Article Text

Download PDFPDF

Activated CD4+CD25+ regulatory T cells inhibit osteoclastogenesis and collagen-induced arthritis
  1. H Kelchtermans1,
  2. L Geboes1,
  3. T Mitera1,
  4. D Huskens2,
  5. G Leclercq3,
  6. P Matthys1
  1. 1
    Laboratory of Immunobiology, Rega Institute for Medical Research, University of Leuven, Leuven, Belgium
  2. 2
    Laboratory of Virology and Chemotherapy, Rega Institute for Medical Research, University of Leuven, Leuven, Belgium
  3. 3
    Department of Clinical Chemistry, Microbiology and Immunology, Ghent University Hospital, Ghent, Belgium
  1. H Kelchtermans, Laboratory of Immunobiology, Rega Institute, Minderbroedersstraat 10, B-3000 Leuven, Belgium; Hilde.Kelchtermans{at}rega.kuleuven.be

Abstract

Objectives: Patients with rheumatoid arthritis (RA) have defective CD4+CD25+ regulatory T (Treg) cells and increased osteoclastogenesis. A similar situation has been described in collagen-induced arthritis (CIA). In this study, it was investigated whether a single transfer of polyclonally activated Treg cells inhibits CIA and osteoclastogenesis.

Methods: Purified Treg cells were expanded in vitro with anti-CD3 and anti-CD28 antibody-coated beads and injected into DBA/1 mice. Mice were immunised with collagen type II (CII) in complete Freund adjuvant (CFA) and scores of arthritis were recorded. In vitro osteoclastogenesis assays were performed on splenocytes by stimulation with macrophage colony-stimulating factor (M-CSF) and receptor activator of nuclear factor (NF)κB ligand (RANKL). Levels of anti-CII antibody and cytokines were determined in the supernatant using ELISA and Bio-Plex protein array system.

Results: It was found that 106 activated Treg cells significantly counteracted the development of CIA, which was accompanied by decreased serum levels of TNFα and IL6, but not by inhibition of autoimmune antibody responses. The differentiation of osteoclasts in splenocyte cultures was significantly reduced in the presence of prestimulated Treg cells. Expression of cytokines that are described to inhibit osteoclastogenesis, including granulocyte macrophage colony-stimulating factor (GM-CSF), interferon (IFN)γ, interleukin (IL)5 and IL10, were dramatically increased upon addition of Treg cells. Furthermore, splenocytes from mice that had been treated with Treg cells displayed an impaired capacity to develop into mature osteoclasts, suggesting that Treg cells abrogated osteoclastogenesis in vivo.

Conclusions: Activated CD4+CD25+ Treg cells improve clinical symptoms of CIA, regulate cytokine production and inhibit osteoclastogenesis in vitro and in vivo.

Statistics from Altmetric.com

CD4+CD25+ regulatory T (Treg) cells control immune responses to self and foreign antigens.1 2 Evidence is accumulating that defects in Treg cell function are important in rheumatoid arthritis (RA).3 In collagen-induced arthritis (CIA), an animal model of autoimmune arthritis reminiscent in several aspects to RA,4 we and others described an accelerated and more severe form of arthritis in mice that were depleted of Treg cells.5 6 The increased disease was not accompanied by increased anti-collagen type II (CII) antibody production.5 Morgan et al found that a single transfer of Treg cells slowed disease progression, that was not associated with losses in CII-specific immune responses.7 These data demonstrate that Treg cells modulate CIA by a mechanism independent of humoral autoimmune responses. One potential mechanism by which Treg cells protect against autoimmune arthritis may be regulation of osteoclasts.

Osteoclasts are formed by fusion of monocytes/macrophages.8 9 They are abundantly present in the inflamed synovium and are responsible for bone destruction in RA and CIA.10 11 Osteoclast precursors (OCP) are recruited from the bone marrow and spleen.12 We and others have described an increased number of OCP in arthritic mice in animal models of RA such as CIA, adjuvant-induced arthritis and tumour necrosis factor (TNF)α transgenic mice.1315 Their differentiation into active osteoclasts is regulated by several cytokines, most importantly receptor activator of nuclear factor (NF)κB ligand (RANKL) and macrophage colony-stimulating factor (M-CSF). Moreover, interleukin (IL)1β, IL6 and IL17 augment RANKL expression stimulating osteoclastogenesis, while IL13 and IL4 inhibit the expression of RANKL.16 Osteoprotegerin (OPG) is a soluble decoy receptor for RANKL. As osteoclastogenesis is known to be regulated by T cells,17 the purpose of our study was to verify whether CD4+CD25+ Treg cells may improve the severity of arthritis by inhibiting the osteoclast differentiation or activity.

MATERIALS AND METHODS

Mice, induction and histological evaluation of CIA

DBA/1 mice were bred at the Animal Centre of the University of Leuven, Leuven, Belgium. Experiments were performed in 6-week-old to 10-week-old mice. Chicken CII (Sigma-Aldrich, St Louis, Missouri, USA) was dissolved at 2 mg/ml in phosphate-buffered saline (PBS) containing 0.1 M acetic acid by stirring overnight at 6°C and emulsified in an equal volume of complete Freund adjuvant (CFA) (Difco Laboratories, Detroit, Michigan, USA) with added heat-killed Mycobacterium butyricum (1.5 mg/ml). Mice were sensitised with a single intradermal injection at the base of the tail with 100 μl of the emulsion on day 0. Clinical and histological disease severity of arthritis were recorded following a scoring system as described.5 18

Cell purification

Splenocytes, CD4+CD25+ cells and CD4+CD25 cells were isolated as described previously.5 CD4+CD25+ cells were activated and expanded with Dynabeads Mouse CD3/CD28 T cell expander (Invitrogen Life Technologies, Carlsbad, California, USA) and 10 ng/ml recombinant IL2 (R&D Systems, Minneapolis, Minnesota, USA). Cells were grown in RPMI 1640 (Bio Whittaker Europe, Verviers, Belgium), supplemented with 10% heat-inactivated fetal calf serum (FCS; Gibco, Paisley, UK), penicillin (100 U/ml) (Continental Pharma, Brussels, Belgium), streptomycin (100 μg/ml; Continental Pharma), 2 mM 1-glutamine, 10 mM 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid (Hepes) (Gibco), 0.1 mM non-essential amino acids (ICN, Asse Relegem, Belgium), 1 mM sodium pyruvate (Gibco) and 50 μM 2-mercaptoethanol (Fluka, Lausanne, Switzerland).

Measurement of serum anti-CII antibodies

Blood samples were taken from the orbital sinus and were allowed to clot at room temperature for about 1 h and at 4°C overnight. Sera were tested for anti-CII antibodies by ELISA as described.5

Quantitative reverse transcriptase (RT)-PCR

RNA was isolated from cells using the Micro-to-Midi Total RNA Purification System (Invitrogen Life Technologies). cDNA was obtained by reverse transcription using Superscript II Reverse Transcriptase and random primers (Invitrogen). For real–time PCR we used a TaqMan Assays-on-Demand Gene expression Product from Applied Biosystems (Foster City, California, USA). Expression levels of Foxp3 were normalised for 18S RNA expression. Analysis was performed in an ABI Prism 7000 apparatus as described previously.18

Osteoclast formation and activity

Splenocytes were resuspended in α-minimal essential medium (α-MEM) containing 10% FCS (Gibco). Cells (2.5×105) in a total volume of 400 μl were seeded in chamber slides (Lab-Tek Brand Products, Nalge Nunc International, Naperville, Illinois, USA) and incubated for 6 days with 20 ng/ml M-CSF (R&D Systems) and 100 ng/ml RANKL (R&D Systems). On day 3, cultures were provided with fresh RANKL and M-CSF, and 5×104 Treg cells, 5×104 effector T (Teff) cells or both were added in the presence of IL2. On day 6 cells were stained for the presence of tartrate-resistant acid phosphatase (TRAP) as described.19 TRAP+ multinucleated cells (three or more nuclei) were defined as osteoclasts.

For the pit forming assay, splenocytes were cultured for 6 days on quartz slides coated with a calcium phosphate film as described.20 On day 3, cultures were provided with fresh RANKL and M-CSF, and 105 Treg cells, 105 Teff cells or both were added in the presence of IL2. Cells were removed and resorption of the film was assessed by light microscopy. In each well, the percentage of resorption pits was assessed in five different regions.

Bio-Plex protein array system and ELISA

Expression of cytokines (ie, IL1β, IL2, IL4, IL5, IL6, IL10, granulocyte macrophage colony-stimulating factor (GM-CSF), interferon (IFN)γ and TNFα) was determined by the Bio-Plex 200 system, Bio-Plex Mouse Cytokine 8-plex assay and Bio-Plex Mouse IL6 Assay (Bio-Rad, Hercules, California, USA). OPG and IL17 levels were measured by ELISA (R&D Systems).

Statistical analysis

Data are expressed as the mean (standard error of the mean (SEM)). Differences were analysed by the Mann–Whitney U test. A p value of ⩽0.05 was considered significant.

RESULTS

CD4+CD25+ Treg cells inhibit the development of CIA

CD4+CD25+ cells were isolated from lymph nodes of naive mice and expanded as described in Material and Methods. At 2 weeks after expansion, Treg cells expressed high levels of Foxp3 mRNA, a molecular marker for CD4+CD25+ Treg cells21 (fig 1A). At 1 day before the induction of CIA, mice received 106 prestimulated Treg cells or PBS. In PBS-treated mice symptoms of arthritis appeared from day 17 and reached a cumulative incidence of 100%. In contrast, mice treated with Treg cells developed a significantly less severe form of arthritis with a lower incidence and a delayed disease onset (fig 1B,C). When analysis was restricted to arthritic mice, the severity of arthritis was still lower in mice treated with Treg cells than in control mice (fig 1C, inset). Mice were killed on day 59 for histological examination of the joints. As shown in fig 1D, reduced severity of arthritis in mice injected with Treg cells was associated with inhibition of infiltration of immunocompetent cells, hyperplasia and pannus formation (measured as the fraction of synovial inflammatory tissue which invaded bone tissue). Interestingly, whereas osteoclasts were abundantly present in joint sections of control mice, they were totally absent in Treg cell-treated mice (data not shown). To verify that the improved clinical outcome upon treatment with activated Treg cells was specific for Treg cells, an additional experiment was performed in which mice were treated with prestimulated CD4+CD25 Teff cells, Treg cells or PBS. As was the case in our previous experiments, Treg cells inhibited CIA and delayed the onset of arthritis, whereas Teff cells rather increased the incidence and score of arthritis (fig 1E,F).

Figure 1

Inhibition of collagen-induced arthritis (CIA) in DBA/1 mice by transfer of CD4+CD25+ regulatory T (Treg) cells. CD4+CD25+ Treg cells were isolated and expanded as described in Materials and methods. Mice were immunised with collagen type II (CII) in complete Freund adjuvant (CFA) on day 0 and injected intravenously with either 106 expanded Treg cells, effector T (Teff) Teff cells or phosphate-buffered saline (PBS) (control) on day −1. A. Foxp3 mRNA levels of expanded Treg cells were compared with expanded CD4+CD25 Teff cells by quantitative PCR. Bars represent the relative quantity of Foxp3 normalised to the quantity of 18S. B,C. The cumulative incidence and the mean arthritic score of control and Treg cell-treated mice are shown. The inset in C shows the mean arthritic score of arthritic mice only. Error bars indicate standard error of the mean (SEM). D. On day 59 after immunisation, histological analysis of the joints was performed. Histograms represent averages (SEM). Results are representative for two independent experiments. E,F. The cumulative incidence of arthritis and the mean arthritic score of control mice and mice treated with Treg cells or Teff cells are shown. Error bars indicate SEM. *p<0.05, for comparison with mice treated with Treg cells (C,F) or control mice (D).

Improved clinical outcome by Treg cells is associated with significant reduced levels of TNFα and IL6

On day 42 (experiment 1) or day 28 and day 38 (experiment 2), sera of mice were tested for the presence of cytokines and total anti-CII IgG antibodies. Levels of GM-CSF, IFNγ, IL1, IL2, IL4, IL5, IL10 and IL17 were below the detection limit of the system. IL6 and TNFα levels on day 42 and 38 were significantly reduced in the sera of Treg cell-treated mice as compared to sera of control mice (data of day 42 are shown in fig 2A, left panel). Addition of Teff cells did not influence production of IL6 and TNFα (data not shown). On day 28, levels of IL6 and TNFα were reduced in the sera of Treg cell-treated mice as compared to sera of control or Teff cell-treated mice, although values did not reach statistical significance (fig 2A, right panel). In both experiments, total anti-CII IgG antibody titres were found not to be significantly lower in the sera of mice treated with Treg cells (fig 2B). Remarkably, on day 28 anti-CII antibody titres were even significantly higher in the sera of Treg cell-treated mice as compared to control-treated mice.

Figure 2

Inhibition of collagen-induced arthritis (CIA) by regulatory T cell (Treg) transfer is associated with reduced tumour necrosis factor (TNF)α and interleukin (IL)6 levels, but not with reduced humoral responses. Mice induced for CIA were injected with expanded Treg cells or effector T (Teff) cells as described in the legend to fig 1. In two experiments (experiments 1 and 2), mice were bled at different time points (day 28, 38 and 42) and sera of individual mice were analysed for IL6 and TNFα (A) and total anti-collagen type II (CII) IgG (B). Histograms represent averages (standard error of the mean (SEM)). *p<0.05, for comparison with Treg cell-treated mice.

Treg cells inhibit osteoclastogenesis

We next tested the hypothesis that Treg cells inhibit the formation and/or activity of osteoclasts. Therefore, splenocytes from wild type mice were cultured in the presence of M-CSF and RANKL. In cultures with Treg cells or Treg cells plus Teff cells, significantly lower numbers of osteoclasts were observed (fig 3A). Addition of Teff cells did not significantly influence the number of osteoclasts. Representative pictures of the TRAP-stained cultures are shown in fig 3C.

Figure 3

CD4+CD25+ regulatory T (Treg) cells inhibit osteoclastogenesis. Treg cells and effector T (Teff) cells were purified and expanded as described in Materials and methods. Splenocytes were isolated 35 days after induction of collagen-induced arthritis (CIA) and cultured for 6 days in chamber slides (A) or on a quartz substrate coated with a calcium phosphate film (B), in the presence of macrophage colony-stimulating factor (M-CSF) (20 ng/ml) and receptor activator for nuclear factor (NF)κB ligand (RANKL) (100 ng/ml). At day 3 Treg cells, Teff cells or a combination of both were added in the presence of IL2. A. After incubation, cells were stained for the presence of tartrate-resistant acid phosphatase (TRAP) and counted. In each experiment, duplicate cultures were used and bars represent the mean of three independent experiments (standard error of the mean (SEM)). *p<0.05, for comparison with medium, **p<0.05, for comparison with medium and cultures with Teff cells. B. Cells were removed and resorption of the film was assessed by light microscopy. Bars represent the mean of two cultures (SEM). C. Representative pictures of TRAP-stained cultures are shown. D. Splenocytes of mice treated with Treg cells, Teff cells or phosphate-buffered saline (PBS) were obtained at day 39, stimulated with RANKL and M-CSF for 6 days and TRAP staining was performed. Bars represent the mean of five separate cultures. *p<0.05, for comparison with control and Teff-treated mice, **p<0.05, for comparison with control and Treg-treated mice.

The effect of Treg cells on osteoclast activity was verified by counting the percentage of resorbed pits on quartz substrates coated with a calcium phosphate film. Whereas addition of Teff cells did not influence osteoclast activity (fig 3B), there was a clear tendency for a decreased osteoclast activity upon addition of Treg cells, although values did not reach statistical significance (p = 0.06 and p = 0.09 for the conditions with Treg cells and Treg cells with Teff cells, respectively, as compared to control cultures). In an additional experiment, ex vivo osteoclastogenesis was analysed on splenocytes of mice treated with PBS, Treg or Teff cells. Interestingly, splenocytes from Treg cell-treated mice developed significantly lower numbers of osteoclasts as compared to control cultures, suggesting that treatment with Treg cells resulted in reduced numbers of OCP. In cultures of Teff cell-treated mice, significantly higher numbers of osteoclasts were observed as compared to control cultures (fig 3D).

Addition of Treg cells increases expression of cytokines inhibiting osteoclastogenesis

To find an explanation for the decreased osteoclast differentiation upon addition of Treg cells, supernatants of osteoclast cultures were analysed for cytokines. First, the concentration of OPG was assessed and was found to be increased in cultures with activated Treg/Teff cells (data not shown). However, since concentrations of OPG were below 0.8 ng/ml and an excess amount of RANKL (100 ng/ml) was added, levels of OPG may have no effect in our system. Subsequently, cytokines known for their stimulatory (IL1, IL6, IL17, TNFα) or inhibitory (GM-CSF, IFNγ, IL4, IL5, IL10) effects on osteoclastogenesis were determined. Addition of Treg cells resulted in an increased expression of cytokines inhibiting osteoclastogenesis, especially GM-CSF, IFNγ, IL5 and IL10 (fig 4). Although addition of activated Treg or Teff cells augmented expression of stimulating cytokines, such as IL1, IL6 and TNFα, these levels were much lower than those of inhibiting cytokines. Levels of IL17 expression were below the detection limit.

Figure 4

Addition of regulatory T (Treg) cells is accompanied by increased expression of cytokines inhibiting osteoclastogenesis. Osteoclast cultures were prepared as described in the legend of fig 3. After incubation, supernatants were collected and the expression of indicated cytokines was determined using the Bio-Plex protein array system. Bars represent the mean expression level of two cultures (standard error of the mean (SEM)) and results are representative of three independent experiments. ND, not detectable (ie, below the detection limit varying from 0.5 to 5 pg/ml for all cytokines).

DISCUSSION

Successful treatment of autoimmune disease with CD4+CD25+ Treg cells has been reported in various animal models of autoimmune diseases.2226 In CIA, we have previously obtained indirect evidence for a role of Treg cells in the pathogenesis of CIA.5 Here, we investigated the effect of a single transfer of Treg cells on the development of CIA. Treg cells were expanded by the method described by Tang et al27 and the retention of regulatory activity after expansion was confirmed by measuring Foxp3 mRNA expression. A single transfer of 106 Treg cells significantly improved the clinical severity of arthritis, confirming the findings of Morgan et al.7 In both studies humoral anti-CII responses were not significantly reduced, suggesting other mechanisms of action of the Treg cells. We found that the adoptive transfer of Treg cells was accompanied with reduced IL6 and TNFα levels. Since Treg cells have been described to inhibit the TNFα release by Teff cells and monocytes,28 and since IL6 and TNFα are well described proinflammatory cytokines in RA and CIA,29 these data may provide an explanation for the protective effect of Treg cells in CIA.

Aside from inhibition of inflammatory cytokines and inflammation (as evident from the reduced cellular influx into the synovium), Treg cells may inhibit osteoclastogenesis. Osteoclast formation is well known to be regulated by T cells.17 Furthermore, in our experiments treatment with Treg cells resulted in significant inhibition of pannus formation (ie, the synovial tissue that covers the articular cartilages that progressively destroys the underlying cartilage and bone tissues). Multinucleated giant cells or osteoclasts were abundantly present in joint sections of control mice, but not in Treg cell-treated mice, even in diseased animals. The reduced pannus formation and osteoclast numbers in Treg cell-treated mice may be explained by the reduced cellular influx in the synovium or by a direct inhibition of Treg cells on osteoclast differentiation. To distinguish between these possibilities, we analysed the effects of Treg cells vs Teff cells on in vitro osteoclastogenesis. Addition of activated Teff cells did not influence osteoclast differentiation, nor activity, although these cells have been shown to express RANKL, thereby endowing them with the capacity to induce osteoclast differentiation.30 Since we have added an excess amount of RANKL to the cultures, RANKL produced by activated T cells probably had little or no effect. By contrast, Takayanagi et al have demonstrated that activated T cells negatively affect osteoclastogenesis through IFNγ production.31 However, we have added T cells 3 days after the induction of the osteoclast culture with RANKL. Suppressor of cytokine signalling (SOCS)-1 was initially identified as a negative-feedback molecule that inhibits Janus kinase-signal transducer and activator of transcription activation initiated by various stimuli, including IFNγ. In addition, SOCS-1 is induced during osteoclastogenesis by RANKL stimulation, indicating that precursor cells are made resistant to IFNγ-mediated inhibition if they encounter RANKL first.32 This may provide an explanation for the fact that addition of activated Teff cells did not result in a lower number of osteoclasts. Addition of prestimulated Treg cells, with or without of Teff cells, resulted in a significantly lower number of osteoclasts. Osteoclast activity was also found to be lower, although not significantly. Importantly, these in vitro data were confirmed ex vivo, as splenocytes from Treg cell-treated mice developed significantly lower number of osteoclasts as compared to control cultures. These data suggest that treatment with Treg cells results in reduced numbers of osteoclast precursor cells.

To find an explanation for the lower number of osteoclasts upon addition of Treg cells in vitro, supernatants of the cultures were analysed for several cytokines. The levels of the decoy receptor OPG were increased in all conditions with activated T cells, but the concentrations were far below the excess amount of RANKL, suggesting that OPG may not be important in our system. Cytokines known to inhibit osteoclastogenesis include IFNγ, GM-CSF, IL4, IL5 and IL10.16 31 33 IL12 and IL18 have been described to inhibit osteoclastogenesis via IFNγ or GM-CSF.16 Concerning stimulators of osteoclastogenesis, TNFα, IL1β, IL6 and IL17 are the most important ones.16 34 35 In the supernatants of cultures with Treg cells, expression of cytokines inhibiting osteoclastogenesis was found to be higher as compared to control cultures or cultures with Teff cells, especially GM-CSF, IFNγ, IL5 and IL10. Our data do not reveal that these cytokines are produced by Treg cells themselves. Although Treg cells are able to produce for example IFNγ themselves,36 it is possible that Treg cells induce such expression in the OCP or T cells present in the osteoclast cultures through cell–cell interaction. For example, GM-CSF and IFNγ are typical Th1-associated cytokines. As Th1 cells have been shown to express RANKL, they are considered to be a pro-osteoclastogenic T cell subset. However, in our experiments we have added an excess amount of RANKL. Although all cultures with activated T cells contain higher levels of osteoclastogenesis stimulating cytokines as compared to control cultures, these levels were much lower than those of inhibitors of osteoclastogenesis (summarised in fig 5A,B). As precursor cells encountered RANKL before IFNγ, they probably were resistant to IFNγ-mediated inhibition.32 37 In addition, Treg cells isolated from IFNγR KO were able to inhibit the differentiation of osteoclasts from splenocytes from IFNγR KO mice (data not shown). Together, these results may suggest that the inhibitory effects of Treg cells on osteoclastogenesis are conferred by GM-CSF, IL5 or IL10.

Figure 5

Addition of regulatory T (Treg) cells favours production of osteoclast inhibiting cytokines. Results represent a summary of the data shown in fig 4.

In a previous report of Sato et al. Treg cells were found to have no effect on osteoclastogenesis.38 However, in this study freshly isolated Treg cells were added, without prestimulation, and in the absence of IL2, a necessary growth factor for Treg cells. Thornton et al have shown that in vitro suppression requires activation of the Treg cells via their receptor39 and Tang et al have provided evidence that Treg cells expanded with beads coated with anti-CD3 and anti-CD28 in the presence of IL2 are more suppressive than freshly isolated Treg cells.27 More recently, Kim et al have provided evidence that human prestimulated Treg cells inhibit the in vitro differentiation of osteoclasts from peripheral blood mononuclear cells.40 In their system, suppression is mediated by cytokine secretion, including IL4 and TGFβ. Similarly, our data suggest an important role for cytokines in the mechanism of osteoclast inhibition of Treg cells. However, cell contact also appears to play a role in the suppressive properties of Treg cells.41 Thus, recently, Zaiss et al have demonstrated that Treg cells inhibit osteoclast formation in a cell contact-dependent mechanism via cytotoxic T lymphocyte antigen 4 (CTLA-4).42

In conclusion, we have demonstrated that an adoptive transfer of Treg cells significantly improves clinical symptoms of arthritis, without affecting systemic humoral responses. Treg cells were shown to inhibit TNFα and IL6 production in vivo and to abrogate in vitro and in vivo osteoclastogenesis, thereby providing an explanation for their beneficial effect in arthritis. Since osteoclastogenesis is an important event in the effector phase, one may speculate on the use of Treg cells for clinical applications in RA.

REFERENCES

Footnotes

  • Competing interests: None declared.

  • Funding: This work was supported by grants from the Fund of Scientific Research Flanders (FWO Vlaanderen), from the Regional Government of Flanders (GOA programme), from the Belgian Federal Government (Interuniversity Network for Fundamental Research, IUAP) and from the Fondation Dormeur. HK received a fellowship from the FWO Vlaanderen.

Request Permissions

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.