Tetracycline-regulated bone morphogenetic protein 2 gene expression in lentivirally transduced primary rabbit chondrocytes for treatment of cartilage defects




Treatment of cartilage defects is still challenging, primarily because of the poor self-healing capacity of articular cartilage. Gene therapy approaches have gained considerable attention, but, depending on the vector system used, they can lead to either limited or unrestrained gene expression, and therefore regulation of gene expression is necessary. This study was undertaken to construct an efficient tetracycline (Tet)–regulated, lentivirally mediated system for the expression of growth factor bone morphogenetic protein 2 (BMP-2) in primary rabbit chondrocytes that will allow for the induction and termination of growth factor gene expression once cartilage regeneration is complete.


Chondrogenic ATDC5 cells and primary rabbit chondrocytes were lentivirally transduced with different tetracycline-on (Tet-On)–regulated, self-inactivating vectors for the induction of expression of enhanced green fluorescent protein (eGFP) or BMP-2, using either a 1-vector system or a 2-vector system.


Expression of eGFP was induced on ATDC5 cells and chondrocytes. The highest induction rate and highest level of gene expression were reached when the spleen focus-forming virus long terminal repeat promoter was used to drive the reverse transactivator expression, after the addition of doxycycline, in chondrocytes. An up to 20-fold induction of Tet-mediated BMP-2 expression was observed on ATDC5 cells. The extent of induction and expression level of BMP-2 in chondrocytes were similar between the 1-vector system– and 2-vector system–infected cells (mean ± SD 15.5 ± 1.1 ng/ml and 14.6 ± 0.4 ng/ml, respectively). In addition, prolonged induction and switching-off of BMP-2 expression, as well as repeated induction, were demonstrated. Production of proteoglycans, as shown by Alcian blue staining, demonstrated the functionality of the lentivirally expressed BMP-2 under induced conditions.


The lentivirally mediated Tet-On system is an effective strategy for efficient, repeatedly inducible expression of BMP-2 in primary rabbit chondrocytes. Therefore, use of this system in in vivo experiments may be a promising approach as a treatment strategy for cartilage defects.

Treatment of cartilage defects is still challenging (1) because it is limited by the poor self-healing capacity of articular cartilage (2), due to the fact that cartilage is an avascular tissue (3–5). Protein application of distinct growth factors, such as members of the bone morphogenetic protein (BMP) family, that are known to enhance/promote cartilage healing (6–9) is an approach that has only a short, transient effect. BMPs are members of the transforming growth factor β superfamily (10). BMP-2 seems to be involved in the growth of mesenchymal stem cells and in the differentiation of mesenchymal stem cells to chondroblasts and osteoblasts (11–13).

Application of recombinant BMP-2 is restricted by the high doses that are required to induce biologic effects, as well as the fact that repeated applications are needed because of the short half-life of the growth factor (14). Repeated applications, however, are associated with a risk of joint infection (15). For these reasons, gene transfer approaches for a stable expression of therapeutic proteins, e.g., growth factors, are promising (16–20).

Viral vectors have been demonstrated to be efficient tools for gene therapy (21). Different viral vectors, i.e., adenoviral vectors (22, 23), adeno-associated virus vectors (24), retroviral vectors (11, 25, 26), and lentiviral vectors (27), have been tested for transgene delivery in chondrocytes and cartilage tissue as a strategy for the treatment of various joint disorders. Lentiviral vectors allow for stable transgene expression, as a result of the integration of the viral genome into the host genome. To avoid continuous gene expression after cartilage healing has occurred, the regulation of gene expression, specifically, the termination of gene expression, is needed. In various gene therapy settings, regulation of gene expression has been achieved using the tetracycline-on (Tet-On) system, which requires the addition of the tetracycline (Tet) derivate doxycycline for induction of transgene expression. It is highly specific and nontoxic (28, 29), and induction of transgene expression is completely reversible (30).

Clinical studies by Bartlett et al (31), Saris et al (2), and Brittberg et al (32) showed that transplantation of non–genetically modified chondrocytes into chondral defects might improve cartilage healing. Type I collagen sponges, fibrin glue, and gels have previously been shown to be suitable carriers for chondrocytes (20, 26, 33, 34). Our aim was to create a combined gene and cell therapy approach for the transplantation of ex vivo–transduced chondrocytes as a treatment strategy for chondral and osteochondral defects. To our knowledge, this is the first report describing VSV.G-pseudotyped lentiviral vectors that allow for inducible and repeated expression of growth factors in primary rabbit chondrocytes. Furthermore, we compare different vector systems for regulation of gene expression, and describe the kinetics of regulation and functionality of the system in vitro.


Cell culture of ATDC5 cells and primary rabbit chondrocytes.

ATDC5 cells were cultured in 50% Dulbecco's modified Eagle's medium (DMEM)/50% Ham's F-12 medium (Biochrom) supplemented with 1% glutamine and 10% fetal calf serum (FCS; PAN Biotech) and subcultured biweekly, at a ratio of 1:5 or 1:6, with 0.05% trypsin/EDTA solution (Biochrom).

Primary rabbit chondrocytes were isolated from the knee joints of female New Zealand white rabbits (mean weight 3.5 kg) as described previously (35). Cells were grown in monolayer in DMEM containing 1% penicillin/streptomycin (Sigma-Aldrich) and 10% FCS. Chondrocytes were fed 2 times per week, and confluent cells were split in a ratio of 1:2 or 1:3. All cells were incubated at 37°C in 5% CO2 in a humidified incubator.

Lentiviral transfer vectors.

All transfer vectors are derivates of plasmid pHIV-7 (36). Sequences expressing enhanced green fluorescent protein (eGFP), BMP-2, or the reverse Tet transactivator (rtTA2S-M2) (29), under control of the immediate-early cytomegalovirus (CMV), spleen focus-forming (SF) virus, or phosphoglycerate kinase (PGK) promoter or the Tet response element (TRE/Tight) (28), were isolated from donor plasmids, filled in by the Klenow enzyme, and inserted at the unique Bam HI restriction site of pHIV-7 by blunt-end ligation. Orientation was determined by restriction enzyme analyses.

For cloning of 1-vector constructs, cassettes consisting of the desired promoter and the transactivator sequence were inserted downstream of the transgene expression cassette (Figure 1). All constructs were verified by DNA sequencing.

Figure 1.

Structure of integrated proviruses for regulated gene expression of enhanced green fluorescent protein (eGFP) or bone morphogenetic protein 2 (BMP-2) under control of the tetracycline-on system. A, Transactivator vector. B, Response vector. C, One-vector construct. R = redundant region; U5 = unique region; rtTA2S-M2 = reverse tetracycline transactivator; WPRE = Woodchuck posttranscriptional regulatory element; Δ3′LTR = deletion at the 3′ long terminal repeat; CMV = cytomegalovirus; PGK = phosphoglycerate kinase; SF = spleen focus-forming (virus); Tre/Tight = tetracycline response element.

Transient transfection for production of VSV.G-pseudotyped lentivirus.

Lentiviral vectors were produced by transient transfection of 293T cells (37), according to the method described by Barry et al (38), with 3 packaging plasmids (pMDLg/pRRE, pRSV-Rev, and pMD.G) (39) and the respective transfer vector. Viral supernatants were filtered through 0.45-μm filters and frozen at −80°C.

Viral titers were measured using a p24 enzyme-linked immunosorbent assay (ELISA) (QuickTiter Lentivirus Quantitation Kit; Cell Biolabs) according to the manufacturer's protocol, or by flow cytometry titration (FACS Vantage; Becton Dickinson) on murine CMS5 cells (40) for dilutions of eGFP-expressing virus. For those vectors not expressing eGFP, viral doses, expressed in transducing units (TU), were estimated on the basis of p24 ELISA data, assuming the same particle: TU ratio for all lentiviral vectors.

Lentiviral infection and induction with the Tet-On system.

Twenty-four hours before infection, 1 × 105 chondrocytes or 3 × 105 ATDC5 cells were seeded on 60-mm dishes. Cells were infected with 1 ml diluted lentiviral supernatant containing 8 μg/ml polybrene, at various multiplicities of infection (MOIs). Two hours after infection, 4 ml medium was added. Approximately 5 days before being analyzed, Tet-dependent gene expression was induced by the addition of 1 μg/ml doxycycline (Sigma-Aldrich).

Analysis of cells for eGFP expression.

The amount of eGFP+ cells and the mean fluorescence intensity (MFI) of the cells were determined by flow cytometric analyses (FACS Vantage; Becton Dickinson) of 2 × 104 cells.

Quantification of BMP-2 production.

Production of BMP-2 protein was measured using a BMP-2 ELISA in supernatants after incubation for up to 72 hours (Quantikine Immunoassay BMP-2; R&D Systems), according to the manufacturer's protocol. For the kinetics studies, supernatants were collected and cells were counted 24 hours after the medium was changed. For analysis of repeated induction of BMP-2, doxycycline was added to cells transduced with the TRE-driven BMP-2 SF rtTA construct (TBS) for 3 days during each cycle. BMP-2 was measured at the beginning, after each induction cycle, and 2 weeks after withdrawal of doxycycline.

Supernatants were centrifuged to avoid influences arising from cells or other particles, and were subsequently stored at −80°C until protein measurements. Levels of BMP-2 production were normalized to the values for 1 × 106 cells and 24 hours.

Detection of proteoglycans.

Proteoglycans were detected by Alcian blue staining. Chondrocytes (4 × 104/12 wells) or ATDC5 cells (1 × 105/6 wells) were seeded on multiwell plates and cultivated with medium containing 37.5 μg/ml ascorbic acid (41) for 21 days or 17 days, respectively. The medium was changed in intervals of 3 or 4 days. Staining and quantification of extracted dye was performed as described previously (42).

Statistical analysis.

Each experiment was performed 3 times (n = 3 each), and results are expressed as the mean ± SD. Comparisons between the induced and uninduced groups were made using the t-test. P values less than 0.05 were considered significant. Statistical evaluations were performed using GraphPad Prism software. Estimations of the half-life of BMP-2 induction and reduction were calculated separately from linear regressions of the respective part of the curve derived from the BMP-2 ELISA.


Lentiviral infection of target cells.

Murine chondrogenic ATDC5 cells and primary rabbit chondrocytes were infected with lentiviral self-inactivating vectors at MOIs of 2.5 and 7.5, respectively. In addition to constitutive transgene expression (achieved with SFeGFP [SE] or SFBMP-2 [SB]), both cell types were infected with different constructs to allow regulated gene expression under control of the Tet-On system. The different constructs produced (CMVrtTA [C], SFrtTA [S], or PGKrtTA [P]) were used to determine the promoter best suited for expression of the rtTA (Figure 1A). Cells were either coinfected with the transactivator vector (C, S, or P) (Figure 1B) and the TREeGFP (TE) or TREBMP-2 (TB) response vector for the 2-vector approach, or were infected with 1-vector constructs (TREeGFPSFrtTA [TES], TREeGFPPGKrtTA [TEP], TREBMP-2SFrtTA [TBS], or TREBMP-2PGKrtTA [TBP]) (Figure 1C), which provide transactivator and response sequences on a single provirus. Tet-dependent gene expression was induced by the addition of 1 μg/ml doxycycline ∼5 days before analysis.

Regulated eGFP expression on infected cells.

Expression of eGFP on ATDC5 cells was used to show inducibility of the Tet-On system. Regarding the percentage of eGFP+ cells, the strongest induction by the Tet-On system on ATDC5 cells was observed on TE + S–infected cells (a 2-vector system; mean ± SD 2.8 ± 0.1–fold induction relative to uninduced cells) (Figure 2A), with a similar extent of induction from TES-infected cells (a 1-vector system; mean ± SD 2.5 ± 0.1–fold induction relative to uninduced cells). However, the overall percentage of induced eGFP+ cells increased from 16.7% when using the 2-vector system of coinfection to 34.8% when using the corresponding 1-vector system. The percentage of detectable eGFP+ cells after induction of TES-infected cells was in the same range as that with the PGK promoter–driven constitutive construct, the latter of which resulted in 39.6% eGFP+ cells.

Figure 2.

Comparison of constitutive enhanced green fluorescent protein (eGFP) expression with tetracycline-induced eGFP expression on ATDC5 cells (A) and primary rabbit chondrocytes (B). ATDC5 cells and chondrocytes were infected with each virus at multiplicities of infection of 2.5 and 7.5, respectively. The tetracycline-on system was induced by the addition of 1 μg/ml doxycycline (Dox). Cells were analyzed by fluorescence-activated cell sorter analysis for the percentage of eGFP+ cells (n = 3) treated with or without doxycycline; results show highly significant differences between the induced and uninduced samples (P < 0.0001) for those samples transduced with the cytomegalovirus (CMV) promoter–driven reverse tetracycline transactivator (rtTA) and Tet response element (TRE) eGFP construct (TE + C), the phosphoglycerate kinase (PGK) promoter–driven rtTA and TRE eGFP construct (TE + P), the spleen focus-forming (SF) virus promoter–driven rtTA and TRE eGFP construct (TE + S), and the TRE-driven eGFP SFrtTA construct (TES), with no significant differences for all control samples (uninfected [−] and those transduced with CMVrtTA [C], PGKrtTA [P], SFrtTA [S], and TREeGFP [TE]). PE = PGKeGFP; SE = SFeGFP.

Induction of the MFI of ATDC5 cells under control of the Tet-On system was also comparable between TE + S–infected and TES-infected samples (mean ± SD 19.6 ± 0.5–fold induction and 20.1 ± 0.5–fold induction, respectively, relative to uninduced cells) (Table 1). After induction, the MFI of TES-infected cells was stronger than that from constitutively expressing PE-infected cells. After induction by doxycycline, the increase in the MFI, as compared with samples without treatment with doxycycline, was not significant for all control groups (uninfected and C-, P-, S-, and TE-infected samples) but was highly significantly different (P < 0.0001) for all other samples.

Table 1. Mean fluorescence intensity (MFI) of transduced ATDC5 cells and primary rabbit chondrocytes, as determined by flow cytometry*
Without DoxWith DoxWithout DoxWith Dox
  • *

    Values are the mean ± SD MFI for enhanced green fluorescent protein (eGFP) expression with or without induction with doxycycline (Dox). ND = not determined; C = cytomegalovirus (CMV) reverse tetracycline transactivator (rtTA); P = phosphoglycerate kinase (PGK) rtTA; S = spleen focus-forming (SF) virus rtTA; TE = Tet response element (TRE) driving eGFP; TE + C = 2-vector TRE driving eGFP + CMV rtTA; TE + P = 2-vector TRE driving eGFP + PGK rtTA; TE + S = 2-vector TRE driving eGFP + SF virus rtTA; TES = 1-vector TREeGFPSFrtTA; PE = constitutive PGKeGFP; SE = constitutive SFeGFP.

Untransduced cells11.0 ± 3.5ND46.5 ± 9.0ND
Transduced cells    
 C11.7 ± 3.310.9 ± 1.544.8 ± 5.444.5 ± 2.5
 P10.2 ± 4.410.1 ± 5.247.0 ± 1.446.9 ± 1.1
 S12.7 ± 4.310.0 ± 1.962.8 ± 25.747.2 ± 3.2
 TE36.9 ± 3.136.9 ± 2.5102.9 ± 23.088.50 ± 7.8
 TE + C36.0 ± 0.7177.8 ± 12.244.6 ± 0.8150.4 ± 7.6
 TE + P37.1 ± 1.8108.2 ± 8.249.3 ± 6.9330.8 ± 12.2
 TE + S36.7 ± 0.3719.8 ± 10.336.4 ± 6.4434.7 ± 57.6
 TES24.7 ± 0.5495.5 ± 3.2143.3 ± 35.7535.1 ± 3.8
 PE218.5 ± 0.9ND792.5 ± 21.0ND
 SE1,513.4 ± 12.3ND6,015.3 ± 133.7ND

The system was then tested on primary rabbit chondrocytes. Comparison of the effects of the different promoters in the 2-vector system on chondrocytes revealed little or no inducibility for eGFP by the CMV promoter–driven rtTA construct (TE + C) (Table 1 and Figure 2B). The PGK promoter–driven construct (TE + P) resulted in 26.1% eGFP+ cells, as compared with 19.2% eGFP+ cells in samples infected with the SF promoter–driven construct (TE + S) (mean ± SD 7.2 ± 0.3–fold induction for TE + P versus 3.8 ± 0.3–fold induction for TE + S, relative to uninduced cells). The highest percentage of induced positive cells, 36.8%, was observed after infection of chondrocytes with the corresponding 1-vector construct (12.1 ± 1.0–fold induction). Constitutive eGFP expression comprised 41.9% eGFP+ cells (in PE-infected samples).

Regarding the MFI of primary rabbit chondrocytes, the SF promoter–driven transactivator in the 2-vector approach was superior to the PGK promoter–driven construct (mean ± SD 6.8 ± 1.0–fold induction for TE + P versus 12.0 ± 0.6–fold induction for TE + S). The highest level of induction of eGFP expression was observed using the corresponding 1-vector construct (TES) (mean ± SD 3.9 ± 1.1–fold induction). Nevertheless, the MFI of constitutive eGFP expression was higher than that from induced TES-infected cells. After induction by doxycycline, the increase in the MFI, as compared with samples without treatment with doxycycline, was not significant in all control groups (uninfected and C-, P-, S-, and TE-infected samples) but was highly significantly different (P < 0.0001) for all other induced samples. Control cells showed no increase in either the percentage of eGFP+ cells or the MFI.

BMP-2 expression in infected cells.

Based on the previous experiments, we then analyzed BMP-2 expression using the 1- and 2-vector systems for induced expression involving the rtTA under control of the SF virus promoter (TBS and TB + S) and the PGK promoter (TBP and TB + P) in infected cells. Secreted BMP-2 was determined by ELISA. The strongest induction rate of Tet-mediated BMP-2 expression was again observed in ATDC5 cells that were infected with the SF-controlled transactivator, both in the 1-vector system and in the 2-vector approach (mean ± SD 20.4 ± 5.4 ng/ml and 11.6 ± 4.6 ng/ml, respectively, normalized to the values for 1 × 106 cells and 24 hours) (Figure 3A). The expression level of BMP-2 produced from TBS-infected ATDC5 cells after induction was in the same range as the BMP-2 expression level produced from SB-infected cells.

Figure 3.

Induction of bone morphogenetic protein 2 (BMP-2) expression on ATDC5 cells (A) and primary rabbit chondrocytes (B and C). Cells were infected with each virus at multiplicities of infection of 2.5 (for ATDC5) or 7.5 (for chondrocytes). The tetracycline-on system was induced by the addition of 1 μg/ml doxycycline. The amount of BMP-2 protein produced was quantified from the supernatants after 24 hours or 72 hours of incubation using a BMP-2 enzyme-linked immunosorbent assay. To assess the kinetics of tetracycline-dependent BMP-2 expression on chondrocytes, doxycycline was added on days 0–4 (C) or days 1–25 (D). BMP-2 production was measured at the time that the medium was changed, in 24-hour intervals. All results are expressed as the mean ± SD, normalized to the values for 1 × 105 or 1 × 106 cells and 24 hours (n = 3 each). A, In ATDC5 cells, significant differences in the induction of BMP-2 expression were observed for those cells treated with doxycycline compared with those left untreated among the samples transduced with the PGK promoter–driven BMP-2 rtTA construct (TB + P) (P < 0.0001), the SF virus promoter–driven BMP-2 rtTA construct (TB + S) (P = 0.025), the TRE-driven BMP-2 PGK rtTA construct (TBP) (P = 0.0009), and the TRE-driven BMP-2 SF rtTA construct (TBS) (P = 0.0005). B, On chondrocytes, the induction of BMP-2 expression was highly significant (P < 0.0001) in all samples except for the TBP-infected samples (P = 0.2325). There was no significant induction on untransduced (Untx) cells. PB = PGKBMP-2; SB = SFBMP-2 (see Figure 2 for other definitions).

On primary rabbit chondrocytes, the extent of induction and overall amount of BMP-2 production were similar between TB + S–infected and TBS-infected chondrocytes (mean ± SD 14.6 ± 0.4 ng/ml and 15.5 ± 1.1 ng/ml, respectively, normalized to the values for 1 × 106 cells and 24 hours). Unlike the findings in ATDC5 cells, BMP-2 production from SB-infected samples of primary rabbit chondrocytes was 30-fold higher than that from induced TB + S–infected or TBS-infected samples (Figure 3B). There was no detectable BMP-2 production in any control cells, nor was there any BMP-2 production in the absence of doxycycline.

Kinetics of Tet-mediated BMP-2 expression.

With regard to further uses for the in vivo application of cells that allow Tet-induced BMP-2 expression for treatment of osteochondral defects, it is important to investigate the kinetics of transgene expression after the addition and withdrawal of doxycycline. Supernatants were collected from TB + S–infected and TBS-infected cells for 9 days to compare the kinetics of the 1- and 2-vector systems. Each day, the medium was changed and cells were counted in separate sets of wells. Doxycycline was added from day 0 to day 4. The strongest BMP-2 expression was observed from cells that were infected with the 1-vector system containing an SF promoter–controlled transactivator (TBS) (Figure 3C). The expression of BMP-2 increased from day 0 to day 4, at which point the maximal induction was reached. Upon withdrawal of doxycycline, the BMP-2 secretion decreased to near background levels on day 8 (Figure 3C).

In addition, supernatants were collected from 2 sets of TBS-infected cells (TBS and TBS + D) (Figure 3D). TBS + D–infected cells were induced by doxycycline from day 1 to day 25, in order to show the kinetics of induction and prolonged Tet-mediated BMP-2 expression. Thereafter, doxycycline was withdrawn from the samples. TBS-infected cells were not induced during the whole experiment. On days 1–7, 10, 15, 18, 22, 25, and 29–33, supernatants were collected and cells were counted, 24 hours after the medium had been changed. The expression of BMP-2 increased from day 1 to day 4, when the maximal induction was reached. Between day 4 and day 25, Tet-mediated BMP-2 expression continued without a significant loss of efficacy (mean ± SD 0.83 ± 0.23 ng/ml for 1 × 105 cells and 24 hours). Upon withdrawal of doxycycline, the BMP-2 secretion decreased to background levels on day 29 (Figure 3D).

Repeated induction of Tet-dependent BMP-2 expression.

The integration of BMP-2–expressing sequences into the host genome potentially allows for repeated induction of transgene expression. Due to the use of a strong viral promoter (SF promoter) for expression of the transactivator, expression might be limited by DNA methylation, leading to silencing of gene activity. Therefore, we induced TBS-infected primary rabbit chondrocytes 3 times over 6 weeks (Figure 4). Each induction cycle included the addition of 1 μg/ml doxycycline for 3 days to samples of infected cells (infected with TBS + D), whereas uninduced cells (TBS-infected cells) served as the control. Between 2 induction cycles, no doxycycline was added. The amount of secreted BMP-2 was measured at the beginning of the experiment, after each induction cycle, and 2 weeks after the withdrawal of doxycycline.

Figure 4.

Repeated induction of tetracycline-dependent bone morphogenetic protein 2 (BMP-2) expression. Primary rabbit chondrocytes were infected at a multiplicity of infection of 10. Three cycles of induction with 1 μg/ml doxycycline (arrows) demonstrated the potential for specific reinduction of BMP-2 expression. One fraction of cells infected with the TRE-driven BMP-2 SF rtTA construct (TBS) was treated with doxycycline for 3 days during each induction cycle, while control samples (TBS-infected) were never induced. Secreted BMP-2 was measured by enzyme-linked immunosorbent assay at the beginning (week 0), after each induction cycle (weeks 1, 4, and 7), and 2 weeks after withdrawal of doxycycline (weeks 3 and 6). All results are expressed as the mean ± SD, normalized to the values for 1 × 105 cells (n = 3 each). Significant differences between the groups treated without or with doxycycline were observed after 1 week (P = 0.0020), 3 weeks (P = 0.0017), 4 weeks (P = 0.0243), and 7 weeks (P = 0.0299), but no significant differences were observed after 6 weeks (P = 0.5039). See Figure 2 for other definitions.

We were able to show that repeated induction cycles over a time period of 6 weeks results in inducible BMP-2 expression without loss of efficacy (mean 1,059.9 pg/ml BMP-2 in the first cycle, 596.1 pg/ml in the second cycle, and 1,555.5 pg/ml in the third cycle) (Figure 4). After the withdrawal of doxycycline, the amount of expressed BMP-2 declined to near background levels (Figure 4). There was no increase in BMP-2 expression from uninduced cells.

Effect of BMP-2 production on synthesis of cartilage-specific proteoglycans.

In order to show that lentiviral regulation of the in situ expression of BMP-2 is sufficient to induce biologic effects, we investigated proteoglycan synthesis, a marker of chondrogenesis of infected cells, by Alcian blue staining. Enhanced proteoglycan synthesis was demonstrated by the observed elevations in the amount of extracted dye on Tet-induced samples or after constitutive expression on ATDC5 cells (Figure 5A).

Figure 5.

Detection of cartilage-specific proteoglycans produced from ATDC5 cells (A) and primary rabbit chondrocytes (B and C). Cells were infected with each virus at multiplicities of infection of 2.5 (for ATDC5) or 10 (for chondrocytes) and incubated for 17 or 21 days, respectively. The medium was changed 2 times per week. After the indicated time periods, proteoglycans were stained using Alcian blue. Light microscopy of chondrocytes after staining (C) and photometric quantification of extracted dye (A and B) were used to assess proteoglycans. Bars show the mean and SD (n = 3). Scale bars in C = 200 μm. A, The increase in extracted dye in ATDC5 cells treated with doxycycline as compared with those left untreated was highly significant for 1-vector–infected samples (P < 0.0004) and significant for samples transduced with the SF virus promoter–driven BMP-2 rtTA construct (TB + S) and the PGK promoter–driven BMP-2 rtTA construct (TB + P) (each P < 0.0185), but was not significant for cells transduced with the CMV promoter–driven BMP-2 rtTA construct (TB + C) or untransduced (−) cells (P > 0.0854). B, On chondrocytes, the increase in the optical density at 630 nm (OD630) was significant for samples transduced with the TRE-driven BMP-2 SF rtTA construct (TBS), the TRE-driven BMP-2 PGK rtTA construct (TBP), and TB + S (P < 0.0068). See Figure 2 for other definitions.

Rabbit chondrocytes infected with Tet-On constructs showed staining for proteoglycans after induction with doxycycline, similar to the staining observed after constitutive BMP-2 expression under control of the PGK or SF promoter (Figure 5C). Extraction of dye confirmed the increase in proteoglycan synthesis after induction of Tet-mediated or constitutive BMP-2 expression. A maximum of 13-fold induction of BMP-2 was achieved using the 2-vector system driven by the SF promoter, whereas a 6-fold induction was detected using the corresponding 1-vector system in chondrocytes (Figure 5B), thus demonstrating the functionality of the Tet-On system.


Successful therapies for chondral and osteochondral defects are still rare, due to the poor self-healing capacity of cartilage (43). Experiments with recombinant BMP-2 have demonstrated that growth factor BMP-2 enhances the healing of osteochondral defects (14). However, for successful treatment with recombinant BMP-2, high doses and repeated applications are required, which leads to high costs and involves the risk of joint infection (15, 44). Novel strategies combine cell and gene therapy approaches (9). Gene transduction of BMP-2 with retroviral vectors is much more efficient and enhances the quality of subchondral bone and cartilage (20). However, complete healing of osteochondral defects has not been seen with this approach; in contrast long-term expression of BMP-2 induced up-regulation of cartilage hypertrophy markers. Therefore, BMP-2 might not be the first choice as a tool for cartilage healing. Nevertheless, this gene is well studied and different gene therapy approaches have been described for BMP-2 (8, 14, 20), allowing for comparison of the data.

An ideal system for treatment of cartilage damage would allow for repeated and local dosing of the growth factor without surgery, thus reducing the potential for joint infection. Growth factor production at the site of cartilage damage would not only reduce potential side effects, but also drastically reduce costs. In addition, this system should be flexible, in order to integrate different genes (e.g., growth factors). Therefore, a regulated lentiviral expression system for the growth factor BMP-2 was established. Viral vectors constitute a suitable delivery system for therapeutic transgenes (9). In particular, integrating vectors based on C-type retro- and lentiviruses allow for stable, long-lasting transgene expression.

In experiments involving retroviral expression of BMP-2, production of up to 150 ng/ml BMP-2 from ATDC5 cells and ∼30 ng/ml BMP-2 from chondrocytes has been observed, and these levels have remained stable for long time periods (8). In this case, gene expression was driven by the long terminal repeat (LTR) promoter. However, infection with retroviruses is limited to dividing cells, whereas lentiviral vectors can also infect nondividing cells. Moreover, with regard to in vivo application of viral gene therapy, the use of self-inactivating vectors improves safety. These vectors have a deletion in the 3′ LTR that comprises the promoter and enhancer elements. Thus, deletion of promoter and enhancer sequences and its duplication into the 5′ end during integration of the provirus into the cellular genome prevents risk of unintentional gene activation (45).

To avoid transgene overexpression after cartilage regeneration, the doxycycline-inducible Tet-On system (28, 29) allows for regulation of gene activity. Ueblacker et al showed efficient induction of lacZ expression using the nonviral Tet-On system on primary rabbit chondrocytes (35). Our aim was to achieve regulable BMP-2 expression in primary rabbit chondrocytes after lentiviral transduction. Transplantation of these transduced chondrocytes might be a promising approach for treatment of osteochondral defects. Different promoters were tested for expression in chondrocytes to determine optimum expression of the rtTA, which would thus indirectly improve the induction of the system. All results showed the most efficient transgene expression with the use of the viral SF promoter, while the weakest promoter tested was the CMV promoter.

In our initial experiments, we were interested in the induction efficiency of the Tet-On system. Experiments using an eGFP reporter gene showed an up to 12-fold induction in the expression of eGFP+ cells and in the MFI on chondrocytes, demonstrating, for the first time, the functionality of lentiviral-mediated Tet regulation in chondrocytes.

Comparable results were gained for regulated, lentiviral-mediated BMP-2 expression on chondrocytes. Both 1-vector system– and 2-vector system–infected samples, both of which contained the SF promoter–controlled transactivators, expressed ∼15 ng/ml BMP-2 (normalized to 1 × 106 cells and 24 hours) after induction, as compared with no detectable BMP-2 expression before the addition of doxycycline. These results demonstrate that under induced conditions, BMP-2 levels that are comparable with the level of onco-retroviral expression driven by the Moloney murine leukemia virus LTR can be achieved (8). In addition, our results demonstrate that gene expression (eGFP, BMP-2) using a 1-vector system that contains transactivator and response components on a single vector is slightly more efficient than a 2-vector approach composed of 2 separate components.

The efficiency of regulated gene expression was below the constitutive gene expression, the latter of which was up to 30-fold stronger on chondrocytes. In addition, our kinetics study demonstrated prolonged Tet-mediated gene expression for a clinically relevant time span, as well as reversibility of Tet-inducible gene expression in a short time period, which is necessary for our intention to withdraw BMP-2 expression after cartilage regeneration. The apparent biologic half-life of BMP-2 expression, comprising the chemical half-lives of the rtTA, BMP-2, and doxycycline, was ∼1.8 days for the 1-vector system and ∼2.5 days for the 2-vector system.

Our experiment also showed that the stably integrated Tet-On system allows for repeated BMP-2 expression on demand, and turning-off is possible at any time point. During the time span of 6 weeks of in vitro culture, there was no loss of efficiency of gene expression due to methylation.

Subsequently, we were able to demonstrate that lentivirally expressed BMP-2, either constitutively expressed or induced by doxycycline, is sufficient to activate downstream processes. Proteoglycan synthesis was visualized by Alcian blue staining, and quantification showed that there was strong production of proteoglycans as a result of BMP-2 production.

These experiments demonstrate, for the first time, a high level of Tet-regulated, lentivirally mediated growth factor expression and functionality in primary rabbit chondrocytes and ATDC5 cells. With regard to its future use in a rabbit model of osteochondral defects, the 1-vector system for regulated growth factor expression deserves special attention. The system presented will not only allow for switching-off of gene expression to avoid unrestrained overexpression of growth factors, but also allows for repeated switching-on in those instances where regeneration is not complete or new degeneration of cartilage should occur. Regulated in situ growth factor gene expression on demand should be attainable soon, and will constitute new and exciting therapeutic options.


All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be published. Wübbenhorst and Anton had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Study conception and design. Wübbenhorst, Imhoff, Gansbacher, Vogt, Anton.

Acquisition of data. Wübbenhorst, Dumler, Wagner, Wexel.

Analysis and interpretation of data. Wübbenhorst, Vogt, Anton.


We thank A. Rothballer for providing technical assistance.