Author contributions: B.L. and D.C.W.: conception and design, administrative support, collection and/or assembly of data, data analysis and interpretation, manuscript writing, and final approval of manuscript; E.N.: administrative support, collection and/or assembly of data, data analysis and interpretation, and manuscript writing; S.L., J.S.H., F.J., M.H., and J.P.G.: collection and/or assembly of data and data analysis and interpretation; A.W.J.: collection and/or assembly of data, data analysis and interpretation, and manuscript writing; D.T.M.: collection and/or assembly of data; M.L.: provision of study material or patients; N.Q.: conception and design, administrative support, collection and/or assembly of data, and data analysis and interpretation; G.C.G., J.C.W. and M.T.L.: conception and design and final approval of manuscript.
Disclosure of potential conflicts of interest is found at the end of this article.
First published online in STEM CELLSEXPRESS October 13, 2011.
An urgent need exists in clinical medicine for suitable alternatives to available techniques for bone tissue repair. Human adipose-derived stem cells (hASCs) represent a readily available, autogenous cell source with well-documented in vivo osteogenic potential. In this article, we manipulated Noggin expression levels in hASCs using lentiviral and nonintegrating minicircle short hairpin ribonucleic acid (shRNA) methodologies in vitro and in vivo to enhance hASC osteogenesis. Human ASCs with Noggin knockdown showed significantly increased bone morphogenetic protein (BMP) signaling and osteogenic differentiation both in vitro and in vivo, and when placed onto a BMP-releasing scaffold embedded with lentiviral Noggin shRNA particles, hASCs more rapidly healed mouse calvarial defects. This study therefore suggests that genetic targeting of hASCs combined with custom scaffold design can optimize hASCs for skeletal regenerative medicine. STEM Cells2011;29:2018–2029.
In United States alone, over 1 million orthopedic procedures were performed in 2006 with an estimated cost exceeding 7 billion dollars . In addition, over 50,000 craniotomies/craniectomies are performed annually, costing over 500 million dollars. A tremendous need thus exists for new strategies to accelerate bone regeneration of skeletal defects . A sophisticated approach to skeletal reconstruction combining gene manipulation with custom scaffold design would be a significant clinical breakthrough [2–4]. Bone morphogenetic protein (BMP)-2 is the most commonly used cytokine with increasing clinical use over the past decade [5–7]. However, along with high cost, outcome studies have found high doses of BMP-2 to be accompanied by worrisome side effects such as ectopic bone formation, which in spinal surgery can cause neural compression [5, 8]. In light of this, a cell-based approach might enable reduction in the amount of BMP-2 required, thereby offering a safer and more cost-effective alternative.
Multiple studies have shown human adipose-derived stem cells (hASCs) to be a readily available osteogenic cell source for tissue engineering purposes. Use of hASCs circumvents the limited cell availability and painful harvest procedures necessary to obtain bone marrow-derived mesenchymal stem cells or bone grafts. Interestingly, BMP-2 has been found to significantly enhance osteogenic differentiation of ASCs [9, 10]. To reduce the total dose BMP-2 used, although, we sought to manipulate levels of a well-described BMP signaling antagonist (Noggin) using lentiviral and nonintegrating minicircle technologies while simultaneously delivering BMP-2 in a slow-release form. This combinatorial approach was theorized to be additive—providing an agonist while also reducing an antagonist—so as to maximally activate the BMP pathway in hASCs and promote bone differentiation.
While lentiviral and adenoviral techniques have been shown to be efficient at gene knockdown, however, the risk of insertional mutagenesis has tempered their clinical potential. In contrast, minicircle plasmids are nonintegrating episomal DNA vectors that possess a circular expression cassette lacking bacterial plasmid DNA. Previous studies have demonstrated the efficiency of these vectors for in vitro and in vivo gene transfer [11–14]. Such methodologies, although, have lacked scalability to clinically relevant levels. Recent work by Kay et al.  has improved on previous minicircle constructs by using a circular integrating plasmid, including a second recombinase and a TPin/9attB.9attP recombination system. This has allowed for site-specific integration of DNA sequences up to 10 kb including both a Noggin (NOG) short hairpin ribonucleic acid (shRNA) and a green fluoresecent protein (GFP) construct.
In this study, we hypothesized that by knocking down Noggin using lentiviral and minicircle techniques, we can increase the osteogenic capability of hASCs. Furthermore, by adding cells with noggin suppression to recombinant human bone morphogenetic protein (rhBMP)-2-loaded biomimetic scaffolds, we can potentially accelerate bone formation in vivo. Such a strategy could be used to maximize the regenerative potential of hASCs for skeletal tissue engineering.
MATERIALS AND METHODS
Chemicals, Supplies, and Animals
Medium, fetal bovine serum (FBS), and penicillin/streptomycin were purchased from GIBCO Life Technologies (Carlsbad, CA). Human Noggin antibody and rhBMP-2 were purchased from R&D Systems (Minneapolis, MN). Unless otherwise specified, all other chemicals were purchased from Sigma-Aldrich (St. Louis, MO). CD-1 nude male mice (Crl:CD1-Foxn1nu) were obtained from Charles Rivers (Wilmington, MA, http://www.criver.com/en-US/Pages/home.aspx).
Human ASCs were harvested from lipoaspirates of six women between the ages of 21 and 53, with an average age of 42 and an average BMI of 27.3. None of the women had any major medical comorbidities . Mouse ASCs (mASCs) were harvested from the inguinal fat pads of transgenic mice with noggin alleles flanked by loxP sites and a tamoxifen inducible Cre-recombinase . Evaluation of Cre-mediated DNA recombination was performed for noggin by gene transcript and protein level analysis. All research involving human tissue has been approved by the Stanford Institutional Review Board, protocol #2188. All research involving vertebrate animals has been approved by Stanford APLAC, protocol #9999.
The minicircle vector was created as previously described and is available on request (Fig. 5A, 5B) . The sequences of primers used for plasmid construction are listed in Supporting Information Table 1. The GFP expressing cassette and shRNA expressing cassette were amplified from PDC316. The resultant plasmid was termed PMC-EGFP-U6. shRNA oligos as shown in Supporting Information Table 1 were synthesized at the Stanford PAN facility. These oligos were subsequently annealed into PMC-EGFP-U6 at restriction sites BamH1 and HindIII.
In Vitro Culture Assays
For osteogenic differentiation, all assays were performed in triplicate wells. After attachment, cells were treated with standard growth medium (Dulbecco's Modified Eagle Medium, 10% FBS) or with osteogenic differentiation medium (ODM; Dulbecco's modified Eagle's medium, 10% FBS, 100 μg/ml ascorbic acid, 10 mM β-glycerophosphate, Invitrogen, Carlsbad, CA). Standard osteogenic assays were performed . Specific gene expression was assayed by quantitative reverse transcription-PCR (RT)-polymerase chain reaction (qRT-PCR; Supporting Information Table 2). For select experiments, rhBMP-2 (200 ng/ml) was added to ODM. Concentrations used were based on data obtained from mASCs and from previous studies . Vehicle control for rhBMP-2 was 0.01% bovine serum albumin (BSA).
Preparation of Scaffolds
Apatite-coated poly(lactic-co-glycolic acid) (PLGA) scaffolds were fabricated from 85/15 poly(lactic-co-glycolic acid) by solvent casting and a particulate leaching process as previously described . For BMP-2-loaded scaffolds, rhBMP-2 (R&D Systems, Minneapolis, MN, http://www.rndsystems.com/) was adsorbed by dropping the protein solution onto scaffolds for 20 minutes followed by further lyophilization in a freeze drier (Labconco, Kansas City, MO). BMP (0.50, 0.75, and 1.25 μg) was applied to our scaffolds at final concentrations of 50, 100, and 200 μg/ml, respectively. For lentiviral particle releasing scaffolds, 10 μl of virus solution was mixed with 1 μl of neutralized 0.25% type I collagen (Vitrogen; Palo Alto, CA, http://www.biocompare.com/CompanyProfile/3653/Vitrogen-Collagen-Corporation.html) at 4°C. The mixture was dropped onto fabricated scaffolds, dried for 20 min, and further lyophilized in a freeze drier.
Creation of Calvarial Defects
Nonhealing, critical-sized (4 mm) calvarial defects were created in the right parietal bone of adult (60-day-old) male CD-1 nude mice as previously described . In preparation for cell engraftment, scaffolds were seeded with hASCs 24 hours prior to implantation. Cells (150,000) were placed on scaffolds in 125 μl of medium in 96-well culture plates and incubated for 24 hours. Before implantation, scaffolds were copiously rinsed with phosphate buffered saline (PBS).
In Vivo Imaging
MicroCT was performed on live animals in a serial manner postoperatively (through 6 weeks healing) using a high-resolution MicroCAT II (ImTek Inc., Knoxville, TN, http://www.imtek.com/) small animal imaging system, as previously described . For In Vivo Imaging System (IVIS), luciferin (150 mg/kg in 200 μl) was injected into the peritoneal cavity. After 10 minutes, animals were then placed in the IVIS 200B imaging system and imaged for 3 minutes.
Animals were sacrificed from 5 days to 6 weeks postoperatively for histology. Histology and histomorphometry were performed as previously described, including Aniline blue, pentachrome, and alkaline phosphatase (ALP) staining . In situ hybridization was performed on select slides for Bmp2, Runx2, Ocn, and Noggin . Nonspecific binding was minimized by high stringency hybridization conditions. For all assays, sense probes were used side-by-side with minimal background. Cellular proliferation was assessed by bromodeoxyuridine (BrdU) immunostaining according to manufacturer specifications (Zymed, South San Francisco, CA, now a part of Invitrogen, Carlsbad, CA, http://www.invitrogen.com/site/us/en/home.html). BrdU labeling reagent was administered 8 hours prior to mouse sacrifice.
Immunohistochemistry of Osteogenic Proteins, Smad Pathway, and Human Nuclear Antigen
Slides were deparaffinized, rehydrated, quenched, and blocked with 10% goat serum. Antibodies used included rabbit polyclonal anti-pSmad1, rabbit polyclonal anti-Ocn, Runx-2, and Bmp-2 (Santa Cruz Biotechnology, Santa Cruz, CA, http://www.scbt.com/) in 1:80 dilution with 1% rabbit serum. Appropriate biotinylated secondary antibodies were used in 1:1,000 dilution (Vector Laboratories, Burlingame, CA, http://www.vectorlabs.com/). Visualization was with diaminobenzidine (Invitrogen, Carlsbad, CA, http://www.invitrogen.com/site/us/en/home.html).
Immunofluoresence for Colocalization of Human Nuclear Antigen and Osteocalcin
Slides were incubated with primary antibodies at 4°C overnight, followed by incubation with an Alexa Fluor 594 goat anti-rabbit secondary antibody (Invitrogen, 1:400, Carlsbad, CA, http://www.invitrogen.com/site/us/en/home.html) and an Alexa Fluor 488 anti-mouse secondary antibody for 1 hour. Coverslips were applied and analyzed with fluorescence. Slides without primary antibody and slides with nonspecific monoclonal antibodies were used as negative controls.
Subconfluent mASCs or hASCs were washed twice with 1× PBS and starved in serum-free medium overnight. Cells were then washed and lysed with cold lysis buffer (50 mmol/l of HEPES, pH 7.5, 150 mmol/l of NaCl, 1 mmol/of EDTA, 10% glycerol, 1% Triton-X-100, and 25 mmol/l of sodium fluoride) containing 1 mmol/l of sodium orthovanadate and Protease Inhibitor Cocktail (Sigma-Aldrich). Aliquots (50–100 μg) of cell lysate were electrophoresed on 12% Tris-HCl sodium dodecyl sulfate (SDS)-polyacrylamide gel electrophoresis (PAGE) gels and transferred onto an Immobilon-P membrane (Millipore Corporation, Bedford, MA, http://www.millipore.com/). Antibodies against the specific phosphorylated and nonphosphorylated proteins of each signaling pathway were chosen as follows: anti-phospho-Smad1/5, Smad1, and Smad5. A horseradish peroxidase-conjugated anti-rabbit antibody (1:8,000) was used as a secondary antibody. Immunoblotted products were visualized by enhanced chemiluminescence substrate (GE Healtcare Life Sciences, http://www.gelifesciences.com). All bands were normalized with the loading controls (α-tubulin) and quantified by densitometry.
Means and standard deviations were calculated from numerical data, as presented in the text, figures, and figure legends. In figures, bar graphs represent means, whereas error bars represent one standard deviation. Statistical analysis was performed using an appropriate analysis of variance (ANOVA) when more than two groups were compared. A Student's t test was used to directly compare two groups, and a Bonferroni correction was used if three or more groups were compared. The exact statistical analysis for each dataset is described in the figure legends. Inequality of standard deviations was excluded by using the Levene's test. p values are included in the figure legends.
Evaluation of Noggin Suppression
To demonstrate the effect of Noggin knockdown in ASCs, we first used primary ASCs harvested from a transgenic mouse homozygous for a noggin allele flanked by loxP sites . Cre-mediated DNA recombination was induced in these cells by treating the mother of the pups with Tamoxifen. Protein analysis demonstrated enhanced pSmad1/5 levels along with reduced noggin levels (Supporting Information Fig. 1a–1d). Similarly, on a gene level we noted noggin knockdown in mASCs was associated with an upregulation of osteogenic genes after 3 days of treatment with ODM (Supporting Information Fig. 1e). The enhanced BMP and osteogenic signaling led to an increase in early osteogenic differentiation as shown by ALP stain and quantification (Supporting Information Fig. 1f, 1g) as well as by extracellular matrix mineralization demonstrated by Alizarin red (Supporting Information Fig. 1h, 1i). To enhance the translational implications of this study, Noggin knockdown was next performed in hASCs (Supporting Information Fig. 2). RNA interference (RNAi) was first performed using a NOG-directed shRNA construct delivered by a lentiviral vector (Santa Cruz Biotechnology, Santa Cruz, CA, www.scbt.com). Infection efficiency was evaluated by three methods including (1) GFP expression in GFP-targeted shRNA control lentiviral particles, (2) NOG gene expression by qRT-PCR after lentiviral infection of Noggin lentiviral particles, and (3) NOG protein expression as assessed by Western blot. By all three methods, greater than 90% knockdown efficiency was verified (Supporting Information Fig. 2).
Noggin Suppression Enhances BMP-2 Signaling
To examine downstream signaling of BMP ligand–receptor interactions, we next examined phosphorylation of Smad1/5 . Suppression of NOG stimulated a significant increase in pSmad1/5 with similar levels of total Smad5 (Fig. 1A–1C). Moreover, NOG knockdown resulted in a striking increase in osteogenic differentiation among hASCs. This included significant increases in osteogenic gene expression of Runt-related transcription factor-2 (RUNX2), ALP, osteocalcin (OCN), and collagen 1A (COL1A) at baseline and after 3 and 14 days of ODM treatment (Fig. 1D–1F). With regards to osteogenic differentiation, at 3 days, NOG knockdown resulted in increased ALP stain and over four times more ALP activity when normalized to protein content (*, p < .05; Fig. 1G, 1H). Similarly, NOG shRNA cells deposited significantly more mineralized extracellular matrix after 7 days of differentiation (Fig. 1I), with a fivefold increase noted on quantification (*, p < .05; Fig. 1J).
Decreased Noggin Expression Enhances Osteogenic Differentiation of hASCs In Vivo
Having demonstrated enhanced in vitro hASC osteogenesis with Noggin inhibition, we next turned to a critical-sized in vivo skull defect model. Human ASCs treated with either Noggin shRNA or GFP control shRNA infection were placed on scaffolds and implanted into these defects [15, 18]. On postoperative day 5, specimens were examined to assess the early wound environment (Fig. 2). Noggin knockdown led to a clear enhancement in BMP ligand expression and BMP signaling activity, as demonstrated by in situ hybridization for Bmp2 and immunohistochemical staining for Bmp2 and pSmad1 (Fig. 2A, 2B). This enhanced BMP activity was accompanied by a global upregulation of osteogenic markers, including Runx2 and Ocn on both gene and protein levels (Fig. 2A, 2B). Collectively, these data suggested that Noggin knockdown augments the osteogenic environment of a skeletal defect, with an increase in BMP signaling and associated osteogenic gene expression. As hASCs undergo osteogenic differentiation, their proliferative capacity often slows. Thus, we also examined the proliferation of Noggin shRNA- and control shRNA-treated cells, demonstrating that the control shRNA cells displayed increased proliferation, by cell counting and BrdU incorporation in vitro (Supporting Information Fig. 3a, b). Similarly, BrdU incorporation was increased among the control shRNA group in vivo (Supporting Information Fig. 3c)
hASCs with Noggin Knockdown Demonstrate Enhanced In Vivo Healing of Calvarial Defects
Having demonstrated increased osteogenic differentiation of hASCs in vitro and increased osteogenic signaling in vivo, we next set out to compare the effect of Noggin on skeletal healing when followed over time. To demonstrate that the implanted hASCs persisted, we performed luciferase imaging and immunofluoresence for GFP using Luciferase/gfp transduced hASCs (data not shown and Supporting Information Fig. 4a). In addition, we performed Immunohistochemistry (IHC) for human nuclear antibody (Supporting Information Fig. 4b). To determine colocalization of human cells and osteogenic proteins, we next performed immunofluoresence for both human nuclear antibody and osteocalcin 6 weeks after defect creation and cell/scaffold implantation. As shown by confocal microscopy, there was strong nuclear staining, indicating that the cells are of human origin, and a positive cytosolic stain for human OCN in the area of new bone formation (Supporting Information Fig. 4d). These studies document persistence of cells over time and differentiation of engrafted cells into an osteogenic lineage. Using our previous in vivo model, we then set out to demonstrate that rhNoggin would inhibit hASC osteogenesis . We first made a 4-mm critical-sized defect in the parietal bone of a nude CD-1 mouse. Subsequently, we placed either the scaffold alone, the scaffold with control GFP-transduced hASCs, or the scaffold with hASCs and rhNoggin injections (200 μg/ml of rhNoggin injected into the calvarial defect for 5 consecutive days following hASC treatment as rhNoggin protein was too large to load onto our scaffold). The rhNoggin-treated defects demonstrated less than 20% healing even when followed up to 6 weeks postoperatively which was over 40% less than our control hASCs (Supporting Information Fig. 5a, 5b).
Once we confirmed the inhibitory effect of Noggin in vivo, we next analyzed the effect of Noggin knockdown on osseous healing (Fig. 3). As expected, the no scaffold group demonstrated minimal healing (Fig. 3A, top row). In contrast, as early as 1 week after implantation, Noggin shRNA hASC-engrafted defects exhibited large amounts of bony regenerate (Fig. 3A, bottom row). By 3 weeks, the majority of the Noggin knockdown hASC-engrafted defects displayed robust healing (Fig. 3A, bottom row) and overall Noggin knockdown hASC-engrafted defects outperformed each control group at all time points examined (Fig. 3B). Consistent with radiographic analysis, Noggin knockdown hASC-engrafted defects showed significant bony regenerate at 6 weeks (as shown by Aniline blue, pentachrome, and ALP staining; Fig. 3C, bottom row). In contrast, groups with no scaffold, scaffold alone, and GFP control shRNA hASCs showed significantly less ossification (Fig. 3C, rows 1–3). In addition to the amount of osteoid, the Noggin shRNA cell-seeded scaffolds demonstrated thick bone throughout the entire expanse of the scaffold (Fig. 3C, bottom row) . Histomorphometry at 6 weeks was performed to quantify the amount of bone formed (as opposed to computed tomography (CT) scans which demonstrate % original defect covered by bone), showing a significant increase in bony regenerate in mice receiving NOG shRNA (Fig. 3D).
BMP-2 Enhances the Osteogenic Potential of hASCs With or Without Noggin Knockdown In Vitro
Our ultimate translational goal is to maximally stimulate the BMP pathway to accelerate osteogenic differentiation of hASCs. Thus, along with suppressing Noggin, we also treated hASCs with rhBMP-2 to determine if we could further augment osteogenic differentiation by simultaneously decreasing a BMP antagonist and providing BMP-2. BMP-2 supplementation corresponded to increased pSmad1/5 signaling by Western blot analysis (data not shown) and increased calcification in vitro as shown by Alizarin red staining in the control shRNA and Noggin shRNA groups (Supplemental Fig. 6a–6c). This was also accompanied by increased expression of osteogenic genes (Supporting Information Fig. 6d–6f). Thus, the addition of BMP-2 to Noggin knockdown further promoted BMP signaling and osteogenic differentiation beyond that seen with noggin shRNA treatment alone.
Slow Released BMP-2 Enhances Osteogenic Healing of a Critical-Sized Calvarial Defect In Vivo
We have previously used rhBMP-2 injection to augment hASC osteogenesis; however, daily injection of rhBMP-2 is not cost effective or clinically practical . Thus, we designed a biomimetic rhBMP-2-releasing scaffold that allows for the slow release of BMP-2. We demonstrated that BMP-2 cumulative release continues to increase through 30 days in vitro (Supporting Information Fig. 7). We then assessed optimal dosing of rhBMP-2 release from our scaffold using an in vivo dose curve and found 200 μg/ml of rhBMP-2 stimulated the most rapid and robust bone formation (Supporting Information Fig. 8). Next we compared the healing seen if hASCs were added to the BMP-2-releasing scaffold at each dosage as previous studies have failed to demonstrate enhanced healing with hASCs over BMP-2 scaffold alone [22, 23]. We noted that BMP-2 scaffolds with 150,000 hASCs had improved healing over scaffold alone at all dosages investigated (Supporting Information Fig. 8b–8e). Thus, although a slow releasing rhBMP-2 scaffold alone does augment healing even without hASCs, the ideal treatment would include cell-based therapy on a maximally osteoconductive/osteoinductive scaffold.
BMP-2 Signaling and Repair of a Critical-Sized Calvarial Defect Is Enhanced by Noggin Suppression
We next set out to combine these two methodologies of noggin knockdown and BMP-2 supplementation to fully drive the BMP-2 pathway and thus skeletal repair. We first assessed BMP and osteogenic signaling by immunohistochemistry. After 5 days in vivo, we found enhanced Smad1 levels and osteogenic transcripts Runx2 and Ocn by immunohistochemical staining (Fig. 4A). Consistent with this increase in osteogenic signaling, there was significantly more skeletal healing in BMP-2 scaffolds seeded with Noggin shRNA transduced hASCs (Fig. 4B–4E). Quantification showed that the BMP-200 μg/ml scaffolds seeded with noggin shRNA-transduced hASCs had 100% percent healing as early as 4 weeks (Fig. 4E). These CT scans correlated with histology and histomorphometry as those defects treated with Noggin shRNA-transduced hASCs on a 200 μg/ml scaffold had the most robust osteoid formation spanning the entire scaffold across the length of the defect (Fig. 4F–4J).
Novel Minicircle Noggin shRNA Knockdown Improves Osteogenesis In Vitro and In Vivo
Although lentiviral Noggin shRNA transduction resulted in a marked increase in osteogenic differentiation of hASCs, our ultimate goal to bring this technology to the bedside requires a nonintegrating technology. To overcome the risk of insertional mutagenesis associated with integrating strategies, we designed a minicircle with three different shRNA constructs (same as those in our lentiviral particles) as well as a GFP construct (Fig. 5A, 5B) [14, 24]. Subsequent sorting for GFP-positive cells (demonstrating Noggin shRNA transfection) allowed for enrichment of the transfected population to over 80% (Fig. 5C, 5D). These cells were next analyzed for expression of Noggin and osteogenic differentiation. Minicircle transfection led to over 70% knockdown in noggin gene transcript and over 40% knockdown of NOG protein (Fig. 5E, 5F). Similar to our lentiviral technique, Noggin shRNA delivery using our minicircle construct also significantly enhanced osteogenic differentiation by ALP and Alizarin Red staining, as well as osteogenic gene expression (Fig. 5G–5J). Finally, we applied the Noggin minicircle-transfected cells to our in vivo model and demonstrate that similar to the lentiviral-infected cells, these cells significantly accelerated healing of a calvarial defect compared to minicircle gfp control cells (Fig. 5K, 5L). Furthermore, on histological sectioning, the osteoid appeared thick and robust, as seen with lentiviral knockdown of Noggin (Figs. 5M and 3C). Thus, we have used a nonintegrating minicircle plasmid to successfully knockdown noggin and enhance the osteogenic capacity of hASCs in vitro and in vivo.
Noggin Lentiviral Delivery by Adsorption to Our Scaffold
As a way to address the shortcoming of having to transduce hASCs ex vivo prior to implantation, we set out to use our scaffold to both knockdown noggin and release BMP-2. To do this, we created a scaffold with embedded Noggin lentiviral particles to allow for noggin knockdown of hASCs at the time of implantation through the scaffold. From a clinical standpoint, this would be ideal as the cells could be implanted directly onto the scaffold immediately after harvest in the operating room. This strategy using in vivo vector delivery would likely affect engrafted cells but could also have a similar effect on surrounding osteocompetent cells such as periosteum and duramater [25, 26]. We adsorbed the same Noggin shRNA and control GFP shRNA particles used to transduce our cells in vitro onto our scaffolds with and without BMP-2 using previously described techniques . Using our GFP control, we found the transduction efficiency to be 86% in vitro (data not shown) and 89% in vivo as the large majority of the cells in the defect with the GFP shRNA-releasing scaffold stained positive for GFP (Fig. 6A, bottom row). Noggin knockdown was confirmed by performing qRT-PCR of the scaffold 1 week after implantation for human NOG and mouse nog as well as mouse and human osteogenic genes (Fig. 6B, 6C). This noggin knockdown was further verified in vivo by performing NOGin situ hybridization, as a greater intensity of stain was noted in defects treated with the control shRNA-releasing scaffold with hASCs compared to the Noggin shRNA-releasing scaffold with hASCs (Fig. 6D). A similar trend was seen when comparing immnohistochemistry for NOG in the region of the defect (Fig. 6E, first column). Along with decreased immunostaining for NOG, we noted an increase in staining for BMP-2, pSmad1, Runx-2, and Ocn, as was previously seen when Noggin shRNA was transduced in vitro prior to implantation (Fig. 6E, second through fifth columns). Interestingly, although, at 6 weeks following implantation, immunostaining for Noggin revealed increased immunoreactivity in defects treated with Noggin shRNA scaffolds with hASCs relative to the 1 week time point (Fig. 6E, far right). This suggests that in vivo noggin knockdown may not persist.
As a final approach to provide BMP-2 and knockdown of Noggin without manipulating the cells in vitro, we placed freshly harvested hASCs on a BMP-2-releasing, NOG shRNA scaffold to treat a critical-sized defect. Similar to our cells that had prior Noggin knockdown in vitro followed by seeding on a BMP-2 scaffold (Fig. 4B, bottom), we found robust healing as early as 1 week and complete healing of all defects by 4 weeks (Fig. 6F, bottom row and 6G). Thus, we believe that the scaffold can be designed to simultaneously release an osteogenic cytokine (such as BMP-2) and to knockdown targeted genes such as Noggin in both host tissues and implanted cells.
Tissue engineering approaches for skeletal regeneration require the use of osteoinductive scaffolds and osteocompetent cells used in a manner that minimizes costs and maximizes safety and efficacy. Our laboratory has previously shown that noggin suppression in osteoblasts enhances osteogenesis in vitro and in vivo . When looking toward clinical applications, however, osteoblasts are limited in their availability and result in large secondary defects when harvested. In contrast, hASCs are an easily accessible and abundant source of mesenchymal stem cells with a robust osteogenic capacity in vitro and in vivo [15, 18, 28]. Clinical studies have demonstrated the use of ASCs to successfully treat craniofacial defects [29–31]. These reports, however, lack consistency both in their application and in the scaffolding materials on which they are implanted. BMPs have also been used clinically to promote bone formation, with Infuse, a collagen sponge with rhBMP-2 protein, frequently used in spine surgeries . Although some have raised concern over BMP's high price, studies have demonstrated improved patient outcomes with a similar total cost following spinal surgery with or without BMP-2 .
Before hASCs can be effectively used in large clinical trials, we believe it is crucial to both maximize their osteogenic potential and the osteoconductive, osteoinductive niche provided by the scaffold. To enhance the osteogenic scaffold, we used our previous Hydroxyappatite (HA)-coated PLGA core scaffold and incorporated a slow release of BMP-2. This BMP-2 scaffold augmented the osteogenic capability of untreated hASCs in a dose-dependant manner, in addition to significantly promoting healing of skeletal defects treated with noggin shRNA hASCs. BMP-2, however, is expensive and has a significant risk profile. It may thus be beneficial to lower the dose of BMP while still stimulating the BMP pathway through other mechanisms such as removing an inhibitor. Noggin is a known antagonist of BMP-2, and we demonstrated that increased osteogenic differentiation can occur by placing cells directly on a scaffold that releases both noggin shRNA and BMP-2. In previous studies, the BMP-2 pathway has been manipulated in cells prior to implantation on a scaffold [33, 34]. In this study, however, we also used the scaffold to simultaneously deliver noggin-directed shRNA and BMP-2. While this resulted in early knockdown of noggin and accelerated bone regeneration, long-term noggin suppression did not occur, as we found increased Noggin levels 6 weeks following implantation of the scaffold. This observation may be secondary to viral inactivation and diminished RNAi activity, which has also been noted in other works . Nonetheless, the ability of a scaffold to manipulate implanted cells and the surrounding host tissue has significant clinical implications as it eliminates the ex vivo period prior to transplantion back into the patient.
Although we have explored the capacity of Noggin to function as a competitive inhibitor for BMP-2, this protein no doubt interacts with several other pathways. For example, BMP inhibition through exogenous noggin supplementation has been shown to interfere with endothelial migration [35, 36]. Previous studies have demonstrated that overexpression of Noggin in human umbilical vein endothelial cells (HUVECs) inhibits angiogenic capacity in vitro and in vivo . Mechanistically, it is believed that Noggin mitigates BMP-initiated ascular endothelial growth factor (VEGF) promotor activation . Thus, it is possible that Noggin suppression in ASCs might further regulate the local vasculogenic niche, which will be the focus of future studies.
Most current studies using shRNA knockdown employ the use of lentiviruses which can become reactivated and are therefore not clinically translatable . Plasmids, although potentially safer, suffer from low transfection efficiencies and labor-intensive production . Localized gene therapy using adenoviruses offers another strategy to manipulate BMP-2 levels [33, 34]. Studies using this approach have shown equivalent effect of adenoviral-mediated BMP expression and exogenous BMP-2 supplementation [33, 34]. However, in adenoviral-mediated gene therapy the viral genome can persist episomally in the infected cells, which could potentially give rise to life-long transgene expression in vivo. Furthermore, adenoviruses interact readily with antigen-presenting cells which could trigger an innate immune response, mitigating vector efficacy. Finally, the potential exists for the development of leukemia due to uncontrolled clonal proliferation following adenoviral infection [40–42]. Based on all of these findings, we thus felt that a BMP-2-releasing scaffold would have the safest risk profile to enhance the BMP signaling pathway. Alternatively, minicircles may also be used to manipulate BMP signaling. As an efficient, nonintegrating technology, minicircle vectors benefit from higher transfection efficiency over traditional plasmids. They also have improved safety over adenoviruses and lentiviruses, as discussed above . Here, we used a scalable minicircle that allows for rapid integration of our shRNA and GFP constructs to allow for transfection, enrichment, and knockdown. We believe this ability to deliver both an shRNA for our gene of interest as well as a GFP label can be further applied to other tissue engineering applications.
We demonstrate that hASCs can undergo enhanced osteogenic differentiation by manipulation of the BMP pathway through noggin. Using integrating lentiviral and nonintegrating scalable minicircle methods to knockdown noggin, we created a population of hASCs that possess superior osteogenic potential in vitro and in vivo. By either enhancing cells ex vivo with a knockdown approach or in vivo with a scaffold designed to release noggin shRNA and BMP, we can optimize the potential utility of cell-based therapies in skeletal regeneration.
This work is supported by grants from the National Institutes of Health, National Institute of Dental and Craniofacial Research grants 1 R21 DE019274-01 and RC2 DE020771-01, the Oak Foundation and Hagey Laboratory for Pediatric Regenerative Medicine to M.T.L. J.C.W. was supported by R01EB009689 and RC1HL099117, R33HL089027. B.L. was supported by the National Institutes of Health, National Institute of Arthritis, and Musculoskeletal and Skin Diseases grant 1F32AR057302-02. The inducible Nog/Cre transgenic mice were a kind gift of Lisa Brunet and the Richard Harland Laboratory at the University of California at Berkeley. Genotyping and colony maintenance was performed with the expert assistance Kristine Rustad, B.S.
DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
The authors indicate no potential conflicts of interest.