SEARCH

SEARCH BY CITATION

Keywords:

  • osteoarthritis;
  • articular cartilage;
  • transforming growth factor alpha;
  • endothelin receptor A;
  • type II collagen, MMP-13

Abstract

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

Previously, our lab identified transforming growth factor-alpha (TGFα) as a novel factor involved in osteoarthritis (OA) in a surgical model of the disease. In the same study, we also observed increased transcript levels for endothelin receptor A (ET(A)R), a known contributor to cartilage pathology. To investigate the connection between TGFα and endothelin signaling in OA, primary articular chondrocytes and osteochondral explants were isolated from Sprague–Dawley rats and treated with vehicle or TGFα. Expression of ET(A)R protein and its encoding gene Ednra was assessed. Chondrocytes and cartilage explants were also treated with the endothelin receptor A/B antagonist Bosentan, in order to determine whether TGFα effects could be blocked. TGFα induced expression of ET(A)R protein and its encoding gene Ednra. In primary chondrocyte cultures, Bosentan did not block TGFα responses of the anabolic genes Sox9, Agc1, and Col2a1, but reduced the induction of Mmp13 and Ednra transcripts by TGFα. In osteochondral explants, the inhibitor partially blocked TGFα reduction of type II collagen, as well as induction of MMP-13 and type II collagen neoepitopes. TGFα induces ET(A)R expression in articular chondrocytes and receptor antagonism appears to block some TGFα-induced catabolic effects in a three-dimensional organ culture system. Thus, TGFα may be a therapeutic target upstream of ET(A)R in OA. © 2012 Orthopaedic Research Society. Published by Wiley Periodicals, Inc. J Orthop Res 30:1391–1397, 2012

Osteoarthritis (OA) is the most prevalent disabling condition in western society and places a serious strain on our healthcare system.1 Clinical symptoms such as joint pain and decreased joint function can severely affect day to day life for patients; however, no cure exists and current treatments only address symptoms without affecting the underlying causes.2 As a result, there is a significant need for further investigation and understanding of OA pathology.

While OA is a multivariable disease affecting the entire joint, one defining feature is the degeneration of articular cartilage.1 To better understand underlying pathological mechanisms, our lab has recently established a genome-wide profile of differentially expressed genes in degrading cartilage in a surgical rodent model of OA.3 Interestingly, transforming growth factor-alpha (TGFα) transcript levels were upregulated in degenerating cartilage when compared to controls.3 Thus TGFα was identified as a novel candidate growth factor involved in cartilage degradation. Further exploration revealed that TGFα treatment had profound effects on the chondrocyte phenotype. For example, treatment of monolayer chondrocyte cell cultures with TGFα resulted in suppressed expression of anabolic genes including aggrecan and type II collagen while matrix metalloproteinase 13 (MMP-13) gene expression increased.4 TGFα treatment also altered chondrocyte morphology and induced stress fiber formation in monolayer cultures.4 Articular cartilage organ culture studies revealed that cartilage treated with TGFα expressed more MMP-13 than controls and also displayed other OA-like characteristics such as cell clusters.4 A subset of human OA patients was also found to have higher than normal TGFα mRNA levels, thus establishing an important correlation between our rodent model and the human disease state.4

TGFα is a member of the epidermal growth factor (EGF) family and contains the characteristic EGF-like domain which binds to the EGF receptor.5 While EGF itself was downregulated in our experimental OA model, EGF receptor phosphorylation was increased, indicating receptor activation.4 Furthermore, a recent study showed that genetically modified mice with enhanced EGFR signaling spontaneously develop OA.6 EGFR signaling is also known to overlap and interact many other signaling systems, including the G-protein-coupled receptor (GPCR) endothelin receptor A (ET(A)R).7 In articular chondrocytes specifically, it has been observed that EGF induces expression of ET(A)R as well as of its ligand, endothelin-1 (ET-1).8, 9 Our interest in interactions between these two signaling systems emerged firstly from the fact that, like TGFα, ET(A)R gene expression was upregulated in our animal model of OA3 and secondly from the body of research implicating both systems in OA disease progression.

Previous studies have indicated that ET(A)R and its ligand play important roles in chondrocyte physiology and pathology. Studies using rodent articular chondrocytes showed that ET-1 production and ET(A)R density increased with age.9, 10 This trend is significant as age is the number one risk factor associated with OA.2 Prolonged exposure to ET-1 has also been shown to inhibit proteoglycan and collagen synthesis in rat articular chondrocytes.11 These data are complemented by reports which indicate that ET-1 increases expression of MMP-1 and MMP-13 in human osteoarthritic chondrocytes.12, 13 Human osteoarthritic chondrocytes also respond to ET-1 with increased expression of inducible nitric oxide synthase (iNOS), as well as elevated nitric oxide (NO) production.13 Elevated levels of iNOS are also characteristic in OA cartilage and studies have shown that inhibitors of iNOS can reduce the progression of experimental OA.14–16 Recently, Kaufman et al.,17 elegantly demonstrated that antagonism of ET(A)R prevents articular cartilage degradation in a surgical rat model of OA. In this study, we hypothesized that TGFα signals upstream of ET(A)R and that inhibiting ET(A)R might block some of TGFα's OA-like changes in articular cartilage.

MATERIALS AND METHODS

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

Surgical Model of Osteoarthritis

Tissues from our original surgical OA model were analyzed in this study. Weight controlled adult male Sprague–Dawley rats underwent anterior cruciate ligament transection (ACL-T) and partial medial meniscectomy (PM) of the right knee joint or sham surgery as described.18, 19 All animals were exercised on a rotating cylinder for 30 min three times a week in order to allow full flexion and extension of their knee joints.18

Histology and Immunohistochemistry

Organ culture explants were fixed overnight in 4% paraformaldehyde and then decalcified in an EDTA solution (0.4 M EDTA, 0.3 N NaOH, 1.5% glycerol, and pH 7.3). Decalcification was determined by a physical end-point test. Tissues were processed, embedded in paraffin wax, and sectioned (6 µm thick) by the Molecular Pathology Laboratory at Robarts Research Institute (London, ON, Canada). Immunohistochemistry (IHC) was performed by first dewaxing sections in xylene and then rehydrating through a series of graded ethanols ending in water. Antigen retrieval was performed in sodium citrate pH 6.0 for cellular proteins or Proteinase K for matrix-specific proteins. Tissues were then blocked in 5% serum and incubated with primary antibody overnight at 4°C. Incubation with horseradish peroxidise-conjugated secondary antibody was followed by colorimetric detection with the substrate diaminobenzidine (DAB). Primary antibodies used include anti-ET(A)R, -ET(B)R, and -iNOS (Abcam, Cambridge, MA), anti-type II collagen (R&D Systems, Minneapolis, MN), anti-MMP-13 (Cedarland, Hornby, ON, Canada), and anti-type II collagen neoepitope.20, 21

Cell Culture and Organ Culture Studies

All cell and organ culture reagents were purchased from Invitrogen (Burlington, Ont., Canada) and Sigma (Oakville, Ont., Canada) while sterile plates were purchased from BD Falcon (Mississauga, ON, Canada). All cells and explants were maintained in a 37°C humidified incubator at 5% CO2 and medium was changed daily.

Primary Articular Chondrocyte Cell Culture

Primary articular chondrocytes were isolated from the distal femoral condyles of neonatal Sprague–Dawley rats as previously described and plated at a density of 5.0 × 104 cells/cm2.22 A 10% fetal bovine serum culture medium was prepared from a 2:3 ratio of DMEM:F12 supplemented with 50 µg/ml ascorbic acid, 0.25% L-glutamine, and 0.25% penicillin/streptomycin. Chondrocytes were serum starved for 24 h prior to treatment and serum-free medium was used throughout the remainder of the time course. Chondrocyte culture medium was supplemented with TGFα (final concentration 10 ng/ml) or an equal volume of vehicle (0.1% bovine serum albumin in PBS) for up to 72 h. Culture medium was changed daily. For our endothelin inhibitor studies, cells were treated with vehicle, TGFα (10 ng/ml), the endothelin receptor inhibitor Bosentan (10 µM; provided by Actelion Pharmaceuticals Ltd, Allschwil, Switzerland) or a combination of both for 48 h. Bosentan is a dual endothelin receptor A/B antagonist which is clinically used for the treatment of pulmonary arterial hypertension.23

E15.5 Tibia Oragan Culture

Time-mated CD1 mice were purchased from Charles River Laboratory (St. Constant, Quebec, Canada) and embryonic E15.5 tibia were dissected as previously described.24, 25 Tibiae were cultured in media containing α-MEM supplemented with ascorbic acid, beta-glycerophosphate, bovine serum albumin, PENSTREP®, and L-glutamine. All tissues were maintained in a 37°C humidified incubator at 5% CO2 and treated with a range of concentrations of recombinant human TGFα (0–1,000 ng/ml) for 6 days. Culture medium was changed every other day.

RNA Isolation and Real-Time PCR

RNA was isolated from rat primary chondrocytes using the Qiagen RNeasy Mini Kit and the manufacturer's animal cell protocol (Qiagen, Mississauga, ON, Canada). Mouse cartilage was dissected from the ends of tibia organ cultures with the aid of a Zeiss Stemi DV4 Stereo microscope as previously described,26 samples were placed in QIAzol solution (Qiagen), and RNA was isolated following the manufacturer's protocol. Real-time PCR was performed for the genes Ednra (endothelin receptor A), Il1b (interleukin-1 beta), Cxcr4 (chemokine (C-X-C motif) receptor 4), Tgfa (TGFα), Agc1 (aggrecan), Col2a1 (type II collagen), Sox9 (Sox9), and Mmp13 (MMP-13). Analysis was performed using the Applied Biosystems 7900HT Real-Time PCR system, the TaqMan® One-step Mastermix Kit, and commercially available probes (Applied Biosystems, Foster City, CA). All samples were normalized to the housekeeping gene Gapdh and day 0 or vehicle-treated controls using the delta-delta cycle threshold (ΔΔCt) method.

Articular Cartilage Organ Culture

Osteochondral explants were isolated from the distal femoral condyles and proximal tibial plateaus of 4- to 5-month-old adult male Sprague–Dawley rats. Explants were then placed in 12-well tissue culture plates and submerged in a 2% bovine serum albumin culture medium made from alpha-minimal essential medium supplemented with 50 µg/ml ascorbic acid, 0.25% L-glutamine, and 0.25% penicillin/streptomycin for 24 h prior to treatment. The same treatment groups described for primary chondrocyte experiments were repeated for organ culture studies. Explants were treated daily for up to 5 days before tissues were prepared for histology.

Rhodamine Phalloidin Staining

Primary chondrocyte cell cultures were isolated as described above and plated on micro cover glasses (VWR, Mississauga, ON, Canada) in 24-well plates at a density of 1.0 × 104 cells/cm2. Cells were treated with 1 of 4 treatments: vehicle, TGFα (10 ng/ml), Bosentan (10 µM), or both for 48 h. Cells were then fixed, permeabilized, and stained with rhodamine phalloidin (Cytoskeleton Inc., Denver, CO) to visualize F-actin and overall cell morphology.

Statistics

All statistical analysis was performed using GraphPad Prism software Version 4.0. Real-time data were analyzed using either a one-way analysis of variance (ANOVA) with a Tukey's post-test or a two-way ANOVA with a Bonferroni's post-test. All graphs show mean values + standard error of the mean (SEM). A minimum of three independent groups was used in all experiments.

RESULTS

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

ET(A)R Expression Increases in a Surgical Model of Osteoarthritis

Our previous studies have shown that ET(A)R gene expression is upregulated in OA-operated animals compared to controls, as demonstrated by microarrays and validated by real-time PCR.3 We next wanted to examine ET(A)R protein expression in histological samples from OA- and sham-operated animals. Immunohistochemical analysis revealed more ET(A)R signal throughout articular surfaces in OA animals compared to controls (Fig. 1).

thumbnail image

Figure 1. Endothelin receptor A (ET(A)R) protein expression is upregulated in degenerating cartilage from an experimental rodent model of osteoarthritis (OA). Histological sections were prepared from knee joint articular cartilage of both sham- and OA-operated rats 4 weeks post-surgery. IHC was performed with an anti-ET(A)R primary antibody followed by secondary antibody and visualization with the substrate DAB (brown precipitate). Nuclei were counterstained with hematoxylin (blue). Representative images show more ET(A)R staining in OA cartilage compared to sham cartilage. This experiment was performed on three separate trials with similar results and one representative trial is shown.

Download figure to PowerPoint

TGFα Treatment Induces Ednra Gene Expression

To examine whether TGFα can induce Ednra expression, primary rat chondrocytes in monolayer culture were treated with vehicle or TGFα (10 ng/ml) for 1–3 days, followed by isolation of RNA and real-time PCR analyses. Ednra mRNA expression increased after 2 and 3 days of TGFα treatment (Fig. 2A). Day 2 showed about a ninefold increase while day 3 showed about a sixfold increase (Fig. 2A). In contrast, transcript levels of several other genes known to be involved in OA or identified in our microarray, including Tgfa itself, Il1b, and Cxcr4, did not change significantly after treatment (Fig. 2B–D). We next wanted to determine whether TGFα-induction of Ednra was species specific. We therefore employed the mouse E15.5 tibia organ culture system. RNA isolated from the cartilage ends of TGFα-treated tibiae also showed an induction of Ednra transcripts, although a higher dose of TGFα was required to reach significance (Fig. 2E).

thumbnail image

Figure 2. TGFα treatment increases expression of the endothelin receptor A gene (Ednra). Neonatal rat primary chondrocytes were treated with or without TGFα (10 ng/ml) for up to 3 days (A–D) while E15.5 CD1 mouse tibiae were treated with a range of TGFα concentrations (0–1,000 ng/ml) for 6 days (E). Real-time PCR was used to determine gene expression relative to Gapdh and day 0 controls. Relative expression means + SEM are shown. (A) Ednra mRNA expression increased in rat primary cultures after days 2 and 3 of treatment while expression of the genes Il1b, Cxcr4, and Tgfa did not change significantly after treatment (B–D). Treated mouse tibiae also showed increased levels of Ednra transcripts (n = 3–4, *p < 0.05).

Download figure to PowerPoint

TGFα Treatment Induces ET(A)R Protein Expression

To study TGFα's effects on ET(A)R expression in a more physiological three-dimensional context, articular organ culture explants were isolated from male Sprague–Dawley rats and treated with or without TGFα. Tissues were immunostained for ET(A)R and ET(B)R and representative images from days 1 to 5 are shown (Fig. 3A). Figure 3 reveals an increase in ET(A)R staining in TGFα-treated explants compared to controls. No evident change in ET(B)R expression was observed between treatments and controls (Fig. 3A).

thumbnail image

Figure 3. Endothelin receptor A (ET(A)R) expression increases in TGFα-treated articular cartilage explants. Tibial and femoral articular cartilage explants were isolated from Sprague–Dawley rats and cultured with or without TGFα (10 ng/ml) for up to 5 days. Histological sections were then immunostained for ET(A)R, ET(B)R, iNOS, and NF-κB (brown precipitate), and counterstained with hematoxylin (blue). (A) Representative images show that ET(A)R expression increased after 5 days of treatment while ET(B)R showed no change with TGFα treatment. (B) TGFα-treated tissues show more staining for both iNOS and the transcription factor NF-κB. This experiment was performed on three separate trials with similar results and one representative trial is shown.

Download figure to PowerPoint

iNOS and NF-κB Expression Increase with TGFα Treatment

Endothelin signaling is known to activate iNOS expression.13 Tissue sections from articular cartilage organ cultures treated with vehicle or TGFa were stained with antibodies against iNOS and the p65 subunit of NF-κB (a transcription factor regulating iNOS expression27). The TGFα-treated tissues appear to have more positively stained cells for iNOS than controls after 5 days of treatment (Fig. 3B). In particular, staining extended deeper into the cartilage in TGFα-treated samples. We saw a similar trend for NF-κB staining (Fig. 3B).

Bosentan Treatment Does Not Block TGFα-Induced Gene Changes

To address whether inhibition of endothelin receptors can block effects of TGFα on gene expression, primary chondrocytes were treated with vehicle, TGFα, the endothelin receptor A/B inhibitor Bosentan, or both. Bosentan treatment did not block TGFα-suppression of Agc1, Col2a1, or Sox9 mRNA levels (Fig. 4A–C); however, Bosentan slightly decreased the induction of Ednra (Fig. 4D) and Mmp13 (Fig. 4E). In the absence of TGFα, Bosentan did not induce any significant changes in chondrocyte gene expression.

thumbnail image

Figure 4. Bosentan treatment does not block TGFα effects on gene expression. Primary articular chondrocytes were cultured with vehicle, TGFα (10 ng/ml), the endothelin receptor A/B inhibitor Bosentan (10 µM), or a combination of both for 48 h. RNA was isolated and real-time PCR was performed for the genes Acg1, Col2a1, Sox9, Ednra, and Mmp13. For all genes, measurements were relative to Gapdh and vehicle-treated controls. Bosentan and had a slight effect of induction of Ednra (D) and Mmp13 (E). Bosentan did not have a significant effect on any other genes (A–C). Relative expression means + SEM are shown. Means with different letters are significantly different (n = 4, p < 0.05).

Download figure to PowerPoint

Bosentan Treatment Does Not Block TGFα Effects on Cell Morphology

TGFα has been shown to alter cell shape and induce stress fiber formation in monolayer articular chondrocyte cultures,4 in agreement with loss of a chondrocyte phenotype.28 As in our earlier studies, TGFα induced stress fiber formation and a more elongated cell shape in primary chondrocytes (Fig. 5). By itself, Bosentan did not appear to have any effects on cell morphology or actin organization (Fig. 5). Moreover, when Bosentan treatment was given in combination with TGFα, no major change was observed when compared to TGFα treatment alone (Fig. 5).

thumbnail image

Figure 5. Bosentan treatment does not alter TGFα effects on actin organization. Primary articular chondrocytes were cultured and treated with vehicle, TGFα (10 ng/ml), the endothelin receptor inhibitor A/B Bosentan (10 µM), or a combination of both for 48 h. Cells were then fixed and stained with rhodamine phalloidin (red) in order to visualize F-actin. Nuclei were stained with DAPI (blue). Representative images of each treatment are shown (n = 3). Bosentan does not block TGFα effects on cell morphology or induction of stress fibers.

Download figure to PowerPoint

Bosentan Treatment Blocks TGFα Effects on Cartilage ECM Turnover

After finding limited effects of endothelin receptor A/B inhibition in our monolayer articular chondrocyte model, we decided to perform inhibitor studies in the more physiologically relevant organ culture model. Explants were cultured with vehicle, TGFα, Bosentan, or both. Tissue sections were then immunostained for type II collagen, MMP-13, and type II collagen neoepitopes that become detectable when type II collagen is cleaved by MMP-13. TGFα-treated tissues showed a decrease in overall type II collagen staining in the extracellular matrix (ECM; Fig. 6A) and an increase in MMP-13 staining (Fig. 6B). Co-treatment with Bosentan appeared to reverse these catabolic effects (Fig. 6A and B). Furthermore, TGFα-treatment appeared to generate more type II collagen neoepitopes in the ECM when compared to controls, but the combination with Bosentan appeared to block neoepitope induction (Fig. 6C).

thumbnail image

Figure 6. Bosentan treatment blocks TGFα reduction of type II collagen, induction of MMP-13, and induction of type II collagen neoepitopes. Articular cartilage explants were treated with vehicle, TGFα (10 ng/ml), Bosentan (10 µM), or both for 5 days. Tissues were then immunostained for type II collagen (A), MMP-13 (B), and type II collagen neoepitopes (C, brown precipitate). TGFα-treated tissues showed a decrease in type II collagen staining, an increase in MMP-13 staining, and an increase in type II collagen neoepitopes. Co-treatment with TGFα and Bosentan appeared to bring expression levels back to control levels. This experiment was performed on four separate trials with similar results and one representative trial is shown.

Download figure to PowerPoint

DISCUSSION

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

We have previously identified TGFα as a growth factor capable of inducing OA-like phenotypic changes in articular chondrocytes. In this study, we demonstrate that TGFα induces ET(A)R expression in articular chondrocytes at both the gene and protein level. Furthermore, we demonstrate that TGFα-induction of the receptor occurs in both rat and mouse models as well as in varying ages of tissue (embryonic, newborn, and 4- to 5-month old). This is consistent with previous findings, in which EGF was able to induce ET(A)R expression in both young and old rats.9 In addition to EGF and TGFα, a number of other growth factors and cytokines are known to modulate ET(A)R receptor density in chondrocytes.9 Furthermore, the literature suggests that the ET(A)R receptor plays a role in both aging and degenerative joint disease.9, 10, 12, 13 Thus, TGFα induction of ET(A)R might be one way through which TGFα mediates its deleterious effects.

To test this hypothesis, we used the dual endothelin receptor A/B inhibitor Bosentan in combination with TGFα treatment. We found that Bosentan treatment was not sufficient to block TGFα-effects on anabolic gene expression in monolayer culture. Bosentan did, however, have a minor effect on Mmp13 and Ednra levels. Bosentan was also unable to reverse TGFα-induced changes to the chondrocyte actin cytoskeleton and cell shape. This leads us to believe that other downstream pathways might be responsible for anabolic gene expression changes and cytoskeletal rearrangement in TGFα-treated chondrocytes. Previous studies done in our laboratory show that TGFα activates many intracellular signaling pathways.21 For example, RhoA/ROCK mediates TGFα-induced morphologic changes in chondrocytes, in agreement with our previous studies demonstrating important roles of this pathway in chondrogenesis.29, 30 The MEK/ERK pathway mediates the downregulation of anabolic gene expression by TGFα,21 again in line with the known roles of this pathway in chondrocytes (reviewed in Refs.31, 32). Both Rho/ROCK and MEK/ERK pathways also regulate type II collagen cleavage and aggrecan breakdown in articular cartilage.21

Bosentan had a stronger effect in the three-dimensional organ culture system where endothelin receptor A/B inhibition was able to suppress TGFα induction of the catabolic factor MMP-13 and subsequent type II collagen break down. There are some possible explanations for the differences observed in our two experimental systems. Firstly, age may play a role: primary cells used in our studies were isolated from neonatal rats while organ culture explants were isolated from adult rats. The baseline density of ET(A)R receptors is known to be dependent on the age of articular chondrocytes and more ET-1 is known to be produced in older chondrocytes.9, 10 In addition, endothelin receptor signaling might be more important in the authentic three-dimensional context of cartilage than in monolayer culture. Recently, Kaufman et al.17 showed that ET(A)R inhibition prevented OA progression in a surgical model of the disease, in agreement with our data shown here.

In our studies, we observed no effect on any parameters with Bosentan treatment alone. While articular chondrocytes produce ET-1, it is possible that its levels are not high enough in the in vitro environment to sufficiently stimulate ET(A)R receptors. ET-1 mRNA levels however, did not change in the TGFα treatment groups (data not shown), suggesting that it is the receptor density itself more so than the concentration of ligand that is responsible for enhanced signaling.

In summary, our data show that ET(A)R expression in articular cartilage is increased in response to both surgical induction of OA and TGFα treatment. We also demonstrate that this pathway is only partially responsible for TGFα-induction of an OA-like phenotype, but that ET(A)R inhibition suppresses catabolic activities in articular cartilage. Further studies should be done to examine the potential benefit of upstream targets such at the EGF receptor itself as well as combinations of downstream targets in OA therapy.

Acknowledgements

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

We would like to thank Dr. Jason Rockel (Toronto, ON) for teaching and assisting with the primary chondrocyte isolation protocol. We would also like to thank Vasek Pitelka (London, ON) for his assistance with the osteochondral explant isolation. We are grateful to Actelion Pharmaceuticals Ltd. for the donation of Bosentan as well as to Dr. Robin Poole (Montreal, QC) for donating the type II collagen cleavage antibodies. This work was funded by an operating grant from the Canadian Institutes of Health Research (CIHR, MOP86574). S.E.U. was supported by a CIHR Doctoral Award, a Graduate Scholarship from the Canadian Arthritis Network and an Ontario Graduate Scholarship in Science and Technology. F.B. is the recipient of a Canada Research Chair.

REFERENCES

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES