Presented by Ms Wehling in partial fulfillment of the requirements for an MD degree, Ludwig-Maximilians-University of Munich Medical School, Munich, Germany.
Interleukin-1β and tumor necrosis factor α inhibit chondrogenesis by human mesenchymal stem cells through NF-κB–dependent pathways†
Article first published online: 26 FEB 2009
Copyright © 2009 by the American College of Rheumatology
Arthritis & Rheumatism
Volume 60, Issue 3, pages 801–812, March 2009
How to Cite
Wehling, N., Palmer, G. D., Pilapil, C., Liu, F., Wells, J. W., Müller, P. E., Evans, C. H. and Porter, R. M. (2009), Interleukin-1β and tumor necrosis factor α inhibit chondrogenesis by human mesenchymal stem cells through NF-κB–dependent pathways. Arthritis & Rheumatism, 60: 801–812. doi: 10.1002/art.24352
- Issue published online: 26 FEB 2009
- Article first published online: 26 FEB 2009
- Manuscript Accepted: 25 NOV 2008
- Manuscript Received: 28 MAR 2008
- AO Foundation, Davos, Switzerland. Grant Number: Research Fund project 04-B86
- Postdoctoral Fellowship from the NIH/National Institute of Biomedical Imaging and Bioengineering. Grant Number: F32-EB-005566
- Top of page
- MATERIALS AND METHODS
- AUTHOR CONTRIBUTIONS
The differentiation of mesenchymal stem cells (MSCs) into chondrocytes provides an attractive basis for the repair and regeneration of articular cartilage. Under clinical conditions, chondrogenesis will often need to occur in the presence of mediators of inflammation produced in response to injury or disease. The purpose of this study was to examine the effects of 2 important inflammatory cytokines, interleukin-1β (IL-1β) and tumor necrosis factor α (TNFα), on the chondrogenic behavior of human MSCs.
Aggregate cultures of MSCs recovered from the femoral intermedullary canal were used. Chondrogenesis was assessed by the expression of relevant transcripts by quantitative reverse transcription–polymerase chain reaction analysis and examination of aggregates by histologic and immunohistochemical analyses. The possible involvement of NF-κB in mediating the effects of IL-1β was examined by delivering a luciferase reporter construct and a dominant-negative inhibitor of NF-κB (suppressor-repressor form of IκB [srIκB]) with adenovirus vectors.
Both IL-1β and TNFα inhibited chondrogenesis in a dose-dependent manner. This was associated with a marked activation of NF-κB. Delivery of srIκB abrogated the activation of NF-κB and rescued the chondrogenic response. Although expression of type X collagen followed this pattern, other markers of hypertrophic differentiation responded differently. Matrix metalloproteinase 13 was induced by IL-1β in a NF-κB–dependent manner. Alkaline phosphatase activity, in contrast, was inhibited by IL-1β regardless of srIκB delivery.
Cell-based repair of lesions in articular cartilage will be compromised in inflamed joints. Strategies for enabling repair under these conditions include the use of specific antagonists of individual pyrogens, such as IL-1β and TNFα, or the targeting of important intracellular mediators, such as NF-κB.
Loss of articular cartilage through injury or disease presents major clinical challenges. Because cartilage has very poor regenerative capacity, various surgical techniques for cartilage repair have been devised (1). While helpful, these procedures fail to provide sustained clinical improvement in most patients, so there is much interest in the development of alternative, biologic approaches to cartilage repair and regeneration (2). Among these, there is considerable research investigating the potential use of mesenchymal stem cells (MSCs) to regenerate cartilage (3).
There are 2 general strategies for harnessing MSCs for this purpose. In a tissue-engineering approach, MSCs are recovered from the patient and used to generate a graft that is subsequently implanted into the site of cartilage damage (4). Although the graft can be developed into mature cartilage in a bioreactor, there is increasing interest in grafting immature tissue, allowing chondrogenesis to occur in situ. The second strategy, which is already in wide clinical use, supplies MSCs to the defect by penetrating the subchondral bone, thereby allowing marrow to enter the lesion. Various related surgical techniques, including microfracture and subchondral drilling, are used for this purpose. These procedures have the convenience of being performed arthroscopically in large joints (1).
Repair strategies that rely on the in situ differentiation of MSCs are attractive but, in many instances, require chondrogenesis to take place within an inflamed environment. Intraarticular inflammation may result from disease, such as arthritis, or from trauma, including the iatrogenic trauma of the cartilage repair surgery itself. Because interleukin-1β (IL-1β) and tumor necrosis factor α (TNFα) are major mediators of local inflammatory processes in the joint, the present study was undertaken to study their effects on chondrogenesis by MSCs derived from human bone marrow.
There is extensive literature dating back over 20 years that describes the inhibition of cartilage matrix synthesis by chondrocytes in response to IL-1β and TNFα (5, 6). In producing these effects, IL-1β and TNFα activate the transcription factor NF-κB, which in turn, inhibits the synthesis of SOX9, another transcription factor required for expression of the chondrocyte phenotype (7); there is evidence that this occurs posttranscriptionally by destabilizing SOX9 messenger RNA (mRNA) (8). However, there is surprisingly little in the literature concerning the influence of mediators of inflammation on the differentiation of MSCs into chondrocytes.
Majumdar et al (9) isolated CD105+ mesenchymal cells from human bone marrow, placed them in alginate culture, and initiated chondrogenesis by the addition of bone morphogenetic protein 2 (BMP-2) and BMP-9. Addition of IL-1 after 14 days of differentiation reduced the abundance of mRNA encoding COL2A1, aggrecan, and SOX9. In a related study, Sitcheran et al (8) used an established murine cell line, MC615, and placed monolayers into a chondrogenic medium. This induced SOX9 mRNA, whose expression was inhibited by TNFα through the induction of NF-κB. These data are consistent with the hypothesis that IL-1 and TNFα inhibit chondrogenesis by MSCs, but this has not been formally demonstrated.
In the present study, human MSCs were obtained from bone marrow and placed into aggregate culture with transforming growth factor β1 (TGFβ1) to induce chondrogenesis. IL-1β and TNFα were found to inhibit cartilage formation very powerfully in a dose-dependent and NF-κB–dependent manner. These findings have important implications for the design of effective, clinically expeditious, cartilage-regeneration strategies. They may also help explain why cartilage regeneration does not occur spontaneously in joints.
MATERIALS AND METHODS
- Top of page
- MATERIALS AND METHODS
- AUTHOR CONTRIBUTIONS
Generation of adenoviral vectors.
First-generation (ΔE1, ΔE3) serotype 5 adenovirus encoding the complementary DNA (cDNA) for a dominant-negative super-repressor IκB (Ad.srIκB) driven by the cytomegalovirus (CMV) immediate early promoter were generously provided by Paul Robbins (University of Pittsburgh School of Medicine, Pittsburgh, PA). Recombinant adenovirus encoding the cDNA for firefly luciferase (Luc) or green fluorescent protein (GFP) and driven by the CMV promoter (Ad.CMV-Luc and Ad.GFP, respectively) were constructed according to the method of Hardy et al (10), as previously described (11). A similar vector encoding the cDNA for firefly luciferase but driven by a synthetic, NF-κB–specific promoter region (Ad.NF-κB-Luc) was acquired from Vector BioLabs (Philadelphia, PA).
To generate high-titer preparations, recombinant vectors were amplified in 293-Cre8 cells and purified over 3 successive CsCl gradients. Following overnight dialysis against a sucrose buffer, the preparations were aliquotted and stored in liquid nitrogen. Viral titers were estimated by optical density (at 260 nm) and standard plaque assay. Using this method, preparations of ∼1012 viral particles/ml were obtained, with a ratio of viral particles to plaque forming units of <100:1.
Isolation and expansion of MSCs.
MSCs were isolated from intramedullary reamings collected from 4 patients (1 man and 3 women; ages 71–78 years) undergoing hip hemiarthroplasty at Brigham and Women's Hospital (Boston, MA), as previously described (12). Tissues were obtained and handled in accordance with a protocol approved by the Institutional Review Board.
Briefly, material generated using a Reamer Irrigator Aspirator (Synthes, Paoli, PA) was filtered of osseous particles, and the filtrate was collected in a sterile vessel. The mononuclear cell fraction within the filtrate was isolated using a Ficoll-Paque Plus density gradient (30 minutes at 400g; StemCell Technologies, Beverly, MA) and cultured at 5 × 107 nucleated cells/flask in 75-cm2 flasks with growth medium, which consisted of low-glucose Dulbecco's modified Eagle's medium (DMEM; Invitrogen, Carlsbad, CA) supplemented with 10% fetal bovine serum (FBS; HyClone, Logan, UT) and 1% antibiotic/antimycotic cocktail (Invitrogen). After 2 weeks in primary culture, adherent colonies were recovered using 0.05% trypsin, 0.5 mM EDTA buffer (Invitrogen) and expanded for 1–2 additional passages using growth medium containing 10 ng/ml of recombinant human fibroblast growth factor 2 (FGF-2; PeproTech, Rocky Hill, NJ), which has been shown to maintain the chondrogenic potential of cultured human MSCs (13).
Transduction of MSCs.
Once sufficient cell numbers were obtained for differentiation cultures, MSCs were rinsed with phosphate buffered saline (PBS) and transduced with Ad.GFP or Ad.srIκB for 2 hours in serum-free DMEM (14). For experiments requiring high transduction efficiency, viral transduction was enhanced by coprecipitation with lanthanum phosphates (15). Briefly, LaCl3 anhydrous salt (Sigma, St. Louis, MO) was dissolved in deionized water to a stock concentration of 0.4M (pH 5.5–6.0). Viral stocks were added to serum-free DMEM (phosphate source) to generate the final concentrations indicated below. To these solutions, sufficient LaCl3 stock was added to produce a final concentration of 0.2 mM La3+, and the mixtures were gently vortexed and incubated at room temperature for 15 minutes prior to infection. After formation of the lanthanum phosphate–adenovirus complex, 5 ml of the suspension was added to each flask, and the flasks were incubated at 37°C for 30 minutes. An additional 5 ml of DMEM plus 10% FBS was added per flask, and incubation was continued for an additional 90 minutes. The transduction solutions were then aspirated and replaced with fresh growth medium with recombinant human FGF-2. After 24–48 hours, cells were harvested for aggregate culture as described below.
To determine the range of MSC transduction using this approach, cell monolayers were transduced with 3 distinct levels of Ad.GFP: low (104 viral particles/cell without La3+), moderate (103 viral particles/cell with La3+), or high (104 viral particles/cell with La3+). After 6 days, cells were trypsinized and rinsed in PBS, and 20,000 cells per group were analyzed on a Cytomics FC 500 flow cytometer (Beckman Coulter, Fullerton, CA) using a band-pass filter at 525 nm.
Aggregate chondrogenesis model.
MSCs were centrifuged into cell aggregates and induced along the chondrogenic lineage as previously described (16, 17). Cells were suspended to a concentration of 1 × 106/ml in serum-free DMEM, and 200-μl aliquots (2 × 105 cells) per well were added to a polypropylene, V-bottom 96-well plate (Corning, Corning, NY), and the plate was spun at 400g for 5 minutes. The supernatant was aspirated and replaced with chondrogenic inductive medium consisting of high-glucose DMEM (containing L-glutamine and sodium pyruvate) with a 1% antibiotic/antimycotic cocktail, 1% ITS+ (insulin–transferrin–selenium) Premix (BD Biosciences, San Jose, CA), 40 μg/ml of proline, 100 nM dexamethasone, and 50 μg/ml of ascorbic acid 2-phosphate (all but ITS+ from Sigma). To a portion of aggregates, 10 ng/ml of recombinant human TGFβ1 (PeproTech) was added to enhance chondrogenesis, and recombinant human IL-1β or TNFα (PeproTech) was added as inflammatory stimulus at the concentrations described below. The cell pellets formed free-floating aggregates within the first 24 hours. Media were changed every other day, and aggregates were collected at various time points for analysis, as described below.
Measurement of NF-κB activity.
NF-κB activity in MSCs was assessed using the Ad.NF-κB-Luc reporter construct. Cells at 70–80% confluence were transduced for 2 hours with 104 viral particles/cell of either Ad.NF-κB-Luc or Ad.CMV-Luc and returned to growth medium overnight. For each luciferase vector group, flasks were transduced again according to 1 of the following 5 subgroups: no virus, high Ad.GFP (see above), low Ad.srIκB, moderate Ad.srIκB, and high Ad.srIκB.
The next day, cells were formed into aggregates and cultured for 5 days in chondrogenic inductive medium with 10 ng/ml of TGFβ1. On day 5, IL-1β (10 ng/ml) was added to a portion of the cultures. After an additional 5 hours, aggregates were collected in 200 μl of Glo Lysis buffer (Promega, Madison, WI), and luciferase activities were determined with a Bright-Glo Luciferase Assay system (Promega). Measurements from Ad.NF-κB-Luc groups were normalized by those from matched Ad.CMV-Luc groups to control for any effects of the various treatments on the general uptake and expression of recombinant adenoviral vector (i.e., independent of NF-κB activity).
For Western blot analysis, aggregates were collected in Laemmli buffer after 6 days of culture in chondrogenic inductive medium with TGFβ1. Protein concentrations were determined by the Lowry method using an RC DC Protein Assay (Bio-Rad, Hercules, CA). Equal amounts of proteins (20–40 μg) were resolved by sodium dodecyl sulfate–polyacrylamide gel electrophoresis on a 10% polyacrylamide gel and transferred to a polyvinylidene difluoride membrane. Immunodetection was performed with rabbit antibodies against human IκBα, phospho-IκBα, and β-actin (all from Santa Cruz Biotechnology, Santa Cruz, CA) and a horseradish peroxidase–conjugated goat anti-rabbit IgG (Chemicon, Temecula, CA). Bands were visualized using a Western Lightning Chemiluminescence System (PerkinElmer, Waltham, MA) and a Kodak Image Station 2000MM (Eastman Kodak, Rochester, NY).
Measurement of pellet size.
After 6 weeks in culture, pictures were taken of the PBS-washed aggregates while they were still in the wells. These images were analyzed with ImageJ software (NIH Image, National Institutes of Health, Bethesda, MD; online at: http://rsb.info.nih.gov/ij/), measuring aggregate cross-sectional areas. Area measurements were converted from pixels squared to millimeters squared using a reference object of known size.
Measurement of DNA and glycosaminoglycan content.
The DNA content of aggregates was determined using the Hoechst Dye 33258 method (18). Samples were digested overnight at 65°C using 100 μg/ml of proteinase K in 30 mM Tris (pH 7.8), 50 mM NaCl, and 10 mM MgCl2. Pellet digests were taken through 3 freeze–thaw cycles, and aliquots were added to 100 ng/ml of Hoechst Dye 33258 (Sigma) in 10 mM Tris (pH 7.4), 1 mM disodium EDTA, and 100 mM NaCl. The fluorescence intensity was measured immediately with a DQ300 Fluorometer (Hoefer Scientific Instruments, San Francisco, CA), and the DNA concentration was determined from a standard curve established with calf thymus DNA (Sigma).
Proteinase K digests were also analyzed for glycosaminoglycan (GAG) content using the dimethylmethylene blue (DMMB) dye binding assay (19). Briefly, aliquots of pellet digest (or serial dilutions) were combined with DMMB solution, and samples were measured at 595 nm. GAG concentrations were interpolated from a standard curve of shark chondroitin 6-sulfate (Sigma), and results were normalized according to the DNA content.
Histologic and immunohistochemical analyses.
Aggregates were fixed for 30 minutes in 4% paraformaldehyde, encapsulated in 0.5% agarose gels (for better handling), embedded in paraffin, and sectioned (5-μm thickness). Sections were mounted onto glass slides, deparaffinized with 3 xylene washes (5 minutes each), and rehydrated in graded alcohol solutions. For detection of matrix proteoglycan, representative sections were stained with 1.0% toluidine blue (Sigma), pH 3.0, for 30 minutes. Slides were rinsed in deionized water, dehydrated in graded alcohol, rinsed 3 times in xylene, and cover-slipped with Cytoseal XYL mounting medium (Richard-Allan Scientific, Kalamazoo, MI).
For immunohistochemistry, endogenous peroxidases were quenched in hydrogen peroxide solution, and aggregate sections were digested with 0.1 units/ml of chondroitinase ABC (Sigma) in PBS with 1% bovine serum albumin (BSA; Sigma) for 1 hour at 37°C. After blocking with 1% BSA plus 10% normal donkey serum in PBS for 1 hour, slides were incubated overnight at 4°C with polyclonal goat anti-human type I or type II collagen antibodies (both from Santa Cruz Biotechnology) or with goat anti–type X collagen (a generous gift from Gary Gibson, Henry Ford Hospital, Detroit, MI) in blocking buffer. Antigens were visualized using a biotin-conjugated donkey anti-goat secondary antibody (Santa Cruz Biotechnology) and a streptavidin–horseradish peroxidase labeling kit (Dako, Carpinteria, CA) using 3,3′-diaminobenzidine. Slides were rinsed, counterstained with hematoxylin, dehydrated, cover-slipped, and imaged on a Leica DM LB microscope (Leica Microsystems, Wetzlar, Germany).
Quantitative reverse transcription–polymerase chain reaction (RT-PCR).
Total RNA was extracted from pooled MSC aggregates (3 per group) following homogenization in a Pyrex tissue grinder (Wheaton, Millville, NJ) and lysis in RLT buffer (Qiagen, Valencia, CA). RNA was purified using RNeasy Micro kit reagents according to the manufacturer's protocol (Qiagen). For cDNA synthesis, 0.2 μg of total RNA for each group was reverse-transcribed using oligo(dT)18 primers and SuperScript III reverse transcriptase (Invitrogen). The cDNA product was diluted 1:3 with deionized H2O, and 2-μl aliquots were used as templates for amplifying partial coding sequences for types II and X collagen, aggrecan core protein, matrix metalloproteinase 13 (MMP-13), and GAPDH by quantitative PCR.
Real-time quantitative RT-PCR was performed using TaqMan technology and an ABI Prism 7300 sequencer (Applied Biosystems, Foster City, CA). PCR reaction parameters were as follows. The reaction mixture consisted of 2 μl of cDNA and 7 μl of deionized H2O mixed with 10 μl of Master Mix 2× Buffer (Applied Biosystems) and 1 μl of TaqMan primer in a final volume of 20 μl. PCR cycles consisted of 10 minutes at 95°C, followed by 40 amplification cycles (95°C for 15 seconds and 60°C for 60 seconds). To minimize the effects of unequal quantities of starting RNA and to eliminate potential sources of inconsistency, the relative expression levels of each gene were normalized to GAPDH using the 2 method (20). Predesigned real-time PCR primers for all gene targets were obtained from Applied Biosystems.
Measurement of alkaline phosphatase (AP) activity.
For measurement of AP activity, pellets were homogenized as described above and resuspended in 0.25 ml of 2% Triton X-100. AP activity was assayed using p-nitrophenyl phosphate (Sigma) as the substrate in a buffer containing 0.15M Tris HCl (pH 9.0), 0.1 mM MgCl2, and 0.1 mM ZnCl2 according to the method of Teixeira et al (21). Hydrolysis of p-nitrophenyl phosphate was monitored as the change in absorbance at 410 nm. AP activity was expressed as nanomoles of product (p-nitrophenol) per minute, where 1 absorbance unit represents 64 nmoles of product. Activity values were normalized according to the total protein content for each sample, as determined using a DC Protein Assay kit (Bio-Rad).
Aggregate experiments were performed at least 3 times using cells from different donors. Representative data sets are shown for all results except for quantitative RT-PCR and AP assays, where the averages from multiple donor populations are presented. Statistically significant differences between treatment groups were determined by Student's unpaired 2-tailed t-tests. P values less than 0.05 were considered significant.
- Top of page
- MATERIALS AND METHODS
- AUTHOR CONTRIBUTIONS
IL-1β and TNFα suppression of the TGFβ-induced chondrogenesis of human MSCs.
We first examined the effects of IL-1β and TNFα on the chondrogenic differentiation of human MSCs induced by TGFβ1. When stimulated with 10 ng/ml of TGFβ1, the size of the cell aggregates increased substantially over 6 weeks (Figure 1A). Cells within these aggregates were surrounded by a GAG-rich extracellular matrix, as demonstrated by purple staining with toluidine blue (Figure 1B). In the presence of IL-1β or TNFα, the aggregate size and GAG staining did not exceed those of cultures lacking TGFβ1. Quantitative analysis of the GAG content of pellet digests confirmed that both cytokines completely blocked the TGFβ1-induced increase in proteoglycan synthesis (Figure 1C). GAG levels in pellet-conditioned media paralleled those of corresponding digest samples, increasing from a mean ± SD of 13.4 ± 3.0 μg/pellet/24 hours to 112.0 ± 26.7 μg/pellet/24 hours (n = 6 pellets from 2 independent experiments) with the addition of TGFβ1, but decreasing to 4.0 ± 2.4 μg/pellet/24 hours (P < 0.0001) with IL-1β costimulation. This suggests that the proinflammatory cytokines inhibited proteoglycan biosynthesis rather than curtailing their deposition (i.e., via increased proteolytic activity).
Testing the dose dependence of inhibition, we found that IL-1β significantly reduced the pellet size (Figure 1D) and GAG content at concentrations as low as 0.1 ng/ml. Quantitative analysis of the GAG content confirmed the histologic results (Figure 1E, top); however, on a per-cell basis, GAG synthesis was significantly reduced only at IL-1β concentrations >0.1 ng/ml (Figure 1E, bottom). Trends with TNFα were similar to those with IL-1β (data not shown).
NF-κB activation and inhibition in human MSC aggregates.
To evaluate the possible role of NF-κB signaling in mediating the effects of IL-1β, we used an adenoviral construct encoding a dominant-negative super-repressor IκB. In this mutant form of IκBα, serines 32 and 36 are replaced by alanine residues (S32A/S36A), preventing the phosphorylation and subsequent degradation of the transgene product (22, 23). Because IκB acts intracellularly, comprehensive effects require its efficient delivery to nearly all cells in culture. Thus, we first studied a similar adenoviral construct encoding the GFP reporter to determine transduction efficiency under different conditions, using the lanthanum method (15) to overcome the otherwise modest susceptibility of MSCs to infection by adenovirus. As shown in Figure 2A, the 3 conditions we selected produced a broad range of GFP+ cells (outer gates). With the aid of lanthanum phosphate precipitation, more high-intensity fluorescence (inner gates) could be achieved using adenoviral transduction.
Using the same viral doses, we next confirmed the expression of the srIκB construct within the chondrogenesis model. Western blot analysis of aggregates collected 6 days posttransduction verified the viral-dose–dependent expression of the super-repressor, which has a higher molecular weight than endogenous IκBα because of the addition of a hemagglutinin tag (Figure 2B). Using an NF-κB–driven firefly luciferase reporter, we then demonstrated quantitatively that the increase in NF-κB activity in response to IL-1β could be inhibited by the srIκB construct in a dose-dependent manner (Figure 2C). As suggested by the phosphorylated IκB band visible in Figure 2B, Ad.GFP administered at the high dose induced NF-κB activity to levels near those produced by IL-1β (Figure 2C). This is consistent with the ability of recombinant adenovirus vectors to activate NF-κB (24) and helps to explain our previous observation (25) that high doses of empty adenovirus vector inhibit chondrogenesis in aggregate culture. The absence of a phosphorylated IκB band in cells transduced with Ad.srIκB confirms the biologic activity of the srIκB encoded by this vector. Based on these results, we chose the lower 2 doses of vector (designated low and moderate; precise transduction conditions are described in Material and Methods) for subsequent chondrogenesis experiments.
Reversal of the IL-1β–induced suppression of human MSC chondrogenesis by NF-κB inhibition.
When srIκB was delivered to human MSCs prior to the addition of IL-1β, aggregate growth in response to TGFβ1 was completely restored by Ad.srIκB in a dose-dependent manner (Figures 3A and B). Toluidine blue staining demonstrated the dose-dependent recovery of proteoglycan synthesis with srIκB delivery (Figure 3C). Interestingly, isolated regions of GAG deposition existed within IL-1-treated pellets given the “low” dose of Ad.srIκB, which is expected to transduce only a fraction of cells. We suggest that these regions contain cells that received sufficient copies of the srIκB gene and that neighboring nonchondrogenic cells did not. The dose-dependent increase in proteoglycan content in response to Ad.srIκB was verified by quantitative assays (Figure 3D, bottom). No significant difference was seen between the IL-1β–positive and IL-1β–negative groups at the moderate dose of Ad.srIκB. In some treatment groups, the DNA content increased with the level of srIκB expression (Figure 3D, top), suggesting a mitogenic response to IL-1β in the absence of a chondrogenic one. This may partly explain the corresponding differences in aggregate cross-sectional area (Figure 3B). Of note, the extent of DNA content and aggregate size responses varied among donor cell populations (data not shown). No differences in aggregate size or proteoglycan content were observed between Ad.GFP and untransduced controls.
The protective effects of NF-κB inhibition on chondrogenesis were further examined by molecular analysis of the 6-week aggregates. Quantitative RT-PCR showed that expression of type II collagen and aggrecan core protein was induced by TGFβ1, inhibited by IL-1β, and restored by Ad.srIκB, but not Ad.GFP (Figure 4A). Levels of type X collagen mRNA generally followed the trends for type II collagen, which is common for this in vitro model (26). MMP-13 expression was substantially up-regulated only in the presence of both TGFβ1 and IL-1β. In this treatment group, MMP-13 expression was down-regulated ∼100-fold when Ad.srIκB was substituted for the Ad.GFP. The large variance in the fold-induction values can be partly attributed to the pooling of intra-experiment aggregates and comparison of expression levels among multiple donor populations. However, statistically significant differences among groups were determined by comparison of the ΔCt values, as shown in Table 1.
|Vector, treatment||Aggrecan core protein||Type II collagen||Type X collagen||MMP-13|
|Without IL-1β, without TGFβ1||7.9 ± 1.5||11.2 ± 0.2||9.2 ± 3.2||3.9 ± 3.6|
|Without IL-1β, with TGFβ1||0.1 ± 0.3||−1.5 ± 0.80||−0.5 ± 1.0||2.8 ± 1.4|
|With IL-1β, without TGFβ1||10.0 ± 3.5||10.3 ± 4.1||10.1 ± 3.9||2.4 ± 2.5|
|With IL-1β, with TGFβ1||6.3 ± 1.8||5.7 ± 2.7||4.3 ± 0.6||−2.7 ± 2.4|
|Without IL-1β, without TGFβ1||7.0 ± 3.2||9.3 ± 4.2||7.8 ± 2.4||6.1 ± 3.2|
|Without IL-1β, with TGFβ1||0.6 ± 0.8||−0.7 ± 1.7||0.5 ± 2.5||3.6 ± 1.6|
|With IL-1β, without TGFβ1||9.7 ± 3.7||11.7 ± 5.3||12.8 ± 5.6||3.8 ± 3.7|
|With IL-1β, with TGFβ1||2.2 ± 2.3||0.8 ± 3.8||3.9 ± 4.4||0.3 ± 4.7|
|Ad.GFP with TGFβ1|
|Without versus with IL-1β||0.004||0.012||0.002||0.024|
|Ad.srIκB with TGFβ1|
|Without versus with IL-1β||0.315||0.561||0.311||0.303|
Immunohistochemical analysis confirmed the synthesis of type II and type X collagen at the protein level, which was lost in the presence of IL-1β but was recovered by transduction with Ad.srIκB (Figure 4B).
No correlation between AP activity and trends in chondrogenic markers.
Increased AP activity is an indicator of osteoblastic differentiation of MSCs as well as hypertrophic differentiation of growth plate chondrocytes. AP activity in aggregate lysates was increased ∼100-fold by TGFβ1 stimulation, but returned to near-control levels with IL-1β costimulation (Figure 5). Surprisingly, the delivery of srIκB did not reverse the effects of IL-1β, in contrast to the expression of chondrogenic markers (Figure 4A and Table 1). This suggests that suppression of NF-κB activity could reduce only certain aspects of the hypertrophic differentiation generated within cartilage by differentiating chondroprogenitors.
- Top of page
- MATERIALS AND METHODS
- AUTHOR CONTRIBUTIONS
For reasons of injury or disease, IL-1β and TNFα are likely to be present in joints in which cartilage is undergoing attempted repair or regeneration. It has been known for a long time that IL-1β and TNFα increase the breakdown of the extracellular matrix of articular cartilage while inhibiting its synthesis (5, 6). As shown by the data from the present study, these cytokines also inhibit the chondrogenic differentiation of human MSCs. These circumstances combine to present enormous biologic challenges to the regeneration of cartilage in human joints and may explain why existing methods fail to provide a reliably successful long-term clinical outcome.
Our data implicate NF-κB in the mechanism through which IL-1β and TNFα inhibit chondrogenesis. This is consistent with the ability of NF-κB to block the expression of SOX9 (8), a transcription factor that is essential for chondrogenesis, and to down-regulate the expression of TGFβ receptor type II (27). There is also evidence that NF-κB inhibits the phosphorylation of stimulatory Smad3/4 in response to TGFβ (28) and enhances the activity of the inhibitory Smad 7 (27, 29).
These observations suggest possible strategies for improving the clinical outcome of cartilage repair procedures. The most immediate are to inhibit the activities of IL-1β and TNFα within the joint space or to block the actions of NF-κB within chondrocytes and cells undergoing chondrogenesis. These two general approaches are not mutually incompatible.
Various antagonists of IL-1β and TNFα are already in clinical use for the treatment of rheumatoid arthritis and other inflammatory diseases (30). They include the IL-1 receptor antagonist (IL-1Ra; anakinra) (31), antibodies directed against TNFα (infliximab, adalimumab), and bivalent soluble TNF receptor–IgG fusion protein (etanercept) (32). Although these are delivered systemically for the treatment of rheumatic diseases, they could be injected intraarticularly into joints undergoing cartilage repair to avoid the side effects and costs associated with their systemic application. However, the intraarticular dwell time of these recombinant proteins is unlikely to be sufficient for the purposes of cartilage regeneration. Gene transfer, either to the synovium (33) or, when using ex vivo strategies, directly to cultured chondrocytes (34) or chondroprogenitors (25), offers one technology for obviating this limitation. The feasibility and safety of transferring IL-1Ra cDNA to the synovial lining of human joints has already been demonstrated (35).
Using an in vitro model of equine cartilage degeneration, Haupt et al (36) demonstrated the protective effect on articular cartilage of delivering IL-1Ra cDNA to synovial cells. IL-1Ra has a molecular weight of 20–25 kd, depending on the degree of glycosylation, and is small enough to diffuse freely into cartilage or sites of cartilage regeneration when synthesized by the synovium (37). Antibody-based drugs, such as infliximab, etanercept, and adalimumab, are probably too large to do so. Other, more general antiinflammatory agents, such as autologous conditioned serum (38), might be more useful in this context, although high doses of steroids (39) and nonsteroidal antiinflammatory drugs (40) are probably contraindicated because they inhibit the synthesis of the cartilaginous matrix.
NF-κB provides an alternative target. Because it is induced by a number of different inflammatory agents, blocking this transcription factor may provide more comprehensive protection for regenerating cartilage within the joint than targeting individual cytokines. The intracellular location of NF-κB restricts the types of inhibitors that can be used for this purpose. Gene delivery is one possibility when using ex vivo strategies, and the adenoviral construct used in the present study confirms the effectiveness of this approach. Alternatively, it is possible to use peptide antagonists of NF-κB that contain peptide transduction domains (41) or to use oligonucleotide decoys (42). A clinical trial on the injection of NF-κB decoys into rheumatoid joints is presently under way. Interestingly, the nutriceutical glucosamine that is commonly taken to treat osteoarthritis shows activity against NF-κB (43) and enhances matrix production by chondrocytes (44). Because of the importance of NF-κB in the host immune response, it is likely that any suppression of this transcription factor to aid cartilage repair will need to be highly localized.
The literature contains 2 articles reporting in vitro data consistent with our conclusions. In the first of these studies (36), IL-1Ra was shown to enhance the ability of insulin-like growth factor 1 to regenerate the matrix of equine articular cartilage. The second study, which was recently published by McNulty et al (45), shows that IL-1 and TNFα impair the repair of meniscus using a simulated autologous plug approach and that this can be reversed with IL-1Ra or neutralizing antibody against TNFα. In vivo studies of the effects of such agents on cartilage repair in suitable animal models seem warranted.
Our data may also be relevant to the observation that articular cartilage has very limited intrinsic ability to repair spontaneously. Increasing data suggest that chondroprogenitor cells exist at the surface of the cartilage (46) and possibly also within the deeper layers (47). The local presence of inhibitors such as IL-1β could prevent these cells from differentiating into chondrocytes following partial-thickness injury or during disease-induced erosion.
- Top of page
- MATERIALS AND METHODS
- AUTHOR CONTRIBUTIONS
Drs. Evans and Porter 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 design. Wehling, Palmer, Evans, Porter.
Acquisition of data. Wehling, Palmer, Pilapil, Liu, Wells, Porter.
Analysis and interpretation of data. Wehling, Palmer, Wells, Müller, Evans, Porter.
Manuscript preparation. Wehling, Evans, Porter.
Statistical analysis. Wehling, Porter.
- Top of page
- MATERIALS AND METHODS
- AUTHOR CONTRIBUTIONS
We are very grateful to Paul Robbins and Gary Gibson for supplying Ad.srIκB and antibody to type X collagen, respectively, and to Mark Vrahas and Mitchell Harris for recovering the intramedullary reamings used in the study.
- Top of page
- MATERIALS AND METHODS
- AUTHOR CONTRIBUTIONS
- 26Premature induction of hypertrophy during in vitro chondrogenesis of human mesenchymal stem cells correlates with calcification and vascular invasion after ectopic transplantation in SCID mice. Arthritis Rheum 2006; 54: 3254–66., , , , , , et al.
- 27Interleukin-1β impairment of transforming growth factor β1 signaling by down-regulation of transforming growth factor β receptor type II and up-regulation of Smad7 in human articular chondrocytes. Arthritis Rheum 2007; 56: 3020–32., , , , , , et al.
- 37Interleukin-1 receptor antagonist. In: ThompsonAW, LotzeMT, editors. The cytokine handbook. London: Academic Press; 2003. p. 669–708., .