Advertisement
Research Article

Prolonged Application of High Fluid Shear to Chondrocytes Recapitulates Gene Expression Profiles Associated with Osteoarthritis

  • Fei Zhu,

    Affiliation: Department of Chemical and Biomolecular Engineering, The Johns Hopkins University, Baltimore, Maryland, United States of America

    X
  • Pu Wang,

    Affiliation: Department of Chemical and Biomolecular Engineering, The Johns Hopkins University, Baltimore, Maryland, United States of America

    X
  • Norman H. Lee,

    Affiliation: Department of Pharmacology and Physiology, The George Washington University Medical Center, Washington, D.C., United States of America

    X
  • Mary B. Goldring,

    Affiliation: Hospital for Special Surgery, New York, New York, United States of America

    X
  • Konstantinos Konstantopoulos mail

    kkonsta1@jhu.edu

    Affiliations: Department of Chemical and Biomolecular Engineering, The Johns Hopkins University, Baltimore, Maryland, United States of America, Johns Hopkins Physical Sciences in Oncology Center and Institute for NanoBioTechnology, The Johns Hopkins University, Baltimore, Maryland, United States of America

    X
  • Published: December 29, 2010
  • DOI: 10.1371/journal.pone.0015174

Abstract

Background

Excessive mechanical loading of articular cartilage producing hydrostatic stress, tensile strain and fluid flow leads to irreversible cartilage erosion and osteoarthritic (OA) disease. Since application of high fluid shear to chondrocytes recapitulates some of the earmarks of OA, we aimed to screen the gene expression profiles of shear-activated chondrocytes and assess potential similarities with OA chondrocytes.

Methodology/Principal Findings

Using a cDNA microarray technology, we screened the differentially-regulated genes in human T/C-28a2 chondrocytes subjected to high fluid shear (20 dyn/cm2) for 48 h and 72 h relative to static controls. Confirmation of the expression patterns of select genes was obtained by qRT-PCR. Using significance analysis of microarrays with a 5% false discovery rate, 71 and 60 non-redundant transcripts were identified to be ≥2-fold up-regulated and ≤0.6-fold down-regulated, respectively, in sheared chondrocytes. Published data sets indicate that 42 of these genes, which are related to extracellular matrix/degradation, cell proliferation/differentiation, inflammation and cell survival/death, are differentially-regulated in OA chondrocytes. In view of the pivotal role of cyclooxygenase-2 (COX-2) in the pathogenesis and/or progression of OA in vivo and regulation of shear-induced inflammation and apoptosis in vitro, we identified a collection of genes that are either up- or down-regulated by shear-induced COX-2. COX-2 and L-prostaglandin D synthase (L-PGDS) induce reactive oxygen species production, and negatively regulate genes of the histone and cell cycle families, which may play a critical role in chondrocyte death.

Conclusions/Significance

Prolonged application of high fluid shear stress to chondrocytes recapitulates gene expression profiles associated with osteoarthritis. Our data suggest a potential link between exposure of chondrocytes/cartilage to abnormal mechanical loading and the pathogenesis/progression of OA.

Introduction

Osteoarthritis (OA) is a chronic disease characterized by the degeneration or destruction of the articular cartilage tissue that covers and protects the moving joints. The clinical correlates of OA are joint pain, dysfunction and restricted motion. The etiologies of OA include joint dysplasia, genetic and developmental joint abnormalities, ageing and joint injuries [1]. Indeed, excessive chronic or repetitive mechanical loading of articular cartilage has been reported to play a key role in the development and progression of OA [1]. Chondrocytes represent the sole cellular component of cartilage, and regulate its fate due to their ability to synthesize matrix-degrading enzymes and matrix proteins such as collagens and proteoglycans, which are responsible for the tensile strength and compressive resistance, respectively, of cartilage to mechanical loading. Mechanical loads produce hydrostatic pressure and shear stress which causes tensile strain in some direction [2], [3]. Elegant modeling studies have shown that, in addition to hydrostatic pressure, chondrocytes of the superficial and transitional zones are subjected to high and low fluid flow, respectively, whereas cells of the middle and deep radial zones experience little to no fluid flow [2], [3]. These observations suggest that fluid flow or fluid shear stress is a pathophysiologically relevant mechanical signal in cartilage biology.

Fluid shear modulates intracellular signaling in a time-, magnitude- and phenotype-dependent manner. In the vasculature, high levels of laminar shear are atheroprotective, whereas low shear oscillatory flow tends to be atherogenic. In contrast, numerous in vitro studies support the concept that low fluid shear (<10 dyn/cm2) is chondroprotective [4], whereas high shear stress (>10 dyn/cm2) elicits the release of pro-inflammatory cytokines such as interleukin-6 (IL-6) [5], and mediates matrix degradation [4], [6] and chondrocyte cell death [7], [8], [9], which represent earmarks of OA. Predicted fluid flow and fluid shear stress values in vivo are lower than those applied in vitro by other investigators and us [4], [5], [6], [7], [8], [9]. We and others have documented that fluid shear affects cell responses in a time- and magnitude-dependent manner. For instance, the reduced antioxidant capacity of chondrocytes was detected after a 24-h exposure to a fluid shear stress level of 40 dyn/cm2 [7]. Quantitatively similar results were obtained when chondrocytes were subjected to a lower shear stress level (20 dyn/cm2) but for an extended (48 h) shear exposure time [7]. As has appropriately been argued in the literature [3], “it is the cumulative influence of loading histories throughout life that governs the biology of the tissue”. It is therefore apparent that detection of chondrocyte responses relevant to OA induced by pathological levels of fluid shear encountered in vivo would require extremely long time scales (equivalent to those associated with the onset of OA), which are infeasible and impractical in a laboratory setting. Of note, the inter-dependence between the magnitude and duration of shear for chondrocytes is not known. We, therefore, strategically chose the standard approach employed by toxicologists to evaluate the potential toxicity of lifetime exposure of man to a chemical substance [10]; that is, the investigation of supra-physiological concentrations of the chemical, in our case supra-physiological shear stress levels, for an experimentally feasible time scale.

Since OA is often a consequence of excessive mechanical forces [1] and given that the application of high fluid shear to chondrocytes recapitulates some of the earmarks of OA [4], [6], [7], [8], [9], we aimed to screen the gene expression profiles of shear-activated chondrocytes and assess potential similarities with OA chondrocytes. Using cDNA microarrays, we found that 42 of the 131 differentially regulated genes in sheared chondrocytes have been reported previously in OA chondrocytes, and are related to extracellular matrix (ECM)/matrix degradation, cell growth/differentiation, inflammation and cell survival/death. Consistent with the critical role of cyclooxygenase-2 (COX-2) in the development and/or progression of OA in vivo [11] and findings on the regulation of shear-induced reactive oxygen species (ROS) [9] and apoptosis in vitro [7], we identified a collection of genes that are regulated by shear-induced COX-2, including genes of the histone and cell cycle families, which may play a critical role in chondrocyte death. Taken together, our data suggest that prolonged application of high fluid shear to human T/C-28a2 chondrocytes recapitulates the earmarks of OA, and illustrate a link between high mechanical forces and the development of OA.

Results

Differentially expressed genes in shear-activated human chondrocytes

OA is often a consequence of excessive mechanical loading of cartilage [12], which produces hydrostatic stress, tensile strain and fluid flow [2], [3]. Exposure of human chondrocytes to high fluid shear elicits the release of pro-inflammatory mediators such as interleukin-6 [5], and mediates matrix degradation [4], [6] and apoptosis [7], [8], [9]. In view of accumulating evidence suggesting that prolonged application of high fluid shear recapitulates some of the earmarks of OA, we aimed to identify the differentially-regulated genes in human T/C-28a2 chondrocytes subjected to high fluid shear (20 dyn/cm2) versus static (control) conditions (0 dyn/cm2) for 48 h and 72 h, using a cDNA microarray technique. In these experiments, total RNA, extracted from control (unsheared) and shear-activated T/C-28a2 cells, was reverse transcribed and labeled with Cy3 and Cy5, respectively, and then hybridized to TIGR 40K human set chips containing 39,936 human expressed sequence tags (ESTs) [7], [9]. As shown in Fig. 1, the expression ratios of 61% of all EST probes between sheared and control genes were statistically significant based on the Student's t-test (p≤0.01). Using SAM with a 5% FDR, 799 probes were found to be differentially regulated between sheared and control specimens. Of these, 98 probes displayed ≥2-fold upregulation, whereas 90 probes showed ≤0.6-fold fold downregulation between sheared and control chondrocytes (Fig. 2, Tables S1 and S2). Of the 98 up-regulated probes, 76, corresponding to 71 non-redundant transcripts, have been sequenced at full-length, whereas the remaining are ESTs (Tables S1). Similarly, of the 90 down-regulated probes, 69, representing 60 non-redundant transcripts, correspond to known genes, whereas the rest are ESTs (Table S2). The differentially-regulated genes with known sequences were classified according to gene ontology (GO), in terms of their involvement in biological processes, and sorted by percentages according to FatiGO (http://www.fatigo.org), a web interface which carries out data mining using GO for DNA microarray data [13], [14] (Fig. S1A and B).

thumbnail

Figure 1. Volcano plot of microarray data.

T/C-28a2 chondrocytes were subjected to fluid shear (20 dyn/cm2) or static control (0 dyn/cm2) conditions for 48 h or 72 h. Three paired samples for each time point were obtained for microarray analysis. The negative log10-transformed p-values of the Student's t-test are plotted against the shear to static ratios of fold change in the six-sample experiment. The horizontal bar represents the nominal significant level 0.01 for the Student's t-test (p≤0.01 for 61% of all ESTs represented by the red and green points). The vertical dashed bars denote ≤0.6-fold downregulation (left) or ≥2.0-fold upregulation (right).

doi:10.1371/journal.pone.0015174.g001
thumbnail

Figure 2. Hierarchical clustering of differentially expressed genes from six sheared and matched static control chondrocyte specimens.

Each horizontal row represents a single gene. Up-regulated genes in shear-activated (20 dyn/cm2 for 48 h or 72 h) relative to matched static control T/C-28a2 chondrocyte samples are shown in red, whereas down-regulated genes are shown in green.

doi:10.1371/journal.pone.0015174.g002

Comparison of the gene expression profiles between sheared and OA chondrocytes

We next investigated the potential similarities in the gene expression profiles of shear-activated chondrocytes determined in this study and OA chondrocytes reported in the literature. Of the 71 shear-up-regulated genes, 32 have previously been reported to be similarly regulated in OA chondrocytes, accounting for 45% similarity. As shown in Table 1, these genes are related to cell adhesion, cell survival/death, cell growth/differentiation, extracellular matrix (ECM)/matrix degradation, inflammatory response, oxidation/reduction and signal transduction. Although prolonged application of fluid shear increased the mRNA synthesis of TCDD-inducible poly(ADP-ribose) polymerase (PARP-1) in human T/C-28a2 chondrocytes (Table 1), a recent microarray study reported this gene to be down-regulated in OA chondrocytes relative to normal controls [15]. Of note, PARP-1 was found to be up-regulated in rheumatoid arthritis (RA) [16]. Moreover, our data are consistent with prior observations suggesting that RHOB, a member of the Rho GTP-binding protein, is overexpressed in OA [17], [18] and the positive association in the expression levels of RHOB and PARP-1 [19]. Our microarray analysis also identified two additional genes, IL-32 and pappalysin, that are up-regulated in shear-activated chondrocytes as well as in RA [20], [21] but not OA [15]. Of the 60 shear-down-regulated genes, only 3 have been reported to be similarly regulated in OA. A previous microarray study identified two members of the histone family, HIST2H2AA and H3F3B, to be mildly down-regulated in OA knees [22]. Here, we identified 6 new genes of the histone family to be significantly down-regulated in shear-activated chondrocytes (Table 2). Moreover, fluid shear down-regulated the mRNA levels of 9 cell cycle-related genes (Table 2), which may be responsible for chondrocyte apoptosis [9]. Three additional genes, vascular cell adhesion molecule-1 (VCAM-1), chitinase 3-like 2 (CHI3L2) and the chemokine CXCL12 were down-regulated in sheared chondrocytes, although these genes have been reported to be up-regulated in the microarray profiling of OA chondrocytes [15], [22], [23].

thumbnail

Table 1. List of differently-regulated genes in shear-activated T/C-28a2 chondrocytes compared to OA chondrocytes obtained from the literature.

doi:10.1371/journal.pone.0015174.t001
thumbnail

Table 2. Identification of genes regulated by COX-2 in shear-activated T/C-28a2 chondrocytes.

doi:10.1371/journal.pone.0015174.t002

Confirmation differential gene expression by qRT-PCR

To validate the expression profiles obtained by microarray analysis, qRT-PCR was used to quantify the mRNA expression levels in sheared and matched static control chondrocytes. We chose to examine the following genes: gremlin in view of consistent literature data suggesting that it is up-regulated in OA chondrocytes [15], [24]; HIST12BD and HIST13H2A, which represent two newly identified genes that are differentially regulated in shear-activated chondrocytes; RHOB in light of conflicting literature data [15]; PAPP-A given their opposite regulation in sheared and OA [15] chondrocytes. qRT-PCR revealed the same gene expression pattern as the microarray analysis in all five genes examined in this work (Table 3).

thumbnail

Table 3. Comparison of the transcript ratios of select genes determined by qRT-PCR versus cDNA microarray.

doi:10.1371/journal.pone.0015174.t003

High fluid shear induces IL-1β expression, matrix degradation and reactive oxygen species in human chondrocytes

Although OA is classified as a non-inflammatory joint disease, prostaglandins and cytokines such IL-1β and IL-6 are believed to play a role in the pathogenesis and progression of disease [11], [25], [26]. In addition to inducing the expression of matrix degrading enzymes, IL-1β also represses the expression of an array of genes associated with the differentiated chondrocyte phenotype, including the type II collagen gene (COL2A1) and aggrecan (AGC) [25], [27], [28]. Degradation of aggrecan is considered an important manifestation of OA. We thus evaluated whether prolonged application of high fluid shear to human T/C-28a2 chondrocytes modulates the expression of key marker genes of OA in a manner similar to that detected in OA relative to healthy chondrocytes. As shown in Fig. 3, high fluid shear increases the mRNA levels of COX-2 and IL-1β and concomitantly suppresses those of COL2A1 and AGC in human T/C-28a2 chondrocytes, which is similar to the gene regulation pattern observed in OA chondrocytes [25], [26], [27], [28].

thumbnail

Figure 3. High shear stress induces gene markers of osteoarthritis in human chondrocytes.

T/C-28a2 chondrocytes were subjected to fluid shear (20 dyn/cm2) or static conditions (0 dyn/cm2) for 48 h. qRT-PCR was used to quantify the mRNA transcript ratios of select genes in sheared compared to static control chondrocytes. Data represent the mean±S.D. of n≥3 independent experiments.

doi:10.1371/journal.pone.0015174.g003

Accumulating evidence suggests that reactive oxygen species (ROS) contribute to the pathophysiology of OA [29]. ROS generation overwhelms the endogenous antioxidant defense system of chondrocytes, as evidenced by the marked downregulation of a battery of antioxidant genes in OA chondrocytes such as superoxide dismutase, gluthione peroxidase 3 and thioredoxin-interacting protein [22]. Using DCFDA in conjunction with flow cytometry, we determined that high shear stress induces ROS generation in human chondrocytes (Fig. 4A). Knockdown of L-prostaglandin synthase (L-PGDS) via RNA interference abrogated the formation of ROS in sheared T/C-28a2 chondrocytes (Fig. 4). Taken altogether, our data suggest that COX-2-derived PGD2 and/or its metabolite 15-deoxy-Δ12,14-PGJ2 (15d-PGJ2) have the ability to generate ROS in sheared T/C-28a2 chondrocytes.

thumbnail

Figure 4. Effects of L-PDGS knockdown on shear-mediated ROS generation and histone regulation in human chondrocytes.

T/C-28a2 chondrocytes were subjected to fluid shear (20 dyn/cm2) or static conditions (0 dyn/cm2) for 48 h. In select experiments, T/C-28a2 cells were transfected with an siRNA oligonucleotide sequence-specific L-PGDS before being subjected to fluid shear. (A) ROS generation was quantified using the DCFDA dye in conjunction with flow cytometry. Histograms are representative of three independent experiments. (B) mRNA transcript ratios for sheared to static control T/C-28a2 chondrocytes (closed bars). In select experiments, the transcript ratio of sheared, L-PGDS knockdown T/C-28a2 cells to static control cells was determined (open bars). Data represent the mean±S.D. of n≥3 independent experiments. *, p<0.05 with respect to shear control.

doi:10.1371/journal.pone.0015174.g004

Genes regulated by COX-2 in shear-activated chondrocytes

In view of the pivotal role of COX-2 in the regulation of shear-induced inflammation and apoptosis in human chondrocytes [7], [9], we next aimed to identify genes regulated by COX-2 in sheared chondrocytes. The gene transcription profile of T/C-28a2 chondrocytes, subjected to high fluid shear (20 dyn/cm2) for 48 h in the presence or absence of the specific COX-2 inhibitor NS398 (50 µM), was determined from microarray experiments using the TIGR MeV software. Our data reveal that the expression pattern of two distinct collections of genes was reversed in sheared chondrocytes incubated with NS398 (Fig. 5). The first collection of genes is positively regulated by COX-2. Thus, inhibition of COX-2 activity by NS398 suppresses the shear-induced, COX-2-dependent upregulation of these genes, which are primarily related to inflammation, matrix degradation and apoptosis (Fig. 5; Table 2; Table S3). The second collection of genes is negatively regulated by COX-2, and as such, inhibition of COX-2 activity restores the shear-induced COX-2-dependent downregulation of these genes back to near basal levels (Fig. 5; Table 2; Table S4). The majority of these genes are histone- and cell cycle- related genes (Table 2; Table S4). To validate the contribution of COX-2 to shear-mediated regulation of histones, T/C-28a2 chondrocytes were transfected, prior to their exposure to high fluid shear, with a siRNA oligonucleotide sequence specific for L-PGDS, which is downstream of COX-2 and responsible for ROS production. This genetic intervention abrogated the shear-mediated downregulation of histone genes (Fig. 4B).

thumbnail

Figure 5. Heat map of genes identified as commonly regulated by COX-2 in shear-activated human T/C-28a2 chondrocytes.

Each horizontal row represents a single gene. Up-regulated genes in shear-activated relative to control chondrocyte specimens are shown in red (left upper part). NS398 (50 µM) suppresses the shear-induced COX-2-dependent upregulation of these genes, which are depicted in green (right upper part). Down-regulated genes in sheared relative to static control chondrocytes are shown in green (left lower part). Inhibition of COX-2 activity by NS398 (50 µM) restores the shear-induced COX-2-dependent downregulation of the genes, which are depicted in red (right lower part).

doi:10.1371/journal.pone.0015174.g005

Discussion

OA is a debilitating disease of the joints characterized by the irreversible erosion of articular cartilage. OA has multiple risk factors including joint dysplasia, genetic and developmental joint abnormalities, ageing and joint injuries [1]. In younger people without genetic/developmental abnormalities, mechanical factors due to trauma are primarily implicated in the initiation and progression of OA lesions [12]. The adult articular chondrocytes, although quiescent in normal cartilage, are able to respond to mechanical forces. Excessive mechanical loading of cartilage producing hydrostatic stress, tensile strain and fluid flow [2], adversely affects chondrocyte function and precipitates OA. The objective of our study was to identify the similarities in the gene expression profiles of shear-activated and OA chondrocytes. Using the cDNA microarray technology, we found that 42 of the 131 differentially regulated genes in sheared chondrocytes have been reported previously in OA chondrocytes, and are related to ECM/matrix degradation, cell growth/differentiation, inflammation and cell survival/apoptosis. It is likely that the 15 histone- and cell cycle- related genes, found to be differentially regulated in sheared chondrocytes, are also involved in OA, since distinct histone [22] and cell cycle [15] related genes were recently reported in microarray studies of OA chondrocytes. In addition, the gene expression patterns of other well-established markers of OA such as COX-2 [11], [30], L-PGDS, IL-1β, COL2A1 and AGC [25], [27], [28], are similar to those detected in sheared chondrocytes. Taken together, at least 60 genes display akin regulation in both sheared and OA chondrocytes.

As shown in Table 1, there were a few genes whose regulation patterns were opposite in shear-activated relative to OA chondrocytes. These differences could be attributed to several reasons such as the distinct etiologies underlying OA, the stage of OA, and the inherent variability of gene expression levels in chondrocytes isolated from different donors. Although high variability might be expected for the disease samples due to different etiology and/or stage of OA, Aigner and coworkers [22] reported a comparable high variability among normal donors. This high variability might also explain why their microarray analysis of OA chondrocytes revealed the downregulation of an array of genes involved in cytokine signaling including IL-1β, IL-8 and leukemia inhibitory factor [22], whereas a recent study showed upregulation of these same genes in OA [15]. Controversy exists among others about whether COL2A1 expression is increased or suppressed in OA cartilage. Aigner and colleagues have suggested that the expression of COL2A1 is suppressed in the upper zones of early OA cartilage, but increased in late-stage OA cartilage relative to normal controls [31], [32]. However, upregulation of collagen genes applies predominantly to those chondrocytes found in the middle and deep zones of OA cartilage, whereas the anabolic phenotype is less obvious in the upper regions [33].

We have demonstrated the critical role of COX-2 in the regulation of shear-induced IL-6 and apoptosis in human chondrocytes [7], [9], [34]. Using cDNA microarrays, we identified genes that were either positively or negatively regulated by COX-2 in shear-activated chondrocytes. The former genes are related to inflammation, matrix degradation and apoptosis. A positive association in the expression levels of COX-2 and caveolin-1 [35], [36] or EPH receptor A2 [37] is supported by findings of other studies employing different cell types. Caveolin-1 and -2 co-localize and form a hetero-oligomeric complex in vivo [38]. Moreover, integrin alpha 2 (ITGA2) is associated with caveolin-1 in tumor cells [38]. Interestingly, our data suggest that EPH receptor A2, caveolins-1 and -2 and ITGA2 are under the control of COX-2 in sheared chondrocytes. Caveolin-1 [39] and FAS [39], also positively regulated by COX-2, have been reported to be up-regulated in OA cartilage. In view of our recent observations suggesting that p53 phosphorylation is regulated by COX-2 in sheared chondrocytes [9], it is not surprising that apoptosis enhancing nuclease (AEN) is also under COX-2 control.

Two major classes of genes were identified to be negatively modulated by COX-2 in shear-activated T/C-28a2 chondrocytes: histone and cell-cycle-related genes. We and others have shown that COX-2 overexpression induces cell cycle arrest in diverse cells including chondrocytes, NIH 3T3 fibroblasts, human embryonic kidney 293 cells [9], [40]. Here, we provide evidence for the first time suggesting that overexpression of COX-2 also negatively regulates histone gene expression in sheared chondrocytes. Downregulation of histone gene expression has been detected after DNA damage induced by ionizing radiation in different cells such as human fibroblasts and osteosarcoma [41]. Endogenous degradation of histones was also observed in K562 human leukemic cells after oxidative challenge [42]. The precise role of histones in OA has yet to be defined. Two histone family genes, H2AFO and H3F3B, were shown to be differentially down-regulated in OA chondrocytes relative to healthy control samples, which is in general agreement with our observations in sheared chondrocytes. Moreover, injection of histone H1 into collagen-induced arthritis (CIA) mice dramatically suppressed CIA [43]. Prior work has shown that transcriptional downregulation of histone occurs in parallel with the inhibition of DNA synthesis by p53 [41]. We recently demonstrated that PGD2 and/or its metabolite 15d-PGJ2 mediate chondrocyte apoptosis via PKA-dependent regulation of p53 phosrphorylation [9]. Indeed, L-PGDS knockdown reverses the shear-mediated histone transcriptional downregulation.

ROS play an important role in the pathogenesis of OA [29]. Excessive levels of ROS generated by abnormal chondrocyte metabolism tip the balance of anabolic and catabolic events, resulting in oxidative stress and loss of homeostasis. We and others have shown that elevated mechanical stress, including shear stress, releases ROS from chondrocytes [9], [12], and that antioxidants repress stress-induced chondrocyte death [7], [12]. L-PGDS knockdown inhibits shear-induced ROS formation, suggesting the involvement of PGD2 and/or its metabolite 15d-PGJ2 in this process.

In summary, we have demonstrated that prolonged application of high fluid shear to T/C-28a2 chondrocytes recapitulates the earmarks of OA, thereby providing further support to the link between exposure of chondrocytes/cartilage to high mechanical loading and the development of OA. Fluid shear is a well-defined biophysical stimulus for in vitro studies of mechanotransduction of articular chondrocytes. Delineating the responses of chondrocytes to high fluid shear may help us understand how OA develops. These studies may also lead to identification of ideal hydrodynamic environments for culturing artificial cartilage in bioreactors.

Methods

Reagents

The specific COX-2 inhibitor NS398 was obtained from Cayman Chemical. All other reagents were from Invitrogen, unless otherwise specified.

Cell Culture and Shear Stress

Human immortalized T/C-28a2 chondrocytes were grown (37°C in 5% CO2) on glass slides in 1:1 Ham's F-12/DMEM medium supplemented with 10% FBS [9], [44]. 24 h prior to the onset of shear stress application, T/C-28a2 cells were incubated in serum-free medium containing 1% Nutridoma-SP (Sigma-Aldrich), a low protein serum replacement that maintains chondrocyte phenotype. T/C-28a2 chondrocytes were subjected to a shear stress level of 20 dyn/cm2 for 48 h or 72 h in medium containing 1% Nutridoma-SP by the use of a streamer gold flow device (Flexcell International). In select experiments, the specific COX-2 inhibitor NS398 (50 µM) was added to the medium just before the onset of shear exposure. T/C-28a2 cells have been shown to behave much like primary human chondrocytes when cultured under appropriate conditions [45]. Further evidence suggesting that T/C-28a2 cells represent an appropriate chondrocyte model stems from the significant similarities between human primary chondrocytes and T/C-28a2 cells in the induction of IL-6 synthesis in response to chemical and shear stimulation [34], [46].

RNA Isolation

Total RNA was isolated using TRIzol, and purified with the RNeasy Mini Kit combined with DNase treatment on a column, according to the manufacturer's protocol (Qiagen).

Microarray Hybridization

Microarray experiments were performed as previously described [7], [9], [47]. Briefly, total RNA (15 µg), isolated from six independent, paired static and shear-activated T/C-28a2 chondrocyte samples, was reverse transcribed in the presence of random primers and aminoallyl(aa)-dUTP with Superscript II Reverse Transcriptase. The aa-dUTP-labeled cDNAs from sheared and static control samples were coupled to NHS-Cy5 and NHS-Cy3 (GE Healthcare), respectively. Cy5- and Cy-3-labeled targets were mixed, and co-hybridized on the microarray slides printed with a set of 39,936 human ESTs (TIGR 40K Human Set).

Microarray Data Analysis

Expression levels from individual genes were determined from the scanned microarray slides using TIGR_SpotFinder, and normalized with the total intensity algorithm of the TIGR Microarray Data Analysis System (MIDAS) [47], [48]. Data are presented as mean ± standard deviation (S.D.) using the TIGR Multiexperiment Viewer (MeV). Comparisons between the expression levels of static control and sheared genes were performed using the unpaired Student's t-test, and considered to be statistically significant if p<0.01. Further microarray data analysis involved only statistically significant genes. Differentially expressed genes were then identified using one-class Significance Analysis of Microarray (SAM) at a 5% false discovery rate (FDR) using TIGR MeV [47], [48]. Average linkage hierarchical clustering analysis with a Euclidean distance metric was performed using TIGR MeV [47], [48]. For pathway and functional category classification of the differentially expressed genes, we used the annotations publicly available from the National Center for Biotechnology Information LocusLink database (http://www.ncbi.nlm.nih.gov/LocusLink/), which classifies a gene according to molecular function, biologic process, and cellular component using Gene Ontology categories (http://www.geneontology.org/).

Quantitative Real-Time PCR (qRT-PCR)

qRT-PCR assays were performed on the iCycler iQ detection system (Biorad) using total RNA, the iScript one-step RT-PCR kit with SYBR green (Biorad) and primers. The GenBank accession numbers and forward (F-) and reverse (R-) primers are as follows:

Gremlin (NM_013372), F-5′-GTATGAGCCGCACAGCCTACA-3′; R-5′-CTCGCTTCAGGTATTTGCGCT-3′

RHOB (NM_004040), F-5′-GGTCCCCTGAGCATGCTTTTCTGA-3′; R-5′-GCCACACTCCCGCGCCAATCTC-3′

PAPP-A (NM_002581), F-5′-CAGAATGCACTGTTACCTGGA-3′; R-5′-GCTGATCCCAATTCTCTTTCA-3′

HIST1H2BD (NM_021063), F-5′-CAAAGAAGGG CTCCAAGAAG-3′; R-5′-TGGTGACGGCCTTGGTGC-3′

HIST3H2A (NM_033445), F-5′-CAGGGTGGCAAGGCGCGCGC-3′; R-5′-TCTTGGGCAGCAGTACGGCC-3′

COX-2 (NM_000963), F-5′-TGAGCATCTACGGTTTGCTG -3′; R-5′-AACTGCTCATCACCCCATTC-3′

Aggrecan (NM_013227), F-5′-ACTTCCGCTGGTCAGATGGA-3′; R-5′-TCTCGTGCCAGATCATCACC-3′

Interleukin-1β (NM_000576), F-5′-ATGGCAGAAGTACCTAAGCTCGC-3′; R-5′-ACACAAATTGCATGGTGAAGTCAGTT-3′

COL2A1 (NM_001844), F-5′-CTGGCTCCCAACACTGCCAACGTC-3′; R-5′-TCCTTTGGGTTTGCAACGGATTGT-3′

L-PGDS (NM_000954), F-5′-GCCTCGCCTCCAACTCGAGC-3′, R-5′-TGCAGCAGCATGGTTCGGGT-3′

GAPDH (NM_002046), F- 5′-CCACCCATGGCAAATTCCATGGCA-3; R-5′- TCTAGACGGCAGGTCAGGTCCACC-3′

GAPDH was used as internal control. Reaction mixtures were incubated at 50°C for 15 min followed by 95°C for 5 min, and then 35 PCR cycles were performed with the following temperature profile: 95°C 15 s, 58°C 30 s, 68°C 1 min, 77°C 20 s. Data were collected at the (77°C 20 s) step to remove possible fluorescent contribution from primer dimers [49].

Transient Transfection

In RNA interference assays, T/C-28a2 cells were transfected with 100 nM of an siRNA oligonucleotide sequence specific for L-PGDS (SC-41640) or control (SC-44240) siRNA (Santa Cruz). Transfected cells were allowed to recover for at least 12 h in growth medium, and then incubated overnight in serum-free medium containing 1% Nutridoma-SP before their exposure to shear or static conditions.

ROS Detection

ROS generation was detected by incubating T/C-28a2 chondrocytes with 5-(and-6)-carboxy-2′,7′-dichlorodihydrof​luoresceindiacetate (carboxy-H2DCFDA; 25 µM in D-PBS containing Ca2+/Mg2+) for 30 min at 37°C. Cells were next washed with D-PBS lacking Ca2+/Mg2+, detached from slides by mild trypsinization, re-suspended in D-PBS and examined by flow cytometry.

Supporting Information

Table S1.

Genes positively regulated by shear stress in human T/C28a2 chondrocytes.

doi:10.1371/journal.pone.0015174.s001

(PDF)

Table S2.

Genes positively regulated by shear stress in human T/C28a2 chondrocytes.

doi:10.1371/journal.pone.0015174.s002

(PDF)

Table S3.

Genes positively regulated by COX-2 in human T/C28a2 chondrocytes.

doi:10.1371/journal.pone.0015174.s003

(PDF)

Table S4.

Genes negatively regulated by COX-2 in human T/C28a2 chondrocytes.

doi:10.1371/journal.pone.0015174.s004

(PDF)

Author Contributions

Conceived and designed the experiments: FZ KK. Performed the experiments: FZ PW. Analyzed the data: FZ PW KK. Contributed reagents/materials/analysis tools: MBG NHL. Wrote the paper: FZ KK.

References

  1. 1. Buckwalter JA, Martin JA, Brown TD (2006) Perspectives on chondrocyte mechanobiology and osteoarthritis. Biorheology 43: 603–609.
  2. 2. Carter DR, Beaupre GS, Wong M, Smith RL, Andriacchi TP, et al. (2004) The mechanobiology of articular cartilage development and degeneration. Clin Orthop Relat Res 427S: S69–77.
  3. 3. Carter DR, Wong M (2003) Modelling cartilage mechanobiology. Philos Trans R Soc Lond B Biol Sci 358: 1461–1471.
  4. 4. Yokota H, Goldring MB, Sun HB (2003) CITED2-mediated regulation of MMP-1 and MMP-13 in human chondrocytes under flow shear. J Biol Chem 278: 47275–47280.
  5. 5. Mohtai M, Gupta MK, Donlon B, Ellison B, Cooke J, et al. (1996) Expression of Interleukin-6 in Osteoarthritic Chondrocytes and Effects of Fluid-Induced Shear on This Expression in Normal Human Chondrocytes In Vitro. J Ortho Res 14: 67–73.
  6. 6. Lee MS, Trindade MC, Ikenoue T, Schurman DJ, Goodman SB, et al. (2002) Effects of shear stress on nitric oxide and matrix protein gene expression in human osteoarthritic chondrocytes in vitro. J Orthop Res 20: 556–561.
  7. 7. Healy ZR, Lee NH, Gao X, Goldring MB, Talalay P, et al. (2005) Divergent responses of chondrocytes and endothelial cells to shear stress: cross-talk among COX-2, the phase 2 response, and apoptosis. Proc Natl Acad Sci U S A 102: 14010–14015.
  8. 8. Lee MS, Trindade MCD, Ikenoue T, Goodman SB, Schurman DJ, et al. (2003) Regulation of nitric oxide and bcl-2 expression by shear stress in human osteoarthritic chondrocytes in vitro. J Cell Biochem 90: 80–86.
  9. 9. Zhu F, Wang P, Kontrogianni-Konstantopoulos A, Konstantopoulos K (2010) Prostaglandin (PG)D2 and 15-deoxy-Δ12,14-PGJ2, but not PGE2, Mediate Shear-Induced Chondrocyte Apoptosis via Protein Kinase A-dependent Regulation of Polo-like Kinases. Cell Death Differentiation 17: 1325–1334.
  10. 10. Greim H (2003) Mechanistic and toxicokinetic data reducing uncertainty in risk assessment. Toxicol Lett 138: 1–8.
  11. 11. Amin AR, Attur M, Patel RN, Thakker GD, Marshall PJ, et al. (1997) Superinduction of cyclooxygenase-2 activity in human osteoarthritis-affected cartilage. Influence of nitric oxide. J Clin Invest 99: 1231–1237.
  12. 12. Martin JA, Buckwalter JA (2006) Post-traumatic osteoarthritis: the role of stress induced chondrocyte damage. Biorheology 43: 517–521.
  13. 13. Al-Shahrour F, Minguez P, Tarraga J, Montaner D, Alloza E, et al. (2006) BABELOMICS: a systems biology perspective in the functional annotation of genome-scale experiments. Nucleic Acids Res 34: W472–476.
  14. 14. Al-Shahrour F, Minguez P, Vaquerizas JM, Conde L, Dopazo J (2005) BABELOMICS: a suite of web tools for functional annotation and analysis of groups of genes in high-throughput experiments. Nucleic Acids Res 33: W460–464.
  15. 15. Karlsson C, Dehne T, Lindahl A, Brittberg M, Pruss A, et al. (2010) Genome-wide expression profiling reveals new candidate genes associated with osteoarthritis. Osteoarthritis and Cartilage 18: 581–592.
  16. 16. Kitamura T, Sekimata M, Kikuchi S, Homma Y (2005) Involvement of poly(ADP-ribose) polymerase 1 in ERBB2 expression in rheumatoid synovial cells. Am J Physiol Cell Physiol 289: C82–88.
  17. 17. Mahr S, Burmester GR, Hilke D, Gobel U, Grutzkau A, et al. (2006) Cis- and trans-acting gene regulation is associated with osteoarthritis. Am J Hum Genet 78: 793–803.
  18. 18. Mahr S, Muller-Hilke B (2007) Transcriptional activity of the RHOB gene is influenced by regulatory polymorphisms in its promoter region. Genomic Med 1: 125–128.
  19. 19. Kim CH, Won M, Choi CH, Ahn J, Kim BK, et al. (2010) Increase of RhoB in gamma-radiation-induced apoptosis is regulated by c-Jun N-terminal kinase in Jurkat T cells. Biochem Biophys Res Commun 391: 1182–1186.
  20. 20. Haas CS, Creighton CJ, Pi X, Maine I, Koch AE, et al. (2006) Identification of genes modulated in rheumatoid arthritis using complementary DNA microarray analysis of lymphoblastoid B cell lines from disease-discordant monozygotic twins. Arthritis Rheum 54: 2047–2060.
  21. 21. Joosten LA, Netea MG, Kim SH, Yoon DY, Oppers-Walgreen B, et al. (2006) IL-32, a proinflammatory cytokine in rheumatoid arthritis. Proc Natl Acad Sci U S A 103: 3298–3303.
  22. 22. Aigner T, Fundel K, Saas J, Gebhard PM, Haag J, et al. (2006) Large-scale gene expression profiling reveals major pathogenetic pathways of cartilage degeneration in osteoarthritis. Arthritis Rheum 54: 3533–3544.
  23. 23. Hu SI, Carozza M, Klein M, Nantermet P, Luk D, et al. (1998) Human HtrA, an evolutionarily conserved serine protease identified as a differentially expressed gene product in osteoarthritic cartilage. J Biol Chem 273: 34406–34412.
  24. 24. Tardif G, Hum D, Pelletier JP, Boileau C, Ranger P, et al. (2004) Differential gene expression and regulation of the bone morphogenetic protein antagonists follistatin and gremlin in normal and osteoarthritic human chondrocytes and synovial fibroblasts. Arthritis Rheum 50: 2521–2530.
  25. 25. Goldring MB, Otero M, Tsuchimochi K, Ijiri K, Li Y (2008) Defining the roles of inflammatory and anabolic cytokines in cartilage metabolism. Ann Rheum Dis 67: Suppl 3iii75–82.
  26. 26. Kobayashi M, Squires GR, Mousa A, Tanzer M, Zukor DJ, et al. (2005) Role of interleukin-1 and tumor necrosis factor alpha in matrix degradation of human osteoarthritic cartilage. Arthritis Rheum 52: 128–135.
  27. 27. Pattoli MA, MacMaster JF, Gregor KR, Burke JR (2005) Collagen and aggrecan degradation is blocked in interleukin-1-treated cartilage explants by an inhibitor of IkappaB kinase through suppression of metalloproteinase expression. J Pharmacol Exp Ther 315: 382–388.
  28. 28. Goldring MB, Birkhead J, Sandell LJ, Kimura T, Krane SM (1988) Interleukin 1 suppresses expression of cartilage-specific types II and IX collagens and increases types I and III collagens in human chondrocytes. J Clin Invest 82: 2026–2037.
  29. 29. Tiku ML, Shah R, Allison GT (2000) Evidence linking chondrocyte lipid peroxidation to cartilage matrix protein degradation. Possible role in cartilage aging and the pathogenesis of osteoarthritis. J Biol Chem 275: 20069–20076.
  30. 30. Zayed N, Li X, Chabane N, Benderdour M, Martel-Pelletier J, et al. (2008) Increased expression of lipocalin-type prostaglandin D2 synthase in osteoarthritic cartilage. Arthritis Res Ther 10: R146.
  31. 31. Aigner T, Vornehm SI, Zeiler G, Dudhia J, von der Mark K, et al. (1997) Suppression of cartilage matrix gene expression in upper zone chondrocytes of osteoarthritic cartilage. Arthritis Rheum 40: 562–569.
  32. 32. Aigner T, Zien A, Gehrsitz A, Gebhard PM, McKenna L (2001) Anabolic and catabolic gene expression pattern analysis in normal versus osteoarthritic cartilage using complementary DNA-array technology. Arthritis Rheum 44: 2777–2789.
  33. 33. Fukui N, Ikeda Y, Ohnuki T, Tanaka N, Hikita A, et al. (2008) Regional differences in chondrocyte metabolism in osteoarthritis: a detailed analysis by laser capture microdissection. Arthritis Rheum 58: 154–163.
  34. 34. Wang P, Zhu F, Lee NH, Konstantopoulos K (2010) SHEAR-INDUCED INTERLEUKIN-6 SYNTHESIS IN CHONDROCYTES: The roles of E prostanoid (EP)2 and EP3 in cAMP/Protein Kinase A- and PI3-K/Akt-dependent NF-κB activation. J Biol Chem 285: 24793–24804.
  35. 35. Kim SR, Park JH, Lee ME, Park JS, Park SC, et al. (2008) Selective COX-2 inhibitors modulate cellular senescence in human dermal fibroblasts in a catalytic activity-independent manner. Mech Ageing Dev 129: 706–713.
  36. 36. Liou JY, Deng WG, Gilroy DW, Shyue SK, Wu KK (2001) Colocalization and interaction of cyclooxygenase-2 with caveolin-1 in human fibroblasts. J Biol Chem 276: 34975–34982.
  37. 37. Siow HC (2004) Seasonal episodic paroxysmal hemicrania responding to cyclooxygenase-2 inhibitors. Cephalalgia 24: 414–415.
  38. 38. Scherer PE, Lewis RY, Volonte D, Engelman JA, Galbiati F, et al. (1997) Cell-type and tissue-specific expression of caveolin-2. Caveolins 1 and 2 co-localize and form a stable hetero-oligomeric complex in vivo. J Biol Chem 272: 29337–29346.
  39. 39. Dai SM, Shan ZZ, Nakamura H, Masuko-Hongo K, Kato T, et al. (2006) Catabolic stress induces features of chondrocyte senescence through overexpression of caveolin 1: possible involvement of caveolin 1-induced down-regulation of articular chondrocytes in the pathogenesis of osteoarthritis. Arthritis Rheum 54: 818–831.
  40. 40. Trifan OC, Smith RM, Thompson BD, Hla T (1999) Overexpression of cyclooxygenase-2 induces cell cycle arrest. Evidence for a prostaglandin-independent mechanism. J Biol Chem 274: 34141–34147.
  41. 41. Su C, Gao G, Schneider S, Helt C, Weiss C, et al. (2004) DNA damage induces downregulation of histone gene expression through the G1 checkpoint pathway. Embo J 23: 1133–1143.
  42. 42. Ullrich O, Grune T (2001) Proteasomal degradation of oxidatively damaged endogenous histones in K562 human leukemic cells. Free Radic Biol Med 31: 887–893.
  43. 43. Jung N, Kim DS, Kwon HY, Yi YW, Kim D, et al. (2000) Suppression of collagen-induced arthritis with histone H1. Scand J Rheumatol 29: 222–225.
  44. 44. Healy ZR, Zhu F, Stull JD, Konstantopoulos K (2008) Elucidation of the signaling network of COX-2 induction in sheared chondrocytes: COX-2 is induced via a Rac/MEKK1/MKK7/JNK2/c-Jun-C/EBPbeta-depe​ndentpathway. Am J Physiol Cell Physiol 294: C1146–1157.
  45. 45. Goldring MB (2004) Culture of immortalized chondrocytes and their use as models of chondrocyte function. Methods Mol Med 100: 37–52.
  46. 46. Wang P, Zhu F, Konstantopoulos K (2010) Prostaglandin E2 induces interleukin-6 expression in human chondrocytes via cAMP/protein kinase A- and phosphatidylinositol 3-kinase-dependent NF-kappaB activation. Am J Physiol Cell Physiol 298: C1445–1456.
  47. 47. Abulencia JP, Gaspard R, Healy ZR, Gaarde WA, Quackenbush J, et al. (2003) Shear-induced cyclooxygenase-2 via a JNK2/c-Jun-dependent pathway regulates prostaglandin receptor expression in chondrocytic cells. J Biol Chem 278: 28388–28394.
  48. 48. Saeed AI, Sharov V, White J, Li J, Liang W, et al. (2003) TM4: a free, open-source system for microarray data management and analysis. Biotechniques 34: 374–378.
  49. 49. Zhu F, Massana R, Not F, Marie D, Vaulot D (2005) Mapping of picoeucaryotes in marine ecosystems with quantitative PCR of the 18S rRNA gene. FEMS Microbiol Ecol 52: 79–92.
  50. 50. Moritani NH, Kubota S, Sugahara T, Takigawa M (2005) Comparable response of ccn1 with ccn2 genes upon arthritis: An in vitro evaluation with a human chondrocytic cell line stimulated by a set of cytokines. Cell Commun Signal 3: 6.
  51. 51. Zhang Q, Wu J, Cao Q, Xiao L, Wang L, et al. (2009) A critical role of Cyr61 in interleukin-17-dependent proliferation of fibroblast-like synoviocytes in rheumatoid arthritis. Arthritis Rheum 60: 3602–3612.
  52. 52. Dunn S, Kolomytkin OV, Waddell DD, Marino AA (2009) Hyaluronan-binding receptors: possible involvement in osteoarthritis. Mod Rheumatol 19: 151–155.
  53. 53. Yatsugi N, Tsukazaki T, Osaki M, Koji T, Yamashita S, et al. (2000) Apoptosis of articular chondrocytes in rheumatoid arthritis and osteoarthritis: correlation of apoptosis with degree of cartilage destruction and expression of apoptosis-related proteins of p53 and c-myc. J Orthop Sci 5: 150–156.
  54. 54. Kim HA, Lee YJ, Seong SC, Choe KW, Song YW (2000) Apoptotic chondrocyte death in human osteoarthritis. J Rheumatol 27: 455–462.
  55. 55. Valdes AM, Van Oene M, Hart DJ, Surdulescu GL, Loughlin J, et al. (2006) Reproducible genetic associations between candidate genes and clinical knee osteoarthritis in men and women. Arthritis Rheum 54: 533–539.
  56. 56. Meng J, Ma X, Ma D, Xu C (2005) Microarray analysis of differential gene expression in temporomandibular joint condylar cartilage after experimentally induced osteoarthritis. Osteoarthritis Cartilage 13: 1115–1125.
  57. 57. Scanzello CR, Plaas A, Crow MK (2008) Innate immune system activation in osteoarthritis: is osteoarthritis a chronic wound? Curr Opin Rheumatol 20: 565–572.
  58. 58. Backman JT, Siegle I, Fritz P (1998) Immunohistochemical localization of metallothionein in synovial tissue of patients with chronic inflammatory and degenerative joint disease. Virchows Arch 433: 153–160.
  59. 59. Gobezie R, Kho A, Krastins B, Sarracino DA, Thornhill TS, et al. (2007) High abundance synovial fluid proteome: distinct profiles in health and osteoarthritis. Arthritis Res Ther 9: R36.
  60. 60. Welch ID, Cowan MF, Beier F, Underhill TM (2009) The retinoic acid binding protein CRABP2 is increased in murine models of degenerative joint disease. Arthritis Res Ther 11: R14.
  61. 61. Yamane S, Ishida S, Hanamoto Y, Kumagai K, Masuda R, et al. (2008) Proinflammatory role of amphiregulin, an epidermal growth factor family member whose expression is augmented in rheumatoid arthritis patients. J Inflamm (Lond) 5: 5.
  62. 62. Lin AC, Seeto BL, Bartoszko JM, Khoury MA, Whetstone H, et al. (2009) Modulating hedgehog signaling can attenuate the severity of osteoarthritis. Nat Med 15: 1421–1425.
  63. 63. Hsieh YS, Yang SF, Lue KH, Chu SC, Li TJ, et al. (2007) Upregulation of urokinase-type plasminogen activator and inhibitor and gelatinase expression via 3 mitogen-activated protein kinases and PI3K pathways during the early development of osteoarthritis. J Rheumatol 34: 785–793.
  64. 64. Appleton CT, Pitelka V, Henry J, Beier F (2007) Global analyses of gene expression in early experimental osteoarthritis. Arthritis Rheum 56: 1854–1868.
  65. 65. Wang WZ, Guo X, Duan C, Ma WJ, Zhang YG, et al. (2009) Comparative analysis of gene expression profiles between the normal human cartilage and the one with endemic osteoarthritis. Osteoarthritis Cartilage 17: 83–90.
  66. 66. Pfander D, Cramer T, Swoboda B (2005) Hypoxia and HIF-1alpha in osteoarthritis. Int Orthop 29: 6–9.
  67. 67. Hansen IB, Ellingsen T, Hornung N, Poulsen JH, Lottenburger T, et al. (2006) Plasma level of CXC-chemokine CXCL12 is increased in rheumatoid arthritis and is independent of disease activity and methotrexate treatment. J Rheumatol 33: 1754–1759.
  68. 68. Bramlage CP, Haupl T, Kaps C, Ungethum U, Krenn V, et al. (2006) Decrease in expression of bone morphogenetic proteins 4 and 5 in synovial tissue of patients with osteoarthritis and rheumatoid arthritis. Arthritis Res Ther 8: R58.
  69. 69. Nzeako UC, Guicciardi ME, Yoon JH, Bronk SF, Gores GJ (2002) COX-2 inhibits Fas-mediated apoptosis in cholangiocarcinoma cells. Hepatology 35: 552–559.
  70. 70. Ali K, Lund-Katz S, Lawson J, Phillips MC, Rader DJ (2008) Structure-function properties of the apoE-dependent COX-2 pathway in vascular smooth muscle cells. Atherosclerosis 196: 201–209.