Introduction

Articular cartilage is a highly specialized tissue that enables almost frictionless motion between the articulating surfaces of diarthrodial joints. Although remarkably durable, it is vulnerable to injury and has a limited capacity for self-repair. Experimental approaches toward treatment of damaged articular cartilage have increasingly focused on cell-based therapies¹. In this regard, adult mesenchymal stem cells (MSCs) provide an attractive alternative to mature chondrocytes that must be isolated from a very limited supply of healthy articular cartilage. MSCs can be obtained relatively easily from bone marrow and other tissue sources and have the capacity for differentiation into the cell types characteristic of various mesenchymal tissues, including cartilage and bone^2,3,4. Under certain culture conditions, MSCs will maintain their multilineage potential with passage, making them amenable to ex vivo applications^5,6.

Delivery of MSCs to cartilaginous lesions has not yielded satisfactory regeneration of articular cartilage. One possible problem is that there is insufficient local stimulation of the implanted cells by the protein factors necessary to drive differentiation in vivo⁷. Gene transfer might be adapted as a means to provide sustained synthesis of bioactive transgene products within cartilaginous lesions; the delivery of the appropriate stimulatory factor(s) in this manner may enable synthesis of an improved cartilaginous repair tissue⁸.

The development of in vitro systems of chondrogenesis has been important to the identification of protein factors that can promote chondrocyte differentiation of adult MSCs and improved cartilage repair in vivo^9,10,11. Johnstone et al. demonstrated that chondrogenesis is induced in MSCs when cultured as aggregates in a defined medium containing dexamethasone and transforming growth factor-1 (TGF-1)⁹. In this system, the aggregates synthesize an extracellular matrix characteristic of cartilage, containing proteoglycan and type II collagen. Related studies have been useful to the elucidation of the chondrogenic potential of other growth factors, including TGF-2, TGF-3, bone morphogenetic protein-2 (BMP-2), BMP-6, and insulin-like growth factor-1 (IGF-1)^{10,11,12,13,14}. For example, administration of recombinant IGF-1 has been found to induce chondrogenesis of limb bud mesenchymal cells¹⁵ as well as periosteal mesenchymal cells from rabbits¹⁶.

Toward the development of gene-based methods for effective delivery of stimulatory proteins, such as those above, several studies have shown that viral vectors, including lentivirus^17,18, retrovirus^19,20,21, and adenovirus^22,23,24, and to a lesser degree nonviral DNA formulations²⁵ can be used to modify MSCs genetically to express various transgene products. Moreover, in animal models, delivery of autologous mesenchymal progenitor cells genetically modified to secrete certain biological factors has been reported to stimulate the formation of bone^26,27 and cartilage^28,29. While these studies indicate that gene transfer can be used to deliver proteins to MSCs and influence their biology, the functional parameters of gene-induced mesenchymal chondrogenesis have not been reported and remain poorly understood.

Using cDNAs encoding TGF-1, BMP-2, and IGF-1, we have begun to delineate the functional considerations important for the use of gene transfer as a protein delivery system for induction of chondrogenesis in primary MSCs. Here, we report that adenoviral-mediated delivery of certain growth factors can induce chondrogenesis of MSCs in aggregate culture; however, the level of transgene expression, its duration, and the viral load influence chondroinduction directly.

Results

Transduction efficiency of primary bone marrow-derived MSCs with adenoviral vectors

For our experiments we used adherent cells cultured from iliac crest bone marrow aspirates of New Zealand White rabbits as a source of MSCs. To determine if these cells can be efficiently modified to express exogenous transgenes, we infected first-passage monolayer cultures with increasing amounts of adenovirus encoding green fluorescent protein, Ad.GFP. We also infected first-passage monolayer chondrocytes in parallel to provide a relative comparison for transduction efficiency. As shown in Fig. 1A, after 72 h a slightly greater percentage of chondrocytes appeared to be transduced by the adenovirus at each dose; however, this difference was not statistically significant. Consistent with the heterogeneous composition of the MSC cultures, green fluorescent protein (GFP)-positive cells appeared with two distinct morphologies, small, fibroblast-like and larger, polygonal cells. At the lower viral doses, GFP⁺ cells appeared predominantly of the polygonal type, but as the dose increased, a greater proportion of GFP⁺ cells appeared fibroblastic (Fig. 1B).

Figure 1.

Adenoviral-mediated transgene expression in primary rabbit MSCs and chondrocytes. (A) Monolayer cultures of MSCs or chondrocytes were infected with increasing doses of Ad.GFP, indicated as viral particles (vp) per cell. At 72 h, GFP⁺ cells were counted in three fields under light and fluorescence microscopy. Results are presented as the mean percentage of fluorescent cells per field at each viral dose. (B) Representative fluorescence of first-passage MSCs infected with 1000 and 10,000 vp/cell as indicated. (C) TGF-1 levels in media conditioned by MSCs or chondrocytes infected with increasing amounts of Ad.TGF-1. At 48 h postinfection the media were changed. Twenty-four hours later the media were collected and TGF-1 production was assayed by ELISA. The data are represented as the means of triplicate experiments. Error bars represent SD.

Full figure and legend (110K)

To determine the relative level of synthesis of a secreted transgene product, we infected parallel cultures of chondrocytes and MSCs with increasing amounts of adenovirus containing the cDNA for TGF-1. At 72 h postinfection, we analyzed the conditioned media for TGF-1 content. As shown in Fig. 1C, consistent with the results for GFP, TGF-1 transgene expression increased with viral dose similarly for both cell types. At the lower viral doses, however, expression from the chondrocyte cultures was somewhat higher, ranging between 1.4- and 3-fold greater than for the MSCs. However, at the maximum dose of virus tested, 10,000 viral particles (vp) per cell, expression from the MSCs was slightly greater.

These results indicated that MSCs were amenable to genetic modification via recombinant adenovirus and showed a modestly reduced capacity of transgenic expression relative to chondrocytes. Based on these results, we used adenoviral doses ranging from 100 to 10,000 vp per cell for subsequent experiments. The predicted levels of secreted transgene product at these doses, between 1 and 100 ng/ml per 24 h, encompass the range of recombinant growth factor concentrations typically used to supplement MSC cultures in models of in vitro chondrogenesis.

Chondrogenic differentiation of MSCs following adenoviral-mediated gene transfer of TGF-1

We performed experiments using MSCs in an aggregate system to determine if gene transfer could be used as an effective means of protein delivery with which to induce chondrogenesis. In this system, the activity of TGF-1 has been well characterized⁹; therefore we compared its ability to induce chondrogenesis when delivered as a recombinant (r) protein or when supplied as the product of a transgene. For the latter, we infected first- or second-passage monolayer cultures of MSCs with doses of Ad.TGF-1 that would be predicted to generate low (<10 ng/ml), medium (10–100 ng/ml), and high (>100 ng/ml) levels of secreted TGF-1 transgene product following aggregate formation. At 24 h postinfection, we seeded monolayer cells into aggregates and maintained them for 21 days in a defined serum-free medium. Control groups consisted of naive and Ad.GFP-infected aggregates maintained in the absence or presence of rTGF-1 protein at a concentration of 10 ng/ml.

Histologic examination indicated distinct evidence of transgene-induced chondrogenesis of the MSCs, the extent of which correlated with the levels of expressed TGF-1 protein (Fig. 2). The genetically modified MSCs that expressed TGF-1 in the medium range at 3 days post-aggregate formation, as well as the unmodified aggregates cultured in 10 ng/ml rTGF-1, were highly cellular and showed extensive metachromatic staining with toluidine blue and corresponding positive immunostaining for type II collagen, characteristic of cartilage matrix. Typical of that shown in Fig. 2, we found the level of chondrogenesis to be reproducibly greater and more consistent when TGF-1 was supplied as a transgene product than as a recombinant protein. For example, quantitative estimates indicated a 75 and 110% increase in relative toluidine blue and type II collagen staining, respectively, in the midlevel-expression (10–100 ng/ml) pellets relative to the recombinant protein.

Figure 2.

Chondrogenesis of MSCs following adenoviral-mediated gene transfer of TGF-1. Monolayer cultures of MSCs were infected with doses of Ad.TGF-1 to generate 1–10, 10–100, and >100 ng/ml of transgene product at day 3 following aggregate formation. At 24 h after infection, MSCs were seeded into aggregates and cultured for 21 days. Nontransduced control groups were cultured in parallel in the absence (0 ng/ml) or presence (10 ng/ml) of recombinant TGF-1. (A) Toluidine blue staining of representative aggregate sections for detection of matrix proteoglycan. (B) Immunostaining for the presence of type II collagen and (C) type I collagen. Regions of positive staining show green fluorescence. (D) Corresponding transgene expression profile for each group for the 21-day culture period. Values represent TGF-1 levels in the conditioned media over a 24-h period at days 3, 7, 14, and 21. Data are shown as the means + SD of triplicate measurements with n = 3 pellets per group per replicate.

Full figure and legend (303K)

In contrast, at day 3 MSCs expressing TGF-1 at less than 10 ng/ml or greater than 100 ng/ml were typically much less chondrogenic and showed only limited, somewhat focal matrix production as evidenced by staining for proteoglycans and collagen type II (Fig. 2). Often, observable chondrogenesis was confined toward the periphery of the aggregates, while the central regions resembled undifferentiated control aggregates. We observed positive immunostaining for collagen type I in all groups and it typically occurred in the outermost regions. Interestingly, aggregates of cells preinfected with Ad.GFP showed no evidence of chondrogenesis either in the presence or in the absence of rTGF-1 (data not shown).

Analyses of media conditioned by genetically modified aggregates revealed a marked decrease in TGF-1 transgene expression in all groups over the course of the experiment (Fig. 2D). For cultures that expressed medium and high levels of TGF-1, the decrease was most prominent, approximately 80%, between day 3 and day 7 (Fig. 2D). Conversely, the low-level TGF-1 group synthesized approximately 5 ng/ml of growth factor up to the first week of culture, but this was not sufficient to induce extensive chondrogenesis within the aggregates.

Chondrogenic differentiation of primary MSCs following adenoviral-mediated gene transfer of IGF-1 and BMP-2

We used the aggregate culture system further to evaluate the chondroinductive activity of two additional transgenes, human BMP-2 and IGF-1. Again, we modified aggregates to express low (<10 ng/ml), medium (10–100 ng/ml), or high (>100 ng/ml) levels of secreted BMP-2 or IGF-1 and cultured them in defined media for 3 weeks. For comparative controls, we cultured similar unmodified aggregates of MSCs in the presence of recombinant BMP-2 or IGF-1 at doses of 25, 50, or 100 ng/ml.

As shown in Fig. 3D, the profiles of transgene expression in Ad.BMP-2-modified aggregates were similar to those of Ad.TGF-1, with a prominent loss of expression at days 3 and 7. As with TGF-1, adenoviral-mediated expression of BMP-2 induced chondrogenesis in a dose dependent manner (Figs. 3A–3C). Aggregates expressing low levels (1–10 ng/ml) of BMP-2 were highly cellular and showed considerable staining for proteoglycan and type II collagen despite a decrease in BMP-2 expression to <1 ng/ml after only 1 week. In aggregates expressing 10–100 ng/ml BMP-2, we observed more abundant staining (greater than twofold) for proteoglycan and type II collagen compared to both the low BMP-2 group and a control Ad.TGF-1 group expressing 10–100 ng/ml TGF-1 (similar to that shown in Fig. 2). These aggregates were also highly cellular, and the majority of cells had the appearance of hypertrophic chondrocytes. Similar to that seen with Ad.TGF-1, the high-expressing group (>100 ng/ml BMP-2) showed little evidence of cartilaginous matrix (Fig. 3), suggesting inhibition of chondrogenesis, perhaps through overproduction of growth factor or excessive adenoviral infection. Aggregates cultured in the presence of rBMP-2 similarly showed a dose-dependent effect, and chondrogenesis increased with dose of protein through the 100 ng/ml level. As seen in Fig. 2, chondroinduction at this dose was similar to that observed with the adenovirus expressing 1–10 ng of BMP-2. Digital analysis of staining confirmed approximately similar levels of toluidine blue and collagen type II reactivity between these two groups.

Figure 3.

Chondrogenesis of MSCs following adenoviral-mediated gene transfer of BMP-2 and IGF-1. Similar to those described for Fig. 2, monolayer cultures of MSCs were infected with doses of Ad.BMP-2 or Ad.IGF-1 to generate 1–10, 10–100, and >100 ng/ml levels of transgene product at day 3 following aggregate formation as indicated. The infected MSCs were seeded into aggregates and cultured for 21 days. Parallel unmodified aggregate cultures were incubated in the presence of 25, 50, or 100 ng/ml rBMP-2 or rIGF-1. The aggregates were then fixed, sectioned, and stained. For rBMP-2, the maximum level of chondrogenesis was observed in the 100 ng/ml group. Since there was no difference in the appearance of sections from any of the rIGF-1 or Ad.IGF-1-infected groups, only sections from the 100 and 10–100 ng/ml groups, respectively, are shown. (A) Toluidine blue staining (B) Immunostaining for type II collagen and (C) type I collagen. (D) Corresponding recombinant protein levels or transgene expression profiles for each group for the 21-day culture period are shown as indicated. Values represent levels of BMP-2 or IGF-1 (shown in yellow on the right) in the conditioned media over a 24-h period at days 3, 7, 14, and 21. Sections shown are representative of a series of three experimental replicates with n = 3 pellets per group per replicate.

Full figure and legend (348K)

Somewhat surprisingly, neither the cultures genetically modified to express IGF-1 nor those incubated in the presence of rIGF-1 showed phenotypic evidence of chondrogenesis in sections stained for type II collagen or proteoglycan (Fig. 3). Sections also showed little staining for collagen type I. Despite the absence of chondrogenesis at the phenotypic level, the profile of expression of the IGF-1 transgene from the modified cultures was similar to that observed with Ad.BMP-2 and Ad.TGF-1 (Fig. 3D).

We determined the expression of cartilage-specific marker genes for the genetically modified aggregates using RT-PCR (Fig. 4). After 21 days, we observed expression of mRNAs for aggrecan and type II collagen in cells modified with Ad.TGF-1, Ad.BMP-2, and Ad.IGF-1, but not preaggregate monolayer MSCs. In Ad.TGF-1- and Ad.IGF-1-modified aggregates RT-PCR products of both splice variants of type II collagen, IIA (432 bp) and IIB (225 bp), were present in approximately equal amounts. However, in Ad.BMP-2-modified aggregates, similar to that seen in articular chondrocytes (lane C), type IIB was the predominant form. Comparison of reaction products between the Ad.TGF-1 and the Ad.BMP-2 aggregates showed approximately equal levels of aggrecan and collagen type IIB transcripts, but Ad.TGF-1-infected aggregates showed approximately 4.7- and 2.1-fold greater levels of collagen type IIA and collagen type I mRNAs, respectively. Despite the apparent expression of mRNAs for aggrecan and type II collagen in the Ad.IGF-1 group at levels nearly equivalent to those of the Ad.TGF-1 aggregates, these cultures showed no histologic evidence of chondrogenesis. This suggests that the rabbit MSCs are capable of recognizing and responding to the IGF-1 protein in a manner consistent with chondrocytic differentiation but may be limited by posttranscriptional mechanisms in this culture system.

Figure 4.

Expression of cartilage-specific genes in genetically modified aggregates. After 21 days, total RNA was extracted from aggregates (6 per group) and expression of cartilage marker genes was determined using RT-PCR. The reaction products were resolved on 1.5% agarose gels and visualized by staining with ethidium bromide. RT-PCR of RNA isolated from rabbit articular cartilage was used as a comparative control (lane C). Reaction products of RNA from preaggregate MSCs are shown in lane 1. RNA from Ad.TGF-1-, Ad.IGF-1-, or Ad.BMP-2-modified aggregates was used in lanes 2–4, respectively. RT-PCR product sizes were as follows: aggrecan, 313 bp; type II collagen, 432 (IIA splice variant) and 225 bp (IIB splice variant); type I collagen, 702 bp; GAPDH, 293 bp. Differences in staining intensities between lanes were normalized using the GAPDH reaction products.

Full figure and legend (44K)

Effect of culture conditions on transgene expression of MSCs

Despite high transduction efficiencies, results from our experiments suggest that transgene expression is rapidly lost from MSCs genetically modified with adenoviral vectors in aggregate culture. However, these results were determined by measuring the level of transgene products released into the conditioned media and do not represent those within the cell or bound to extracellular matrix components. To generate a more accurate profile of transgene expression by the cells, we transduced MSCs to express a nonchondrogenic gene, luciferase (Ad.Luc), and cultured them either as aggregates or as monolayers in the presence of rTGF-1 at 10 ng/ml.

Luciferase transgene expression in both aggregate and monolayer cultures was characterized by a marked decrease by day 7 and followed a pattern similar to that of growth factor expression in the earlier experiments (Fig. 5). Despite the presence of rTGF-1, there was no visible evidence of chondrogenesis in Ad.Luc-infected cultures (not shown). The similar expression profiles in monolayer and aggregate culture indicate that loss of transgene expression in MSCs is not related to events associated with chondrocytic differentiation, but more likely occurs through the loss of adenoviral genomes to cell division or cell death.

Figure 5.

Luciferase transgene expression in monolayer and aggregate MSC cultures. Monolayer MSCs were infected with 500 vp/cell Ad.Luc and after 24 h were cultured as aggregates or reseeded into monolayers. Aggregates and monolayers were cultured with or without rTGF-1 10 ng/ml as indicated. Aggregates were harvested and analyzed for luciferase activity at days 3, 7, 14, and 21. Values shown are mean levels of luciferase activity for triplicate samples in relative light units (RLU) + SD.

Full figure and legend (57K)

Effect of adenoviral load on chondrogenesis of MSCs

Noting a reduction in chondrogenesis in pellets transduced to express high levels of TGF-1 or BMP-2, we wanted to determine whether high levels of adenoviral infection could inhibit differentiation of MSCs. For these experiments, we used the parent adenoviral vector, 5, which does not contain a transgene. Previously, viral doses of 100, 500, and 5000 vp/cell were used to establish 1–10, 10–100, and >100 ng/ml expressed protein product, respectively. Since optimal chondrogenesis was observed in the 10–100 ng/ml range of expressed TGF-1 or BMP-2 following infection with up to 500 particles/cell, the differentiation potential of primary MSCs is not compromised at this level of virus. Based on this, we first infected primary MSCs with 500 particles/cell Ad.TGF-1 and after 24 h infected them with increasing amounts of 5 to yield cumulative viral doses of 1000, 5000, and 10,000 vp/cell. After a further 24 h we seeded the cells into aggregates and cultured them as before.

Histologic examination of the aggregates after 21 days revealed that 5 overinfection decreased chondrogenesis in a dose-dependent manner (Fig. 6A). Control aggregates modified with 500 vp/cell Ad.TGF-1, or 500 vp/cell 5 cultured in the presence of rTGF-1, were similarly chondrogenic as evidenced by metachromatic staining with toluidine blue. In aggregates infected with cumulative viral doses of 1000 and 5000 vp/cell there was a progressive decrease in matrix staining with toluidine blue and aggregate size. Type II collagen immunostaining correlated with toluidine blue in all aggregates (not shown). After initially forming aggregates, MSCs infected with 10,000 vp/cell disintegrated during the 3-week culture period and thus are not shown. Analysis of media conditioned by the aggregates revealed similar levels of expressed TGF-1 at 3 days after aggregate formation and throughout the culture period, in aggregates either with or without 5 infection. This indicated that the inhibitory effect of 5 was not due to altered expression of the TGF-1 transgene, but rather to excessive adenoviral infection.

Figure 6.

Chondrogenesis of MSCs following exposure to increasing adenoviral loads and paracrine delivery of a transgenic growth factor. (A) MSCs were infected with 500 vp/cell Ad.TGF-1 as indicated, followed 24 h later by increasing concentrations (0, 500, 4500, and 9500 vp/cell) of the 5 adenoviral vector to yield cumulative viral doses of 500, 1000, 5000, and 10,000 vp/cell. (Aggregates of MSCs infected with a total of 10,000 vp/cell disintegrated during culture and are not shown.) A control group (rTGF-1 + 5) was transduced with 5 only (500 vp/cell) and cultured in the presence of 10 ng/ml rTGF-1. Images show toluidine blue staining of representative aggregate sections after 21 days. Transgene expression (tx exp) levels at day 3 (mean SD for triplicate samples) for each group are shown below the images. (B) Donor MSC aggregates were genetically modified with Ad.BMP-2 to express either 10–100 or >100 ng/ml transgene product. Every 24 h, the media conditioned by these aggregates were removed and fed to Recipient, untransduced aggregates. Control, untransduced aggregates were cultured in the presence of rBMP-2 at a final concentration of 25 ng/ml (10–100 lane) or 100 ng/ml (>100 lane).

Full figure and legend (102K)

Effect of paracrine delivery of transgenic BMP-2 on chondrogenesis of MSCs

Although we had evidence that high-level infection of MSCs was inhibitory to chondrogenesis, we performed additional experiments to determine if excessive growth factor production may also contribute. For this study we transduced "donor" MSC cultures with doses of Ad.BMP-2 that would generate medium and high levels of transgene product following aggregate formation. At 24 h after initiation of aggregate culture, we removed the conditioned media from both groups and added them to untransduced, "recipient" aggregates cultured in parallel. Donor cultures were refed with chondrogenic media and the process was repeated every 24 h for 21 days. An additional group of aggregates cultured in the presence of rBMP-2 at doses ranging from 25 to 100 ng/ml were also generated for comparison.

ELISA of the conditioned media from donor cultures after 3 days confirmed that aggregates modified to express medium- and high-dose BMP-2 were within the predicted range (23 and 152 ng/ml at day 3, respectively). Consistent with Fig. 3, genetically modified donor aggregates expressing medium-dose BMP-2 showed extensive chondrogenesis following toluidine blue staining (Fig. 6B). Similarly, chondrogenesis was inhibited in the high-dose BMP-2-modified donor group. Somewhat surprisingly, secreted, transgenic BMP-2 from medium-dose cultures failed to stimulate chondrogenesis in recipient aggregates (Fig. 6B). However, we observed partial chondrogenesis in recipient aggregates receiving transgenic BMP-2 from the high-dose donor cultures. The chondrogenic response of aggregates treated with rBMP-2 followed a similar trend, with the extent of chondrogenesis increasing with BMP-2 protein concentration.

These results suggested that BMP-2 provided to cells in aggregate is less potent as a soluble protein than as a gene product. They also indicated that levels of BMP-2 in culture medium of >100 ng/ml are not necessarily inhibitory to chondrogenic differentiation in this system.

Discussion

In the present study, we demonstrate that adenoviral-mediated expression of certain chondrogenic growth factors can serve as an effective means of protein delivery to induce chondrogenesis of rabbit, primary, bone marrow-derived MSCs in aggregate culture. Indeed, for TGF-1 and BMP-2 this method was typically superior to delivery of recombinant protein. Important differences, however, were noted between the two transgenes with regard to the histologic appearance of the pellets and transgene expression. Within the range of 1–100 ng/ml expressed transgene product, Ad.BMP-2-infected aggregates were larger, had increased cellularity, and showed more intense staining for proteoglycan and collagen type II than Ad.TGF-1 aggregates. This pattern was consistent over three experiments, each with preparations of MSCs from different rabbits. In addition, RT-PCR revealed different mRNA splice variants of type II collagen from each growth factor, with Ad.BMP-2 aggregates expressing a greater percentage of the type IIB form typically associated with maturing chondrocytes^14,30.

The failure of IGF-1 to stimulate differentiation of rabbit MSCs effectively in aggregate culture is largely inconsistent with the literature. For example, Oh and Chun report the ability of recombinant IGF-1 to drive chondrogenesis of chicken limb bud mesenchymal cells in micromass culture¹⁵; Fukumoto et al. describe IGF-1-mediated mesenchymal chondrogenesis in periosteal explants from rabbits¹⁶ and Gelse et al. the use of rat MSCs genetically modified to overexpress IGF-1 in a cartilage repair model in vivo²⁹. It could be argued that adenoviral infection may interfere with the activity of this particular protein, but we were unable to observe chondrogenesis with the recombinant protein even at high doses (100 ng/ml). Ad.IGF-1 was found to stimulate transcription of genes associated with chondrogenesis in the aggregate system, indicating that rabbit cells are capable of recognizing and responding to the IGF-1 transgene product. These results suggest that IGF-1 is capable of initiating certain biological pathways associated with chondrocytic differentiation, but as a chondroinductive agent it likely relies on the presence of other factors not present in the context of the aggregate culture system.

Typical of those observed with nonintegrative vectors³¹, transgene expression levels in the adenovirally transduced MSC cultures was transient and showed a marked decrease after 7 days. This decline was observed with both growth factor and marker transgenes and occurred independent of chondrogenesis or aggregate culture. This transient profile may limit the types of genes/factors that can be effectively delivered to MSCs via an adenovirus. Alternatively, given that many growth factors are pleiotropic, a vector that provides transient transgene expression may naturally limit overproduction and the detrimental side effects that can occur when delivering these agents to articular tissues^29,32,33.

We found that overinfection of MSCs with adenovirus was inhibitory to chondrogenic differentiation. Along these lines, we also observed an absence of chondrogenesis in MSCs modified with adenovirus to express marker genes, GFP and luciferase, despite prolonged culture in media containing rTGF-1. This type of response has not been reported previously and is inconsistent with the observations of others. Mosca et al. reported that human MSCs retrovirally transduced to express GFP were fully capable of chondrogenic differentiation in pellet culture³⁴. Similarly, transgenic mice that express GFP are viable and develop mesenchymal tissues normally^35,36. Thus, it is possible that the lack of differentiation we observed is not a direct consequence of marker gene expression per se, but more due to the route of gene delivery via recombinant adenovirus.

An interesting observation from these studies is the increased potency of certain growth factors when supplied as transgene products rather than as recombinant protein. This may reflect differences in the presentation of the protein in the microenvironment of the aggregate. It is possible that within the aggregate the functional concentration of the transgene product is manifold higher than is reflected by ELISA measurements of the conditioned media. In the experiments described here, a single aggregate of 2 10⁵ cells represents approximately 1/1000 of the volume of the culture fluid, yet is capable of raising the concentration of the total fluid volume to greater than 100 ng/ml transgene product. Therefore, within the immediate microenvironment of the aggregate, the concentration of growth factor at certain times must be far greater than that of the medium. Further, the almost direct presentation of synthesized growth factor to immediately neighboring cells may enhance its stimulatory capacity. An additional consideration is that the transgene product is being released from numerous cells within the aggregate, while the recombinant protein must diffuse throughout the matrix of the aggregate to stimulate the cells in the interior. Thus, although there is clear evidence that high viral loads will inhibit chondrogenic differentiation, because of the functional differences in the presentation of growth factors when provided as recombinant protein supplement to the culture medium or as a transgene product within the aggregate cultures, we were unable to resolve completely the possibility that excess growth factor may inhibit chondrogenesis.

In conclusion, our experiments provide a further demonstration of the capacity of gene transfer as a delivery system for bioactive proteins and, within certain parameters, the ability to use it to guide the chondrogenic differentiation of multipotent cells. Indeed the system described here should provide a useful tool to examine the chondrogenic potential of other candidate transgenes in vitro. Thus far, our work has entailed the examination of gene products on an individual basis. However, to fulfill the potential of gene therapy for stem-cell-based cartilage repair, it is possible that far more sophisticated strategies will be required to reproduce faithfully the complex molecular events of chondrocyte differentiation and then long-term maintenance of the articular cartilage phenotype. These may require the coordinate expression of multiple genes using complex regulatory systems. How gene-induced chondrogenesis may be translated into therapeutic applications has yet to be determined. There are several experimental approaches currently under investigation that may prove useful, such as the local implantation into cartilage defects of MSCs genetically modified ex vivo^28,29 or, as we have recently described, the delivery of genetically modified bone marrow coagulates to cartilaginous lesions³⁷. Regardless of the method, a critical first step in development is the characterization of candidate gene products in vitro to enable selection of viable reagents to take forward into animal studies.

Materials and methods

Generation of vectors

First-generation, E1, E3-deleted, serotype 5 adenoviral vectors carrying the cDNAs for firefly luciferase, GFP, human TGF-1, human IGF-1, and human BMP-2 were constructed using the method of Hardy et al.³⁸. The resulting vectors were designated Ad.Luc, Ad.GFP, Ad.TGF-1, Ad.IGF-1, and Ad.BMP-2, respectively. To generate high-titer preparations, the recombinant vectors were amplified in 293CRE8 cells and purified over three successive CsCl gradients. Following dialysis the preparations were aliquoted and stored at -80°C. Viral titers were estimated by optical density and standard plaque assay. Using these methods preparations of 10¹²–10¹³ particles/ml were obtained.

Cell harvest

Rabbit bone marrow was harvested from the iliac crests of young adult New Zealand White rabbits and plated in monolayer culture at 2 10⁷ cells/75-cm² flask in DMEM supplemented with 10% FBS and 1% mesenchymal stem cell stimulatory supplements (Stem Cell Technologies). After 2–3 weeks, adherent colonies of cells were trypsinized and replated in 25-cm² tissue culture flasks or 24-well plates depending on the procedure.

Adenoviral transduction of MSCs in monolayer

Confluent monolayer cultures of MSCs in 24-well plates were incubated with various doses of recombinant adenovirus as indicated in the text and figure legends, in 100 l of serum-free DMEM for 2 h in a tissue culture incubator. Following infection, the culture fluids were aspirated and replaced with 1 ml DMEM with 10% FBS. After 24 h, the media were replaced and were collected following an additional further 24 h incubation. Harvested media were stored at -20°C until analysis by specific ELISA. Ad.GFP-transduced cultures were viewed for fluorescence at 48 h following infection.

Aggregate culture of MSCs

Following the initial plating, the adherent cultures of MSCs were seeded into 25-cm² flasks and grown to confluence, generating approximately 6 10⁵ cells per flask. Individual flasks of cells were infected with low, medium, or high doses of Ad.TGF-1, Ad.IGF-1, or Ad.BMP-2. Afterward, the supernatant was aspirated and replaced with 5 ml DMEM containing stem cell supplements. After 24 h, the cells were trypsinized and seeded at 2 10⁵ cells/ml in 15-ml polypropylene tubes. The cells were centrifuged for 5 min at 500g to form a pellet. The pelleted cells were maintained in 0.5 ml of chondrogenic medium consisting of serum-free DMEM containing pyruvate (1 mM), 1% ITS (Sigma) ascorbate 2-phosphate (37.5 g/ml), and dexamethasone (10^-7 M)⁹. In parallel nontransduced cultures, the medium was supplemented with or without recombinant human TGF-1 (R&D Systems) at 10 ng/ml. The pelleted cells formed free-floating aggregates within the first 24 h of culture. The media were changed every 2–3 days.

RNA isolation and RT-PCR

RT-PCR was used to evaluate qualitatively transcription of cartilage-specific genes following infection of MSCs with medium doses of Ad.TGF-1, Ad.BMP-2, or Ad.IGF-1. Total RNA was isolated from two 25-cm² flasks of MSCs grown in monolayer or from six MSC aggregates cultured for 21 days. For cDNA synthesis, 1 g of total RNA from each group was reverse transcribed using random hexamer primers and M-MLV reverse transcriptase (Invitrogen). Two microliters of reaction product containing approximately 80 ng of cDNA was used as a template for PCR amplification. The primer sets used have been previously described: human type II collagen a1(II) chain, human type I collagen a1(I)⁹ rabbit aggrecan, and rabbit glyceraldehyde-3-phosphate dehydrogenase (GAPDH)³⁹. PCR products were visualized on 1% agarose gels containing 0.1 mg/ml ethidium bromide. The relative intensity of the individual RT-PCR products within the gels was determined from digital images using the Scion Image program, version 1.63. Reaction products of GAPDH were used to normalize the intensities between lanes.

Quantitation of transgene expression

Conditioned media were collected at various time points and assayed for concentrations of the respective growth factors using the appropriate Quantikine Immunoassay kits (R&D Systems) for human TGF-1, IGF-1, and BMP-2.

To measure luciferase activity, monolayer MSC cultures were infected with 1.5 10⁸ particles of Ad.Luc. At 24 h postinfection, the cells were trypsinized and reseeded at 4 10⁵ cells/ml in either aggregate or monolayer cultures as described. For aggregate cultures, samples were incubated with 3 mg/ml collagenase and 0.1% trypsin in serum-free DMEM for 2 h at 37°C. The digests were then transferred to Eppendorf tubes and mixed with an equal volume of 2 reporter lysis buffer (Promega) and homogenized using a motorized pestle homogenizer. Homogenates were centrifuged briefly to remove cellular debris, and aliquots of the supernatant were assayed for luciferase activity by mixing with an equal volume of Bright-Glo luciferase assay buffer and measuring light emitted with a Autolumat Plus luminometer. In monolayer cultures, cells were lysed by incubation with reporter lysis buffer for 15 min at room temperature and mixed with an equal volume of Bright-Glo luciferase assay buffer prior to measurement.

Histology and immunohistochemistry

Aggregates were embedded in 0.8% agarose for ease of handling and then fixed in 10% neutral-buffered formalin for 1 h at room temperature. After dehydration in graded alcohols, the aggregates were paraffin embedded and sectioned to 5 m. Representative sections were stained using toluidine blue (Sigma) for the detection of matrix proteoglycan, and alternate sections were used for immunohistochemistry.

For immunohistochemistry, sections were deparaffinized and treated with 0.1 U/ml chondroitinase ABC in PBS with 1% BSA at room temperature for 30 min. Sections were then blocked with 5% BSA in PBS for 30 min. Afterward, the sections were incubated for 1 h at room temperature with rabbit polyclonal anti-collagen type I and II primary antibodies (Rockland, Inc., Gilbertsville, PA, USA) diluted in 1% BSA in PBS. After three PBS washes to remove unbound primary antibody, sections were incubated with a fluorescein-conjugated anti-rabbit IgG secondary antibody for 45 min at room temperature. The slides were washed again and mounted in Fluoromount G, coverslipped, and analyzed using fluorescence microscopy.

For each experiment described, three replicates were performed, with n = 3 pellets for each group and replicate. From each pellet three sections taken at positions throughout were stained and analyzed. To quantify relative toluidine blue or immunofluorescent staining, digital images of individual sections were taken, and the mean density of staining across the entire section was determined using the Scion Image program version 1.63. For the respective methods, sections of aggregates cultured in the absence of specific growth factor stimulation served as negative controls and were used to establish baseline levels of staining against which the sections from the experimental groups were evaluated.

References

1. Caplan, A. I. and Bruder, S. P. (2001). Mesenchymal stem cells: building blocks for molecular medicine in the 21st century. Trends Mol. Med. 7: 259–264. | Article | PubMed | ISI | ChemPort |
2. Ashton, B. A., Allen, T. D., Howlett, C. R., Eaglesom, C. C., Hattori, A. and Owen, M. (1980). Formation of bone and cartilage by marrow stromal cells in diffusion chambers in vivo. Clin. Orthop. 151: 294–307. | PubMed |
3. Haynesworth, S. E., Goshima, J., Goldberg, V. M. and Caplan, A. I. (1992). Characterization of cells with osteogenic potential from human marrow. Bone 13: 81–88. | Article | PubMed | ISI | ChemPort |
4. Prockop, D. J. (1997). Marrow stromal cells as stem cells for nonhematopoietic tissues. Science 276: 71–74. | Article | PubMed | ISI | ChemPort |
5. Conget, P. A. and Minguell, J. J. (1999). Phenotypical and functional properties of human bone marrow mesenchymal progenitor cells. J. Cell. Physiol. 181: 67–73. | Article | PubMed | ISI | ChemPort |
6. Colter, D. C., Class, R., DiGirolamo, C. M. and Prockop, D. J. (2000). Rapid expansion of recycling stem cells in cultures of plastic-adherent cells from human bone marrow. Proc. Natl. Acad. Sci. USA 97: 3213–3218. | Article | PubMed | ChemPort |
7. Trippel, S. B., Ghivizzani, S. C. and Nixon, A. J. (2004). Gene-based approaches for the repair of articular cartilage. Gene Ther. 11: 351–359. | Article | PubMed | ChemPort |
8. Evans, C. H., Ghivizzani, S. C., Smith, P., Shuler, F. D., Mi, Z. and Robbins, P. D. (2000). Using gene therapy to protect and restore cartilage. Clin. Orthop. 379: S214–S219. | PubMed |
9. Johnstone, B., Hering, T. M., Caplan, A. I., Goldberg, V. M. and Yoo, J. U. (1998). In vitro chondrogenesis of bone marrow-derived mesenchymal progenitor cells. Exp. Cell Res. 238: 265–272. | Article | PubMed | ISI | ChemPort |
10. Mackay, A. M., Beck, S. C., Murphy, J. M., Barry, F. P., Chichester, C. O. and Pittenger, M. F. (1998). Chondrogenic differentiation of cultured human mesenchymal stem cells from marrow. Tissue Eng. 4: 415–428. | PubMed | ISI | ChemPort |
11. Pittenger, M. F., et al. (1999). Multilineage potential of adult human mesenchymal stem cells. Science 284: 143–147. | Article | PubMed | ISI | ChemPort |
12. Sekiya, I., Colter, D. C. and Prockop, D. J. (2001). BMP-6 enhances chondrogenesis in a subpopulation of human marrow stromal cells. Biochem. Biophys. Res. Commun. 284: 411–418. | Article | PubMed | ISI | ChemPort |
13. Hanada, K., et al. (2001). BMP-2 induction and TGF-beta 1 modulation of rat periosteal cell chondrogenesis. J. Cell. Biochem. 81: 284–294. | Article | PubMed | ChemPort |
14. Barry, F., Boynton, R. E., Liu, B. and Murphy, J. M. (2001). Chondrogenic differentiation of mesenchymal stem cells from bone marrow: differentiation-dependent gene expression of matrix components. Exp. Cell Res. 268: 189–200. | Article | PubMed | ChemPort |
15. Oh, C. D. and Chun, J. S. (2003). Signaling mechanisms leading to the regulation of differentiation and apoptosis of articular chondrocytes by insulin-like growth factor-1. J. Biol. Chem. 278: 36563–36571. | Article | PubMed | ChemPort |
16. Fukumoto, T., et al. (2003). Combined effects of insulin-like growth factor-1 and transforming growth factor-beta1 on periosteal mesenchymal cells during chondrogenesis in vitro. Osteoarthritis Cartilage 11: 55–64. | Article | PubMed | ChemPort |
17. Totsugawa, T., et al. (2002). Lentiviral transfer of the LacZ gene into human endothelial cells and human bone marrow mesenchymal stem cells. Cell Transplant 11: 481–488. | PubMed |
18. Morizono, K., et al. (2003). Multilineage cells from adipose tissue as gene delivery vehicles. Hum. Gene Ther. 14: 59–66. | Article | PubMed | ChemPort |
19. Allay, J. A., et al. (1997). LacZ and interleukin-3 expression in vivo after retroviral transduction of marrow-derived human osteogenic mesenchymal progenitors. Hum. Gene Ther. 8: 1417–1427. | PubMed | ChemPort |
20. Marx, J. C., et al. (1999). High-efficiency transduction and long-term gene expression with a murine stem cell retroviral vector encoding the green fluorescent protein in human marrow stromal cells. Hum. Gene Ther. 10: 1163–1173. | Article | PubMed | ChemPort |
21. Thalmeier, K. and Huss, R. (2001). Highly efficient retroviral gene transfer into immortalized CD34(-) cells and organ distribution after transplantation into NOD/SCID mice. Cytotherapy 3: 245–251. | Article | PubMed | ChemPort |
22. Riew, K. D., Wright, N. M., Cheng, S., Avioli, L. V. and Lou, J. (1998). Induction of bone formation using a recombinant adenoviral vector carrying the human BMP-2 gene in a rabbit spinal fusion model. Calcif. Tissue Int. 63: 357–360. | Article | PubMed | ChemPort |
23. Conget, P. A. and Minguell, J. J. (2000). Adenoviral-mediated gene transfer into ex vivo expanded human bone marrow mesenchymal progenitor cells. Exp. Hematol. 28: 382–390. | Article | PubMed | ISI | ChemPort |
24. Tsuda, H., et al. (2003). Efficient BMP2 gene transfer and bone formation of mesenchymal stem cells by a fiber-mutant adenoviral vector. Mol. Ther. 7: 354–365. | Article | PubMed | ISI | ChemPort |
25. Corsi, K., Chellat, F., Yahia, L. and Fernandes, J. C. (2003). Mesenchymal stem cells, MG63 and HEK293 transfection using chitosan–DNA nanoparticles. Biomaterials 24: 1255–1264. | Article | PubMed | ISI | ChemPort |
26. Lou, J., Tu, Y., Ludwig, F. J., Zhang, J. and Manske, P. R. (1999). Effect of bone morphogenetic protein-12 gene transfer on mesenchymal progenitor cells. Clin. Orthop. 369: 333–339. | PubMed |
27. Lieberman, J. R., et al. (1999). The effect of regional gene therapy with bone morphogenetic protein-2-producing bone-marrow cells on the repair of segmental femoral defects in rats. J. Bone Joint Surg. Am. 81: 905–917. | PubMed | ChemPort |
28. Mason, J. M., Grande, D. A., Barcia, M., Grant, R., Pergolizzi, R. G. and Breitbart, A. S. (1998). Expression of human bone morphogenic protein 7 in primary rabbit periosteal cells: potential utility in gene therapy for osteochondral repair. Gene Ther. 5: 1098–1104. | Article | PubMed | ChemPort |
29. Gelse, K., von der Mark, K., Aigner, T., Park, J. and Schneider, H. (2003). Articular cartilage repair by gene therapy using growth factor-producing mesenchymal cells. Arthritis Rheum. 48: 430–441. | Article | PubMed | ChemPort |
30. Chimal-Monroy, J. and Diaz de Leon, L. (1997). Differential effects of transforming growth factors beta 1, beta 2, beta 3 and beta 5 on chondrogenesis in mouse limb bud mesenchymal cells. Int. J. Dev. Biol. 41: 91–102. | PubMed | ChemPort |
31. Oligino, T. J., Yao, Q., Ghivizzani, S. C. and Robbins, P. (2000). Vector systems for gene transfer to joints. Clin. Orthop. 379: S17–S30. | PubMed |
32. Bakker, A. C., et al. (2001). Overexpression of active TGF-beta-1 in the murine knee joint: evidence for synovial-layer-dependent chondro-osteophyte formation. Osteoarthritis Cartilage 9: 128–136. | Article | PubMed | ChemPort |
33. Mi, Z., Ghivizzani, S. C., Lechman, E., Glorioso, J. C., Evans, C. H. and Robbins, P. D. (2003). Adverse effects of adenovirus-mediated gene transfer of human transforming growth factor beta 1 into rabbit knees. Arthritis Res. Ther. 5: R132–R139. | Article | PubMed | ChemPort |
34. Mosca, J. D., et al. (2000). Mesenchymal stem cells as vehicles for gene delivery. Clin. Orthop. 379: S71–S90. | PubMed |
35. Frank, A. C., Meyers, K. A., Welsh, I. C. and O'Brien, T. P. (2003). Development of an enhanced GFP-based dual-color reporter to facilitate genetic screens for the recovery of mutations in mice. Proc. Natl. Acad. Sci. USA 100: 14103–14108. | Article | PubMed | ChemPort |
36. Yang, M., Reynoso, J., Jiang, P., Li, L., Moossa, A. R. and Hoffman, R. M. (2004). Transgenic nude mouse with ubiquitous green fluorescent protein expression as a host for human tumors. Cancer Res. 64: 8651–8656. | Article | PubMed | ChemPort |
37. Pascher, A., et al. (2004). Gene delivery to cartilage defects using coagulated bone marrow aspirate. Gene Ther. 11: 133–141. | Article | PubMed | ChemPort |
38. Hardy, S., Kitamura, M., Harris-Stansil, T., Dai, Y. and Phipps, M. L. (1997). Construction of adenovirus vectors through Cre–lox recombination. J. Virol. 71: 1842–1849. | PubMed | ISI | ChemPort |
39. Le Graverand, M. P., Eggerer, J., Vignon, E., Otterness, I. G., Barclay, L. and Hart, D. A. (2002). Assessment of specific mRNA levels in cartilage regions in a lapine model of osteoarthritis. J. Orthop. Res. 20: 535–544. | Article | PubMed | ChemPort |

Acknowledgements

This work was supported by Grants AR48566 and AR50249 from the National Institute of Arthritis and Musculoskeletal and Skin Diseases.