Generation of Small RNA-Modulated Exosome Mimetics for Bone Regeneration

Abstract

Administration of exosomes is considered an attractive cell-free approach to skeletal repair and pathological disease treatment. However, poor yield for the production technique and unexpected therapeutic efficacy of exosomes have been obstacles to their widespread use in clinical practices. Here, we report an alternative strategy to produce exosome-related vesicles with high yields and improved regenerative capability. An extrusion approach was employed to amass exosome mimetics (EMs) from human mesenchymal stem cells (hMSCs). The collected EMs had a significantly increased proportion of vesicles positive for the exosome specific CD-63 marker compared with MSC-derived exosomes. EMs were further obtained from genetically modified hMSCs in which expression of noggin, a natural bone morphogenetic protein antagonist, was downregulated to enhance osteogenic properties of EMs. Moreover, the administration of hMSCEMs in conjunction with an injectable chitosan hydrogel into mouse non-healing calvarial defects demonstrated robust bone regeneration. Importantly, mechanistic studies revealed that the enhanced osteogenesis by EMs in which noggin was suppressed was mediated via inhibition of miR-29a. These findings demonstrate the great promise of MSC-mediated EMs and modulation of small RNA signaling for skeletal regeneration and cell-free therapy.

Mesenchymal stem cells (MSCs) are multipotent progenitor cells capable of differentiating into mature cells of various mesenchymal tissues such as bone, cartilage, and adipose, and have been widely applied to cell-mediated tissue repair and regeneration.^1–3 However, evidence suggests that, while the direct contribution of MSCs to regenerated tissues is limited, the stimulation of local healing processes via paracrine secretion exerts more important roles.^4–6 Recent studies revealed the therapeutic potency of unsegregated MSC secretomes as well as MSC-derived exosomes in skeletal tissue regeneration in varying animal-based models.⁷ Unlike direct stem cell transplantation, the application of exosomes possesses several advantages – an intrinsic homing effect, high stability in circulation, low immunogenicity, and effective molecular signaling stimulation.⁷ Because of their ideal native structure and characteristics, exosomes are also increasingly employed as a favorable nanoscale drug carrier for tissue regeneration and pathological disease treatment.^8,9 Nonetheless, the implementation of exosomes secreted from mammary cells has been significantly compromised by a difficult purification process with low yield.¹⁰ The development of a scalable approach like generation of exosome mimetics (EMs) with substantial production yields is essential for translation of exosome use to routine clinical practice.^10,11

Exosomes contain multiple bioactive molecules (protein, nucleic acid, lipid and RNAs) that can be transferred to target cells though ligand-receptor interaction, endocytosis, or direct membrane fusion.¹² Of these bioactive molecules, small RNAs (sRNAs) particularly like microRNAs were revealed to exert crucial roles in enabling intercellular signal communication to mediate related cellular functions.^13,14 However, the use of exosomes alone was not adequate for complete tissue regeneration, especially in challenging environments (e.g. large skeletal defects) due to their limited inductive capacity.^13–15 Exosome-mediated delivery of extra inductive or therapeutic factors like small molecule drugs and siRNA has been shown to enhance exosome-based tissue regeneration and skeletal disease treatment.^13–15 The exogenous transport of these molecular factors also raises concerns of high cost, poor pharmacokinetics, and inefficiency, as well as potential adverse effects.^13–15 Therefore, the enhancement of intrinsic inductive or therapeutic molecules within exosomes could present an attractive therapeutic strategy for promoting exosome-mediated treatment.

Skeletal homeostasis is attributed to multiple bioactive factors that reciprocally regulate the metabolism of local bone-forming cells.¹⁶ Skeletal cells secrete several important growth factors that induce osteogenic differentiation of marrow MSCs and subsequent mineral deposition by osteoblasts; one such growth factor is BMP2.¹⁷ In response to BMP stimulus, MSCs or osteoblasts are observed to significantly elevate BMP antagonist like noggin, suggesting a negative feedback mechanism to prevent excess BMP signaling in cells.^18,19 Previous studies by our group and others have shown that the introduction of exogenous noggin impaired osteogenesis in vitro and bone formation in vivo, while inhibition of endogenous noggin augmented bone regeneration by activating endogenous BMP/Smad signaling.^20–23 Moreover, knockdown of noggin in MSCs triggers key mediators of BMP signaling (Smad 1/5/8) and osteogenic molecules (Runx2 and OCN).²³ These observations further emphasize the potency of noggin suppression on up-regulation of endogenous BMP activity and subsequently increased osseous deposition. Additionally, exosomes are known to contain cytoplasmic contents and the membrane components of the parental cells from which they are obtained. Together, we hypothesize that suppression of BMP antagonist in MSCs may enhance the endogenous accumulation of osteogenic molecules in exosomes, and further augment the exosome-mediated treatment of bone deficiencies, a significant medical problem as the aging population is dramatically rising.²⁴

To improve current exosome-based skeletal regeneration, we developed a combination approach for scalable production yields with enhanced osteogenic capability (Figure 1). First, a simple extrusion approach was created to generate EMs from human MSCs (hMSCs), resulting in a high yield of exosomes. Second, EMs were generated from hMSCs in which noggin was suppressed; these EMs exhibited a significant increase of osteogenic induction. Third, implantation of EMs derived from noggin-suppressed hMSCs and encapsulated in an injectable chitosan (MeGC) hydrogel displayed substantial bone healing in rodent critical-size calvarial defect models. Mechanistically, inhibition of miR-29a was observed to be involved in the elevated osteogenesis of EMs from the noggin-suppressed hMSCs. Collectively, these investigations offer a promising cell-free approach using EMs and modulation of endogenous sRNAs for treatment of skeletal injury and tissue resection.

Figure 1:

Schematic diagram of generation of EMs from hMSCs with noggin suppression for calvarial bone regeneration through modulating BMP signaling-related small RNA expression.

RESULTS AND DISCUSSION

Production, characterization and differentiation of EMs

To obtain exosomes, exosome-free culture medium for hMSCs was collected and undergone with density gradient ultracentrifugation. A convenient extrusion approach was developed to collect and purify EMs by extruding hMSCs into polycarbonate membrane filters with progressively reduced pore size, followed by a 100 kDa centrifugal filter. Cryogenic transmission electron microscopy (TEM) images of EMs displayed 100–150 nm spherical particles with an intact membrane structure, similar to the recognized characteristics of exosomes (Figure 2A). Nanoparticle tracking analysis (NTA) of the purified EMs and exosomes demonstrated a size distribution with a peak diameter of around 100–150 nm (EMs: 141.8±18.3 nm, exosomes: 122.41±14.2 nm), which is consistent with the outcomes of TEM analysis (Figure 2B). When compared to exosomes from hMSC-growth medium, EMs from hMSC-growth medium (EM-GM) displayed ~20-fold increase of protein concentration in terms of a constant number of hMSCs used (Figure 2C). Similar increases in protein production were observed for EMs from hMSCs with osteogenic medium (EM-OM) as well as EMs from noggin-suppressed hMSCs with osteogenic medium (EM-OMN) compared to their corresponding exosomes (Figure S1). Additionally, an approximately 35-fold increase of particle number was detected in the EMs as compared to exosomes (Figure 2D). When normalized to equal amount of protein, ELISA analysis of CD63 (a specific marker for exosomes) revealed an approximately 7-fold increase in the hMSC-derived EMs compared to hMSC-derived exosomes, indicating that the use of an extrusive approach significantly scales up the production of exosomes (Figure 2E). To investigate the internalization of EMs into cells, hMSCs were further treated with DiI-labelled EMs or exosomes respectively for 6h. Confocal imaging demonstrated that both EMs and exosomes exhibited similar biological internalization and were efficiently transferred to the target cells (Figure 2F). Moreover, treatment with exosomes from hMSC-osteogenic medium revealed the enhanced osteogenic differentiation of hMSCs compared to exosomes from hMSC-growth medium, as shown by increased early-stage osteogenic gene expression (Runx2 and ALP) and ALP staining (Figure 2G,H). Of note, treatment with EM-OM increased osteogenic differentiation compared to treatment with exosomes from hMSC-osteogenic medium, as shown by a higher level of ALP expression. Collectively, these data indicate that the distinctive extrusion approach can efficiently produce a high yield of exosomes. Both hMSC-derived EMs and exosomes exhibit similar characteristics and biological internalization, while EMs demonstrate a better osteogenic effect.

Figure 2: Production, characterization and differentiation of EMs and EXOs from hMSCs.

(A): SEM images of MSC-derived EXOs and EMs. Scale bar: 50 nm. (B): Size distribution of MSC-derived EXOs and EMs was assessed by a nanoparticle tracking analysis. (C,D): The yields of EXOs and EMs from hMSCs with growth medium were assessed by the protein concentration (C) and particle number (D). (E): ELISA assay of CD63 for analysis of EXOs present in MSC-derived EXOs and EMs. (F): Internalization of DiI-labelled EXOs and EMs into hMSCs was visualized by confocal images. Scale bar: 10 μm. (G, H): Osteogenic effect of MSC-derived EXOs and EMs on hMSCs. hMSCs were respectively treated by Ctl, EXO-GM, EXO-OM and EM-OM at a dosage of 5 μg/mL. (G) Expression of osteogenic genes (Runx2 and ALP) was measured by a real-time PCR assay at day 2. (H) ALP expression was detected by ALP stain at day 3. Scale bar: 100 μm. * p < 0.05, ** p < 0.01, *** p < 0.001. EXOs, exosomes; EMs, exosome mimetics; Ctl, Control; EXO-GM, EXOs from hMSCs with growth medium; EXO-OM, EXOs from hMSCs with osteogenic medium; EM-OM, EMs from hMSCs with osteogenic medium.

Unlike the conventional isolation methods, this extrusion approach is convenient as well as scalable. Secreted exosomes are often mixed with large amount of tiny materials like debris, dead cells, and other proteins, raising more challenges to purify them.^10,25 By using the extrusion approach reported here, EMs collected from hMSCs exerted the improved purity with encompassment of increased concentration of exosomes when compared to the same amount of secreted exosomes collected by conventional isolation methods. Importantly, functional studies showed that treatment using EMs stimulated osteogenesis more effectively than the same amount of exosomes. Based on levels of CD-63 expression, it appears the increased osteogenesis by EMs may be due to the elevated proportion of exosomes present in EMs. Although our approach is capable of obtaining high yields of exosomes, EMs contained non-exosome substances including cytochrome c, GM-130 and calnexin as observed by western blot analysis (Figure S2).

Enhanced osteogenic induction by EMs from noggin suppressed cells

To investigate whether the osteogenic potency of EMs from conditioned osteogenic medium can be further strengthened through abrogation of the BMP antagonist-noggin in parental cells, EMs generated from the noggin-knockdown hMSCs treated with conditioned osteogenic medium were used to treat hMSCs. By a real-time PCR assay, expression of noggin siRNA was demonstrated to be increased ~2.5-fold in cells treated with EM-OMN when compared to cells treated with EM-OM (Figure 3A). Furthermore, noggin protein expression was obviously decreased in EM-OMN through a western-blot assay and colorimetric analysis (Figure 3B,C). When EM-OMN was applied to hMSCs in vitro, the expression of major osteogenic genes including Runx2, Osterix, ALP and OCN was significantly elevated compared to application of EM-OM or control treatment (Figure 3D). Particularly, treatment of hMSCs with EM-OMN at high dosage (30 μg/mL) displayed a significant increase of osteogenic gene expression as compared to treatment with EM-OMN at low dosage (5 μg/mL). Moreover, the increased osteogenic differentiation stimulated by EM-OMN was shown with increased expression of ALP and mineralization as measured by ALP stain/activity and Alizarin red stain/quantification, respectively (Figure 3E–H). The treatment with high dose EM-OMN also stimulated higher level of ALP and mineralization expression than low dose EM-OMN treatment. Overall, the osteogenic potency of EMs from conditioned osteogenic medium was synergistically enhanced by suppression of endogenous noggin in the hMSCs from which the EMs were derived.

Figure 3: Enhanced osteogenesis mediated by EM-OMN.

The osteogenic potency of EM-OMN was detected in the treated hMSCs in vitro. (A): Expression of noggin siRNA within EM-OM and EM-OMN was measured by a real-time PCR assay. (B,C): Expression of noggin protein within EM-OM and EM-OMN was assessed by a western-blot assay (B) and colorimetric analysis (C). (D): Expression of osteogenic genes including Runx2, Osterix, ALP and OCN was measured by a real-time PCR assay at day 2. (E,F): ALP expression in hMSCs was examined by ALP stain (E) and activity (F) at day 3 and 7, respectively. (G,H): Mineralization of hMSCs was detected by Alizarin red stain (G) and quantification (H) at day 14 and 21, respectively. Scale bar = 100 μm. _* p < 0.05, _** p < 0.01, _*** p < 0.001. EMs, exosome mimetics; Ctl, Control; EM-OM, EMs derived from hMSCs with osteogenic medium; EM-OMN, EMs derived from noggin-suppressed hMSCs with osteogenic medium; L-EM-OM, low dose EM-OM; H-EM-OM, high dose EM-OM; L-EM-OMN, low dose EM-OMN; H-EM-OMN, high dose EM-OMN; AR, Alizarin red staining.

In addition to lack of scalable production, low functionality of exosomes could prevent the widespread clinical use of exosome-mediated treatment. Growing evidence from prior experimental or pre-clinical trials demonstrated that the only employment of exosomes alone lacked sufficient competence for tissue regeneration or disease treatment, particularly under challenging conditions.^26,27 With a natural lipid bilayer membrane structure, exosomes are thought to be an ideal drug carrier. Major efforts have been made to promote the functionality of exosomes through exosome-mediated carrier of extra protein, siRNA or small molecular drug; however, common restrictions like low efficiency, poor pharmacokinetics, high cost, or potential side effect prevented loading exosomes with cargo of exogenous factors, and significantly barred exosomes from extensive use.^13–15,28 Thus, enhancement of endogenous inductive or regenerative factors within exosomes represents a promising therapeutic approach. Given that exosomes are composed of molecules present in the parent cell cytoplasm and membranes, genetic manipulation of parental cells has good potential to promote accumulation of therapeutic factors within exosomes.¹⁴ In our studies, we down-regulated the BMP antagonist noggin in hMSCs using a shRNA transduction approach. The harvested EMs were projected to contain high levels of noggin siRNA. Indeed, using a customized PCR assay, our results showed that the isolated EMs comprised high amounts of noggin siRNA. Although we haven’t compared the level of noggin siRNA within exosomes between our approach and other methods like direct siRNA transduction, the outcomes of our current work highlight the possibility of endogenous enhancement of molecular factors via a genetic approach in parent cells. Moreover, noggin is increasingly studied as an important therapeutic target for bone regeneration as noggin is closely involved in BMP signaling and bone development/regeneration. Our results also demonstrated that noggin suppression significantly enhanced the osteogenic induction potential of EMs. In addition to this demonstration of a functional outcome, the current approach is cost-effective, offering a promising alternative to current growth factor-based tissue engineering therapy.^29,30

Enhanced osteogenesis of MeGC hydrogels with encapsulation of EM-OMN in vitro

In order to translate EMs to potential bone repair in vivo, injectable photocrosslinking MeGC hydrogels were developed to deliver EMs. First, the EMs were obtained from hMSCs treated with conditioned osteogenic medium, and then laden into MeGC hydrogels for testing their osteogenic induction potential in vitro. Consistent with monoculture outcomes in vitro, the increased osteogensis of hMSCs encapsulated into MeGC hydrogels in the presence of EM-OM was detected by elevated ALP expression and mineralization as well as increased osteogenic gene expression (Runx2, Osterix, ALP and OCN) when compared to controls (Figure 4A–E). The immunohistochemical stain also revealed an apparent increase of OCN expression, further confirming the osteogenesis stimulated by EM-OM in the three dimensional (3D) setting of hydrogels with the encapsulation of EM-OM (Figure 4F). Importantly, a further increase of osteogenesis was observed in the hMSC-laden hydrogel in the presence of EM-OMN, as shown by a significant increase of ALP, mineralization, osteogenic gene expression, and OCN stain. Of note, the encapsulation of high dose EM-OMN (30 μg) in 3D hydrogels stimulated a higher level of osteogenic maker expression than low dose EM-OMN (5 μg). In addition to observation of osteogenesis, the biocompatibility of EM-laden hydrogels was also assessed by live/dead assay and AlamarBlue assay (Figure S3). The results demonstrated that MeGC-based hydrogels coupled with EM-OM or EM-OMN were able to maintain high cellular viability.

Figure 4: Enhanced osteogenesis of EM-OMN-laden hydrogels.

An injectable crosslinking MeGC-based hydrogel was developed to deliver EM-OMN, EM-OM or control. The osteogenic induction of the complex was analyzed in in vitro 3D setting as encapsulated with hMSCs. (A,B): ALP expression was examined by ALP staining (A) and activity (B) at day 7 and 14, respectively. (C,D): Mineralization was assessed by Alizarin red stain (C) and quantification (D) at day 28. (E): Osteogenic gene expression (Runx2, Osterix, ALP and OCN) was measured by a real-time PCR assay at day 3. (F): Immunohistochemical staining of OCN was performed on tissue-sections from hydrogels at day 28 and images were taken from the center of the hydrogel sample. Scale bar = 100 μm. ** p < 0.01_{, ***} p < 0.001. EMs, exosome mimetics; Ctl, Control; L-EM-OM, hydrogel encapsulated with low dose EM-OM; H-EM-OM, hydrogel encapsulated with high dose EM-OM; L-EM-OMN, hydrogel encapsulated with low dose EM-OMN; H-EM-OMN, hydrogel encapsulated with high dose EM-OMN; AR, Alizarin red staining.

Due to the properties of tissue-like water content, easy implantation, and high biocompatibility, hydrogels have been increasingly used as cell or growth factor carriers for tissue engineering-based skeletal repair.³¹ The porous structure of hydrogels is able to release the encapsulated bioactive factors in a sustained manner. Especially, injectable biopolymer hydrogels are suitable for repair of bone defects since they can be directly administrated into the osseous defect and provide a spatial fit between the formed implant and a randomly shaped defect.^32,33 The MeGC-based hydrogels we previously developed are a favorable carrier for cells or growth factors, providing excellent biological activity.³² Similarly, the cargo of EM-OMN with the MeGC hydrogel displayed high cellular viability and osteogenic induction in current studies. Overall, the injectable MeGC hydrogel with encapsulated EM-OMN may facilitate the repair of critical-sized bone defects in vivo.

In vivo delivery of EM-OMN promotes calvarial bone healing

According to in vitro two dimensional (2D) and 3D outcomes, we further investigated the efficiency of EM-OMN at stimulating in vivo bone repair in a rodent calvarial defect model (Figure 5). EM-OMN (EMs collected from the genetically-engineered hMSCs in which noggin expression was suppressed by transduction of noggin shRNA) were laden into the MeGC hydrogel and then injected into critical-size calvarial defects created in adult mice. We first investigated the transfection efficiency of EMs in vivo. Five days post-operation, treatment with hydrogel containing DiI-labelled EM with low (5 μg) or high dosage (30 μg) was detected with high transfection efficiency, as displayed with high intensity of fluorescence visualized in the infiltrated cells within graft and quantitatively evaluated by image analysis (Figure 5A,B). By contrast, the control treatment group (hydrogel only) showed no obvious fluorescence. These results illustrated that EMs are consistent with exosomes that possess native capability on cell transfection.

Figure 5: EM-OMN-laden hydrogel promotes calvarial bone repair in vivo.

(A,B): DiI-labelled EMs were encapsulated into the MeGC-based hydrogel, and the complex was then injected into the critical-sized (3 mm-diameter) calvarial defects created in mice. Five days post-operation, the implant was collected from surgical animals. The transfection efficiency of EMs in vivo was observed in whole hydrogel implants retrieved by a fluorescence microscopy (A). The dotted lines indicate the border of implant. The dashed box indicates the high magnification of image below. Scale bar = 1000 μm (top), Scale bar = 200 μm (blow). Fluorescence intensity was quantified by image analysis (B). (C-E): After 8 weeks postoperatively, the implant was further collected from mice, and undergone with microCT scan, quantitative analysis and histological analysis: (C) Micro-CT images of mouse calvarial bone regeneration. Scale bar = 1 mm; (D) Quantitative analyses of new bone area, bone volume/tissue volume (BV/TV), Trabecular number (Tb.N) (n=6 per group); (E) H&E stain. The vertical dashed lines indicate the relative original defect area. Black arrowheads indicate new bone tissues. Blue arrowheads indicate hydrogel. Green arrowheads indicate native bone. The dashed box (black) indicates the high magnification of image below. Scale bar = 1000 μm (top), Scale bar = 200 μm (blow). (F): Masson trichrome stain, Picrosirius stain and OCN immunohistochemistry. Red arrowheads indicate new bone tissues. White arrowheads indicate deposition of collagen matrix normally existing in native bone. Scale bar = 200 μm. ** p < 0.01 and *** p < 0.001. EMs, exosome mimetics; Ctl, Control (hydrogel only); L-EM-OM, hydrogel encapsulated with low dose EM-OM; H-EM-OM, hydrogel encapsulated with high dose EM-OM; L-EM-OMN, hydrogel encapsulated with low dose EM-OMN; H-EM-OMN, hydrogel encapsulated with high dose EM-OMN.

In order to examine EM-OMN-mediated bone healing, the mouse calvarial tissues were extracted at 8 weeks postoperatively, and imaged by micro-CT scan. Micro-CT analysis demonstrated that the implant of EM-OMN/hydrogel incited significant bone healing as compared to EM-OM/hydrogel or control-treated groups (hydrogel only) (Figure 5C). In particular, the complex of MeGC hydrogel with high dose EM-OMN (30 μg) indicated nearly complete bone healing, while only partial bone healing was observed in EM-OM/hydrogel or control-treated groups. Moreover, quantitative analysis of micro-CT scan revealed up to 62% bone healing appearing in the treatment using high dose EM-OMN in contrast to only ~3% in control, ~8% in low dose EM-OM and ~16% in high dose EM-OM (Figure 5D). In addition to bone volume, bone microarchitecture is considered important in determining mechanical strength of trabecular bone.^34,35 The assessment of structural indices such as trabecular number or thickness in micro-CT images is widely used to predict the quality of regenerated bone.^34,35 In our studies, a significant increase of new bone formation in the high dose EM-OMN treatment quantitatively demonstrated increased trabecular number (Figure 5D).

The substantial bone healing in the high-dose EM-OMN treatment was also validated by histological and immunohistochemical analyses (Figure 5E,F). Hematoxylin and eosin (H&E) stain showcased that defects treated with hydrogels containing high dose EM-OMN resulted in new lamellar bone formation nearly bridging the defect area at 8 weeks postoperatively, while defects treated with hydrogel alone (control) were filled with fibrous tissues and residual hydrogel materials with minimal bone formation (Figure 5E). Complete ossification was not obvious in defects treated with hydrogels containing EM-OM or low dose EM-OMN. In contrast to other treatment groups, a large amount of new bone was present throughout the calvarial defects treated with hydrogels with high dose EM-OMN, as detected by Masson’s trichrome stain (Figure 5F). There was no apparent bone formation detected in the area of defects treated with control. Furthermore, organization of collagen matrix in the defect area was stained by Picrosirius red and visualized using polarized microscopy. Higher levels of collagen deposition were detected in the area of defects treated with hydrogels with high dose EM-OMN compared with other treatment groups (Figure 5F). Birefringence of these collagen fibers can resemble collagen fibers present in native calvarial bone. Immunohistochemical stain for OCN, a protein associated with osteoblasts and matrix mineralization, demonstrated the significantly elevated expression of OCN in defects treated with hydrogels with EM-OM or EM-OMN when compared with control groups (Figure 5F). This suggests that the increased osteoblastic differentiation and mineralization occurred in these treatment groups.

Collectively, these in vivo observations display that the delivery of EM-OMN through an injectable MeGC hydrogel can induce robust bone repair in a mouse non-healing calvarial defect model. The efficacy of our system to induce bone regeneration during the early phase of healing and the involved molecular mechanisms of osteogenesis will be further investigated in our future studies.

Noggin suppression promotes EM-mediated osteogenesis by inhibition of miR-29a

Exosomes have been demonstrated to serve as a mediator of intercellular communication mainly through sRNAs especially like miRNAs.¹⁴ We predicted that noggin shRNA in parental cells not only would lead to elevation of noggin siRNA but also may be associated with relevant miRNA modulation. To elucidate the underlying mechanisms of how noggin suppression triggered EM-mediated osteogenic induction, we further investigated the profile of miRNA expression within EM-GM, EM-OM and EM-OMN via conducting a microRNA sequencing assay (miRNA-seq). The miRNA-seq data identified a bundle of miRNAs that were up-regulated or down-regulated in EM-OM vs EM-GM, EM-OMN vs EM-OM and EM-OMN vs EM-GM (Figure 6A). Both Venn diagram and heatmap further revealed that several miRNAs commonly expressed in pairwise comparison exerted synergistic up-regulation or down-regulation (Figure 6B,C). As suppression of noggin, a natural BMP antagonist, can stimulate the endogenous BMP activity in MSCs, we screened those miRNAs particularly in relation to BMP/Smad signaling using molecular approaches in vitro. Interestingly, we found that expression of miR-29a was significantly decreased in EM-OMN compared to EM-OM or EM-GM (Figure 6D). Unexpectedly, the level of miR-29a expression in EM-OM was also lower than EM-GM, while the extent of reduced miR-29a was greater in EM-OMN vs EM-GM than in EM-OM vs EM-GM. These outcomes also suggested the potency of miR-29a in EM-mediated osteogenesis. In addition, the expression of BMPR1A and ID1 (a key downstream target of BMP signaling) was detected to significantly increase in the EM-OMN when compared to EM-GM or EM-OM (Figure 6E). To further investigate whether miR-29a regulates BMP/Smad signaling, hMSCs were treated with EM-OMN or miR-29a. Interestingly, treatment with EM-OMN significantly increased expression of BMPR1A and pSmad 1/5/8 while treatment with exogenous miRNA-29a decreased expression of BMPR1A and pSmad1/5/8 (Figure 6F), as measured western-blot assay. Lastly, the elevated osteogenesis of hMSCs mediated by EM-OMN was observed to be abrogated by transporting additional miRNA-29a, as measured by ALP stain and activity (Figure 6G,H). Taken together, these data show that miR-29a exerts crucial roles in the enhanced osteogenic induction of EM-OMN. Inhibition of miR-29a stimulates endogenous BMP/Smad signaling via enhancement of BMPR1A.

Figure 6: Noggin suppression mediates the elevated osteogenesis of EMs through inhibition of miR-29a.

miRNA-seq assay was used to examine the profile of miRNA expression in EMs. (A): Volcano plots for down-regulation and up-regulation of miRNAs between EM-GM, EM-OM and EM-OMN, respectively. (B): Venn diagrams displayed the distribution of the common miRNAs in EM-OM vs EM-GM, and EM-OMN vs EM-OM. (C): Heatmap was adopted to assess the expression of common miRNAs in EM-GM, EM-OM and EM-OMN. Red and blue denote high and low expression, respectively. (D): Expression of miR-29a was measured by a miRNA PCR assay. (E): Expression of BMPR1A and ID1 was measured by a real-time PCR. (F): Inhibition of miR-29a enhanced BMP/Smad signaling via elevating BMPR1A, as measured by a western-blot assay. (G,H): Treatment of EM-OMN stimulated ALP expression of hMSCs, which was abrogated by exogenous carrier of miR-29a mimics (miR-29a). ALP expression at day 3 was evaluated by ALP staining (G) and image analysis (H). Scale bar = 100 μm. ns, no significant difference, * p < 0.05, ** p < 0.01, _*** p < 0.001. EMs, exosome mimetics; EM-GM, EMs derived from hMSCs with growth medium; EM-OM, EMs derived from hMSCs with osteogenic medium; EM-OMN, EMs derived from noggin-suppression hMSCs with osteogenic medium.

BMPs are known to bind to BMPRI/II to induce BMP/Smad signaling.³⁶ Activity of BMPs is also antagonized by cognate binding proteins like noggin.³⁷ Therefore, suppression of noggin initiates endogenous BMP activity, which triggers subsequent osteogenic differentiation. In addition, miRNAs were shown to exert a crucial role in mediating the functions of exosomes.^12,14,38 In the past decades, there have been a bundle of miRNAs identified to regulate BMP signaling and bone regeneration.^39,40 Additionally, noggin was reported to be associated with miRNA regulation of BMP signaling.^40,41 Based on these prior findings, we predicted noggin suppression would be associated with the regulation of miRNAs, especially in relation to BMP signaling. To test this hypothesis, our mechanistic studies incorporating miRNA-seq and molecular analysis clearly displayed that miR-29a was significantly down-regulated in EM-OMN. Further molecular observations demonstrated that inhibition of miR-29a elicited BMPR1A expression, which in turn increased BMP/Smad signaling. The elucidation of this mechanism is thought to be highly rational in terms of aforementioned noggin regulation and exosome characteristics. Interestingly, the miR-29a was also found to be decreased in the EM-OM when compared to EM-GM. Noggin suppression and osteogenic medium mediate a synergistic inhibition of miR-29a expression, further indicating that miR-29a may be an important therapeutic target in exosome treatment and related skeletal regeneration. Nonetheless, the details of how noggin interacts with miR-29a may be further investigated in future. In general, the findings of endogenous BMP signaling-related sRNAs and their roles in EM-mediated osteogenesis may offer important knowledge for development of efficient therapeutic modalities for bone regeneration or other skeletal disease treatment.

CONCLUSIONS

While cell-derived exosomes present promise for skeletal tissue regeneration, exosome-based therapy has not yet been translated to clinical practice due to several major obstacles like low production yield and limited regenerative ability. To conquer these barriers, we developed an alternative strategy that not only increases the production yield of EMs but also strengthens their regenerative ability. Generation of EMs from the noggin-suppressed hMSCs significantly increased osteogenic induction. Moreover, the delivery of EM-OMN by an injectable MeGC hydrogel revealed robust bone regeneration in mouse non-healing calvarial defect model. Mechanistically, the elevated osteogenesis by EM-OMN was mediated through the increased noggin siRNA and reduced miRNA-29a. These mechanistic findings may lay out a significant basis for developing effective therapeutic modalities by modulating intrinsic BMP signaling-related sRNAs. Along with application of injectable hydrogels, our work also exhibits potentials to broaden translational use of EM-mediated clinical bone regeneration.

METHODS

Cell culture and shRNA transduction

hMSCs purchased from Lonza, Vancouver were grown in human MesenCult™ medium (STEMCELL Technologies, Vancouver). For noggin suppression, hMSCs were transduced by lentivirus particles encoding noggin shRNA (Santa Cruz Biotechnology, CA) following manufacture’s protocol. Briefly, cells at passage 3 were seeded in a 6-well plate. After 24 hours, cells at 70% confluency were incubated with the lentiviral particles targeting either noggin shRNA or control, and polybrene (5 μg/mL). After overnight incubation, the medium was replaced with regular culture medium. Puromycin dihydrochloride (Sigma) was applied to select clone with stable expression of shRNA.

Preparation of MSC-derived exosomes and EMs

Collection of MSC-derived exosomes utilized the conventional density gradient ultracentrifugation method. In brief, hMSCs with and without noggin suppression were cultured at 5×10⁵ cells/T-150 flask with growth medium containing 10% exosome-free FBS (Fisher) or osteogenic medium containing 10% exosome-free FBS, 50 μg/mL L-ascorbic acid (Sigma), 10 mM β-glycerophosphate (Sigma), and 100 nM dexamethasone (Sigma). After 48 hour, 100 mL cell culture media collected from each group was centrifuged at 300 g for 10 min at 4°C in order to pellet dead cells and debris. The remaining supernatant then underwent a series of centrifugation - at 2000 g for 10 min, 10,000 g for 30 min, and finally 100,000 g for 70 min. The remaining pellet from the last centrifugation was resuspended in PBS.

Furthermore, MSC-derived EMs were collected and purified using a serial extrusive approach (Figure 1). Briefly, hMSCs (1×10⁶ cells) were collected and resuspended in 1 mL of PBS (pH 7.4). The cell suspensions were extruded sequentially through 5- and 1-μm polycarbonate membrane filters (Nuclepore; Whatman Inc., Clifton, NJ) using a mini-extruder (Avanti Polar Lipids, Birmingham, AL). Cell debris and microvesicles of the final extruded sample (1 mL) were removed by centrifugation at 10,000 g for 10 min at room temperature. The exosome mimetics were purified and concentrated with a 100 kDa centrifugal filter (EMD Millipore, Temecula, CA, USA) at 1,000 g for 15 min at room temperature with PBS (pH 7.4), repeated three times with an equivalent volume of PBS. The final sample was stored at 80°C. Size of nanoparticles (500 μg/mL total protein) was analyzed by NTA and dynamic light scattering using Nano ZS (Malvern). The morphology of nanoparticles was further observed using a TEM.

Cellular uptake assay

hMSCs were labeled with DiI (Sigma) for 2 hour and subsequently washed three times with PBS. Exosomes and EMs were separately collected from DiI-labeled hMSCs following the above methods. DiI-labeled exosomes and EMs were re-suspended in FBS-free DMEM medium, and incubated with the cultured hMSCs up to 6 hour. After incubation, hMSCs were washed, stained with Hoechst 33342, and imaged by confocal microscopy.

ELISA assay

For measuring total protein content of vesicles, the collected exosomes and EMs were lysed using RIPA buffer (Thermo Scientific), sonicated for 5 min to obtain whole proteins from the vesicles and measured by a BCA protein assay (Fisher). The protein samples were then subjected to ExoELISA assay (System Biosciences, CA) following the manufacture’s protocol. Briefly, 50 μL of freshly prepared protein standards or exosome samples was added to an appropriate well of a micro-titer plate. The plate was incubated with CD63 primary antibody for 1 hour, colored with Super-sensitive TMB ELISA substrate, and read by a spectrophotometric plate reader at 450 nm.

RNA extraction, qRT-PCR, and miRNA-seq

The total RNA from exosomes, EMs, and cells was extracted using Trizol reagent (Life Technologies) and RNeasy Mini kit (Qiagen, CA) according to the manufacture’s protocol. A 100 ng aliquot of RNA from each sample was reverse transcribed to cDNA by a SuperScript III First-Strand Synthesis System (Life Technologies). qRT-PCR analysis was subsequently performed with a 20 μL of SYBR Green reaction system in a LightCycler 480 PCR instrument (Roche, IN). GAPDH was adopted as an internal control to normalize signal for each target gene. The primer sequences are detailed in Table S1. To measure the expression level of noggin siRNA in EMs, a series of synthetic noggin siRNA oligonucleotides with a gradient of concentrations were also reverse transcribed, amplified with stem-loop RT primers (Forward: CTCAACTGGTGTCGTGGA, Reverse: TCGGCAGGTGGATCCCGGAGGAAGT) and results plotted to form the standard curve. Since there is no consensus on the application housekeeping genes for real-time PCR in exosomes, the expression levels of noggin siRNA in EMs were normalized to the total protein content of EM. To examine the microRNA expression profile in EMs, a high-throughput miRNA-seq was performed at the UCLA sequencing core facilities. RNA-Seq libraries were prepared by Small RNA-Seq Library Preparation Kits (Lexogen, NH). The pooled libraries were subsequently sequenced by an Illumina HiSeq 3000 machine. The heatmap was generated with Morpheus.

Western blot assay

After total protein of EMs or cells was determined by a BCA assay, protein samples were separated on a 10% SDS-PAGE gel, transferred onto an immobilon transfer membrane (Millipore, MA) and blocked at 4°C overnight in 5% Non-Fat Milk. Membranes were then incubated with primary antibodies against Noggin, CD63, Cytochrome c, GM-130, Calnexin, pSmad 1/5/8, BMPR1A or GAPDH (Santa Cruz Biotechnology, CA) for 2 hour. Membranes were then washed with PBST buffer and incubated with secondary antibodies for 1 hour at room temperature. The membranes were washed again with PBST buffer for 5 min and imaged with the Clarity Western ECL Substrate (Bio-Rad, CA) and a VersaDoc 4000 MP (Bio-Rad).

Preparation of MeGC hydrogel

Injectable MeGC-based hydrogel was prepared according to the methods we previously established.³² Glycidyl methacrylate was mixed with glycol chitosan solution, and then dissolved in ddH₂O (2% w/v) to obtain a 1:1.1 M ratio solution. The solution was adjusted to pH 9.0 with 0.5 N HCl and reacted for 48 hours with the gentle shaking at room temperature. After dialysis in 50 kDa dialysis tubes against ddH₂O for 24 hours, the purified MeGC solution was lyophilized for at least 48 hours. The lyophilized MeGC was finally dissolved in PBS to form the 2% MeGC hydrogel for further experiments.

Cell proliferation in MSC-EM-laden hydrogel

To detect cellular proliferation in EM-laden hydrogels, hMSCs encapsulated with MeGC hydrogels were measured by an AlamarBlue assay (Bio-Rad). Cell-laden hydrogels were incubated in 10% (v/v) Alamarblue reagent for 2 hours, and then fluorescence intensity was measured at 530 (excitation) and 590 (emission) nm. Additionally, live/dead fluorescence stain was conducted to measure the viability of the encapsulated cells in the hydrogels. Cell-laden hydrogels were stained using a live/dead assay kit (Invitrogen) for 30 min at 37°C and 5% CO₂ environment. The fluorescence images were taken by an Olympus IX71 fluorescence microscope (Olympus, Japan).

ALP stain, ALP activity, Alizarin red stain and qualification

For investigating the extent of osteogenic differentiation in 2D and 3D environment, hMSCs with and without hydrogel were cultured in osteogenic medium for a fixed time interval. To measure ALP expression, cells were fixed with 10% formalin, stained using an ALP colorimetric kit (Sigma) including 5-Bromo-4-chloro-3-indoxylphosphate, Nitro Blue tetrazolium, and AP buffer, and imaged by an Olympus IX71 fluorescence microscopy. To quantitate ALP activity, cells were digested in NP-40 lysis buffer (Life technologies) and incubated in the ALP buffer including p-nitrophenol phosphate substrate (Sigma). Absorbance was measured at 405 nm, and signal normalized to whole DNA content of each sample detected by PicoGreen dsDNA kit (Life Technologies). To investigate mineral deposition, cells were fixed with 10% formalin and stained with 2% alizarin red solution (Sigma) about 20 min and imaged by an Olympus IX71 fluorescence microscopy. Mineralization was quantified by dissolving the stained cells in 10% (v/v) acetic acid and measurement of absorbance at 405 nm.

Calvarial defect model

The animal experiment procedure was strictly conducted under the supervision, approved by Chancellor’s Animal Research Committee of the Office for Protection of Research Subjects at UCLA and followed the instruction of the UCLA Office of Animal Research Oversight. Ten-week old nude mice were purchased from Charles River Laboratories (Wilmington, MA) and were cared under the direction of the Guidelines for the Care and Use of Laboratory Animal of the National Institutes of Health. The surgical procedure for creation of mouse calvarial defects followed the protocol previously established.^42,43 After the mouse underwent inhalational anesthesia with isoflurane, 3-mm full-thickness calvarial defects were created on the parietal bone using a trephine drill with continuous irrigation. The dura mater was protected to avoid damage during surgery. 20 μL of EM-laden MeGC hydrogel was then injected into each defect region. Injection of 20 μL MeGC hydrogel without EMs was regarded as control group. Postoperatively, all of experimental mice were allowed to fully recover from anesthesia and then transported to the vivarium for later care. Buprenorphine (0.1 mg/kg) was administered for up to 3 days and drinking water included trimethoprim-sulfamethoxazole for 7 days in order to prevent infection.

Micro-computerized tomography (μCT) scanning

Entire calvarial bone tissues were collected from experimental mice 8 weeks post-operation. The extracted calvarial tissues were fixed with 4% formaldehyde for 2 days at room temperature and then rinsed with PBS three times. The fixed tissue samples were immersed in 70% ethanol before scanning with a high–resolution MicroCT machine (μCT40; SkyScan 1172; SkyScan, Kontich, Belgium) at 57 kVp, 184 μA, 0.5 mm filtration, and 10 μm resolution. Reconstruction of 3D images was accomplished with Dolphin 3D software (Dolphin Imaging & Management Solutions, CA). Volume of interest (VOI) was chosen following blow procedures. Dimension of 3 mm region of interest (ROI) was selected on the surface of bone, where the initial defect was produced. Then, the ROI was overlaid on the center of the defect. Black and white conversion of the images visualized the boundary of new bone in ROI and VOI was further selected by adding frames to reach 0.5 mm thickness. Quantitative analyses including bone volume/tissue volume (BV/TV %) and trabecular number (Tb.N.) were performed by the CTan software (Skyscan). Relative bone growth surface area (Bone growth area %) was determined by Image J software (NIH).

Histological analysis

After completing the micro CT scan, the fixed tissue samples were decalcified with 10% EDTA solution at room temperature and gentle shaking for up to 1 week and then embedded in paraffin. Samples were sectioned at 5 μm. Histological analyses including H&E, Masson-Goldner trichrome, and Picrosirius red staining were conducted on paraffin-embedded sections and imaged by an Olympus IX71 microscope. Immunohistochemical staining for OCN was done on remaining sections. After being processed with citric acid antigen retrieval, slides were incubated in the first antibody against OCN (sc-365797, Santa Cruz Biotechnology, CA) and stained by HRP/DAB detection kit (ab64259, Abcam, MA). In brief, deparaffinized slide were initially incubated in sodium citrate solution (10 mM, pH 6.0) for 20 min at 100–120 °C. Then, slides were cooled down to room temperature and washed with PBS. The slides were incubated in hydrogen peroxide for 10 min, washed with PBS, and incubated with protein block solution (1% BSA in DW) for 10 min. Primary antibody against OCN was diluted at 1:200 ratio and dropped on the slides. The slides were incubated at 4 °C overnight under humid condition. The following day, slides were incubated with biotinylated anti-mouse secondary antibody for 10 min, washed with PBS, incubated with streptavidin peroxidase for 10 min, and washed with PBS. For color development, the slides were finally incubated with DAB chromogen solution for 8 min. Immunohistochemical stain was also employed to measure the expression of OCN in the encapsulated cells in hydrogels with EM-OM or EM-OMN in vitro. Briefly, hMSCs and EM-OM/OMN were co-encapsulated in the hydrogels and the composites were cultured in the osteogenic medium. The collected hydrogels were fixed with 10% formalin at day 28, embedded in paraffin and cut into 5 μm section for OCN immunohistochemistry as described above. Images were taken from the center of tissue sections.

Statistical analysis

Quantitative data were expressed as mean ± SEM, with * p < 0.05, ** p < 0.01, and *** p < 0.001 considered statistically significant. One-way analysis of variances (ANOVA) with Tukey’s post hoc test was used when comparing more than two groups, and for comparison between two groups, two-tailed Student’s t-test analysis was used.