Objectives In osteoarthritis (OA) chondrocytes surrounding lesions express elevated bone morphogenetic protein 2 (BMP2) levels. To investigate the functional consequence of chondrocyte-specific BMP2 expression, we made a collagen type II dependent, doxycycline (dox)-inducible BMP2 transgenic mouse and studied the effect of elevated BMP2 expression on healthy joints and joints with experimental OA.
Methods We cloned a lentivirus with BMP2 controlled by a tet-responsive element and transfected embryos of mice containing a collagen type II driven cre-recombinase and floxed rtTA to gain a mouse expressing BMP2 solely in chondrocytes and only upon dox exposure (Col2-rtTA-TRE-BMP2). Mice were treated with dox to induce elevated BMP2 expression. In addition, experimental OA was induced (destabilisation of the medial meniscus model) with or without dox supplementation and knee joints were isolated for histology.
Results Dox treatment resulted in chondrocyte-specific upregulation of BMP2 and severely aggravated formation of osteophytes in experimental OA but not in control mice. Moreover, elevated BMP2 levels did not result in alterations in articular cartilage of young healthy mice, although BMP2-exposure did increase VDIPEN expression in the articular cartilage. Strikingly, despite apparent changes in knee joint morphology due to formation of large osteophytes there were no detectible differences in articular cartilage: none with regard to structural damage nor in Safranin O staining intensity when comparing destabilisation of the medial meniscus with or without dox exposure.
Conclusions Our data show that chondrocyte-specific elevation of BMP2 levels does not alter the course of cartilage damage in an OA model in young mice but results in severe aggravation of osteophyte formation.
Statistics from Altmetric.com
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.
Osteoarthritis (OA) is a joint disease with ageing being the dominant risk factor. The main characteristics are damage of the articular cartilage, synovial activation, fibrosis and osteophyte formation.
Members of the transforming growth factor (TGF) superfamily are crucial in maintenance of articular cartilage. Loss of TGF-β signalling is suggested as a prerequisite for cartilage damage in OA.1 Whereas TGF-β receptors and subsequent signalling go down, we found that bone morphogenetic protein 2 (BMP2) is elevated near OA lesions.2 Like TGF-β, it is capable of inducing cartilage formation in vitro and in vivo.3–5 Exposing chondrocytes to BMP2 results in elevation of proteoglycan (PG) production and collagen type II expression, crucial components of articular cartilage.6 However, we found previously that BMP2 can elevate aggrecan degradation in articular cartilage without inducing structural damage in murine knee joints, pointing towards a potential role in cartilage remodelling.7
BMP2 plays a major role in bone formation. In endochondral bone formation, it stimulates chondrocytes to become hypertrophic, accompanied by matrix degradation and matrix metalloproteinase (MMP) 13 expression.8 ,9 In this process of bone formation, BMP2 plays an important role in inducing transient cartilage that is replaced by bone. This is seen in new bone formation in embryos,10 in tissue engineering11 and during fracture healing.12 ,13 During osteophyte formation in OA, BMP2 might be involved in a similar process.14
The elevated levels of BMP2 near lesions are intriguing. Is this a reparative response, trying to repair the lesions using the cartilage-inducing and/or remodelling properties of BMP2? Or is its presence contributing to or even aggravating the OA process by stimulating chondrocytes to become hypertrophic? In addition, does the presence of BMP2 in an OA joint contribute to osteophyte formation or not?
Therefore our research question is: do elevated levels of BMP2 in chondrocytes lead to a reparative response in articular cartilage or to deleterious effects in healthy and/or OA joints? Until now either protein injections or viral overexpression systems are generally used to study the effects of a protein in vivo. However, this results in high BMP2 levels for a relatively short period, which is in contrast with the relative low continuously elevated levels that will be produced by chondrocytes. In addition, adenoviruses primarily target the synovium. To gain prolonged, low level expression in a tissue-specific manner, a transgenic mouse is an obvious choice. However, meddling with factors involved in skeletal development will result in malformation of joints during growth and development, ultimately leading to misinterpretation of results. Therefore we made an inducible chondrocyte-specific transgenic mouse for BMP2.
Using this novel inducible chondrocyte-specific BMP2 transgenic mouse, we show that in the presence of chondrocyte-specific BMP2 overexpression even for prolonged periods of time, healthy articular cartilage is able to maintain its stable phenotype. In addition, we show that chondrocyte-specific BMP2 overexpression can result in osteophyte formation but that this seems to be dependent on an additional trigger. Moreover, the course of articular cartilage damage induced during experimental OA (destabilisation of the medial meniscus (DMM)-model) remains unaltered by BMP2 overexpression despite BMP2-induced aggravation of osteophyte formation.
STOCK Gt(ROSA)26Sortm(rtTA,EGFP)Nagy/J (floxed-rtTA) and B6;SJL-Tg(Col2a1-cre)1Bhr/J mice (Col2a1-cre) (The Jackson Laboratory, Bar Harbor, USA) were used to generate Col2a1-rtTA mice. These were parents for Col2a1-rtTA-BMP2 mice. Mice were kept in filter-top cages with woodchip bedding under standard pathogen-free conditions and fed standard diet and tap water ad libitum. In experiments including doxycycline (dox) exposure standard diet (non-dox) was replaced by food containing dox hyclate (Sigma) (500 mg/kg) which was refreshed twice a week. Male and female mice were used for each experiment and no consistent differences were found between the sexes. This study was approved by the local animal experimentation committee, Nijmegen, The Netherlands.
Construction of inducible chondrocyte specific BMP2 transgenic mice
Generating the inducible chondrocyte-specific BMP2 overexpressing mouse: Col2a1-rtTA-BMP2
We generated a lentivirus containing a minimal cytomegalovirus (CMV) promoter with a tet-responsive element and human BMP2 (Lv-TRE-BMP2) and tested expression and functionality upon dox exposure in vitro (see online supplementary data). We crossed the Col2a1-cre and floxed-rtTA mice to generate homozygotes for both transgenes (Col2a1-rtTA). Col2a1-rtTA embryos were transfected with Lv-TRE-BMP2 as previously described15 Briefly, female mice were superovulated with 5 units Pregnant Mares Serum Gonadotropin (Sigma) followed by 5 units of human chorionic gonadotropin 46 h later, followed by mating with Col2a1-rtTA male mice. After 48 h female oviducts were isolated to collect two-cell embryos. Zona pellucida was removed in acidic tyrodes solution followed by washing and incubation in medium containing Lv-TRE-BMP2 (1, 0.3 or 0.1 ng/μL in 20 μL) for 48 h. Upon morula stage embryos were transplanted into pseudo-pregnant female mice. To determine hBMP2 integrations we performed linear amplification-mediated PCR as previously described.16 For the three different integration sites that were found specific primers for each integration and wild type counterpart were designed for quantitative PCR (Q-PCR) (see online supplementary text). Integrations were tested for functionality (see online supplementary figure S1).
Testing cartilage-specific BMP2 inducibility in Col2a1-rtTA-BMP2 mice
We exposed Col2a1-rtTA-BMP2 mice to dox (n=8) or non-dox (n=6) for 72 h. Knee articular cartilage was isolated for mRNA isolation and Q-PCR. For six mice (three dox and three non-dox), additional mRNA was isolated from spleen and liver.
RNA was isolated using the RNeasy Mini Kit (Qiagen), followed by RT-PCR. The RT reaction mix contained 1.9 μL dH2O2, 2.4 μL 10× DNAse buffer, 2.0 μL 0.1 M Dithiothreitol (DTT), 0.8 μL 25 mM deoxyribonucleotide triphosphate (dNTP), 0.4 μg oligo dT primer, 1 μL 200 U/μL Moloney Murine Leukemia Virus Reverse Transcriptase (M-MLV RT) (all Life Technologies, USA) and 0.5 μL 40 U/μL RNAsin (Promega, the Netherlands).The Q-PCR mix contained 1.5 μL forward and reverse primer (5 μM) (see online supplementary table S1), 4.5 μL dH2O, 12.5 μL Sybr Green PCR master mix (Applied Biosystems, Foster City, California, USA) and 5 μL cDNA. Q-PCR was performed by an ‘ABI/PRISM 7000 sequence detection system’ (Applied Biosystems, Foster City, California, USA).
Chondrocyte-specific overexpression of BMP2 in vivo
To explore the effect of elevated chondrocyte-specific BMP2 on articular cartilage we exposed Col2a1-rtTA-BMP2 mice, with at least two out of three integrations, to non-dox or dox. After 6 weeks knee joints were isolated for histology.
We investigated the effect of elevated chondrocyte-specific BMP2 during OA by exposing Col2a1-rtTA-BMP2 mice to dox or non-dox. After 1 week a DMM model was performed in the right knee joint as previously described.17 Left knee joints served as non-OA controls. Due to osteophyte formation in the spine, some dox treated mice had to be sacrificed earlier than intended resulting in a range of time points between 4 weeks and 8 weeks after DMM. Separate analysis showed no differences in any of the parameters between time points. Therefore all time points were pooled.
Isolated knee joints of mice were fixed for 7 days in phosphate-buffered formalin followed by decalcification in 10% formic acid for 1 week. Tissue was dehydrated with an automated tissue processing apparatus (Miles Scientific Tissue-Tek VIP tissue processor; Miles Scientific, now part of Bayer Corp, Emeryville, California, USA) and embedded in paraffin. Seven µm frontal sections were made and used for staining with Safranin O and Fast Green (Brunschwigh chemie, Amsterdam, the Netherlands) or immunohistochemistry for NITEGE or VDIPEN as previously described.7 For NITEGE, neoepitope NITEGE (1 : 1000) (Aggrecan Neo) (Acris, Hiddenhausen, Germany) was incubated overnight followed by biotin-labelled secondary antibody (DAKO, Denmark), a biotin-streptavidin detection (Vector Laboratories, Burlingame, California, USA) and diaminobenzidine reagent (Sigma-Aldrich, St Louis, Missouri, USA) and counterstaining with haematoxylin (Merck & Co, Whitehouse Station, New Jersey, USA). For VDIPEN, affinity-purified rabbit anti-VDIPEN immunoglobulin G18 ,19 was used after chondroitin ABC treatment, and biotinilated goat-antirabbit which was detected using biotin-streptavidin-peroxidase (Elite kit; Vector Laboratories). Peroxidase staining was developed with nickel enhancement and counterstained with orange G (2%).
OA and osteophyte score
To measure OA we adapted the Pritzker score using the same grade, but modified stage 4 into 50–75% and added a stage 5 which is>75%.20 Final OA score per joint was calculated by grade multiplied by stage, using the mean of three sections per joint.
As a measure of osteophytes the total number of chondrophytes and (chondro-) osteophytes in each joint was counted in Safranin O/Fast green stained sections.
Histological scores of cartilage
Articular cartilage of the medial tibia was selected manually using Leica Application Suite software (Leica Microsystems, Germany). To measure PG content, mean blue was measured in Safranin O stained sections as previously described.21 Briefly, the surface area of the cartilage was selected and the amount of blue light passing through the region of interest was measured by the software. The higher the amount of light, the lower was the PG content. Controls were set at 100%. For VDIPEN and NITEGE, the surface area staining positive for each antibody was measured and corrected for the total area that was measured. The mean of three sections was representative for each knee joint.
Digital radiography imaging of knee joints and spinal osteophytes
We performed digital radiography imaging with a Faxitron MX-20 (Faxitron Bioptics, Tucson USA) at 26 kV to visualise the osteophytes in spines and knee joints that had been isolated from DMM and/or dox-treated Col2a1-rtTA and Col2a1-rtTA-BMP2 mice. Representative images were selected for display.
Data was analysed using Graph Pad Prism V.5 Software, in which an unpaired two-tailed t test was performed or a one-way analysis of variance (ANOVA) with a post hoc Bonferroni for comparison between groups (figure 4). p Values of less than 0.05 were considered statistically significant. Error bars display the SE of the mean.
Chondrocyte-specific BMP2 expression in Col2a1-rtTA-BMP2 mice
After having confirmed expression and functionality of BMP2 by the lentivirus used to generate our transgenic mice, as well as functionality in ex vivo (see online supplementary data) we wanted to confirm expression in articular cartilage. Therefore Col2a1-rtTA-BMP2 mice were fed dox or non-dox for 72 h after which articular cartilage was isolated for mRNA detection of hBMP2. Mice with non-dox had hBMP2 mRNA levels near the Q-PCR detection limit. After dox-treatment hBMP2 mRNA was significantly higher (4.98 ct ±1.18) (p<0.001) (figure 1).
Chondrocyte-specific overexpression of BMP2 does not lead to significant changes in naive murine knee joints
We exposed Col2a1-rtTA-BMP2 mice to dox versus non-dox food for 6 weeks to investigate alterations in articular cartilage. In both groups cartilage stained intense and similar with Safranin O indicating an abundance of PGs (figure 2). No cartilage damage was found in either group. We did find that 4 out of 10 dox-treated Col2a1-rtTA-BMP2 mice had an osteophyte score higher than controls, but the mean was not significantly different (figure 2).
Chondrocyte-specific overexpression of BMP2 during experimental OA (DMM model) does not result in structural alterations in articular cartilage
To investigate whether BMP2 altered the course of OA in articular cartilage we exposed Col2a1-rtTA-BMP2 mice to dox or non-dox food starting 1 week prior to induction of experimental OA (DMM model). Overall the OA score after DMM was relatively low, which can be explained by the early time of sacrifice between 4 weeks and 8 weeks after surgery. Non-dox treated and dox treated groups contained severe and very mild cases of OA (figures 3 and 4). Elevation of hBMP2 levels did not change the OA score, nor the PG content (figure 4).
To investigate in more detail whether the presence of BMP2 had an effect on cartilage degradation we stained for aggrecan-breakdown neoepitopes VDIPEN and NITEGE (figures 3 and 4). NITEGE staining did not show any significant differences between any of the groups, but did show a trend where dox treated groups seem higher than controls. DMM treated animals without dox exposure had a lower expression of VDIPEN compared with those treated with dox. Despite these elevated VDIPEN levels no structural changes in cartilage were observed.
Chondrocyte-specific overexpression of BMP2 during experimental OA (DMM model) induces severe aggravation of osteophyte formation
Whereas BMP2 did not induce changes in articular cartilage, we did find striking BMP2-induced differences in osteophyte formation (figure 3Q–T and 5C,D–G,H). We found a significant increase in the number of osteophytes in mice treated with DMM and dox when compared with DMM non-dox (figure 3Q–T, 4E and 5C,D–G,H). This score did not include a measure for the size of the osteophytes which was much larger in the group with elevated levels of chondrocyte-specific hBMP2 (figure 3T). In contrast, the number of osteophytes was extremely low to absent in non-DMM groups, even when treated with dox (figure 3S,T, 4E and 5C,D) similar to the prior 6-week exposure experiment. These results imply that the DMM treatment provides a trigger crucial for BMP2 to be able to aggravate osteophyte formation.
BMP2 induces osteophyte formation ventrally of the spine
Despite the clear necessity for a prior trigger for BMP2 to induce osteophytes in murine knee joints, we found massive osteophytes ventrally in the spine of dox groups irrespective of a DMM treatment (figure 5A–D). The location of the spinal osteophytes varied greatly from cervical to thoracic and lumbar areas and there did not seem to be a preferential site. Even though there was no exogenous trigger added to induce the osteophytes in addition to the elevated chondrocyte-specific BMP2, there might have been a trigger prior to BMP2 exposure that we were not aware of such as a small injury.
During experimental OA we found that BMP2 is elevated near OA lesions.2 However, it is unclear what the effect of BMP2 is: a reparative response or a deleterious side effect contributing to OA. To overcome potential side effects of bolus injections or adenoviral transfection of the synovium, we switched to a more elegant approach: letting the chondrocytes produce BMP2 themselves by using a collagen type 2a1 promoter as a driving force. Iwai et al previously used the Col11a2 promoter in a conditional transgenic mouse for condensed mesenchymal cells and chondrocytes. This elegantly showed expression in a cartilage-specific pattern.22 Bridgewater et al showed that the Col11a2 resembles the Col2a1 tissue-specific enhancer.23 We used the Col2a1-cre mouse, which Ovchinnikov showed had expression in differentiating chondrocytes.24 Many others used the same promoter/enhancer to drive chondrocyte-specific overexpression, but almost exclusively for expression throughout life.25–27 We required expression only upon adulthood after a normal skeleton had been formed. Therefore we used the inducible tet-on system providing us flexibility to produce BMP2 on demand, thereby enabling the mice to reach adulthood without interfering with tissue build up.
BMP2 has been found to upregulate the production of glycosaminoglycan (GAG) en collagen type II,7 ,28 ,29 both crucial components of the articular cartilage extracellular matrix (ECM). Strikingly, despite its cartilage-inducing capability, elevated chondrocyte-specific BMP2 for 6 weeks did not result in any detectable alteration in the articular cartilage. However, the fact that we did not find any changes in PG content might have been obscured by the already high endogenous PG levels in the healthy young mice that could mask minor changes. Moreover, young healthy cartilage has a stable phenotype and might be able to resist mild insults such as high levels of growth factors.30 ,31 We have previously shown a difference between young and old mice in their responsiveness to deleterious agents like interleukin 1.30
We have shown that chondrocytes in young cartilage have a dominant ALK5 versus ALK1 receptor level for TGF-β signalling resulting in Smad2/3 being more dominant over Smad1/5/8 in young mice.32 This Smad2/3 signalling route is crucial for maintenance of the articular cartilage and is known to block terminal differentiation.33 ,34 We have to keep in mind that in young mice, the solid Smad2/3 levels might prevent BMP2 from having an effect on cartilage that might be lost in old individuals.
Our data show that even having OA does not generate a condition for BMP2 to have an effect on articular cartilage, as we did not find a BMP2-induced change in OA-score during a DMM model. We did find a BMP2-induced increase in VDIPEN expression during DMM, but that did not result in increased structural damage.
In contrast with the articular cartilage, BMP2 induced a striking difference in osteophyte formation. In non-predisposed animals we could incidentally find minor chondrophytes, but in the presence of DMM, osteophyte formation was severely aggravated by BMP2, in number and size. This implies that an additional trigger might be a prerequisite for BMP2 to induce osteophytes. We previously showed that TGF-β is crucial for osteophyte development as blocking TGF-β during OA completely prevented osteophyte formation.21 In contrast, using the BMP inhibitor gremlin, during OA could not inhibit the onset of osteophyte formation.14 ,21 Similar to the lack of osteophyte formation in dox-treated mice, this shows that moderate BMP levels are not sufficient to start osteophyte formation. However, the aggravated osteophyte formation by BMP2 in our DMM model shows an important synergistic role in osteophyte formation. We hypothesise (assume) that TGF-β is the crucial trigger in the present experiments driving periosteal cells to differentiate into chondrocytes. Once they are chondrocytes the Col2a1 promoter activates the dox-dependent BMP2 production which amplifies osteophyte formation.
Mechanistically it is interesting that the articular chondrocytes are seemingly uninfluenced by BMP2 whereas the cells that will become the osteophytes are highly reactive. Apparently the cells are truly different. Whereas the articular chondrocytes are resistant, the cells developing the osteophytes go into an almost self-perpetuating loop. The latter is most likely due to the fact that the cells undergo differentiation towards chondrocytes, which then due to our transgenic insert start producing BMP2, in turn affecting adjacent cells. In a recent study comparing articular chondrocytes with those developing into osteophytes, they found that articular chondrocytes expressed increased transcript levels of antagonists and inhibitors of the BMP-signalling and Wnt- signalling pathways of which gremlin was one of the two most prominent differentially expressed . Gremlin is a natural inhibitor of BMPs and might therefore provide a potential explanation why the articular chondrocytes are resistant whereas the osteophyte-developing chondrocytes are not.35
An interesting finding is that in otherwise seemingly unchallenged mice, elevated BMP2 levels did result in ventral osteophyte formation in the spine. We speculate that there was a trigger that we were unaware of. Mice having been in a fight could easily have a small, otherwise unnoticed trauma that presensitises the cells, which when combined with elevated levels of BMP2 could result in osteophyte formation. The lack of a particular site of preference supports the idea that an additional trigger might be required for osteophyte formation. That might also provide a possible explanation for why ectopic bone formation occurs only in some of the patients when BMP2 is used in a clinical setting, while others have no side effects.36
Our current results confirm our prior experiments using the Ad-BMP2 virus. There BMP2 did not induce damage to the articular cartilage despite the elevated levels of VDIPEN and NITEGE.7 We found similar effects on VDIPEN, whereas NITEGE staining showed a trend towards higher expression in dox-treated groups. Lack of elevated degradational epitopes in the DMM-treated groups are most likely due to the short time after DMM and subsequent mild OA. Clearly in healthy young mice the BMP2 effect on articular cartilage is minor irrespective of the administration. The major difference in results between Ad-BMP2 and the transgenic mouse model was that Ad-BMP2 induced massive osteophytes without the presence of OA. However, from our current experiment we identified that an additional trigger might be key for BMP2 being able to induce osteophyte formation. Therefore besides the apparent difference in dose and duration of BMP2 exposure, bolus injection as well as use of an adenovirus could already provide the necessary trigger prior to BMP2 exposure.
In conclusion, our data show that chondrocyte-specific elevation of BMP2 levels is not deleterious for healthy cartilage and does not alter the course of cartilage damage in an OA model in young mice but results in severe aggravation of osteophyte formation.
The authors would like to express gratitude to Professor Dr Inder Verma for the opportunity to learn the technique of embryo transfection with lentivirus in his lab at the Salk Institute, La Jolla, USA.
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Files in this Data Supplement:
- Data supplement 1 - Online supplement
Handling editor Tore K Kvien
Contributors ENBD conceived this study, was involved in experimental setup and performance and data collection of every experiment, performed all analysis and wrote the article. ELV was involved in all experimental set-up and data collection. MBB, PLEMvL, APMvC and ABB were all involved in subsets of experiments, in set-up, performing experiments and data acquisition. WBvdB and FAJvdL were involved in experimental set-up, critically evaluated data analysis and critically reviewed the article. PMvdK conceived this study, was involved in experimental set-up, critically evaluated data analysis and was involved in writing the article.
Funding This work was funded by the Dutch arthritis association (Reumafonds) (Reumafonds grant number 11-1-402) and the Nederlandse Vereniging Reumatologie (NVR) (Rheumatology Grant).
Competing interests None.
Provenance and peer review Not commissioned; externally peer reviewed.