve the therapeutic effects of their derived EVs in bone regeneration and other bone diseases. Abstract Mesenchymal stromal cells (MSCs) have long been used in research for bone regeneration, with evidence of their beneficial properties. In the segmental area of MSC-based therapies, MSC-derived extracellular vesicles (EVs) have also shown great therapeutic effects in several diseases, including bone healing. This study aimed to assess whether the conditioning of MSCs improves the therapeutic effects of their derived extracellular vesicles for bone regeneration. Electronic research was performed until February 2021 to recover the studies in the following databases: PubMed, Scopus, and Web of Science. The studies were screened based on the inclusion criteria. Relevant information was extracted, including in vitro and in vivo experiments, and the animal studies were evaluated for risk of bias by the SYRCLE tool. A total of 463 studies were retrieved, and 18 studies met the inclusion criteria (10 studies for their in vitro analysis, and 8 studies for their in vitro and in vivo analysis). The conditioning methods reported included: osteogenic medium; dimethyloxalylglycine; dexamethasone; strontium-substituted calcium silicate; hypoxia; 3D mechanical microenvironment; and the overexpression of miR-375, bone morphogenetic protein-2, and mutant hypoxia-inducible factor-1α. The conditioning methods of MSCs in the reported studies generate exosomes able to significantly promote bone regeneration. However, heterogeneity regarding cell source, conditioning method, EV isolation and concentration, and defect model was observed among the studies. The different conditioning methods reported in this review do improve the therapeutic effects of MSC-derived EVs for bone regeneration, but they still need to be addressed in larger animal models for further clinical application.


Search Results
A total of 463 studies were retrieved in the initial database search. After duplicate removal, 237 studies were screened by title and abstract, resulting in the removal of 143 studies. After full-text examination, a total of 18 studies met the inclusion criteria and were included in this systematic review ( Figure 1). Among these studies, three articles that reported in vitro and in vivo experiments did not meet the inclusion criteria for the in vivo experiments [28][29][30]; however, the in vitro analysis was included in this review. For this reason, 10 studies were extracted for in vitro analysis [28][29][30][31][32][33][34][35][36][37], and 8 studies were extracted for in vitro and in vivo analysis [17,18,20,21,[38][39][40][41].

Study Characteristics
The studies included were published from 2016 to 2021; the year 2020 was the year with the most publications, indicating that conditioning MSCs before EV collection is becoming a more common approach. Several journals published these studies, from general science to specific regenerative medicine journals, and China was the country with the

Study Characteristics
The studies included were published from 2016 to 2021; the year 2020 was the year with the most publications, indicating that conditioning MSCs before EV collection is becoming a more common approach. Several journals published these studies, from general science to specific regenerative medicine journals, and China was the country with the highest number of published studies. Detailed information can be found in the supplementary material (Table S1).

EV Isolation and Characterization
A summary of the EV isolation and characterization methods reported by the included studies can be found in Table 2.

In Vitro Studies
All studies included in this review reported some experiments in vitro ( Table 3). The most frequent cell type used as the subject of treatment were human BMSCs [21,29,32,35,[37][38][39], followed by HUVECs [17,20,28,41], hMSCs [36,40], rat BMSCs [18,33], rabbit BMSCs [28], human ASCs [34], DPSCs [31], MC3T3 [30], hFOB 1.19 [17], and RAW 264.7 [35]. The type of cell was chosen according to the focus of the conditioning method, e.g., hypoxic MSCderived exosomes were used to treat HUVECs and analyze the potential for angiogenesis, and osteogenically conditioned exosomes were used in the osteogenic differentiation of BMSCs, ASCs, or PDSCs.     The presence of a group with unconditioned EVs (unconditioned compared to the principal method in the study) was mandatory in the studies included in this review, so that a fair comparative analysis would be possible. Generally, the control EV groups in the studies were isolated from untreated cells [17,20,29,30,33,36,[38][39][40][41]. However, in some studies, when serum deprivation was applied, the control EV group was isolated from cells with serum deprivation, varying from 12 to 72 h of additional culturing [21,28,32,34,35,37]. Only one study reported the conditioning of MSCs with osteogenic medium without serum deprivation, but the EV control was collected from the serum-free medium [31], and one study did not report the culture method of the control EVs [18].
In the majority of the studies, osteogenic differentiation was detected by alizarin red staining [18,28,[30][31][32][33][34]36,[38][39][40]. Another common analysis was ALP activity and osteogenic marker expression. To evaluate angiogenesis, cell proliferation, migration, and tube formation assays were performed in all studies that used HUVECs [17,20,28,41]. Only one study evaluated the immunoregulatory role of exosomes and did so by analyzing the expression of cytokines in RAW 264.7 cells [35]. It is important to note that some studies focused on the evaluation of angiogenesis in vitro, not in osteogenesis (Liang et  To analyze osteogenic differentiation, cells were cultured in proliferation medium (PM) or osteogenic medium (OM) and treated with EVs. Only three studies analyzed the osteogenic induction of EVs both in OM and PM [29,33,38]; five studies performed the osteogenic induction of EVs only in PM [21,30,31,36,40]; and five studies analyzed the effects of EVs in OM only [18,32,34,35,39]. One study did not report the type of medium in which the cells were cultured during the treatment [37].
Regarding the treatments in the in vitro analysis, the concentration of EVs ranged from 10 to 200 µg/mL. The duration of the treatments varied between studies; however, ALP activity was analyzed from 7 to 14 days of culture; ARS up to 21 days of culture; and osteogenic markers generally from 3 to 21 days of culture.

In Vivo Studies
A total of eight studies with in vivo experiments met the inclusion criteria for this review ( Table 4). Seven of them reported the use of rats as animal models, of which six studies used Sprague-Dawley rats [18,20,[38][39][40][41] and one study did not report the strain [21]. Only one study used mice as the animal model, though it also did not report the strain [17]. Regarding the defect model, four studies performed bilateral calvaria defects [18,20,21,38]; one study performed a femoral fracture model in mice [17]; one study performed an alveolar bone defect [39]; one study performed a segmental radius defect [40]; and one study performed a distal femur defect [41]. Notably, the animal models did not vary in size, and the defect models among the studies did not explore critical-sized defects in long bones or larger animals.
The treatment groups were homogeneous among studies regarding the presence of a conditioned exosome-treated group, an unconditioned exosome-treated group, and a control group (generally with the scaffold alone, except for [17], which used PBS as a carrier). One study reported a blank defect in all treated animals [38], one study reported a positive control group treated with recombinant human BMP2, and one study reported a group treated with hMSCs seeded in a titanium scaffold [40].

In Vitro and In Vivo Outcomes
The outcomes for osteogenic differentiation, angiogenesis, bone regeneration, and vascularization are summarized in Tables 3 and 4.

Osteogenic Differentiation
Osteogenic differentiation was performed in several studies by analyzing ALP activity as an early marker, ARS for mineral deposition, and the expression of osteogenesis-related genes. ALP activity [29,[31][32][33][34]36,40] and mineral deposition [32][33][34]36,40] in cells treated with osteogenically induced exosomes were higher in most of the studies compared to non-osteogenic exosomes. The exception was reported by Wei et al. [35]; non-osteogenic exosomes were able to improve ALP activity and osteogenic marker expression (OPN, ALP, RUNX2, BMP2, and BMP7) in BMSCs cultured in OM compared to exosomes derived from 3, 7, and 14 days of osteogenic induction. Narayanan et al. [37] reported the expression of several osteogenesis-related genes and showed that 2-and 4-week osteogenically induced exosomes were able to upregulate when compared to untreated cells. However, when cultured in a 3D environment, cells treated with 4-week and non-osteogenic exosomes had similar results. Unfortunately, this study did not provide a comparison between exosome treatments.
The best day for collecting exosomes under osteogenic induction differed among studies. By analyzing the optimal exosome induction time, Li et al. [29] showed that exosomes derived from non-osteogenically induced ASCs could not enhance ALP activity either in OM or PM. However, exosomes derived from osteogenically induced ASCs significantly enhanced the ALP activity of BMSC culture in OM irrespective of the time span (2, 4, 7, and 14 days), but not in PM. Further analysis showed that ALP activity; mineral deposition; and RUNX2, ALP, and COL1A1 expression were higher in BMSCs with two-day induced exosomes cultured in OM compared to standard osteogenic medium alone. On the other hand, Wang et al. [36] showed that 3, 6, 9, and 12 days of osteogenic induction of hMSCs did not provide exosomes with the capacity to significantly induce the osteogenic differentiation of hMSCs when cultured in PM. However, 15-, 18-, and 21-day derived exosomes showed a higher mineral quantification of hMSCs. Yang et al. [34] only reported 14 days of the osteogenic induction of ASC-derived exosomes. The results showed that treatment improved ALP activity, gene and protein expression (RUNX and ALP), and mineral deposition of ASCs (cultured in OM) in comparison to non-osteogenic exosomes. Liu et al. [33] showed that exosomes from 3, and 14 days of osteogenic induction were able to enhance ALP activity in PM; when cultured in OM, 3-, 7-, and 14-day exosomes were able to enhance ALP activity. However, mineral deposition was significantly enhanced only by 14-day exosomes in PM, while in OM, 3-, 7-, and 14-day exosomes were capable of a significant improvement. Regarding gene expression, 14-day exosomes in PM and OM culture upregulated RUNX2, ALP, and osterix compared to non-osteogenic exosomes. Pishavar et al. [32] compared exosomes from osteogenically induced PSCs and BMSCs in the treatment of BMSCs cultured in OM. All EV-treated BMSCs showed higher levels of ALP activity at 14 days of culture in comparison to OM alone, except for non-osteogenic BMSC-derived exosomes. Calcium deposition was significantly higher after 21 day of induction by all EVs in comparison to OM alone, except for non-osteogenic PSC-derived exosomes. Xie et al. [31] reported only 7 days of osteogenically induced exosomes and showed qualitatively that they were able to enhance ALP activity and mineral deposition by DPSCs. The quantification of RUNX2, COL-1, and OCN showed that exosomes derived from 7-day induced DPSCs were able to upregulate their expression compared to non-osteogenic exosomes. Zhai et al. [40] evaluated exosomes from 4-, 10-, 15-, and 20-day osteogenically induced hMSCs in PM-cultured cells. Results showed that COL-1 expression was higher in 10-day-exosome-treated cells and OPN expression, ALP activity, and mineral deposition were higher in 10-and 15-day-exosome-treated cells-all compared to non-osteogenic exosomes. In summary, exosomes derived from late osteogenically induced MSCs can trigger osteogenic differentiation in cells treated in basal/growth media compared to early induced exosomes. However, when cells are cultured in osteogenic media treated with exosomes derived from osteogenically induced MSCs, a great improvement in ALP activity, osteogenic marker expression, and mineral deposition is seen.
Other conditioning methods of MSCs showed great results. Dexamethasone-activated BMSCs provided exosomes with the ability to improve the migration, proliferation, mineral deposition, and osteogenic marker expression (RUNX2, ALP, and OPN) of MC3T3-treated cells in comparison to non-conditioned exosomes [30]. Similarly, Ying et al. [18] reported that mutant BMSC-HIF1α-derived exosomes generated higher levels of BMSC proliferation, ALP activity, mineral deposition, and osteogenic marker expression (ALP, RUNX2, and COL1A1) compared to exosomes derived from non-mutant BMSCs. Li et al. [28] also used mutant BMSC-HIF1α-derived exosomes that resulted in the upregulation of OCN and ALP expression and enhanced mineral deposition by treated BMSCs. Exosomes derived from miR-375-overexpressing ASCs showed improvement in the ALP activity, mineral deposition, and osteogenic marker expression (RUNX2 and OCN) of BMSCS cultured in PM and OM compared to non-modified exosomes; ALP and COL1A1 were upregulated only when BMSCs were cultured in OM [38]. Huang et al. [21] performed genetic modification as well, by overexpressing BMP2 in BMSCs. The derived exosomes were able to upregulate the expression of BMP2, RUNX2, osterix, and BMP9 in BMSCs cultured in collagen sponge, but only when compared to untreated cells. A three-dimensional mechanical microenvironment was reported by Yu et al. [39]. The culture of PDLSCs in these conditions resulted in exosomes that were able to increase the mineral deposition and upregulate the expression of ALP, RUNX-2, OCN, and COL-1 in BMSCs when compared to exosomes derived from the non-mechanical 3D microenvironment.

Angiogenesis
Four studies analyzed the effects of exosomes on HUVEC proliferation, migration, and tube formation [17,20,28,41]. DMOG-stimulated BMSC-derived exosomes were able to improve cell migration and tube formation but did not significantly improve cell proliferation compared to non-stimulated exosomes [20]. All exosome treatments evaluated by Liu et al. [17] and Liu et al. [41] were able to improve cell proliferation, migration, and tube formation; however, exosomes derived from hypoxic [17] and strontium-substitutedcalcium-silicate-conditioned [41] cells showed significantly better results. In addition, the expression of VEGF was higher in HUVECs treated with hypoxic and strontium-substitutedcalcium-silicate-conditioned exosomes. Exosomes from mutant-HIF-1α BMSCs were also able to promote angiogenesis when compared to exosomes derived from wild-type BM-SCs [28]. In summary, all the above-cited conditioning methods were able to promote angiogenesis by increasing HUVEC proliferation, migration, and tube formation and were better than unconditioned exosomes.

Bone Regeneration
Micro-CT and histology were the principal methods reported to describe bone regeneration in the studies. In all in vivo studies, treatment with exosomes derived from preconditioned MSCs led to a more advanced bone formation in comparison with unconditioned derived exosomes.
Among the in vivo studies, only one reported treatment with exosome derived from osteogenically induced MSCs [40], which contrasted with the in vitro studies, wherein standard osteogenic medium was the prevalent method of preconditioning. In this study, a more advanced bone formation was observed when the bone defects were treated with exosomes derived from 10-and 15-day osteogenically induced MSCs. The results described in these two groups included the presence of more Haversian canal-like structures, more bone cells, and more blood vessels. Additionally, more collagen formation and osteoblasts were observed, which were comparable to the MSC-seeded group. However, only descriptive analysis was presented in this study.
Other chemical additions to the MSC culture included DMOG [20] and strontiumsubstituted calcium silicate [41]. Both proved to be a good conditioning method for MSCs, generating exosomes with the ability to promote greater bone formation in a calvaria defect model [20] and a distal femur defect model [41] when compared to other exosome treatments.
In the category of genetic modifications, three studies reported exosome treatment in the rat calvaria defect model. Mesenchymal stem cells were modified to overexpress miR-375 [38] and BMP2 [21] and to express mutant HIF-1α [18]. Chen et al. [38] reported significantly greater bone formation in the defects treated with exosomes derived from miR-375-overexpressing MSCs compared to negative control exosomes. Compared to the blank group, both exosome types enhanced bone formation. The expression of OCN and BMP2 was analyzed by IHC and the range and intensity were higher in the miR-375-Exo group. Huang et al. [21] also observed a better performance of EVs derived from BMP2-overexpressing MSCs in bone formation compared to control EVs. The results of BMP2-EV-treated defects were comparable to recombinant human BMP2; however, the presence of a fatty marrow was observed in the latter but not the former. Additionally, the early expression of BMP2, BSP, DMP1, and OCN was more pronounced in BMP2 EVs, indicating a more rapid turn in the bone microenvironment. Similarly, Ying et al. [18] showed that mutant HIF-1α MSCs generate exosomes capable of effectively promoting bone regeneration. The association of exosomes with β-TCP enhanced bone formation compared to β-TCP alone, and exosomes carrying mutant HIF-1α were better than control exosomes. Furthermore, pronounced positive staining for OCN was observed in HIF-1α exosomes.
Hypoxic MSC-derived exosomes were reported by Liu et al. [17] and showed great improvement in bone formation. In a mice femur fracture model, a greater callus volume was observed in both exosome-treated groups; however, hypoxic exosomes showed a significant difference compared to exosomes derived from normoxia. A three-dimensional mechanical microenvironment was used to condition MSCs and was reported by Yu et al. [39] in an alveolar bone defect model. Exosomes derived from MSCs in these culture conditions were able to enhance the new bone area and volume as well as the expression of RUNX-2 and OCN.
Collectively, the studies' results show that exosomes from both conditioned and unconditioned MSCs can enhance bone formation compared to untreated defects. However, the conditioning methods of the MSCs used in all the reported studies generated exosomes with superior potential to promote bone regeneration when compared to exosomes derived from unconditioned MSCs.

Vascularization
Four studies reported a vascularization assessment in vivo, mostly with microfil perfusion by micro-CT analysis and immunohistochemistry. Ying et al. [18] showed that exosomes derived from mutant HIF-1α cells promoted a greater vessel area, vessel number, and CD31 expression in a rat calvaria defect model. Similarly, Liu et al. [17] showed that hypoxic MSC-derived exosomes could enhance vessel volume, vessel number, and the expression of CD31 and endomucin in a mice femur fracture model. In a calvaria defect model, DMOG-stimulated MSC-derived exosomes promoted a significantly greater vessel area and expression of CD31 [20]. The expression of CD31, VEGF, and VE-cad was significantly enhanced in a distal femur defect model treated with exosomes derived from strontium-substituted-calcium-silicate-stimulated MSCs.

Secondary Outcomes
Twelve studies reported secondary outcomes, related mostly to signaling pathways and miRNA profiles ( Table 5). Table 5. Secondary outcomes related to microRNA/circRNA profile and signaling pathway/gene expression analysis.
Potential signaling pathways: TGF-beta signaling pathway, Hippo signaling pathway, Map kinase, and Wnt signaling pathway, among others. [32] NA ↓ PTEN in DMOG-MSC-Exos. The deficiency of PTEN is related to increased migration and invasion of HUVECs. The downstream target of PTEN, AKT/mTOR, was blocked, and DMOG-MSC-Exos lost their superior pro-angiogenic abilities. [20] ↑ of miR-146a in BMSCs by Sr-CS extracts The miR-146a inhibition led to the downregulation of miR-146a in both BMSCs and BMSC-Exos. Treatment with Sr-CS + 146I-Exo (derived from miR-146a inhibition) diminished its angiogenic ability in HUVECs. Prediction targets showed that miR-146a directly targets Smad4 and NF2 (reported to inhibit angiogenesis). [41] ↓ of IGFBP3 associated with miR-375 overexpression. ↑ cell differentiation in IGFBP3-deficient cells cultured in OM. Furthermore, the increase in osteogenic differentiation induced by Exo (miR-375) treatment was reversed by IGFBP3 recombinant treatment. [38] NA BMP2 FEEs were able to trigger SMAD 1/5/8 phosphorylation, and control EVs had no effect beyond the resting cell state. [21] ↑ of 94 miRNAs and ↓ of 39 miRNAs in the Hypo-Exos group when compared to the Exos group.
Knockdown of miR-126 inhibited Hypo-Exos-mediated proliferation, migration, and angiogenesis in vitro and in vivo. SPRED1 is a target of miR-126 and was increased in miR-126 knockdown Hypo-Exos-treated cells. Silencing of SPRED1 promoted cell proliferation, migration, and tube formation during treatments with miR-126 knockdown Hypo-Exos. Hypo-Exos suppressed SPRED1 by activating the Ras/Erk pathway.
One study showed that the pro-osteogenic effects of exosomes derived from 0-day osteogenically induced MSCs (undifferentiated cells) were mediated by protein phosphorylation [35]. An increase in the phosphorylation of proteins implicated in bone metabolism (STAT6, GSK-3α/β, STAT5b, and STAT5a/b) was observed in a BMSC culture. Furthermore, SMAD-related genes (SMAD4 and BMPR2), related to the TGF-β signaling pathway, were upregulated.
Three studies provided secondary outcomes related to angiogenesis [17,20,41]. Exosomes from DMOG-stimulated MSCs were able to downregulate the expression of PTEN, and its deficiency is related to the increased migration and invasion of HUVECs [20]. Liu et al. [17] reported the upregulation of miR-126 in exosomes derived from hypoxic MSCs, and they showed that the knockdown of miR-126 inhibited the ability of these exosomes to mediate proliferation, migration, and angiogenesis in vitro and in vivo. Similarly, the stimulation of MSCs with strontium-substituted calcium silicate generated exosomes with upregulated miR-146a, and its inhibition led to diminished migration, tube formation, and VEGF and ANG1 expression in HUVECs [41].
A different analysis was reported by Xie et al. [31], whereby they searched for changes in circRNA expression during the osteogenic differentiation of DPSCs. The levels of circLPAR1 (hsa_circ_0003611) increased in exosomes derived from 7-day induced MSCs, and it was predicted to bind to hsa-miR-31, a miRNA that showed a significant inhibitory effect against osteogenic differentiation. Both the downregulation of hsa-miR-31 and the upregulation of circLPAR1 improved osteogenic differentiation.

Risk of Bias Assessment
The studies included for in vivo experiments were assessed for their risk of bias by the SYRCLE tool, and the results are illustrated in Figure 2.
A different analysis was reported by Xie et al. [31], whereby they searched for changes in circRNA expression during the osteogenic differentiation of DPSCs. The levels of circLPAR1 (hsa_circ_0003611) increased in exosomes derived from 7-day induced MSCs, and it was predicted to bind to hsa-miR-31, a miRNA that showed a significant inhibitory effect against osteogenic differentiation. Both the downregulation of hsa-miR-31 and the upregulation of circLPAR1 improved osteogenic differentiation.

Risk of Bias Assessment
The studies included for in vivo experiments were assessed for their risk of bias by the SYRCLE tool, and the results are illustrated in Figure 2. Regarding selection bias, six studies reported the randomization of group allocation and were assigned a low risk, but the randomization process was not specified [17,18,20,38,40,41]. Two studies did not describe the baseline characteristics of the animals; therefore, they were assigned an unclear risk [17,21]. Allocation concealment was not mentioned in any study included; thus, they were assigned an unclear risk.
Poor information was reported by the studies regarding performance, detection, and attrition bias (random housing, research blinding, random outcome assessment, outcome blinding, and incomplete outcome data); therefore, an unclear risk was assigned in most cases. Only one study described the housing conditions and was classified as low risk for random housing [38]. Furthermore, only one study reported the random outcome assessment and was classified as low risk, although it did not mention the randomization process [18].
Six studies were classified as having a high risk of bias for selective outcome reporting [20,21,[38][39][40][41]. They all failed to clearly mention the methodology for histology, histomorphometry, or IHC results, e.g., not mentioning the software used for quantification or the methodology for qualitative or semi-quantitative results. Furthermore, Zhai et al. [40] described in the Section 2 seven groups for in vivo experiments; however, in the Section 3 only a few of them were reported.
Other risks of bias included a lack of clear information regarding treatment distribution along with the defects and animals [40,41], e.g., the existence of a blank group without indication as to whether it was a separate group or the contralateral limb [40] or if one animal received more than one treatment in different limbs [41]. Furthermore, one study reported a very low number of animals included in the experiment (n = 8), resulting in only one animal for time point/treatment analysis [39].

Discussion
This review showed that all approaches used to conditioned MSCs enhanced the therapeutic ability of exosomes for bone regeneration. However, what do these approaches have in common?
In general, the conditioning or priming of MSCs are performed to mimic or magnify a response that they would have in the natural microenvironment and produce a targeted message. However, for MSCs to promote osteogenesis, angiogenesis, immunomodulation, and several other effects in the natural in vivo microenvironment, the niche is involved. The three-dimensional space that MSCs are inserted into provides cell-to-cell contact and biochemical and biomechanical signals which will determine the function and actions of these cells [42]. Therefore, the functions of exosomes will be affected by the niche and the microenvironment into which the exosome-secreting cells are inserted.
In this sense, the osteogenic induction method is the more obvious approach. The addition of several chemicals in the culture medium leads to a commitment of the cells towards the osteogenic lineage. Once the cells are committed to that lineage, they start producing targeted information, which is partially packed within their respective EVs [32][33][34]36,40]. When MSCs in a mimetic osteogenic microenvironment (cultured in OM) were submitted to exosomes derived from osteogenically induced MSCs, an enhancement was observed in the osteogenic differentiation, showing that these cells were able to assimilate the information carried by the exosomes and generate a response. When the cells were not cultured in a mimetic osteogenic microenvironment (PM), only the exosomes from highly differentiated cells were able to barely generate a response towards osteogenic commitment [29,33,36,40]. This shows that the effects of exosomes will depend on the microenvironment they are applied to. However, in the in vivo scenario, an interplay of events known as the inflammatory phase occurs at the moment of injury, wherein the cascade of events will begin [1], which strongly differs from the in vitro environment. Only one study included in this review applied exosomes from osteogenically induced MSCs in a bone defect model [40], and although the miRNA profile indicated differently expressed miRNAs that could be connected to the improvement observed in the in vivo results, only a descriptive analysis was performed of the histology findings. Therefore, there is still a lack of evidence regarding the benefits of exosomes from osteogenically induced MSCs in bone regeneration and how they are involved in the extensive events occurring at the beginning of the healing phase.
Dexamethasone is routinely used as part of the osteogenic medium components to differentiate MSCs. Zhao et al. [30], by only using a low concentration of dexamethasone to enhance the commitment of MSCs, showed that the derived microvesicles could promote the osteogenic differentiation of preosteoblasts even more than the standard osteogenic medium. However, there is still a lack of evidence for the effects of EVs derived from this type of conditioned method in vivo, since the reported in vivo experiments did not include a control group of unconditioned exosomes. The content of exosomes derived from this type of conditioning should be further analyzed, along with the effects of the treatment in bone defect models. Furthermore, dexamethasone in vivo is related to pathological conditions such as osteoporosis [43], therefore reinforcing the need for a more detailed analysis.
Two studies focused on the genetic modification of MSCs by overexpressing two components that play a role in bone regeneration: BMP2 [21] and miR-375 [38]. Bone morphogenetic proteins belong to the TGF-β superfamily and have an important osteoinductive role in the bone microenvironment [44]. Specifically, BMP2 is produced by osteoblasts and osteoprogenitors [1] and initiates the repair cascade, promoting the differentiation of MSCs in osteoblasts or chondroblasts [44,45]. The overexpression of BMP2 in BMSCs generated EVs that stimulated bone formation significantly more than control EVs and were comparable to rhBMP2 treatment. However, instead of forming bone with a fatty marrow, as in the rhBMP2 group, EVs derived from BMP2-overexpressing BMSCs promoted a dedicated intramembranous bone regeneration process [21], a well-determined part of bone regeneration. This strategy for MSC conditioning is promising for bone regeneration, due to the well-established role of BMP2.
MicroRNAs are a class of small, non-coding RNAs that negatively regulate gene expression at the mRNA level [46]. Their biological functions have been discovered throughout miRNA-knockout models and overexpression experiments [47] and include differentiation and development, metabolism, proliferation, apoptotic cell death, viral infection, and tumorigenesis [48]. The role of miRNA-375 in the osteogenic differentiation of ASCs was confirmed by Chen and coworkers [19], and the exosomes derived from miRNA-375overexpressing ASCs enhanced bone formation in vivo [38]. Thus, miRNA in osteogenesis is a field that could unravel potential targets for bone regeneration.
Lineage commitment is not the only thing needed for MSCs to exert their positive effects. For example, to activate MSCs in a fracture microenvironment, cells present in the early inflammatory phases will send signals, such as growth factors and pro-inflammatory cytokines [49]. Then, upon activation, MSCs will also function as secretory cells, including angiogenic factors [5,50,51]. Angiogenesis and osteogenesis are tightly related, and bone health is entirely dependent on vascular supply [52]. Five articles included in this review analyzed the role of MSC-conditioning methods in angiogenesis and bone healing [17,18,20,28,41].
Mesenchymal stem cells in natural physiological conditions are exposed to a low concentration of oxygen, ranging from 2 to 9% (or lower), while for in vitro cultures of MSCs, the oxygen concentration is around 21% [53]. Studies have demonstrated that culturing MSCs in hypoxic conditions improves their therapeutic effects upon direct and indirect application in several scenarios, such as myocardial repair [54,55], cerebral ischemia [56,57], skin wound healing [58,59], and bone healing [17,60,61]. Hypoxically conditioned MSCs produced exosomes that enhanced proliferation, migration, and tube formation by HU-VECs, and the bone formation and vascularization of a mice fracture model; additionally, the protein concentration of these exosomes was higher than unconditioned ones, indicating that hypoxia can improve MSC-derived exosome yield. Interestingly, the mechanisms behind these results are related to the enriched levels of miR-126, which was possibly mediated by hypoxia-inducible factor-1α (HIF-1α) [17].
In hypoxic conditions, the HIF-1α is upregulated. Genes related to angiogenesis and osteogenesis, such as VEGF and RUNX2, are regulated by this transcription factor [62][63][64]. Therefore, genetic modification was performed to generate mutant-HIF-1α-modified BM-SCs, and in vitro results showed that exosomes derived from these cells greatly improved HUVEC proliferation, migration, and tube formation [28]. In a calvarial defect model, significantly greater bone and neovascular formation was observed when treated with exosomes derived from mutant-HIF-1α BMSCs. It was also shown that these exosomes could promote proliferation, osteogenic differentiation, and RUNX expression by BM-SCs [18]. When taking into consideration the microenvironment of a fracture site, HIF-1α kick-starts several other mechanisms, and it is an ideal target and good approach for bone regeneration.
Also aiming at the angiogenic properties of MSCs, Liang and coworkers [20] conditioned these cells using DMOG, instead of genetically modifying them or culturing them in hypoxia. The HIF-1α transcription factor remains stable only under hypoxic conditions [18], and DMOG is an angiogenic molecule that inhibits the degradation of HIF-1α, stabilizing its expression by cells under normal oxygen conditions [65,66]. This approach has shown improvements in neurodegenerative diseases [67], cardiac ischemia [68], and bone regeneration [69,70]. The exosomes derived from DMOG-stimulated MSCs significantly promoted bone and vessel formation in a calvarial defect model and HUVEC proliferation, migration, and tube formation. Interestingly, the in vitro results were related to the downregulation of PTEN in HUVECs treated with DMOG exosomes, which lead to the activation of the AKT/mTOR pathway [20].
The capacity to both promote osteoblast and inhibit osteoclast functions led to the extensive use of strontium in bone research [71]. Strontium (Sr) is a bone-seeking element, and the introduction of strontium ranelate in clinical trials showed its role in preventing fractures in osteoporotic patients [72]. The mechanism by which Sr acts in an organism relies on its actions towards cellular targets similar to those of calcium (Ca), and so it interacts with signaling pathways related to calcium-sensing receptors [73]. Studies have shown that the partial substitution of Ca by Sr in ceramic-and cement-based biomaterials leads to an upregulation in the expression of osteogenesis-and angiogenesis-related genes [74] and an improvement in osteointegration [75]. The in vitro stimulation of BMSCs by strontiumsubstituted calcium silicate (Sr-CS) upregulated the expression of RUNX2, BMP-2, VEGF, and ANG1-all genes that are strongly related to bone regeneration. The derived exosomes of Sr-CS-stimulated BMSCs were able to promote the in vitro angiogenesis of HUVECs and angiogenesis and bone formation in vivo by differentially expressed miR-146a [41]. Thus, beyond the application of Sr in vivo in association with biomaterials, its use to condition MSCs and their derived exosomes is a viable and promising approach for bone regeneration. Further studies in other bone defect models should elucidate even more the therapeutic effects of Sr-CS.
The bone fracture site is filled with cells and biological factors that are organized in a timed and spatially coordinated performance, and their response to the mechanical stimuli will guide the migration, proliferation, and differentiation of progenitor cells [76]. The mechanical stimuli can act by external or intracellular forces that generate changes in the expression of several biological factors by a mechanism called mechanical transduction [77]. Using collagen as a three-dimensional environment with applied mechanical strains, Yu and coworkers [39] cultured PDLSCs and extracted their exosomes. The derived exosomes enhanced mineralization and upregulated the expression of ALP, RUNX2, OCN, and COL-1 by BMSCs in vitro. Furthermore, the exosomes were able to improve bone formation in a rat alveolar bone defect, although the number of animals was very small. A 3D mechanical environment is an option for the enhancement of the therapeutic effects of MSC-derived exosomes; however, more in vivo studies should further elucidate its mechanism.
The several conditioning methods reported in this review provided results that place EVs in the race for cell-free bone regeneration therapies. The authors' current research involves the preconditioning of canine umbilical cord perivascular cells with transforming growth factor (TGF)-β1 and the characterization of their derived EVs. This growth factor is part of the TGF-β superfamily, which is known to play an important role during bone repair [78]. Furthermore, it has been documented that this growth factor improves the migration of bone-marrow-derived MSCs [79] and optimizes the differentiation of these cells toward the osteogenic lineage [80]. Preliminary results showed that EVs were successfully isolated from the conditioned medium (CM) by differential centrifugation and ultracentrifugation ( Figure 3). Nanoparticle tracking analysis (Nanosight) demonstrated the isolation of EVs with a mode size of 121.6 and 120.2 nm for control and TGF-β1 EVs, respectively. Additionally, the particle distribution showed the superior production of vesicles by the TGF-β1 group (1.76 × 109 +/− 5.13 × 107 particles/mL) in comparison to the control group (1.61 × 108 +/− 7.58 × 106) (Figure 4). The low yield of EVs isolated from the conditioned medium is a limiting aspect for their clinical application; therefore, it is an important factor to consider when choosing a preconditioning method.
Strong findings were reported by the majority of studies, strengthening the urge for new and complementary research, such as: (i) the well-described conditioning and culturing methods; (ii) the well-characterized EVs; (iii) in vitro assays well-described with concise and complete results; (iv) in vivo bone regeneration well-reported by microcomputed tomography; and (v) the complementary analysis of miRNA profiles, signaling pathways, and gene expression. However, the in vivo studies were conducted only in small animals with varied bone defect models, consequently leaving a lack of clear evidence. Larger animals with critical bone defects or non-union models should be the next step to evaluate the conditioning methods of MSCs and their derived exosomes. Biology 2022, 11, x FOR PEER REVIEW Despite the evidence of the therapeutic potential of EVs derived from conditioned MSCs, the heterogeneity of the cell sources, EV concentrations, and the scaffolds used limits the evidence supporting a particular conditioning method as the best option for bone regeneration so far. Additionally, the reporting of outcomes could be better addressed in further studies, advocating for the quantitative analysis of standardized methods for in vivo bone regeneration assessment.
Limitations were found mostly regarding the unclear risk of bias in most of the SYR-CLE tool's domains. The lack of reporting information pertaining to allocation concealment, housing randomization/conditions, blinding of performance, and blinding outcome assessment meant that the studies were at risk of bias. Furthermore, the lack of reports of the methodology, i.e., the specific method of assessment and how it was performed, diminished the strength of the studies' evidence. For further studies focusing on the therapeutic potential of EVs derived from conditioned MSCs, this review suggests critically analyzing the best conditioning method according to the specific goal of the study, including the most suitable cell source, an approach that increases the yield of EVs, diminishes the concentration for clinical application, and is financially viable. Additionally, following the guidelines for studies of extracellular vesicles [81] and the ARRIVE guidelines for reporting in vivo studies [82] can improve the reproducibility of a study and translate the science from bench to bedside. Strong findings were reported by the majority of studies, strengthening the urge for new and complementary research, such as: (i) the well-described conditioning and culturing methods; (ii) the well-characterized EVs; (iii) in vitro assays well-described with concise and complete results; (iv) in vivo bone regeneration well-reported by microcomputed tomography; and (v) the complementary analysis of miRNA profiles, signaling pathways, and gene expression. However, the in vivo studies were conducted only in small animals with varied bone defect models, consequently leaving a lack of clear evidence. Larger animals with critical bone defects or non-union models should be the next step to evaluate the conditioning methods of MSCs and their derived exosomes.
Despite the evidence of the therapeutic potential of EVs derived from conditioned MSCs, the heterogeneity of the cell sources, EV concentrations, and the scaffolds used limits the evidence supporting a particular conditioning method as the best option for bone regeneration so far. Additionally, the reporting of outcomes could be better addressed in further studies, advocating for the quantitative analysis of standardized methods for in vivo bone regeneration assessment.
Limitations were found mostly regarding the unclear risk of bias in most of the SYR-CLE tool's domains. The lack of reporting information pertaining to allocation concealment, housing randomization/conditions, blinding of performance, and blinding outcome assessment meant that the studies were at risk of bias. Furthermore, the lack of reports of the methodology, i.e., the specific method of assessment and how it was performed, diminished the strength of the studies' evidence. For further studies focusing on the therapeutic potential of EVs derived from conditioned MSCs, this review suggests critically analyzing the best conditioning method according to the specific goal of the study, including the most suitable cell source, an approach that increases the yield of EVs, diminishes the concentration for clinical application, and is financially viable. Additionally, following the guidelines for studies of extracellular vesicles [81] and the ARRIVE guidelines for reporting in vivo studies [82] can improve the reproducibility of a study and translate the science from bench to bedside.

Conclusions
This review systematically assessed the application of EVs derived from conditioned MSCs in bone regeneration in vitro and in vivo, evidencing that different conditioning methods do improve the therapeutic effects of MSC-derived EVs for bone regeneration. These findings, despite being promising, still rely on a few heterogeneous studies (in terms of the conditioning approaches, animal and defect models, and therapeutic dosages) and still need to be addressed in larger animal models for further clinical application.
Author Contributions: F.C.H.: conception of the study, systematic search, screening for eligibility, data extraction, bias and quality assessment, interpretation of data. A.S.d.S.: systematic search, screening for eligibility, data extraction, bias and quality assessment. A.d.P.S.: interpretation of data, critical review of the manuscript. F.L.V.: interpretation of data, critical review of the manuscript. E.C.C.R.: conception of the study, bias and quality assessment, interpretation of data, critical review of the manuscript. The first draft of the manuscript was written by F.C.H., and all authors commented on previous versions of the manuscript. All authors have read and agreed to the published version of the manuscript.