Article Text

Extended report
Monosodium urate monohydrate crystals inhibit osteoblast viability and function: implications for development of bone erosion in gout
  1. Ashika Chhana1,
  2. Karen E Callon1,
  3. Bregina Pool1,
  4. Dorit Naot1,
  5. Maureen Watson1,
  6. Greg D Gamble1,
  7. Fiona M McQueen2,
  8. Jillian Cornish1,
  9. Nicola Dalbeth1
  1. 1Bone Research Group, Department of Medicine, University of Auckland, Auckland, New Zealand
  2. 2Department of Molecular Medicine and Pathology, University of Auckland, Auckland, New Zealand
  1. Correspondence to Dr Nicola Dalbeth, Bone Research Group, Department of Medicine, Faculty of Medical and Health Sciences, University of Auckland, 85 Park Rd, Grafton, Auckland, New Zealand; n.dalbeth{at}auckland.ac.nz

Abstract

Background Bone erosion is a common manifestation of chronic tophaceous gout.

Objectives To investigate the effects of monosodium urate monohydrate (MSU) crystals on osteoblast viability and function.

Methods The MTT assay and flow cytometry were used to assess osteoblast cell viability in the MC3T3-E1 and ST2 osteoblast-like cell lines, and primary rat and primary human osteoblasts cultured with MSU crystals. Quantitative real-time PCR and von Kossa stained mineralised bone formation assays were used to assess the effects of MSU crystals on osteoblast differentiation using MC3T3-E1 cells. The numbers of osteoblasts and bone lining cells were quantified in bone samples from patients with gout.

Results MSU crystals rapidly reduced viability in all cell types in a dose-dependent manner. The inhibitory effect on cell viability was independent of crystal phagocytosis and was not influenced by differing crystal length or addition of serum. Long-term culture of MC3T3-E1 cells with MSU crystals showed a reduction in mineralisation and decreased mRNA expression of genes related to osteoblast differentiation such as Runx2, Sp7 (osterix), Ibsp (bone sialoprotein), and Bglap (osteocalcin). Fewer osteoblast and lining cells were present on bone directly adjacent to gouty tophus than bone unaffected by tophus in patients with gout.

Conclusions MSU crystals have profound inhibitory effects on osteoblast viability and differentiation. These data suggest that bone erosion in gout occurs at the tophus–bone interface through alteration of physiological bone turnover, with both excessive osteoclast formation, and reduced osteoblast differentiation from mesenchymal stem cells.

Statistics from Altmetric.com

Bone erosion is a frequent manifestation of chronic tophaceous gout and can lead to joint damage and deformity and, eventually, musculoskeletal disability.1 2 Tophus infiltration into bone has been strongly implicated in the development of bone erosion in gout,1 and monosodium urate monohydrate (MSU) crystal deposits have been observed within subchondral bone, suggesting that bone cells have direct contact with MSU crystals in erosive gout.3 4

During bone remodelling, osteoblast-mediated bone formation and osteoclast-mediated bone resorption are coupled in order to preserve bone mass.5 Our previous work has implicated the osteoclast in the pathogenesis of bone erosion in chronic gout.6 The role of the osteoblast in erosive gout is uncertain. We have reported that culture of osteoblast-like cells with MSU crystals alters the receptor activator for the nuclear factor κB ligand/osteoprotegerin balance to promote osteoclastogenesis and bone resorption in vitro.6 Bouchard et al7 reported that human osteoblasts stimulated with MSU crystals and interleukin (IL) 1 had decreased osteocalcin formation and alkaline phosphatase activity, indicating reduced osteoblast differentiation.7 These observations suggest that the osteoblast may be an important cellular mediator of bone erosion in chronic gout.

The aim of this study was to understand the effects of MSU crystals on osteoblast viability and function.

Materials and methods

Reagents and cell lines

Unless specified, all reagents were purchased from Sigma-Aldrich (St Louis, Missouri, USA). The culture medium, fetal bovine serum (FBS) and antibiotics used in tissue culture were from Life Technologies (Gibco, Invitrogen Corp, Auckland, New Zealand), and bovine serum albumin (BSA) was obtained from Immuno-chemical Products Ltd (Auckland, New Zealand). All cell lines were purchased from the American Type Culture Collection.

Ethical approval

Collection of human samples was approved by the northern regional ethics committee, and all patients provided written informed consent. All protocols involving animals were approved by the University of Auckland animal ethics committee.

MSU crystal synthesis

Endotoxin-free MSU crystals were prepared by recrystallisation from uric acid as previously described.8

Osteoblast-like cell culture

Primary rat osteoblasts (rOBs) were isolated from 20-day fetal rat calvariae, as previously described.9 Cells were grown in Dulbecco's modified Eagle's medium (DMEM) with 5 μg/ml L-ascorbic acid 2-phosphate (AA2P) for 2 days and then changed to minimum essential medium (MEM) with 5 μg/ml AA2P and grown to 90% confluency. Cultures of primary human osteoblasts (hOBs) were prepared using normal human trabecular bone from 50–70-year-old patients undergoing knee or hip arthroplasty, using a modified method of Robey and Termine.10 Cells were cultured to 90% confluency in DMEM with 5 μg/ml AA2P. MC3T3-E1 cells were maintained in MEM with 1 mM sodium pyruvate, and ST2 cells were cultured in DMEM. All cultures contained 10% FBS to promote cell attachment, growth and proliferation, and to maintain cell viability. All cells were maintained in 75 cm2 flasks (Corning Inc, Lowell, Massachusetts, USA) and incubated at 37°C with 5% CO2.

MTT assay for cell viability

Osteoblast-like cells were seeded for 24 h onto 24-well plates (Cellstar; Greiner Bio-One, Frickenhausen, Germany) at 2.5 × 103 cells/well in 5% FBS in αMEM (MC3T3-E1); 5% FBS in DMEM (ST2); 5% FBS in MEM with 5 μg/ml AA2P (rOBs); and 5% FBS in DMEM with 10 μg/ml AA2P (hOBs). Growth of cells was arrested in 0.1% BSA medium for 24 h to ensure all cells were synchronised in the G0 phase of the cell cycle. Fresh medium containing 0.1% BSA and various concentrations of MSU crystals were then added for a further 26 h. Preliminary optimisation experiments demonstrated that MSU crystals alone interfered with endpoint values for all viability assays assessed (data not shown). In the MTT (3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide) assay, MSU crystals (in the absence of cells) interfered with the endpoint absorbance values showing false increases in absorbance and therefore viability; this effect was entirely prevented by dissolution of the crystals with 0.5 U/ml uricase for 2 h. Therefore, uricase was added to cell cultures for the final 2 h of MSU crystal incubation and also to control wells. Uricase had no effects on endpoint values for all assays used (data not shown). Cells were washed and then incubated with 0.5 mg/ml MTT (Invitrogen) for 4 h at 37°C. Resulting formazan crystals were dissolved with dimethyl sulphoxide and the absorbance read at 570 nm using a Synergy 2 multi-detection microplate reader (BioTek Instruments Inc, Winooski, Vermont, USA). There were six wells in each treatment group.

Flow cytometry

MC3T3-E1 cells were cultured with MSU crystals and uricase as described for the MTT assay. All cells were collected and washed twice with cold phosphate-buffered saline. Cells (104) were then resuspended in binding buffer and incubated in the dark with saturating amounts of fluorescein isothiocyanate-labelled annexin V and propidium iodide (PI) staining solution (both from Becton Dickinson, Mountain View, California, USA) for 15 min. A control sample of MSU crystals alone was included and samples were gated to exclude any residual MSU crystals from the analysis. All samples were analysed on an LSRII flow cytometer using FACSDiva software (Becton Dickinson).

Mineralisation assay

MC3T3-E1 cells were seeded onto six-well plates (Greiner Bio-One) at 5 × 103 cells/well in 10% FBS in MEM with 1 mM sodium pyruvate. When cells were confluent (after approximately 4 days), the medium was changed to 15% FBS in αMEM supplemented with 50 μg/ml AA2P and 10 mM β-glycerophosphate, and MSU crystals were added. These supplemented media were changed twice weekly and fresh MSU crystals were added. After 21 days the cells were treated with 0.5 U/ml uricase (including the control) for up to 6 h, and then fixed in neutral buffered formalin overnight. Cultures were stained for mineral using von Kossa stain. The area of mineralisation was quantified using BIOQUANT OSTEO software (BIOQUANT Image Analysis Corp, Nashville, Tennessee, USA). There were six wells in each treatment group.

Quantitative real-time PCR

RNA for real-time PCR was prepared from MC3T3-E1 cell cultures. The culture conditions were as previously described for the mineralisation assay. MSU crystals were added on day 0, when the medium was first changed to 15% FBS in αMEM supplemented with 50 μg/ml AA2P and 10 mM β-glycerophosphate, and cells were harvested at the indicated time points after the start of treatment. Purification of total cellular RNA, removal of genomic DNA, synthesis of cDNA and multiplex real-time PCR was performed as previously described.6 The ΔΔCt method was used to quantify mRNA gene expression and expression levels were normalised to those at day 0. Expression from all samples was normalised to 18S rRNA, an endogenous control which corrects for differences in cell number between samples.

Histology

Bone samples (two from the first metatarsophalangeal joint and one each from the knee and mid-foot) were obtained from three patients with gout undergoing orthopaedic surgery and from one cadaveric donor with aspirate-proven gout. Paraffin-embedded demineralised bone slides were prepared, dewaxed and rehydrated as previously described.6 Once hydrated, sections were briefly left in 5% acetic acid and then stained with 1% toluidine blue (BDH, Poole, UK) for 30 min at room temperature. Slides were then dehydrated with ethanol and xylol. Slides were mounted with DPX (BDH) and analysed by light microscopy. The number of osteoblast-like cells and lining cells was determined by counting cells on all bone surfaces directly adjacent to tophus in up to 10 microscopic fields, and the number on bone surfaces unaffected by tophus in 10 microscopic fields (both at ×40 objective) for each patient. Only bone surfaces without evidence of adjacent cartilage damage or osteophytes were included in the analysis. Osteoblasts were defined as cuboidal cells stained blue sitting on the osteoid seam of bone, while lining cells were defined as thin cells sitting on the osteoid seam. Observations were made by two independent scorers (AC and ND), who were blinded to each other's results and patient data. Interobserver intraclass correlation coefficients were 0.88 (0.83–0.91) for osteoblast scores, 0.77 (0.70–0.83) for lining cell scores and 0.91 (0.87–0.93) for osteoblast and lining cells combined scores. For each patient, the combined mean number of cells at each site was used for analysis.

Statistical analysis

Data from the viability assays were analysed using SAS Software (SAS Institute Inc, Cary, North Carolina, USA). Data were pooled from multiple plates and inter-plate variation was corrected for using the median control value from each plate, and treatment to control ratios were calculated. All other data were analysed using GraphPad Prism Software (GraphPad Software, San Diego, California, USA). Data were analysed using one-way or two-way analysis of variance with post hoc Dunnett's and Bonferroni's multiple comparison tests when there were more than two groups; or by Student's t test for two groups. Each experiment was repeated at least three times for each assay.

Results

MSU crystals reduce osteoblast cell viability in a dose-dependent manner

MSU crystals reduced osteoblast viability in a dose-dependent manner in the MTT assay, with maximal effects observed at the higher concentrations of 0.3 mg/ml and 0.5 mg/ml MSU crystals (figure 1A–D). When observed under light microscopy, cells cultured with these higher concentrations of MSU crystals appeared unhealthy with fewer cells present, and those cells that were alive had a shrunken appearance (figure 1E). The MC3T3-E1 and ST2 osteoblast cell lines, primary rOBs and primary hOBs all showed comparable reductions in viability in the MTT assay. For this reason, MC3T3-E1 cells and primary rOBs were used in subsequent experiments. The reduction of osteoblast cell viability following culture with MSU crystals was confirmed using alamarBlue and trypan blue assays (supplementary text and figure 1).

Figure 1

Monosodium urate monohydrate (MSU) crystals reduce osteoblast-like cell viability in a dose-dependent manner. Cell viability was assessed using the MTT assay after 26 h of culture with MSU crystals in (A) MC3T3-E1 cells and (B) ST2 cells; (C) primary rat osteoblasts and (D) primary human osteoblasts. Data shown are mean (SEM); one-way analysis of variance (p<0.0001) with post hoc Dunnett's test *p<0.001, **p<0.0001 versus control (no MSU crystals). (E) Photomicrographs showing MC3T3-E1 cell morphology following incubation with MSU crystals at indicated concentrations for 26 h. Scale bar indicates 100 µm.

Characteristics of the osteoblast–MSU crystal interaction

MSU crystals rapidly induced cell death in MC3T3-E1 osteoblast-like cells after 2.5 h of culture (figure 2A). The effects on osteoblast viability did not alter with different crystal lengths (figure 2B) and were not inhibited by the addition of serum (figure 2C). MC3T3-E1 cells cultured in standard conditions of 10% FBS throughout the experiment showed comparable reductions in viability to cells cultured in lower serum conditions. The effects on osteoblast viability were confirmed using different batches of MSU crystals synthesised in our laboratory and from another institution (kind gift of Professor Dorian Haskard, Hammersmith Hospital, UK) (data not shown).

Figure 2

Characteristics of the osteoblast–monosodium urate monohydrate (MSU) crystal interaction. (A) MSU crystals rapidly reduced MC3T3-E1 cell viability after 2.5, 3, 3.5 and 4 h of treatment. (B) Effects of MSU crystals on MC3T3-E1 cell viability do not alter with different MSU crystal lengths. Cell viability was assessed after 26 h of culture with MSU crystals. The indicated sizes of MSU crystals were obtained by sonicating the crystals for 0, 5, 10 and 15 min, respectively (from left to right). (C) Inhibition of MC3T3-E1 cell viability by MSU crystals did not alter with the addition of serum. Cell viability was assessed after 26 h of culture with MSU crystals in the presence of serum. (D) Soluble uric acid does not reduce osteoblast-like cell viability. MC3T3-E1 and primary rat osteoblast-like cells were cultured with various concentrations of uric acid for 24 h before cell viability was assessed. The MTT assay was used to assess cell viability. Data shown are mean (SEM); one-way analysis of variance (p≤0.0001) with post hoc Dunnett's test *p<0.05 versus control (no treatment).

Soluble uric acid did not have major effects on MC3T3-E1 and primary rOB viability (figure 2D). A small decrease in viability was seen with 0.5 mg/ml uric acid in the MC3T3-E1 cells; however, the reduction observed (26%) was minimal in comparison with the reduction observed with 0.5 mg/ml MSU crystals (78%).

MC3T3-E1 cells did not show evidence of phagocytosis of 0.1 mg/ml MSU crystals after 24 h of culture as determined by polarising and fluorescent microscopy (supplementary methods and supplementary figure 2B). However THP-1 cells, a monocytic cell line, did show evidence of MSU crystal phagocytosis using the same method (supplementary figure 2A). Addition of cytochalasin D, an inhibitor of phagocytosis, did not prevent reduction of MC3T3-E1 cell viability following culture with 0.1 mg/ml MSU crystals (supplementary methods and supplementary figure 2C).

MSU crystals induce cell death but not apoptosis in osteoblast-like cells

In flow cytometry assays, there was no change in the percentage of early apoptotic cells (PI negative and annexin V positive) following culture with various concentrations of MSU crystals for 26 h (figure 3A). Furthermore, there was no change in the percentage of early apoptotic cells following culture with 0.1 mg/ml MSU crystals for shorter time periods of culture (figure 3B). However, there was a steady rise in the percentage of dead cells (PI positive) and a steady decline in the percentage of live cells (annexin V negative and PI negative) over time with MSU crystal culture, and over 90% of cells were dead (PI positive) at the end of the 26 h incubation (figure 3B). Representative flow cytometry plots are shown in figure 3C.

Figure 3

Monosodium urate monohydrate (MSU) crystals rapidly induce cell death but not apoptosis in osteoblast-like cells. (A) Changes in the percentage of dead (propidium iodide (PI) positive), live (PI negative, annexin V negative) and apoptotic MC3T3-E1 cells (PI negative, annexin V positive) are shown following culture with various concentrations of MSU crystals after 26 h, and (B) 0.1 mg/ml MSU crystals over the course of 26 h as determined by flow cytometry. (C) Plots showing 20% of gated MC3T3-E1 cells following culture with or without 0.1 mg/ml MSU crystals for 3 h. Numbers represent the percentage of cells in each quadrant. Graphs and plots are representative of three separate experiments.

MSU crystals inhibit differentiation and reduce function of osteoblast-like cells

In 3-week mineralisation assays, MSU crystals inhibited the area of mineralised bone formed with 0.1 mg/ml MSU crystals showing a significant reduction in mineralisation (figure 4). Repeated exposure of the cells to the higher doses of MSU crystals (0.3 and 0.5 mg/ml) resulted in an absence of viable cells and mineralisation (figure 4).

Figure 4

Monosodium urate monohydrate (MSU) crystals inhibit bone mineralisation. MC3T3-E1 cells were cultured over a period of 3 weeks with MSU crystals added twice weekly. (A) Representative photomicrographs showing mineralised areas (black) and cells (yellow) with indicated concentrations of MSU crystals (mg/ml). Scale bar indicates 10 mm. (B) Changes in the percentage of mineralised area. Data shown are mean (SEM); one-way analysis of variance (p<0.0001) with post hoc Dunnett's test *p<0.001 versus control (no MSU crystals). Results are representative of three separate experiments.

Real-time PCR was used to determine changes in the relative gene expression levels of early expressed osteoblast transcription factors Runx2 and Sp7 (osterix); as well as changes in the gene expression levels of Ibsp (bone sialoprotein), Bglap (osteocalcin) and Dmp1 (dentin matrix protein 1), factors usually expressed during later stages of osteoblast differentiation that are required for new bone formation; and lastly, Col1a1 (collagen, type I, α 1). Differences in cell numbers between samples were corrected for using 18S rRNA. The presence of 0.1 mg/ml MSU crystals in 3-week cultures of MC3T3-E1 cells significantly reduced the relative mRNA expression of all genes examined at one or more time points (figure 5).

Figure 5

Monosodium urate monohydrate (MSU) crystals reduce the relative mRNA expression of osteoblast differentiation genes in cultures of MC3T3-E1 cells. Changes in the relative mRNA expression levels of (A) Runx2; (B) osterix; (C) bone sialoprotein; (D) osteocalcin; (E) dentin matrix protein 1 and (F) Collagen, type I, α 1, are shown following culture with (▲) or without (□) 0.1 mg/ml MSU crystals. Data were pooled from two or more repeats and are shown as mean (SD); two-way analysis of variance (p≤0.05) with post hoc Bonferroni's test *p<0.05, **p<0.01, ***p<0.001 versus control (no MSU crystals) at that time point.

Fewer osteoblasts are present on bone directly adjacent to gouty tophus in patients with gout

In bone samples from patients with gout, the number of osteoblasts, lining cells, and the combined number of both cell types present was significantly reduced in bone adjacent to tophus compared with bone unaffected by tophus (figure 6A–C). Bone adjacent to tophus often had no osteoblasts or lining cells present (figure 6D). However, osteoclasts were frequently observed at sites adjacent to tophus and bone appeared to be considerably eroded (figure 6D).

Figure 6

Fewer osteoblasts are present on bone directly adjacent to gouty tophus in patients with gout. Number of (A) osteoblast cells; (B) lining cells and (C) osteoblasts and lining cells combined, quantified in human bone from patients with gout, at sites unaffected by or adjacent to gouty tophus. Data are shown as the median number of cells/microscopic field (×40 objective). p Value determined by Student's paired t test. (D) Representative photomicrographs of paired samples from two separate patients showing toluidine blue stained bone unaffected by or adjacent to gouty tophus. *Indicates bone, osteoblasts (OB) and osteoclasts (OC). Scale bar indicates 50 µm.

Discussion

This study shows that MSU crystals have a profound negative effect on osteoblast viability, function and differentiation. The potential clinical relevance of these findings is confirmed by analysis of human bone samples affected by gout, demonstrating relative paucity of osteoblasts at sites affected by MSU crystal deposition. These data suggest that bone erosion in gout occurs at the tophus–bone interface through alteration of physiological bone turnover, with both excessive osteoclast formation, and reduced osteoblast differentiation from mesenchymal stem cells.

The observation that MSU crystals induce osteoblast death differs from our previous study of osteoclasts, in which culture with MSU crystals did not alter the number of osteoclasts.6 MSU crystals have also been shown to enhance macrophage survival11 and to activate dendritic cell membranes through cholesterol interactions and lipid raft movements which lead to initiation of the phagocytosis machinery.12 Our findings in osteoblasts have some similarities to previous reports that MSU crystals induce cell death in neutrophils.13,,16 However, MSU crystal-induced neutrophil death is mediated by induction of apoptosis,17 and can be prevented by addition of serum to the cultures.14 16 18 Furthermore, blocking crystal phagocytosis reduces MSU crystal-induced neutrophil death.19 In the current study, addition of serum to osteoblast cultures failed to block MSU crystal-induced cell death, even when relatively high levels of serum were present. Similarly, inhibition of MSU crystal phagocytosis did not alter the effects on viability seen in osteoblast-like cells. These results suggest a different mechanism of MSU crystal induced osteoblast death, most probably induction of cell necrosis following direct contact and subsequent disruption of the osteoblast cell membrane.

The results of our in vitro studies are supported by analysis of bone samples from patients with gout. We acknowledge that this sample set is small, consistent with the difficulty of obtaining bone samples from patients with gout. However, this analysis has clearly demonstrated a paucity of osteoblasts and lining cells within bone adjacent to the site of MSU crystal deposition, the gouty tophus, suggesting a lack of functional osteoblasts within the erosive lesion in chronic tophaceous gout. Of particular interest is the observation that this is a localised effect, and that osteoblasts are present in those areas of bone unaffected by tophus within the same sample. This finding may be relevant when considering the radiographic appearance of gout, which is characterised by both bone erosion and new bone formation. It is possible that new bone formation might still occur at distant sites in the absence of MSU crystal deposits. The mechanisms that contribute to new bone formation in some gouty joints remain unclear.

The findings of this study can be compared with other studies of osteolytic conditions, such as wear-debris osteolysis. Consistent with our study of MSU crystals, osteoblast viability, differentiation and function are impaired by wear-debris particulates.20,,23 Osteoblast-mediated osteoclastogenesis is also enhanced at the bone–implant interface.24 25 However, the effects of wear-debris particulates on osteoblasts and their progenitors are mostly dependent on phagocytosis, and experiments blocking phagocytosis of wear-debris particulates show a reduction in their inhibitory effect.21 26 This is in contrast to our findings that osteoblasts do not phagocytose MSU crystals and that blocking of phagocytosis does not alter the inhibition of osteoblast viability by MSU crystals.

Osteoblastogenesis is also impaired in rheumatoid arthritis (RA), another erosive arthropathy. Osteoblasts on bone adjacent to inflammation in a mouse model of RA have abundant protein and mRNA expression of the early osteoblast lineage marker Runx2, but not later markers such as alkaline phosphatase and osteocalcin.27 Patients with RA have also been shown to have an increased percentage of active osteoid surface compared with patients with osteoarthritis.28 These results suggest that in RA, osteoblasts are fixed in an early differentiation phase and are unable to mature into bone mineralising osteoblasts. This differs from our current study where MSU crystals inhibited mRNA expression of genes related to both early and late osteoblast differentiation, implying that suppression of osteoblast maturation occurs at all stages of osteoblastogenesis in erosive gout.

This study is consistent with some of the findings of Bouchard et al, which showed reduced osteoblast activity in response to MSU crystals.7 That study focused on the effects of MSU crystals on osteoblast-like cells in conjunction with IL-1, and, in particular, the ability of MSU crystals to upregulate inflammatory factors such as cyclo-oxgenase-2 and IL-6, which in turn might have indirect inhibitory effects on bone formation. This study has expanded these observations to assess the direct effects of MSU crystals on osteoblast viability, differentiation and function, and to determine the relevance of these findings in bone affected by chronic tophaceous gout.

In summary, this study indicates that MSU crystals contribute to bone erosion in gout through promotion of osteoclast formation and also through reduction of osteoblast viability, function and differentiation. The data provide further rationale for studies of intensive urate-lowering treatment to determine the impact of MSU crystal dissolution in bone remodelling in chronic tophaceous gout.

Acknowledgments

The authors acknowledge Professor Rocco Pitto, Dr Alan King, Dr Michael Dray and Sharita Meharry who assisted with sample collection and processing.

References

Supplementary materials

Footnotes

  • Funding This study was funded by a University of Auckland Doctoral Scholarship (AC) and the Auckland Medical Research Foundation.

  • Competing interests None.

  • Ethics approval This study was conducted with the approval of the New Zealand Ministry of Health Northern Regional Ethics Committee.

  • Provenance and peer review Not commissioned; externally peer reviewed.

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.