Lentiviral Delivery of PPARγ shRNA Alters the Balance of Osteogenesis and Adipogenesis, Improving Bone Microarchitecture

Abstract

Introduction: Skeletal aging is associated not only with alterations in osteoblast (OB) and osteoclast (OC) number and activity within the basic metabolic unit, but also with increased marrow adiposity. Peroxisome proliferator-activated receptor gamma (PPARγ) is commonly considered the master transcriptional regulator of adipogenesis, however, it has known roles in osteoblast and osteoclast function as well. Here, we designed a lentiviral delivery system for PPARγ shRNA, and examined its effects in vitro on bone marrow stromal cells (BMSC) and in a mouse intramedullary injection model.

Methods: PPARγ shRNA was delivered by a replication-deficient lentiviral vector, after in vitro testing to confirm purity, concentration, and efficacy for Pparg transcript reduction. Next, control green fluorescent protein lentivirus or PPARγ shRNA expressing lentivirus were delivered by intramedullary injection into the femoral bone marrow of male SCID mice. Analyses included daily monitoring of animal health, and postmortem analysis at 4 weeks. Postmortem analyses included high resolution microcomputed tomography (microCT) reconstructions and analysis, routine histology and histomorphometric analysis, quantitative real time polymerase chain reaction analysis of Pparg transcript levels, and immunohistochemical analysis for markers of adipocytes (PPARγ, fatty acid binding protein 4 [FABP4]), osteoblasts (alkaline phosphatase [ALP], osteocalcin [OCN]), and osteoclasts (tartrate-resistant acid phosphatase [TRAP], Cathepsin K).

Results: In vitro, PPARγ shRNA delivery significantly reduced Pparg expression in mouse BMSC, accompanied by a significant reduction in lipid droplet accumulation. In vivo, a near total reduction in mature marrow adipocytes was observed at 4 weeks postinjection. This was accompanied by significant reductions in adipocyte-specific markers. Parameters of trabecular bone were significantly increased by both microCT and histomorphometric analysis. By immunohistochemical staining and semi-quantification, a significant increase in OCN+osteoblasts and decrease in TRAP+multinucleated osteoclasts was observed with PPARγ shRNA treatment.

Discussion: These findings suggest that acute loss of PPARγ in the bone marrow compartment has a significant role beyond anti-adipose effects. Specifically, we found pro-osteoblastogenic, anti-osteoclastic effects after PPARγ shRNA treatment, resulting in improved trabecular bone architecture. Future studies will examine the isolated and direct effects of PPARγ shRNA on OB and OC cell types, and it may help determine whether PPARγ antagonists are potential therapeutic agents for osteoporotic bone loss.

Introduction

Skeletal aging is associated not only with alterations in osteoblast (OB) and osteoclast (OC) number and activity within the basic metabolic unit, but also with an increase in marrow adiposity. Historically, intramarrow adipocytes have been considered a passive, inert cell type, however, recent literature has identified such cells as elaborators of “adipokines”–cytokines that negatively affect bone health.^1–3 For example, high levels of adiponectin have been associated with low bone mineral density in those with disuse bone loss after spinal cord injury.⁴ Therefore, negative regulation of intramarrow adipogenesis may be a novel mechanism in the promotion of bone maintenance and prevention of age-related bone loss.

Peroxisome proliferator-activated receptors (PPAR) are members of the steroid/thyroid hormone receptor gene superfamily⁵ and were initially named for PPARα.⁶ Subsequently, structural analogs, PPARδ and PPARγ have been discovered and all are activated by polyunsaturated fatty acids.⁷ All three PPARs are found in mammals and interact with binding sites on targeted genes by forming heterodimers with the retinoid X receptor to recruit transcriptional co-activator proteins.⁸ While both PPARα and PPARδ are expressed during adipogenesis, PPARγ is adipocyte cell type restricted and rapidly increases in expression during early adipogenesis.^9,10 Indeed, PPARγ is considered the master regulator of adipogenesis, for no other factor can rescue adipocyte formation in the event of PPARγ knockout, and generally all cell signaling pathways leading to adipogenesis converge with PPARγ.^11–13 In fact, PPARγ haploinsufficient mice have reduced ability to differentiate into adipocytes, depicting PPARγ as an essential factor in adipogenesis.^14,15

Moreover, PPARγ also has important roles in cell types other than adipocytes, including both osteoblasts and osteoclasts.¹⁶ In terms of osteoblasts, ectopic or increased expression of PPARγ is required to convert immature osteoblasts, and a subset of mature osteoblasts, into adipocytes.^17,18 Conversely, suppression of PPARγ inhibits adipogenesis and stimulates osteoblastogenesis.¹⁹ Similarly, PPARγ haploinsufficiency results in increased bone mass and increased bone marrow stromal cells (BMSC) osteoblastic differentiation.²⁰ Thus, PPARγ activity modulates osteoblastic maturation as seen in changes in alkaline phosphatase (ALP) activity, Runx2/Cbfa1 activity, and osteocalcin (OCN) expression.²⁰ In terms of osteoclasts, gain in PPARγ function via ligand activator rosiglitazone promotes both osteoclast lineage commitment and maturation by maintaining levels of c-fos in monocyte precursors and osteoclasts.²¹ Conversely, loss of function through PPARγ deletion impairs osteoclast differentiation and bone resorption.²¹ Wei et al. showed that the pro-osteoclastogenic effects are mediated by PPARγ coactivator 1β, and estrogen-related receptor α along with PGC1beta.²² In a subsequent study, Wei et al. identified GATA2 as a novel and critical PPARγ target gene in osteoclast progenitors.²³ In aggregate, these data clearly demonstrate that PPARγ transcriptional activity has a significant role beyond the confines of adipogenic differentiation.

In the present study, we designed a lentiviral delivery system for PPARγ shRNA to examine its effects on both adipogenesis and bone metabolism in vitro and in vivo, using a mouse intramedullary injection model. PPARγ shRNA delivery not only led to a dramatic reduction in intramarrow adiposity, but also led to a significant improvement in trabecular bone architecture and alteration in osteoblast and osteoclast numbers.

Materials and Methods

Antibodies and reagents

Primary antibodies used in this study were anti-PPARγ (sc-7273; Santa Cruz Biotechnology), anti-OCN (sc-18322; Santa Cruz Biotechnology), anti-Cathepsin K (ab19027; Abcam), anti-fatty acid-binding protein 4 (anti-FABP4, AF1443; R&D Systems), anti-proliferating cell nuclear antigen (anti-PCNA, (M0879; Dako), and anti-ALP (sc-166261; Santa Cruz Biotechnology). Antibodies used for western blot analysis were anti-PPARγ (sc-7273; Santa Cruz Biotechnology) or anti-GAPDH (GTX-627408; GeneTex). All other reagents were purchased from Dako unless otherwise specified.

Viral production and purification

The lentiviral vector encoding PPARγ shRNA or green fluorescent protein (GFP) was generated by co-transfection of 293T cells with the PPARγ-shRNA plasmid vector (SC-29455-SH; Santa Cruz Biotechnology) or the FG12 vector (14884; Addgene),²⁴ with helper plasmids pCMV-dR8.2-vprX and pCMV-VSVG. The viral vector was collected from transfected cell cultures and cell debris was removed through 0.22 μm filtration and concentrated by ultrafuge at 4°C, 17,000 rpm for 60 min using Beckman SW32 rotors. The pellets that contained the lentiviral vector were resuspended with Dulbecco's Modified Eagle's Medium (DMEM) medium to bring the concentration to 5×10⁷ TCID₅₀/mL. Lentiviral vector titer was estimated by measuring the gag p24 protein, 1 pg of p24 reading was assigned as 10 tissue culture infective dosage (TCID₅₀) for freshly isolated lentiviral vectors.

BMSC isolation and analysis

Femurs from 8-week-old C57BL/6 mice were dissected and marrow was harvested by inserting a syringe needle (27-gauge) into one end of the bone and flushing with DMEM (Gibco). The bone marrow cells were filtered through a 70 mm nylon mesh filter (BD, Falcon). Cells were plated into six-well plastic cell culture plate at a density of 25×10⁶ cells per well in DMEM containing 15% fetal bovine serum (FBS; Sigma-Aldrich), 2 mm L-glutamine (Gibco BRL), 100 U/mL penicillin, and 100 U/mL streptomycin (Gibco). Cultures were kept at 37°C in a humidified atmosphere containing 95% air and 5% CO₂. Cells from passage 2 only were used for further experiments. The generated lentiviral vector was used to infect mouse BMSC cells with multiplicity of infection of 10 (MOI=10). Adipogenic differentiation was performed using Mouse Adipogenic Stimulatory Supplements in conjunction with MesenCult™ Basal Medium (Stemcell Technologies). Next, Oil Red O staining was performed. Monolayers were rinsed with PBS, fixed with buffered formalin for 60 min at room temperature, and stained with 3% Oil Red O solution in 60% isopropyl alcohol/40% ddH₂O for 10 min. The intensity of staining was analyzed using commercial software Image-Pro Plus 6 and presented as relative mean optical density.

Protein isolation and western blot

Total protein was extracted from BMSC with RIPA lysis buffer. Lysate was separated by 10% sodium dodecyl sulfate polyacrylamide gel electrophoresis, and the gel was blotted onto PVDF membrane (Millipore). The membrane was blocked in 5% nonfat milk, and then incubated with either anti-PPARγ or anti-GAPDH antibody at a 1:1000 dilution. The bands were quantified using the Image J software by measuring the band intensity (area×optical density) for each group, and normalized with GAPDH. The final results are expressed as fold changes by normalizing the data to the GFP group.

Intramedullary injection

Animals were housed in a light and temperature-controlled environment and given food and water ad libitum. All animal procedures were conducted in accordance with the Guide for the Care and Use of Laboratory Animals of the University of California and approved by the Chancellor's Animal Research Committee. Male SCID mice (8-week-old, six animals per group) were kept and handled under pathogen-free conditions and fed on sterile food and acidified water. Knee arthrotomy and intramedullary injection of the femur were performed for virus delivery.

Briefly, a 5 mm-longitudinal incision was made along the medial aspect of the quadriceps–patellar complex. The patella was dislocated laterally to expose the intercondylar groove. A 0.9 mm k-wire on a trephine drill was used to create a trephination defect. A 26-gauge needle was inserted through the defect, and 15 μL of PPARγ lentivirus (4×10⁷ TCID₅₀/mL) was injected into the intramedullary cavity. The quadriceps–patellar complex was then repositioned, and the medial arthrotomy was carefully repaired with 5-0 vicryl suture. Fifteen microliters of GFP lentivirus (5×10⁷ TCID₅₀/mL) as control was injected into the contralateral femur, following the same surgical procedure as above.

Microcomputed tomography analysis

Femur samples were harvested 4 weeks after viral injection, fixed in formalin, and scanned using high-resolution microcomputed tomography (microCT; Skyscan 1171F) at an image resolution of 17.81 μm (55 kV and 181 mA radiation source, using a 0.5 mm aluminum filter). For analysis of trabecular bone formation, regions of interest were drawn to select only metaphyseal intramedullary trabecular bone from the distal end, with a length of 1.225 mm (35 slices at 0.035 mm each). Trabecular bone formation was analyzed using CTAn software, with a threshold value of 75. The microCT scans were reconstructed using CTVol software.

Histological and immunohistochemical analyses

Tissue from the femur and its overlying periosteum adjacent to the injection site were harvested after sacrifice, fixed in formalin, and decalcified using 19% ethylenediaminetetraacetic acid (EDTA). Five-μm-thick paraffin sections, of decalcified samples cut along the coronal plane, were stained with hematoxylin & eosin (H&E) and Masson's Trichrome staining. Using H&E sections, histomorphologic assessments of random 200 cell counts were performed to assess the treatment effects on bone marrow composition (n=6 samples per group). Histomorphometric analysis of serial H&E stained sections was performed using random 400×H&E images with Adobe Photoshop quantitation, with parameters including mean bone area (B.Ar), bone perimeter (B.Pm), trabecular width (Tb.Wi), trabecular number (Tb.N), and trabecular separation (Tb.Sp). Marrow adiposity of random 400×H&E sections was quantified using Adobe Photoshop, by comparing the area of adipose tissue to the area of the whole bone marrow. Additional sections were analyzed by indirect immunohistochemistry. Briefly, unstained sections were deparaffinized and incubated with anti-OCN, anti-Cathepsin K, anti-PPARγ, anti-PCNA, anti-FABP4, and anti-ALP primary antibodies and then by the appropriate biotinylated secondary antibodies (Dako). PCNA primary antibody was used at a 1:1000 dilution. All other primary antibodies were used at a dilution of 1:100, while all secondary antibodies were used at a dilution of 1:200. Positive immunoreactivity was detected following ABC complex (PK-6100, Vectastain Elite ABC Kit; Vector Laboratories, Inc.) incubation and development with AEC chromagen (K346911-2; Dako). Controls for each antibody consisted of incubation with secondary antibody in the absence of primary antibody. Sections were counterstained with hematoxylin. Photomicrographs were acquired using Olympus BX51 (400×magnification lens, UPLanFL; Olympus). The relative intensity of PPARγ, FABP4, and Cathepsin K staining was analyzed using commercial software Image-Pro Plus 6 and quantified by the mean optical density of staining signal x per percent area positively stained×100.²⁵ For ALP and OCN immunohistochemical staining, the number of positively stained osteoblasts along the bone perimeter were determined by three blinded observers. For PCNA immunohistochemical staining, the number of positively stained cells along the bone perimeter was determined by three blinded observers. Reported results were average of data collected from six random fields per sample.

Tartrate resistant acid phosphatase staining

For tartrate-resistant acid phosphatase (TRAP) staining, the acid phosphatase, leukocyte (TRAP) kit (387A; Sigma-Aldrich) was used per the manufacturer's protocol. Sections were immersed in TRAP stain solution (12.5 mg/mL Naphthol AS-BI phosphoric acid+2.5 M acetate buffer+diazotized Fast Garnet GBC solution+deionoized water) at 37°C, for 60 min protected from light. Subsequently, sections were thoroughly rinsed in deionized water and counterstained in hematoxylin. TRAP staining was analyzed by three blinded observers and quantified by the number of osteoclasts per trabecular bone perimeter in millimeters. The observers analyzed and averaged the results from six random fields per sample.

TUNEL assay for detection of apoptosis

A terminal deoxynucleotidyl transferase dUTP nick end labeling (TUNEL) method was used for the detection of apoptotic cells. For this purpose, we used the DeadEnd™ Colorimetric TUNEL Assay kit (G7360; Promega) according to the manufacturer's instructions. Briefly, paraffin-embedded slides were labeled with biotinylated nucleotide at the 3′-OH DNA ends, using the terminal deoxynucleotidyl transferase, recombinant (rTDT) enzyme, for 60 min in a humidified chamber. The slides were incubated with streptavidin horseradish peroxidase and diaminobenzidine was used as the chromagen. TUNEL staining was analyzed by three blinded observers and quantified by the percentage of positively stained cells per field from six random fields per sample.

Real-time polymerase chain reaction

To analyze the PPARγ expression in SCID mice with intramedullary injection of PPARγ shRNA or GFP lentivirus, BMSC were isolated 1 week postinjection using the aforementioned method. Total RNA from tissue or cells for real-time polymerase chain reaction (PCR) evaluation was extracted using TRIzol reagent (Invitrogen). Bone tissue was homogenized in ∼2 mL of TRIzol reagent per femur using mortar with liquid nitrogen. After homogenization, insoluble material from the homogenate was removed by centrifuging at 10,000 rpm for 10 min at 4°C. n=3 mice per treatment group were used for RNA studies. To determine osteogenic differentiation marker expression levels in PPARγ shRNA or GFP transfected BMSC, BMSC was cultured in osteogenic differentiation medium (alpha MEM+15% FBS with 50 μg/mL ascorbic acid and 3 mM β-glycerophosphate) for 3, 6, and 9 days. Total RNA was isolated using TRIzol reagent. Following isopropyl alcohol precipitation, total RNA was treated with DNase per the manufacturer's protocol. cDNA was synthesized from 1 μg of total RNA using the SuperScript III Reverse-Transcriptase Kit (Invitrogen) in a final volume of 20 μL. Real-time PCR reactions were routinely performed in triplicate in 96-well plates using the 7300 Real-Time PCR System instrument (Applied Biosystems). The reactions were incubated in 96-well optical plates at 95°C for 10 min, followed by 40 cycles at 95°C for 15 s, and at 60°C for 10 s. The threshold cycle (Ct) data were determined using default threshold settings. The Ct was defined as the fractional cycle number at which the fluorescence passes the fixed threshold. Mix primer sequences of mouse PPARγ were sense: GGAAAGACAACGGACAAATCA and anti-sense: TACGGATCGAAACTGGCAA; those of mouse Runx2 were sense: CGGTCTCCTTCCAGGATGGT and anti-sense: GCTTCCGTCAGCGTCAACA, those of mouse ALP were sense: GTTGCCAAGCTGGGAAGAACAC and anti-sense: CCCACCCCGCTATTCCAAAC, and those of mouse OCN were sense: GGGAGACAACAGGGAGGAAAC and anti-sense: CAGGCTTCCTGCCAGTACCT. The relative expression level of the transcripts was compared to GAPDH. The primer sequences of the mouse GADPH control were sense: TGCACCACCAACTGCTTAGC and anti-sense: CCACCACCCTGTTGCTGTAG.

Statistical analysis

Statistical analysis was performed using an appropriate Student's t-test when two values were being compared. No collected data or experimental animals were excluded from analysis. All analyses were performed in a blinded fashion when feasible. All “center values” represent the median, while all error bars represent one standard deviation. In general, *p<0.05 and **p<0.01 were considered significant. A Bonferroni correction was performed for tests with multiple comparisons.

Results

Lentiviral PPARγ shRNA inhibits BMSC adipogenesis in vitro

We first sought to verify the effectiveness of lentiviral delivered PPARγ shRNA in vitro. To do this, primary mouse BMSC were cultured in the presence of lentivirus encoding PPARγ shRNA or control (GFP). Oil red O staining and quantification of BMSC adipogenic differentiation confirmed a significant reduction in lipid droplet accumulation with lentiviral delivered PPARγ shRNA (Fig. 1A, B). Similarly, a significant reduction in Pparg transcript expression and PPARγ protein was confirmed in PPARy shRNA-treated samples by qRT-PCR analysis (Fig. 1C) and western blot (Fig. 1D, E). Furthermore, the PPARγ shRNA construct used in this study had no significant effects on Pparα or Pparδ transcript levels, as confirmed by qRT-PCR analysis (Supplementary Fig. S1; Supplementary Data are available online at www.liebertpub.com/tea). Significant cytotoxicity was not observed, as assayed by Trypan blue staining (not shown). In summary, these findings verified the in vitro efficacy, specificity, and anti-adipogenic effects of lentiviral delivered PPARγ shRNA.

FIG. 1.

BMSC adipogenic differentiation with PPARγ shRNA. (A) Oil Red O staining performed after 12 days differentiation of mouse BMSC infected with PPARγ shRNA lentivirus or GFP control lentivirus. (B) Photometric quantification of relative staining from (A). (C) Relative Pparg mRNA expression levels in BMSC after 3 days treatment. (D) Western blot analysis to determine expression levels of PPARγ protein in BMSC after 3 days treatment. The two bands correspond to PPARγ1 and PPARγ2. (E) Quantification of protein bands using densitometry. Experiments were independently run three times. The data presented in this figure are from one representative experiment. **p<0.01. BMSC, bone marrow stromal cells; GFP, green fluorescent protein; PPARγ, peroxisome proliferator-activated receptor gamma. Color images available online at www.liebertpub.com/tea

Lentiviral PPARγ shRNA significantly alters intramedullary histology

Next, the ramifications of in vivo lentiviral delivery of PPARγ shRNA were assessed. To directly affect BMSC, an intramedullary injection of lentivirus was performed in SCID mice. This was achieved via an intramedullary femoral injection, using a transpatellar (distal) approach. No postsurgical morbidity or mortality was encountered. Four weeks postinjection, animals were sacrificed and femurs with accompanying soft tissue were harvested for analysis. By routine H&E staining, a dramatic difference in marrow contents was observed at low and high magnifications (Fig. 2A, B). GFP-treated limbs showed a mixture of marrow elements, including mature adipocytes, and a combination of erythroid, myeloid, and megakaryocytic cell types. In marked contrast, PPARγ shRNA-treated limbs exhibited an essential absence of mature adipocytes. This was quantified, showing a mean marrow adiposity of 43.3% in control limbs as compared to 0.67% in PPARγ shRNA-treated limbs (Fig. 2C). To quantify the hematopoietic marrow elements, morphologic assessments were performed using high-powered analysis of 200 random cells per sample (Table 1). Little statistically significant differences were observed in hematopoietic cell content between GFP and PPARγ shRNA-treated samples.

FIG. 2.

Histologic appearance after PPARγ shRNA intramedullary injection. (A, B) H&E stained sections of the intramedullary space of the femur at low magnification (A) and high magnifications (B) demonstrate dramatic histologic changes after PPARγ shRNA lentivirus injection. GFP lentivirus-injected limbs showed a mixture of marrow elements, including mature adipocytes, and a combination erythroid, myeloid, and megakaryocytic cell types. In marked contrast, with PPARγ shRNA there was an essential absence of mature adipocytes in all samples. (C) Quantification of marrow adiposity among GFP control or PPARγ shRNA lentiviral treatment. Representative 100×and 400×images shown. n=6 samples per treatment group. **p<0.01. H&E, hematoxylin & eosin. Color images available online at www.liebertpub.com/tea

Table 1.

Bone Marrow Differential Count After PPARγ shRNA Intramedullary Injection

Cell type	GFP (mean%±SD)	PPARγ shRNA (mean%±SD)	p-Value
Erythroblast	0.51±0.87	0.66±0.63	0.819
Orthochromatic erythroblast	9.86±3.96	13.46±2.41	0.263
Polychromatic erythroblast	8.30±2.09	9.66±6.82	0.767
Myeloblast	1.84±1.90	1.14±1.09	0.618
Promyelocyte	4.01±4.34	6.45±3.91	0.510
Myelocyte	9.76±2.00	7.15±7.25	0.601
Metamyelocyte	11.20±2.50	8.90±3.12	0.378
Bands	14.37±0.77	14.62±3.36	0.913
Polymorphonuclear leukocyte	30.84±0.68	30.38±3.51	0.842
Lymphocyte	5.57±2.09	0.83±1.44	0.038
Megakaryocyte	3.75±0.90	6.76±0.66	0.012

Bone marrow differential counts were performed after GFP or PPARγ shRNA intramedullary injection, using 200 cell counts in random high-powered fields of n=3 samples from each treatment group. Overall, a similar cellular differential was observed between treatment groups. Numbers expressed as mean percentage±SD and p-value.

GFP, green fluorescent protein; SD, standard deviation.

Lentiviral PPARγ shRNA inhibits intramedullary adipogenesis

Having observed a striking reduction in marrow lipid content with PPARγ shRNA treatment, we next sought to extend these findings to adipocyte specific cytokines. First, reduction of Pparg transcript abundance was confirmed, using qRT-PCR analysis of total marrow cells 1 week postlentiviral injection (Fig. 3A). Immunohistochemical staining was next performed, confirming a near total reduction in observed positive staining for PPARγ at 4 weeks postinjection (Fig. 3B, C). Similarly, FABP4 showed a marked reduction in immunohistochemical staining and semi-quantification (Fig. 3D, E). In summary, lentiviral delivered PPARγ shRNA resulted in a significant reduction in the intramarrow lipid content, associated with reduction of multiple adipocyte specific markers.

FIG. 3.

Adipogenic marker expression after PPARγ shRNA intramedullary injection. (A) Relative intramarrow Pparg mRNA expression levels 1 week postinjection, assessed by qRT-PCR. (B, C) Immunohistochemical staining and semi-quantification of PPARγ. Semi-quantification is expressed as the relative integrated intensity of intramarrow staining. (D, E) Immunohistochemical staining and semi-quantification of fatty acid-binding protein 4 (FABP4). Semi-quantification is expressed as the relative integrated intensity of intramarrow staining. Representative images shown at 400×. All semi-quantitation based on six random images per sample. n=6 samples per treatment group. **p<0.01. qRT-PCR, quantitative real time polymerase chain reaction. Color images available online at www.liebertpub.com/tea

Lentiviral PPARγ shRNA increases intramedullary trabecular bone

Recent studies have documented the anti-osteoblastic effects of PPARγ signaling,¹⁶ and the negative effects of intramarrow adipocytes on bone health.³ This led us to examine the effects of PPARγ shRNA treatment on femoral bone formation. High-resolution microCT cross sections of the distal femoral metaphysis showed a relative increase in trabecular bone among PPARγ shRNA-treated limbs (Fig. 4A). In contrast, cortical bone thickness appeared unaffected. Three-dimensional microCT reconstructions of the distal femur trabecular bone again showed a qualitative increase in trabecular bone with PPARγ shRNA treatment (Fig. 4B). Next, microCT quantifications of metaphyseal trabecular bone were performed (Fig. 4C–H). Significant increases in parameters including bone volume (BV) and percentage BV/tissue volume (BV/TV) were observed with PPARγ shRNA treatment (Fig. 4D, E). Trabecular analyses were likewise significantly improved with PPARγ shRNA treatment, including a significant increase in trabecular thickness (Tb.Th) and Tb.N, and a nonsignificant trend toward reduced trabecular separation (Tb.Sp) (Fig. 4F–H). In summary, these data suggested that lentiviral delivered PPARγ shRNA treatment had not only anti-adipocytic effects, but also exerted significant and positive effects on intramedullary trabecular bone.

FIG. 4.

Trabecular bone formation after PPARγ shRNA intramedullary injection. (A) Representative two-dimensional microCT images of femoral transverse sections from GFP and PPARγ shRNA treatment groups. Images were taken from the distal femur. (B) Representative three-dimensional microCT reconstructions of distal femoral intramedullary trabecular bone from the GFP and PPARγ shRNA treatment groups. (C–H) microCT quantification data, including bone marrow density (BMD), bone volume (BV), Percent BV/tissue volume (BV/TV), trabecular thickness (Tb.Th), trabecular number (Tb.N), and trabecular separation (Tb.Sp). n=6 samples per treatment group. *p<0.05; **p<0.01. microCT, microcomputed tomography. Color images available online at www.liebertpub.com/tea

Lentiviral PPARγ shRNA alters the balance of osteoblast: osteoclast numbers

To further define the positive effects of PPARγ shRNA treatment on intramedullary bone, histologic analyses were performed (Fig. 5A). Histomorphometric analyses of serial histologic sections confirmed microCT findings, including significant increases in Trabecular B.Ar, trabecular perimeter (T.Pm), Tb.Wi, and Tb.N, and a significant reduction in Tb.Sp with PPARγ shRNA treatment (Fig. 5B–F). Alterations in trabecular bone could be the result of increased osteoblastic activity, decreased osteoclast-mediated bone resorption, or both. To address this, we first examined osteoprogenitor/osteoblast specific markers, by performing immunohistochemical staining for OCN (Fig. 5G) and ALP activity (Fig. 5I). Semi-quantification of immunohistochemical stains revealed a significant increase in both the number of OCN+and ALP+osteoblasts per B.Pm (Fig. 5H, J). Next, an in vitro correlate was performed in which BMSC were cultured under osteogenic conditions. To assess osteogenic differentiation, qRT-PCR was used to measure the mRNA expression levels of osteogenic differentiation markers, including Runx2, OCN, and ALP. Results showed that at 3, 6, and 9 days, PPARγ shRNA significantly upregulated the expression of all three osteogenic gene markers (Fig. 6A–C).

FIG. 5.

Osteoblast markers after PPARγ shRNA intramedullary injection. (A) Representative images of Masson's Trichrome staining after GFP or PPARγ shRNA treatment. (B–F) Trabecular bone histomorphometric quantification performed on serial H&E sections of GFP or PPARγ shRNA-treated samples, including bone area (B.Ar), bone perimeter (B.Pm), Tb.Wi, Tb.N, and Tb.Sp. (G, H) Immunohistochemical staining and semi-quantification of osteocalcin (OCN). Semi-quantification is expressed as the number of OCN+osteoblasts per B.Pm. (I, J) Immunohistochemical staining and semi-quantification of alkaline phosphatase (ALP). Semi-quantification is expressed as the number of ALP+osteoblasts per B.Pm. Arrowheads indicate positively stained cells. Images shown at 400×. All semi-quantitation based on six random images per sample. n=6 samples per treatment group. **p<0.01. Color images available online at www.liebertpub.com/tea

FIG. 6.

Osteogenic differentiation of BMSC after PPARγ shRNA treatment. (A) Relative Runx2 mRNA expression levels at 3, 6, and 9 days in osteogenic differentiation medium, as assessed by real-time PCR. (B) Relative ALP mRNA expression levels at 3, 6, and 9 days in osteogenic differentiation medium, as assessed by qRT-PCR. (C) Relative osteocalcin mRNA expression levels at 3, 6, and 9 days in osteogenic differentiation medium, as assessed by qRT-PCR. The cell culture experiments were independently run three times. The data presented in this figure are from one representative experiment. **p<0.01 compared to GFP control at that time point.

Next, the effects of PPARγ shRNA treatment on osteoclastic activity were assessed. Osteoclast-specific markers were examined by assessment of TRAP activity (Fig. 7A, B) and immunohistochemical staining for Cathepsin K (Fig. 7C, D). Both osteoclast markers showed a significant reduction in activity or immunohistochemical staining when treated with PPARγ shRNA. Lastly, cellular proliferation and apoptosis was assessed via PCNA and TUNEL staining, respectively. No difference in either parameter was assessed by staining and semi-quantitation (Fig. 7E–H).

FIG. 7.

Osteoclast markers and proliferation/apoptosis after PPARγ shRNA intramedullary injection. (A, B) Histochemical staining and semi-quantification of tartrate-resistant acid phosphatase (TRAP). Semi-quantification is expressed as the number of TRAP+osteoclasts per B.Pm. (C, D) Immunohistochemical staining and semi-quantification of Cathepsin K. Semi-quantification is expressed as the relative intensity of Cathepsin K staining per high powered field. (E, F) Immunohistochemical staining and semi-quantification of PCNA. Semi-quantification is expressed as the number of PCNA+cells per B.Pm. (G, H) Histochemical staining and semi-quantification of TUNEL. Semi-quantification is expressed as the percentage of TUNEL stained cells per high powered field. Arrowheads indicate positively stained cells. Images shown at 400×. All semi-quantitation based on six random images per sample. n=6 samples per treatment group. **p<0.01. PCNA, proliferating cell nuclear antigen; TUNEL, terminal deoxynucleotidyl transferase dUTP nick end labeling. Color images available online at www.liebertpub.com/tea

Discussion

In summary, we have observed that PPARγ shRNA delivery via a lentiviral vector not only led to a dramatic reduction in intramarrow adiposity, but also led to a significant improvement in trabecular bone architecture and alteration in osteoblast and osteoclast numbers. These findings suggest that loss of PPARγ in the bone marrow compartment has a significant role beyond anti-adipose effects. These effects were acute in onset, with the development of a long bone phenotype in only 4 weeks post-treatment.

As mentioned, the reduction in intramarrow adipocytes was quite dramatic and seen across all animals examined. This led us to inquire as to the mechanism of such a relatively rapid change in marrow composition. We examined in vivo intramarrow proliferation and apoptosis, finding no difference in either cell parameter with PPARγ shRNA treatment. Likewise, cell toxicity was not observed under any condition or in any cell type examined with PPARγ shRNA. Several potential explanations for this striking observation exist. First, although unlikely, PPARγ shRNA treatment may have led to a brisk and resolving pro-apoptotic effect on intramarrow adipocytes that was not seen in our 4 week time point. We consider this unlikely as cell death such as this would have likely resulted in residual inflammation, edema, or fibrosis—none of which were observed with PPARγ shRNA treatment. Along these lines, most sources consider PPARγ signaling to have pro-apoptotic effects itself,²⁵ and so PPARγ silencing as in our study would unlikely have the same effect. A second and more likely explanation for this relatively rapid disappearance of intramarrow differentiation is transdifferentiation of mature fat cells induced by PPARγ shRNA treatment. Theoretically, the transdifferentiation of mature mesenchymal cells to that of another lineage has been repeatedly described. For example, mature white adipocytes have been described to transdifferentiate into osteoblastic cells²⁶ or adopt a brown adipocyte phenotype under specific conditions.²⁷ Other examples include the transdifferentiation of fully differentiated osteoblasts²⁸ or mature myocytes into mature adipocytes.²⁹ Certainly a similar phenomenon of transdifferentiation of mature adipocytes may have been observed in our study, although the elucidation of this phenomenon requires further study.

Interesting similarities and differences exist between the phenotype induced by PPARγ shRNA treatment and the skeletal phenotype induced by deficiency in PPARγ. The effects of complete knockout of PPARγ has been difficult to study, as homozygous knockout mice demonstrate lethality in the midembryonic stage.³⁰ However, heterozygosity in PPARγ (PPARγ^+/− mice) or lineage-specific PPARγ knockout mice have been studied in terms of both skeletal and adipose tissue phenotypes. For example, culture of PPARγ^+/− BMSC revealed not only reduced adipogenic differentiation (as anticipated), but also increased ALP activity and von Kossa staining.³¹ An increased expression of Ruxn2, Osterix, and Lrp5 was also observed.³¹ This is in direct agreement with our findings, in which PPARγ shRNA treatment led to increased expression of Runx2, ALP, and OCN expression in BMSC. However, this increase in osteogenic differentiation was not observed with PPARγ^+/− calvarial osteoblasts.²⁰ Thus, it appears that loss of PPARγ signaling has its osteoblastogenic effects on less differentiated osteoprogenitor cells. Interesting future studies would involve using PPARγ shRNA treatment on various osteoprogenitor cell types, to confirm the hypothesis that only uncommitted or early osteoprogenitor cells are responsive to PPARγ loss.

Studies regarding PPARγ deficiency in osteoclasts have been less comprehensive. For example, Kawaguchi et al. found no difference in PPARγ^+/− osteoclastic activity either when isolated or when co-cultured with PPARγ^+/− osteoblasts.²⁰ Wan et al. used osteoclast lineage-specific deletion of PPARγ with Tie2Cre mice to show that PPARγ deficiency results in increased bone density and reduced osteoclastic number and activity.²¹ Our results with PPARγ shRNA mediated inhibition of osteoclast numbers parallel and confirm these results from osteoclast-specific PPARγ deletion. Interestingly, the skeletal phenotype observed in our study, after only 4 weeks post- shRNA-mediated PPARγ inhibition, was remarkably similar to that observed by other investigators after lifelong PPARγ depletion. Further study is required to more fully ascertain the differences in acute versus chronic loss of PPARγ expression.

Clear limitations exist in the model studied herein. For example, intramedullary lentiviral delivered PPARγ shRNA is a relatively impractical approach for clinical translation, involving both the surgical risks from an invasive procedure, and the risks associated with insertional mutagenesis. However, we have shown that acute onset loss of PPARγ activity in the bone marrow niche has dramatic repercussions on intramarrow adipocytes, osteoblasts, and osteoclasts. Whether these effects of PPARγ shRNA can be sustained to improve bone strength have yet to be investigated. For example, these experiments were performed in young mice, while the clinical need for improved bone microarchitecture is primarily in the aged skeleton. Kawaguchi et al. found that the high bone mass phenotype of PPARγ haploinsufficient mice becomes more pronounced with age,²⁰ potentially suggesting age-related changes in the response to PPARγ loss. Future studies will plan to elucidate the differential responsive of young and aged mice to PPARγ shRNA treatment.

An alternative approach would be the pharmacologic inhibition of PPARγ, which indeed has been pursued. Duque et al. used the PPARγ antagonist BADGE (bisphenol-A-diglycidyl ether) to improve BV and decrease intramarrow adiposity in skeletally mature mice.³² However, other research groups have used BADGE treatment in mice without a discernable benefit to bone parameters.³³ Thus, the efficacy of pharmacologic PPARγ inhibition for improved bone health is not yet known.

Acknowledgments

This work was supported by the CIRM Early Translational II Research Award TR2-01821, NIH/NIDCR (grants R21 DE0177711 and RO1 DE01607), UC Discovery Grant 07-10677, and the Eli and Edythe Broad Center of Regenerative Medicine and Stem Cell Research at UCLA Innovation Award.

The authors thank L. Chang, K. Le, M.A. Scott, and A.S. James for their excellent technical assistance.

Disclosure Statement

No competing financial interests exist.