Nonvirally Engineered Porcine Adipose Tissue-Derived Stem Cells:Use in Posterior Spinal Fusion

DIMA SHEYN,a GADI PELLED,a YORAM ZILBERMAN,a FARAHNAZ TALASAZAN,b JONATHAN M. FRANK,b

DAN GAZIT,a,b ZULMA GAZITa,b

aSkeletal Biotechnology Laboratory, Hebrew University-Hadassah Medical Center, Jerusalem, Israel; bDepartment ofSurgery, Cedars Sinai Medical Center, Los Angeles, California, USA

Key Words. Adipose tissue-derived stem cells • Nonviral gene delivery • Genetic engineering • Bone morphogenetic protein •Posterior spinal fusion

ABSTRACT

Multiple factors alter intervertebral disc volume, structure,shape, composition, and biomechanical properties, oftenleading to low back pain. Spinal fusion is frequently per-formed to treat this problem. We recently published resultsof our investigation of a novel system of in vivo bone for-mation, in which we used nonvirally nucleofected humanmesenchymal stem cells that overexpress a bone morphoge-netic protein gene. We hypothesized that primary porcineadipose tissue-derived stem cells (ASCs) nucleofected withplasmid containing recombinant human bone morphogeneticprotein-6 (rhBMP-6) could induce bone formation andachieve spinal fusion in vivo. Primary ASCs were isolatedfrom freshly harvested porcine adipose tissue. Overexpres-sion of rhBMP-6 was achieved ex vivo by using a nucleofec-tion technique. Transfection efficiency was monitored byassessing a parallel transfection involving an enhanced

green fluorescent protein reporter gene and flow cytometryanalysis. rhBMP-6 protein secreted by the cells was mea-sured by performing an enzyme-linked immunosorbent as-say. Genetically engineered cells were injected into the lum-bar paravertebral muscle in immunodeficient mice. In vivobone formation was monitored by a quantitative microcom-puted tomography (�CT). The animals were euthanized 5weeks postinjection, and spinal fusion was evaluated usingin vitro �CT and histological analysis. We found formationof a large bone mass adjacent to the lumbar area, whichproduced posterior spinal fusion of two to four vertebrae.Our data demonstrate that efficient bone formation andspinal fusion can be achieved using ex vivo, nonvirally trans-fected primary ASCs. These results could pave the way to anovel biological solution for spine treatment. STEM CELLS2008;26:1056–1064

Disclosure of potential conflicts of interest is found at the end of this article.

INTRODUCTION

Low back pain is a common complaint voiced by the adultpopulation [1–3]. Multiple elements, including aging and ge-netic and environmental factors, contribute to disc degeneration.As people age, the intervertebral disc undergoes alterations involume, structure, shape, composition, and biomechanical prop-erties [2]. Although the etiology and pathophysiology of discdegeneration remain unknown [2, 3], disc degeneration can bedescribed clinically as a loss of two major disc functions:stability and mobility [4, 5]. Pathological conditions, such asspondylosis, scoliosis, spondylolisthesis, tumor, infection, post-traumatic fracture, and instability, can produce an abnormalrelationship between adjacent vertebral structures. Fusion mod-els have been designed to stabilize the spinal column by remov-ing intervertebral articulations and repositioning vertebral seg-ments in appropriate, mechanically advantageous alignment.Spinal fusion ranks as the second most common lumbar spineprocedure, with approximately 46,500 lumbar spinal arthrodesesperformed each year in the United States. Autologous bone graftderived from the iliac crest is the standard procedure used forspinal fusion; nevertheless, the number of complications involv-

ing surgical harvest of this graft material is high, and theincidence of morbidity is estimated at 7%–25% [6].

Adipose tissue-derived stem cells (ASCs) [7] are excellentcandidates for stem cell-based therapy for bone regenerationbecause they have the potential to differentiate into variousmesenchymal lineages, including bone-forming cells [8, 9]. Ourpreliminary studies showed the ability of ASCs to differentiatein vitro into osteogenic and chondrogenic lineages (data notshown). Both human bone marrow-derived mesenchymal stemcells (BM-MSCs) and ASCs have proved effective in regener-ating bone defects in animals. Genetically engineered BM-MSCs were shown to regenerate nonunion fractures in nudemice [10, 11]. Quarto et al. [11] used autologous BM-MSCsdelivered on a macroporous hydroxyapatite carrier to heal largebone defects (�4.0 cm) and reported preliminary success inachieving callus formation after 2 months. Dragoo et al. [12, 13]achieved bone induction in NOD/SCID mice by using humanASCs expressing recombinant human bone morphogenetic pro-tein-2 (rhBMP-2). The ability of rhBMP-2-expressing humanASCs to heal critically sized bone defects in the nude rat femurhas also been demonstrated [14]. Recently, autologous humanASCs were shown to induce bone formation and regeneration ina 7-year-old girl with multiple calvarial fractures [15]. The

Correspondence: Zulma Gazit, Ph.D., Skeletal Biotech Laboratory, Hebrew University of Jerusalem–Hadassah Medical Campus, Jerusalem91120, Israel. Telephone: 972-2-6757625; Fax: 972-2-6757628; e-mail: zulmag@ekmd.huji.ac.il Received November 1, 2007; accepted forpublication January 10, 2008; first published online in STEM CELLS EXPRESS January 24, 2008. ©AlphaMed Press 1066-5099/2008/$30.00/0doi: 10.1634/stemcells.2007-0858

TRANSLATIONAL AND CLINICAL RESEARCH

STEM CELLS 2008;26:1056–1064 www.StemCells.com

simplicity of the surgical procedure, the easy and repeatableaccess to subcutaneous adipose tissue, and the uncomplicatedenzyme-based isolation procedures make this source of autolo-gous adult MSCs most attractive to researchers and clinicians[16]. Only local anesthesia is necessary to harvest ASCs, thepatient is subjected only to minimal discomfort, and large quan-tities of these cells can be obtained repeatedly.

Bone morphogenetic proteins (BMPs) are members of thetransforming growth factor-� superfamily and are known fortheir potential to induce bone formation and osteogenic differ-entiation of MSCs [10, 17–20]. Various BMPs exert differenteffects on bone formation in vivo following adenovirus-medi-ated direct gene transfer [21–23]. Recently, it was shown thatrhBMP-2, -6, and -9 are the most potent osteogenic inducers ofMSCs among the BMP family [24]. Previous studies haveindicated that human BM-MSCs nonvirally transfected with therhBMP-9 gene were able to induce prominent bone formation invivo [6].

Nonviral gene delivery approaches hold great promise forthe development of efficient and safe cell-mediated gene ther-apy for clinical orthopedic applications. Viral gene delivery isefficient but less safe than nonviral gene delivery. There is thusan enormous need to develop efficient novel nonviral vectors sothat we may advance gene and cell therapy to the clinicalsetting. Furthermore, nonviral vectors have greater potential toprovide nucleic acid-based drugs that closely resemble tradi-tional pharmaceutical agents. Nucleofection, a novel form ofelectropermeabilization, is an efficient physical method of de-livery of plasmid DNA into primary cells and established celllines [25, 26]. Recently, it was shown that nucleofection, devel-oped by Amaxa Biosystems (Cologne, Germany, http://www.amaxa.com), efficiently transfects human BM-MSCs with areporter gene [26] and a therapeutic gene from the BMP familyand that the transfected cells can be used to create ectopic boneformation in vivo [6].

On the basis of our earlier findings, we hypothesized thatprimary porcine adipose tissue-derived stem cells that have beennucleofected with a plasmid encoding the rhBMP-6 gene willexpress and secrete biologically effective levels of rhBMP-6 andthat these cells can induce posterior spinal fusion after theirdirect injection into the vicinity of the intervertebral disc. Thisinjectable treatment is minimally invasive and more effectivethan bone marrow transplantation.

MATERIALS AND METHODS

Isolation of Porcine ASCsPorcine ASCs were isolated from adipose tissues obtained from thesubcutaneous fat pads of euthanized adult female Yorkshire pigs19–20 weeks old. The adipose tissues were cut into small pieces,placed on a shaker, and treated with 0.075% collagenase (Sigma-Aldrich, St. Louis, http://www.sigmaaldrich.com) in phosphate-buffered saline (PBS) at 37°C for 40–60 minutes. The collagenasewas inactivated by addition of complete growth medium (Dulbec-co’s modified Eagle’s medium [DMEM] containing 10% fetal calfserum [FCS]; Biological Industries, Beth Haemek, Israel, http://www.bioind.com). The mononuclear cells were plated in tissue-culture dishes at a density of 10 � 106 to 15 � 106 cells per 100-mmculture plate, in 5% CO2/95% air at 37°C. The medium waschanged after 72 hours and again every 3–4 days. The ASCs wereexpanded in culture up to the fifth passage.

Nucleofection and the Establishment of a NonviralGene Delivery ProtocolWith the aid of nucleofection (Amaxa Biosystems) technology,including a Nucleofector device and an MSC-specific nucleofection

buffer, 106 or 2 � 106 porcine ASCs were transfected with 5, 10, or15 �g of pCMV-EGFP-N1 (hereafter referred to as pEGFP), whichencodes the enhanced green fluorescent protein (EGFP) gene, ac-cording to the manufacturer’s protocol for human MSCs. Thenucleofection procedure was performed using program G22, whichwe found to be optimal in our previously reported preliminarystudies [6]. Immediately after nucleofection, the cells were trans-ferred to 100-mm plates, which contained 10 ml of complete growthmedium (including 20% FCS), and were maintained in culture for24 hours. Viable cells were counted after trypan blue staining byusing light microscopy 24 hours post-transfection. Cell viability wascalculated as the ratio of surviving cells to cells initially nucleo-fected, as previously reported [6]. Nucleofected cells were analyzed24 hours after transfection by performing a flow cytometry analysisfor the expression of green fluorescent protein (GFP) and by deter-mining the percentage of nucleofected cells among the total viablecells. After the nucleofection conditions were calibrated, the ASCswere nucleofected with pCMV-cDNA3-rhBMP-6 (hereafter re-ferred to as prhBMP-6).

Real-Time Reverse Transcriptase-Polymerase ChainReactionTo estimate the rhBMP-6 gene expression profile, ASCs from threedifferent donors were nucleofected by following the same methoddescribed earlier and grown in confluent conditions in culture for 21days. Total RNA was extracted from nucleofected porcine ASCs byusing Trizol reagent on days 1, 2, 4, 7, 10, 14, and 21 afternucleofection. Total RNA was first treated with DNase I (Promega,Madison, WI, http://www.promega.com) and then retrotranscribedusing random hexamers and reverse transcriptase (RT) (Promega);rhBMP-6 gene expression was quantitatively analyzed using real-time RT-polymerase chain reaction (PCR). Real-time quantitativePCR was performed with the aid of an ABI 7300 Prism system(Applied Biosystems, Foster City, CA, http://www.appliedbiosystems.com) by using an absolute quantification method and Assay-on-Demand gene expression assays (Applied Biosystems). The numberof transcript copies was determined using a standard curve.

In Vitro Protein Secretion AssayTo estimate the amount of rhBMP-6 that was secreted, ASCs fromthree different donors were nucleofected using the method de-scribed above and grown in confluent conditions in culture for 21days. Media from the flasks were collected on days 1, 2, 4, 7, 10, 14,and 21 after nucleofection. To quantify the rhBMP-6 secretedduring 24 hours, the media were changed 24 hours before samplingat every time point tested. An enzyme-linked immunosorbent assay(R&D Systems Inc., Minneapolis, http://www.rndsystems.com) wasperformed to measure the amount of rhBMP-6 protein secreted intothe culture media by the nucleofected porcine ASCs. Aftersampling the media, we lifted and counted cells so that we couldnormalize the secreted protein to 106 cells.

Alkaline Phosphatase Activity In VitroTo induce osteogenic differentiation in vitro, adipose tissue-derivedASCs were nucleofected with prhBMP-6 and plated at a density of7 � 104 cells per cm2 in maintenance DMEM containing 10% FCS,0.05 mM ascorbic acid-2-phosphate, and 10 mM �-glycerophos-phate (Sigma-Aldrich). To provide a positive control of osteogenicinduction, non-nucleofected cells were cultured in an inductionmedium consisting of maintenance medium plus 0.1 �M dexameth-asone as an induction agent. To provide negative controls, ASCswere nucleofected with the pEGFP reporter, and non-nucleofectedcells were cultured in maintenance medium only. Six days afternucleofection or addition of osteogenic supplements, an alkalinephosphatase (ALP) colorimetric assay was performed as previouslydescribed [27]. Briefly, the cells were lysed with alkaline buffersolution (Sigma-Aldrich) containing 0.5% Triton X-100 and 10 mMMgCl2. For the ALP assay, the cell lysates were incubated withassay buffer containing 0.75 M 2-amino-2-methyl-1-propranolol(pH 10.3) for 10 minutes at 37°C with p-nitrophenylphosphate as asubstrate.

1057Sheyn, Pelled, Zilberman et al.

www.StemCells.com

Animal StudiesOur institutional animal care and use committee approved all theprocedures used in this study and agreed that care of the animalswas consistent with that outlined in the United States NationalInstitutes of Health Guide for the Care and Use of LaboratoryAnimals. For the spinal fusion procedure, immunodeficient (NOD/SCID) mice were anesthetized by administration of an intraperito-neal injection of a xylazine-ketamine mixture [28] (100 mg/kgketamine and 3.3 mg/kg xylazine). Nucleofected porcine ASCs(5 � 106) were resuspended in 50 �l of fibrin gel (Tisseel kit;Baxter, Vienna, Austria, http://www.baxter.com) and injected intothe lumbar paravertebral muscle of each mouse as previously de-scribed [20]. As a control group, nontreated ASCs were injected inthe same manner. The mice were housed in a specific pathogen-freeanimal facility for 5 weeks. In the donor cell contribution experi-ment, the samples were harvested after 2 and 5 weeks. At the end ofthis period, the mice were euthanized, their fused spines wereharvested, and bone formation and spinal fusion in these animalswere analyzed.

Microcomputed TomographyTo obtain a detailed qualitative and quantitative three-dimensional(3D) analysis, formation of the fusion mass was evaluated usingmicrocomputed tomography (�CT). Five weeks after cells injec-tion, the mice were euthanized, and their lumbar vertebrae, intactmuscles, and soft tissue were dissected and measured using aDesktop Cone-Beam �CT Scanner (�CT 40; Scanco Medical AG,Bassersdorf, Switzerland, http://www.scanco.ch). Microtomo-graphic slices were acquired at 1,000 projections and reconstructedat a nominal spatial resolution of 16 �m. A constrained 3D gaussianfilter (� � 0.8; support � 1) was used to suppress some noise foundin the volumes. Bone tissue was cut away from the marrow and softtissue by using a global thresholding procedure [29]. Newly formedbone tissue was separated from intact spines by applying a manualcontouring method. For reference spinal bone tissue, we used pos-terior portions of lumbar vertebrae that were contoured in the samemanner. In addition to the visual assessment of structural images,morphometric indices were determined from the microtomographicdata sets by using direct 3D morphometry [30]. Structural metricsmeasured using �CT are closely correlated with those measuredusing standard histomorphometry [31]. The following morphomet-ric indices were determined for the newly formed bone and verte-bral bone control tissue: (a) total volume of bone (TV; in mm3),including new bone and soft tissue cavities; (b) volume of miner-alized bone tissue (BV; in mm3); (c) bone volume density based onthe BV/TV ratio; (d) average bone thickness (in mm); (e) degree ofanisotropy determined from the ratio of maximal to minimal radii ofthe mean intercept length ellipsoid [30]; and (f) connectivity density(1/mm3). The connectivity (C) of a two-component system—boneand marrow—is derived directly from the Euler number (E) byfollowing the formula C � 1 � E, if all trabeculae and bone marrowcavities are connected without isolated marrow cavities inside thebone [32]. Connectivity was normalized by examining tissue vol-ume and is reported as connectivity density.

Histological AnalysisThe harvested lumbar spines, intact muscle, and soft tissue were fixedin 4% formalin for at least 24 hours, decalcified by 0.5 M EDTAsolution (pH 7.4) for 7 days, passed through an increasing grade seriesof ethanol baths, and embedded in paraffin. Five-micrometer-thicksections were cut from each paraffin block with the aid of a motorizedmicrotome (Leica Microsystems, Nussloch, Germany, http://www.leica.com). The slides were heated at 65°C for 45 minutes; this processwas followed by deparaffinization. Hematoxylin and eosin (H&E)staining and bone matrix-specific Masson’s trichrome staining wereperformed as previously reported [10].

Contribution of Injected ASCs to Spinal FusionGenetically engineered ASCs were labeled with Vybrant CM-DiI(Invitrogen, Eugene, OR, http://www.invitrogen.com) cell-labelingsolution 24 hours after nucleofection. Labeling was performed byresuspending the cells in serum-free DMEM at a concentration of

106 cells per milliliter, mixing the cell suspension with VybrantCM-DiI solution (15 �l per milliliter of the 106-cell suspension),and incubating the cells in the dark on a shaker for 30 minutes in anatmosphere of 95% air/5% CO2 at 37°C. The CM-DiI-labeledgenetically engineered ASCs were injected into the mouse paraver-tebral muscle in the manner described earlier. The implants wereharvested 2 or 5 weeks after the injections, fixed, decalcified using0.5 M EDTA solution, and embedded in paraffin. Five-micrometer-thick sections were deparaffinized and rehydrated by passing themthrough a descending grade series of ethanol baths, after which theywere counterstained with 4�,6-diamidino-2-phenylindole (DAPI;catalog no. D9542; Sigma-Aldrich) for 5 minutes and mounted withglycerol-polyvinyl-alcohol (GVA; Zymed Laboratories, San Fran-cisco, http://www.invitrogen.com). The CM-DiI and DAPI stainswere visualized using fluorescent microscopy.

Vimentin Immunohistochemical AnalysisAn immunohistochemical assay was performed to detect the expressionof porcine vimentin on paraffin sections of fused spines by using aHistoMouse-SP kit (catalog no. 95-9541; Zymed Laboratories). Sec-tions of tissue were deparaffinized with xylene, rehydrated in a de-scending grade series of ethanol baths, and rinsed in PBS. The antigenswere retrieved enzymatically by using an antigen retrieval kit (catalogno. ab8212; Abcam, Cambridge, U.K., http://www.abcam.com) En-dogenous peroxidase activity was removed by treatment with 0.1%H2O2 for 10 minutes. A primary antibody that reacts with human,porcine, and equine vimentin but not with mouse vimentin (catalog no.ab8069; Abcam) was diluted 1:10 in PBS and applied to the slides for1 hour at room temperature. After incubation with the primary anti-body, the slides were rinsed in PBS, and a secondary goat anti-mouseIgG antibody (biotin-conjugated; Zymed Laboratories) was applied tothe slides at room temperature for 10 minutes. After they had beenwashed with PBS, the slides were incubated with horseradish peroxi-dase conjugated to streptavidin and then stained with 3-amino-9-ethyl-carbazole dye. The slides were counterstained with hematoxylin,washed, attached with GVA (Zymed Laboratories), and visualized withthe aid of light microscopy.

Statistical AnalysisAssays were performed using ASCs obtained from three differentdonors. For each donor, the samples within each assay were assayedin three repeated experiments. All mean values in Results andfigures are displayed with their SEs. Statistical tests for significancewere performed using Student’s t test, and the minimal criterion forsignificance was determined to be a probability level less than 0.05.

RESULTS

Porcine MSC Transfection and Evaluations ofTransfection Efficiency and Cell ViabilityPorcine ASCs were transfected with pEGFP by applying thenucleofection technique under different nucleofection condi-tions to quantitatively evaluate the transfection efficiency andviability of the cells. The conditions tested included differentcell numbers per sample and different amounts of plasmid DNA(Fig. 1A). Transfection efficiency was evaluated by performinga flow cytometric analysis. The results are presented as thepercentage of GFP-expressing cells among total live cells. Thecells were considered transfected if the fluorescent signal inten-sity was greater than the autofluorescent signal of nontrans-fected cells. These results do not reflect the expression intensityof the transfected cells but rather the ratio of transfected cells tototal viable cells. The optimal nucleofection conditions werefound to comprise 2 � 106 cells per sample and 10 �g ofplasmid DNA, which resulted in approximately 50% cell via-bility and more than 60% transfection efficiency (Fig. 1B). Alladditional experiments were performed using these conditions.

1058 Engineered Porcine ASCs Form Spinal Fusion

Nucleofection of ASCs with rhBMP-6—In VitroEvaluationsASCs were transfected with rhBMP-6 using the nucleofectionsystem, and the optimal conditions were found. The transfectedrhBMP-6 gene expression was evaluated using real-time quan-titative RT-PCR. Gene expression was assessed at several timepoints during the first 3 weeks after transfection and was foundto reach a maximal expression of 5.2 � 107 � 1.6 � 107 copiesof mRNA per 1 �g of RNA on day 2 after nucleofection (n �3). The gene expression of rhBMP-6 gradually declined, and onday 21, it reached a level of 1.4 � 105 � 2.2 � 104 copies ofmRNA per 1 �g of RNA (n � 3), which was close to the basallevel of rhBMP-6 detected in pEGFP-nucleofected ASCs (Fig.2A). This level of gene expression is very similar to previouslyreported gene expression in human MSCs nucleofected withplasmids encoding rhBMP-2 and rhBMP-9 [6].

To estimate the efficiency of protein synthesis afterrhBMP-6 gene transfer, we performed an immunoassay of con-ditioned media collected from cultures of porcine ASCs nucleo-

fected with prhBMP-6. These porcine ASCs secreted rhBMP-6on day 1 postnucleofection at a level of 4.78 � 0.97 ng/106 cellsper 24 hours (n � 3), whereas ASCs transfected with pEGFPsecreted rhBMP-6 at levels of only 0.56 � 0.07 ng/106 cells per24 hours. These results correlate with findings in our previousstudies, in which prhBMP-2 was transfected using the samemethod in human MSCs [6]. The profile of protein secretionfollowed the profile of gene expression, and protein secretionpeaked on day 2 at a level of 45.4 � 2.89 ng/106 cells per 24hours (n � 3). After 3 weeks in culture, rhBMP-6 proteinsecretion declined to 0.2 � 0.02 ng/106 cells per 24 hours,which was similar to the level, detected in the control pEGFP-transfected ASCs (Fig. 2B).

Osteogenic Potential of rhBMP-6 In VitroFollowing the evaluation of gene expression and protein secre-tion of rhBMP-6, we tested the osteogenic potential of thenucleofected ASCs. The genetically modified ASCs were grownin vitro in maintenance conditions (without differentiation sup-

Figure 1. Nucleofection condition optimi-zation. (A): The optimal conditions ofnucleofection of ASCs were calibrated usingnucleofection of plasmid enhanced GFP.The viability of the cells was calibrated bycounting the surviving cells 24 hours post-nucleofection. The efficiency of the nucleo-fection was evaluated by flow cytometry.(B): Flow cytometric analysis of the chosenconditions. Bars indicate SE (n � 3). Abbre-viations: FSC-H, forward scattering; GFP,green fluorescent protein.

Figure 2. ASCs nucleofected with rhBMP-6 expressed and secreted rhBMP-6. The gene expression was evaluated using quantitative reversetranscription-polymerase chain reaction (A), and the protein secretion was quantified using an enzyme-linked immunosorbent assay during the 3weeks after nucleofection (B). The gene expression and protein secretion by ASCs nucleofected with EGFP were set as day 0. The osteogenicdifferentiation of genetically modified ASCs in vitro was evaluated using colorimetric alkaline phosphatase assay 7 days postnucleofection (C). Barsindicate SE (n � 3). Abbreviations: ALP, alkaline phosphatase; ASC, adipose tissue-derived stem cell; GFP, green fluorescent protein; h, hours;rhBMP, recombinant human bone morphogenetic protein.

1059Sheyn, Pelled, Zilberman et al.

www.StemCells.com

plements), and the osteogenic differentiation was evaluated us-ing a quantitative colorimetric alkaline phosphatase assay. TheASCs nucleofected with prhBMP-6 showed the highest level ofosteogenic activity, 1.53 � 0.22 nmol of ALP/minute per �g ofprotein (n � 3), compared with the control group transfectedwith pEGFP and the control group of non-nucleofected cells,which had 0.57 � 0.09 nmol of ALP/minute per �g of protein(n � 3) and 0.11 � 0.04 nmol of ALP/minute per �g of protein(n � 3), respectively (Fig. 2C).

Spinal Fusion Formation Evaluated Using �CT andHistological AnalysisGenetically engineered porcine ASCs overexpressing the rhBMP-6gene were injected into the paraspinal muscles of immunodeficientmice. Bone formation was analyzed using �CT (Figs. 3, 4) andhistological analysis (Fig. 5). As a control, nonengineered ASCswere injected in the same manner; however, no bone formation wasobserved (data not shown). We were able to demonstrate that thebone formation bridged two or three spine segments. Extensivebone formation on and adjacent to posterior elements of the spineactually constituted a bridging mass of bone fusion above andcovering several spinal segments. The structural parameters mea-sured using �CT showed that the bone fusion mass that formed denovo in the mice did not differ from intact vertebral bone tissue(Fig. 3). The TV, including cavities within the tissue, was found tobe 13.3 � 3.3 mm3 (n � 5). The BV, not including cavities,reached 8.5 � 2.4 mm3 (n � 5). To evaluate the morphological andstructural properties of the ectopically formed bone tissue andcompare them with those of native bone, we compared ectopicallyformed bone tissue to the posterior part of intact vertebrae that hadnot fused (Fig. 4). The bone volume density, which is the BV/TVratio, was found to be 0.64 � 0.04 (n � 5) in the fusion bone massand 0.8 � 0.04 (n � 3) in the posterior part of the vertebrae.Average bone thickness indicates the solidness of bone tissuecompartments [30]. The average bone thickness of the fusion masswas very similar to that of control tissue: 0.16 � 0.01 mm (n � 5)

and 0.19 � 0.01 mm (n � 3), respectively. The parameter ofconnectivity density is used to describe the porosity of the bonesample and to show how branched the bone tissue structure is [30].The fusion mass displayed a connectivity density of 24.19 � 3.191/mm3 (n � 5), whereas the control tissue exhibited a connectivitydensity of 16.97 � 3.87 1/mm3 (n � 3). The degree of anisotropywas also found to be similar in both tissues: 1.78 � 0.11 (n � 5) in thefusion mass and 1.52 � 0.08 (n � 3) in the control bone tissue.Overall, none of the structural parameters differed significantly be-tween the two groups according to Student’s t test (p � 05).

Standard H&E and bone matrix-specific Masson’s trichromestaining revealed a well-organized fusion mass on the posterioraspect of the mouse spines (Fig. 5A, 5C). Light microscopy of thebone architecture revealed an organized fusion mass with a lamel-lar microstructure, which contained large compartments of bonemarrow within it and highly resembled the intact vertebral bonetissue morphologically (Fig. 5A–5C). To illustrate the histologicalstaining, we have included a schematic diagram in Figure 5B, whichdepicts the three main components: intact vertebral bone tissue, newlyformed bone tissue, and bone marrow. Conversely, the control non-nucleofected cells that were injected in the same manner were found inrather small amounts, mostly undifferentiated (Fig. 6E). Several cellshad an adipocyte-like morphology, indicating that some cells couldhave undergone adipogenic differentiation.

Contribution of Injected ASCs to the Newly FusedBone MassThe genetically modified cell therapeutic approach described inthis report raises the question, what was the contribution ofinjected genetically engineered cells to the newly fused bonemass? We chose two methods to pursue this question. First, westained the genetically engineered porcine ASCs with the fluo-rescent, biologically inactive tracing dye CM-DiI. Second, weperformed immunohistochemical staining for the intracellularmesenchymal marker protein vimentin by using an antibody that“recognizes” the porcine molecule but not the mouse one. CM-

Figure 3. Spinal fusion bone mass forma-tion imaged by microcomputed tomography(�CT). The spines were harvested 5 weeksafter the injection of genetically engineeredadipose tissue-derived stem cells and ana-lyzed by �CT. The new bone formation wascontoured manually and is depicted in or-ange color on the three-dimensional (3D)reconstructed images. Shown are a represen-tative lateral view of 3D image of fusedspine on the left side and coronal sections intwo-dimensional and 3D views on the right.

1060 Engineered Porcine ASCs Form Spinal Fusion

DiI-stained porcine ASCs were shown to form cartilage-likeislands in the paraspinal muscle after 2 weeks in vivo and inbone tissue after 5 weeks in vivo (Fig. 6). Most of the cellsforming the cartilage-like islands were stained with CM-DiI,indicating that these islands were composed of the geneticallymodified ASCs that had been injected. However, the bonestructures formed after 5 weeks were not composed of CM-DiI-labeled cells alone (Fig. 6), indicating that the bone tissue is achimera of donor (CM-DiI-labeled) cells and host (nonlabeled)cells. These findings were confirmed by examining paraffin-embedded tissue sections that had been subjected to an immu-nohistochemical stain designed to detect porcine vimentin (Fig.6G–6J). The immunohistochemical staining technique producedan intracellular (red) stain in the fusion bone mass (Fig. 6G, 6H)

but no stain in the osteocytes that formed intact mouse vertebrae(Fig. 6J). These results show that most of the fusion bone masswas formed by the injected cells, although there are indicationsthat some bone was formed of donor-host chimerical tissue.

DISCUSSION

Our data demonstrate, for the first time, the following: (a)primary ASCs that have been nonvirally transfected with plas-mid rhBMP-6 ex vivo can secrete biologically active rhBMP-6;and (b) after these cells have been injected into the lumbarparavertebral muscle of immunodeficient mice, the cells caninduce functional bone tissue formation and efficient posterior

Figure 4. Quantitative analysis of structural properties of new bone formation using microcomputed tomography. (A): Representative three-dimensional image of the new bone formed in the spinal fusion (in orange) on the left side was compared with posterior part of intact vertebra, arepresentative image of which is depicted on the right side. (B): The total tissue volume and actual bone volume. Bone volume density is calculatedas the ratio of the bone volume and total volume (C), average bone thickness (D), connectivity density (E), and degree of anisotropy (F) are structuralparameters of bone. Bars indicate SE (spinal fusion mass, n � 5; control vertebra, n � 3).

Figure 5. Spinal fusion bone mass formation imaged by histology. Histological analysis revealed that bone fusion mass adjusted to the intactvertebrae 5 weeks after injecting the nucleofected cells. The bone formation included morphologically normal bone structures and bone marrow. Thesections are presented in H&E standard staining (A); a schematic map of the histological sections in which green indicates new bone, red indicatesbone marrow, and blue indicates the intact vertebrae (B); and Masson’s trichrome staining (C).

1061Sheyn, Pelled, Zilberman et al.

www.StemCells.com

spinal fusion. We optimized the nucleofection conditions of theporcine ASCs so that we could gain maximal viability andefficiency. The outcome of this optimized protocol was approx-imately 50% cell viability and more than 60% transfectionefficiency, as detected using flow cytometry 24 hours afternucleofection. These results are close to what we obtained usinghuman bone marrow-derived MSCs [6] and human ASCs [25].The gene expression and protein secretion profiles demonstratedhigh peaks on day 2 postnucleofection; however, the levels oftransgene expression and protein secretion declined until theyreached basal levels. These profiles were observed in ASCs thatwere also nucleofected with pEGFP. The profiles were previ-ously observed in human ASCs nucleofected with the EGFPreporter gene [25]. In this stem cell-mediated therapeutic ap-proach, the transient expression of the transgene is advantageousbecause induction of osteogenic differentiation is needed onlyfor a short period. The genetically modified ASCs that ex-pressed and secreted rhBMP-6 were deemed competent to in-duce massive bone formation and vast spinal fusion in a mini-mally invasive injection approach. The quality and structuralproperties of the newly formed bone, as tested by performingquantitative �CT, were comparable to those of native bone;nevertheless, in a comparison with intact vertebrae, the newbone formation appeared much younger and less mature.

The bone fusion mass contained donor porcine ASCs; thiswas verified by CM-DiI and immunohistochemical staining.The expression of porcine vimentin by most cells forming thenew bone formation indicated the vast contribution of these cellsto de novo bone tissue formation. Genetic engineering of ASCsto produce rhBMP-6 had a major effect on tissue formation; thenonengineered ASCs did not induce bone formation.

In an attempt to produce spine fusion in rodents, directnonviral delivery of osteogenic genes efficiently induced ec-topic bone formation in vivo [33]. However, the bone tissue thatdeveloped was insufficient to bridge the vertebrae and inducespinal fusion. That was probably due to the low number ofprogenitor cells in the region that were able to respond to thesecreted BMP and induce major bone formation. In other studiesin which MSCs were used, we and others relied on adenoviralvectors to genetically modify the cells [22, 34]. The use ofadenoviral vectors for clinical applications remains doubtful,however, given the relatively high safety of ex vivo nonviralgene delivery [21]. To date, autograft bone implantation inconjunction with added rhBMP-2 protein is the only way toinduce ectopic bone formation and spinal fusion in the clinicalsetting [21, 22].

In this study, we created conditions that come as close aspossible to those required for clinical applications. First, in

Figure 6. Contribution of the donor adi-pose tissue-derived stem cells (ASCs) to thespinal fusion bone mass formation. PorcineASCs nucleofected with bone morphoge-netic protein 6 were stained with CM-DiI,and 5 � 106 cells were injected into paraspi-nal muscle of NOD/SCID mice. The muscleswere harvested 2 weeks after the injection(A, B) and 5 weeks after the injection (C–J).The sections were counterstained with 4�,6-diamidino-2-phenylindole (DAPI) (A–D, F).The control non-nucleofected cells were im-aged using H&E stain (E) and fluorescentCM-DiI and DAPI stains (F). The sectionsof fused spines were stained with immuno-histochemical staining for the porcine mes-enchymal marker gene vimentin (G–J). Newbone tissue (G), new cartilage-like tissue(H), and mouse intact vertebrae are depictedat magnifications of �10, �20, and �40.(I): Schematic map indicating the regions ofthe spine. In (A–F), white arrows indicateintact bone, and green arrows indicate newbone formation.

1062 Engineered Porcine ASCs Form Spinal Fusion

adults, the reservoir of available adipose tissue-derived stemcells is greater than that of bone marrow-derived MSCs, andthese cells can be obtained with less risk of morbidity [8].Second, the nonviral gene delivery method we used to geneti-cally engineer the ASCs ex vivo is safe and transient, limitingoverexpression of the osteogenic gene to a short period of a fewweeks. There is a potential advantage in the transient expressionof the osteogenic transgene: short-term expression may be pre-ferred for skeletal regeneration applications. This advantage waspreviously exhibited only by adenoviral vectors, which inducedtransient overexpression of BMPs [34–36]. Third, the therapeu-tic transgene of BMP is gradually overexpressed, translated, andsecreted by the injected mammalian cells, rather than beingproduced by prokaryotes, as is recombinant BMP that is locallydelivered in high nonphysiological doses with a high immuno-genicity. We used porcine adipose tissue-derived ASCs to ex-plore the osteogenic potential of these cells within the context ofspinal fusion. The data generated in this study will enable us totake the next essential step to preclinical studies involving largeanimals, because the porcine model for spine fusion is clinicallyrelevant [37] and porcine ASCs have been extensively investi-gated [38].

An injection-based therapeutic system carries great advan-tages. First, it saves the patient from the invasive and traumaticsurgical treatments currently practiced for degenerative disc

diseases. Second, the injection system enables the clinician tocontrol bone formation in vivo and offer subsequent treatmentsto shape the newly formed bone and produce incremental fusion.Genetically engineered ASCs have great clinical potential com-pared with autologous bone grafts, as does endogenously pro-duced BMP compared with recombinant BMP with respect tothe immunogenic response.

CONCLUSION

The proposed therapeutic model involving the use of geneticallymodified adipose tissue-derived adult stem cells has great po-tential in various applications of bone tissue engineering. In thepresent study, we identified engineered bone tissue that mor-phologically and structurally resembled native tissue. In futurestudies, we intend to validate the biomechanical properties ofthe engineered tissue, because the fused section of spine shouldbear the load of the degenerated spine.

DISCLOSURE OF POTENTIAL CONFLICTS

OF INTEREST

The authors indicate no potential conflicts of interest.

REFERENCES

1 Kraemer J. Natural course and prognosis of intervertebral disc diseases.International Society for the Study of the Lumbar Spine Seattle, Wash-ington, June 1994. Spine 1995;20:635–639.

2 Buckwalter JA. Aging and degeneration of the human intervertebral disc.Spine 1995;20:1307–1314.

3 Nishida K, Kang JD, Gilbertson LG et al. Modulation of the biologicactivity of the rabbit intervertebral disc by gene therapy: An in vivo studyof adenovirus-mediated transfer of the human transforming growth factorbeta 1 encoding gene. Spine 1999;24:2419–2425.

4 Frick SL, Hanley EN Jr, Meyer RA Jr et al. Lumbar intervertebral disctransfer. A canine study. Spine 1994;19:1826–1834; discussion 1834–1825.

5 Bao QB, McCullen GM, Higham PA et al. The artificial disc: Theory,design and materials. Biomaterials 1996;17:1157–1167.

6 Aslan H, Zilberman Y, Arbeli V et al. Nucleofection-based ex vivononviral gene delivery to human stem cells as a platform for tissueregeneration. Tissue Eng 2006;12:877–889.

7 Gimble JM, Katz AJ, Bunnell BA. Adipose-derived stem cells forregenerative medicine. Circ Res 2007;100:1249–1260.

8 Zuk PA, Zhu M, Ashjian P et al. Human adipose tissue is a source ofmultipotent stem cells. Mol Biol Cell 2002;13:4279–4295.

9 Zuk PA, Zhu M, Mizuno H et al. Multilineage cells from human adiposetissue: Implications for cell-based therapies. Tissue Eng 2001;7:211–228.

10 Turgeman G, Pittman DD, Muller R et al. Engineered human mesenchy-mal stem cells: A novel platform for skeletal cell mediated gene therapy.J Gene Med 2001;3:240–251.

11 Quarto R, Mastrogiacomo M, Cancedda R et al. Repair of large bonedefects with the use of autologous bone marrow stromal cells. N EnglJ Med 2001;344:385–386.

12 Dragoo JL, Choi JY, Lieberman JR et al. Bone induction by BMP-2transduced stem cells derived from human fat. J Orthop Res 2003;21:622–629.

13 Dragoo JL, Lieberman JR, Lee RS et al. Tissue-engineered bone fromBMP-2-transduced stem cells derived from human fat. Plast ReconstrSurg 2005;115:1665–1673.

14 Peterson B, Zhang J, Iglesias R et al. Healing of critically sized femoraldefects, using genetically modified mesenchymal stem cells from humanadipose tissue. Tissue Eng 2005;11:120–129.

15 Lendeckel S, Jodicke A, Christophis P et al. Autologous stem cells(adipose) and fibrin glue used to treat widespread traumatic calvarialdefects: Case report. J Craniomaxillofac Surg 2004;32:370–373.

16 Schaffler A, Buchler C. Concise review: Adipose tissue-derived stromalcells—Basic and clinical implications for novel cell-based therapies.STEM CELLS 2007;25:818–827.

17 Gazit D, Turgeman G, Kelley P et al. Engineered pluripotent mesenchy-

mal cells integrate and differentiate in regenerating bone: A novel cell-mediated gene therapy. J Gene Med 1999;1:121–133.

18 Moutsatsos IK, Turgeman G, Zhou S et al. Exogenously regulated stem cell-mediated gene therapy for bone regeneration. Mol Ther 2001;3:449–461.

19 Turgeman G, Aslan H, Gazit Z et al. Cell mediated gene therapy for boneformation and regeneration. Curr Opin Mol Ther 2002;4:390–394.

20 Hasharoni A, Zilberman Y, Turgeman G et al. Murine spinal fusioninduced by engineered mesenchymal stem cells that conditionally ex-press bone morphogenetic protein-2. J Neurosurg Spine 2005;3:47–52.

21 Kimelman N, Pelled G, Gazit Z et al. Applications of gene therapy andadult stem cells in bone bioengineering. Regen Med 2006;1:549–561.

22 Kimelman N, Pelled G, Helm GA et al. Review: Gene- and stemcell-based therapeutics for bone regeneration and repair. Tissue Eng2007;13:1135–1150.

23 Li JZ, LI H, Sasaki T et al. Osteogenic potential of five differentrecombinant human bone morphogenetic protein adenoviral vectors inthe rat. Gene Ther 2003;10:1735–1743.

24 Cheng H, Jiang W, Phillips FM et al. Osteogenic activity of the 14 typesof human bone morphogenetic proteins (BMPs). J Bone Joint Surg Am2003;85-A:1544–1552.

25 Zaragosi LE, Billon N, Ailhaud G et al. Nucleofection is a valuabletransfection method for transient and stable transgene expression inadipose tissue-derived stem cells. STEM CELLS 2007;25:790–797.

26 Hamm A, Krott N, Breibach I et al. Efficient transfection method forprimary cells. Tissue Eng 2002;8:235–245.

27 Aslan H, Zilberman Y, Kandel L et al. Osteogenic differentiation of non-cultured immunoisolated bone marrow-derived CD105� cells. STEMCELLS 2006;24:1728–1737.

28 Calve S, Dennis RG, Kosnik PE 2nd et al. Engineering of functionaltendon. Tissue Eng 2004:755–761.

29 Muller R, Ruegsegger P. Micro-tomographic imaging for the nondestruc-tive evaluation of trabecular bone architecture. Stud Health TechnolInform 1997;40:61–79.

30 Hildebrand T, Laib A, Muller R et al. Direct three-dimensional morphomet-ric analysis of human cancellous bone: Microstructural data from spine,femur, iliac crest, and calcaneus. J Bone Miner Res 1999;14:1167–1174.

31 Muller R, Van Campenhout H, Van Damme B et al. Morphometricanalysis of human bone biopsies: A quantitative structural comparison ofhistological sections and micro-computed tomography. Bone 1998;23:59–66.

32 Muller R, Hildebrand T, Ruegsegger P. Non-invasive bone biopsy: Anew method to analyse and display the three-dimensional structure oftrabecular bone. Phys Med Biol 1994;39:145–164.

33 Bright C, Park YS, Sieber AN et al. In vivo evaluation of plasmid DNAencoding OP-1 protein for spine fusion. Spine 2006;31:2163–2172.

34 Peterson B, Iglesias R, Zhang J et al. Genetically modified humanderived bone marrow cells for posterolateral lumbar spine fusion in

1063Sheyn, Pelled, Zilberman et al.

www.StemCells.com

athymic rats: Beyond conventional autologous bone grafting. Spine2005;30:283–289; discussion 289–290.

35 Hidaka C, Goshi K, Rawlins B et al. Enhancement of spine fusion usingcombined gene therapy and tissue engineering BMP-7-expressing bonemarrow cells and allograft bone. Spine 2003;28:2049–2057.

36 Dumont RJ, Dayoub H, Li JZ et al. Ex vivo bone morphogeneticprotein-9 gene therapy using human mesenchymal stem cells induces

spinal fusion in rodents. Neurosurgery 2002;51:1239–1244; discussion1244–1235.

37 Drespe IH, Polzhofer GK, Turner AS et al. Animal models for spinalfusion. Spine J 2005;5(suppl):209S–216S.

38 Qu CQ, Zhang GH, Zhang LJ et al. Osteogenic and adipogenic potentialof porcine adipose mesenchymal stem cells. In Vitro Cell Dev Biol Anim2007;43:95–100.

1064 Engineered Porcine ASCs Form Spinal Fusion