Contribution of calcium-containing crystals to cartilage degradation and synovial inflammation in osteoarthritis

Osteoarthritis and Cartilage (2009) 17, 1333e1340 ª 2009 Osteoarthritis Research Society International. Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.joca.2009.04.022

International Cartilage Repair Society

Contribution of calcium-containing crystals to cartilage degradation and synovial inflammation in osteoarthritis Y. Z. Liu*, A. P. Jackson and S. D. Cosgrove AstraZeneca R&D Charnwood, Bakewell Road, Loughborough, UK Summary Objectives: The role of calcium phosphate and pyrophosphate crystals in osteoarthritis (OA) is unclear: are they a symptom of the damage that occurs to the joint or a key intermediate in the progression of inflammation and joint damage that occurs in OA? The proinflammatory and catabolic response of synthetic calcium phosphate and pyrophosphate crystals and crystals extracted from human osteoarthritic knee cartilage has been investigated. The crystal forms eliciting a response have been characterised allowing a comparison of the biological effects of synthetic and native calcium crystals on human osteoarthritic chondrocytes and synoviocytes to be carried out. Methods: Calcium phosphate and pyrophosphate crystals were synthesised in vitro and their crystal forms characterised by X-ray powder diffraction (XRPD). The inorganic crystalline material present in human osteoarthritic cartilage was extracted and its structural composition elucidated by XRPD. These crystals were applied to human primary osteoarthritic chondrocytes and synoviocytes and the production of proinflammatory and catabolic mediators measured. Results: The crystals extracted from human osteoarthritic knee cartilage were identified as consisting of a mixture of monoclinic and triclinic calcium pyrophosphate dihydrate (m-CPPD and t-CPPD). These crystals elicited an inflammatory and catabolic response in human primary osteoarthritic chondrocytes and synoviocytes as measured by an increase in nitric oxide (NO), matrix metalloproteinase 13 (MMP-13) and prostaglandin E2 (PGE2) production. NO, MMP-13 and PGE2 production was also increased when the synthetic calcium hydrogen phosphate dihydrate (DCPD) and calcium pyrophosphate hydrates were applied to the cells. Conclusions: Crystals extracted from human osteoarthritic knee cartilage induce the production of proinflammatory and catabolic mediators (NO, MMP-13 and PGE2) in human primary chondrocytes and synoviocytes. Synthetic calcium phosphate and pyrophosphate crystals elicit a similar response in those cells. Our findings suggest that these crystals could contribute to cartilage degradation and synovitis in OA. ª 2009 Osteoarthritis Research Society International. Published by Elsevier Ltd. All rights reserved. Key words: Osteoarthritis, CPPD, CPPT, BCP, DCPD, MMP-13, NO, PGE2, Chondrocyte, Synoviocyte, Pyrophosphate.

CPPD exists in two forms: triclinic and monoclinic, generally abbreviated to t-CPPD and m-CPPD, respectively. A tetrahydrate also exists and is dimorphic, denoted mCPPTa and m-CPPTb16. Characterisation of these forms in the context of their in vivo prevalence has been focussed primarily on microscopy and X-ray powder diffraction (XRPD) studies17,18. Crystal structures have been determined for t-CPPD19, m-CPPTa20 and m-CPPTb16. In addition, the crystal growth and dissolution kinetics of the calcium pyrophosphate hydrate phases have been investigated21e25, as has the dehydration26. Analysis of synovial fluid and menisci often reveals the presence of both triclinic and monoclinic forms of CPPD, t-CPPD predominating27,28. BCP is a broad term used to describe a number of crystallographically and chemically distinct phases. Therefore BCP encompasses hydroxyapatite Ca10(PO4)6(OH)2, octacalcium phosphate Ca8(HPO4)2(PO4)45H2O and tricalcium phosphate Ca3(PO4)2. BCP is also synonymous with biphasic calcium phosphate, a mixture of hydroxyapatite and tricalcium phosphate29. In general, BCP is often insufficiently characterised in the literature. Other phosphate phases of interest are dicalcium phosphate dihydrate (DCPD, CaHPO42H2O) and its anhydrate (monatite, CaHPO4). These phases are known to be intermediates in the synthesis of hydroxyapatite30,31. DCPD has been observed in synovial tissues and cartilage, and in menisci at necropsy32,33.

Introduction Osteoarthritis (OA) is characterised by articular cartilage degeneration, synovitis and subchondral bone remodelling1. Chondrocytes play a key role in cartilage degradation by releasing matrix metalloproteinases (MMPs) and nitric oxide (NO)2,3. Synovial inflammation is implicated in the signs and symptoms of OA4e6. Synoviocytes release inflammatory mediators, including prostaglandin E2 (PGE2), contributing to signs and symptoms of the disease. Calcium pyrophosphate dihydrate (CPPD) and basic calcium phosphate (BCP) are commonly deposited in osteoarthritic joints. These crystals have been found in the synovial fluid of 60% of all patients with knee effusions in OA and over 90% of a small group of patients with grade-4 OA7. They are associated with cartilage lesion and a severe form of OA8e10. These crystals have been demonstrated to induce inflammatory response in cells11e13. Further, intra-articular injection of CPPD crystals can exacerbate the OA progression in rabbit knee joints14. On the basis of those observations, CPPD and BCP crystals have been assumed to contribute to OA disease progression15. *Address correspondence and reprint requests to: Dr Yu-Zhen Liu, Department of Bioscience, AstraZeneca R&D Charnwood, Bakewell Road, Loughborough LE11 5RH, UK. Tel: 44-1509645815; Fax: 44-1509-645557; E-mail: yu-zhen.liu@ astrazeneca.com Received 6 February 2009; revision accepted 28 April 2009.

1333

1334

Y. Z. Liu et al.: Contribution of calcium-containing crystals in OA

Most previous in vitro or in vivo studies with CPPD and BCP crystals used the synthetic crystal materials. Studies of the biological effects of synthetic calcium crystals must also take into account the existence of several crystalline phases in natural pathological conditions. The effects of endogenous crystals from diseased joints have not been reported. This work was designed to characterise the forms of crystals from crystalline osteoarthritic joints and compare the proinflammatory and catabolic effects of synthetic and native calcium crystals in human osteoarthritic chondrocytes and synoviocytes. Materials and methods SYNTHESIS OF CALCIUM-CONTAINING CRYSTALS Attempted synthesis of BCP followed the method of McCarthy et al.34. 50 ml 0.2 M NaPi, pH 6.7 was added to 42 ml 0.2 M CaCl2 and stirred for 1 h at room temperature. The precipitate was harvested by centrifugation and washed with water. The crystals were resuspended in 6.7 ml water and 67 ml of 0.5 M HCl added. The pH was adjusted to between 8.5 and 10 and the suspension stirred overnight. It was then washed with water, sterilised by washing in 70% ethanol and freeze dried. Polymorphs of CPPD were synthesised using methods described by Cheng and Pritzer35. For t-CPPD, the initial solution contained 1 mM CaCl2, 0.1 mM MgCl2, 0.24 mM NaPPi and 140 mM NaCl, pH 7.4. For m-CPPD, the initial solution contained 1 mM CaCl2, 0.8 mM MgCl2, 0.34 mM NaPPi and 140 mM NaCl, pH 7.4. m-CPPTb was prepared using the starting solution containing 1.6 mM CaCl2, 0.48 mM NaPPi and 140 mM NaCl, pH 7.4. One litre of each of the above solutions was prepared and shaken at 37 C for 3 weeks. The precipitate was harvested by centrifugation, washed with water, sterilised by washing in 70% ethanol and freeze dried for approximately 16 h. Where indicated, synthetic crystals were reduced in size through crushing with a pestle and mortar. This process tended to form loose aggregates that were dispersed by sonicating for 30 s. Initially samples were sterilised and rendered lipopolysaccharide (LPS) free by heating at 200 C for 2 h36. This baking process was found to change the crystal form hence was only carried out for the indicated samples. PURIFICATION OF CRYSTALS FROM OSTEOARTHRITIC KNEE JOINTS Crystals were extracted from the osteoarthritic cartilage using a method based on Swan et al.37. Human OA cartilage samples were obtained from patients undergoing total knee replacement operations for OA. Five human OA cartilage samples had been identified visually as containing crystals and used for crystal purification. All five patients had no history of acute pseudogout. Chondrocalcinosis was present on radiographs from two patients. The cartilage samples were chopped finely and resuspended in 2.5 volumes (w/v) of 100 mM 2-(N-morpholino)ethanesulfonic acid (MES), 1 mM dithiothreitol (DTT), pH 6.5. Papain was added to 10 mg/ml and the suspension incubated at 37 C overnight. The suspension was filtered and the filtrate was spun at 4000 rpm for 5 min and the supernatant discarded. The pellet was washed with water followed by 1% sodium hypochlorate for 2 min and a further water wash. The pellet was finally resuspended in 70% ethanol, and freeze dried for 4 h. Yields of crystals obtained by this method were up to 10 mg crystals per gram wet weight tissue. Processing normal cartilage in parallel with OA tissue did not generate any detectable crystalline material. CHARACTERISATION OF SYNTHETIC AND PURIFIED ENDOGENOUS CRYSTALS The synthetic and extracted samples were characterised using a PANalytical CubiX PRO X-ray diffractometer of wavelength 1.5418 A˚ (Cu source, voltage 45 kV, filament emission 40 mA). Samples were scanned from 2 to 40 2q using a 0.02 step width and a 100 s time count with an X’celerator detector (active length 2.54 2q). As most previous diffraction studies have cited diffraction reflections in d-spacing, rather than 2q, we continue this convention. It can be noted that d-spacing (d ), and theta (q), are interrelated by the Bragg equation, nl ¼ 2d sin q, where l is the wavelength of the X-rays.

CULTURE AND TREATMENT OF PRIMARY CHONDROCYTES AND SYNOVIOCYTES Human OA cartilage and synovium were obtained from patients undergoing total knee replacement operations. The patients were taking pain relief medication (e.g., paracetamol or ibuprofen). Full ethical consent was obtained from all donors and families.

To isolate primary chondrocytes, cartilage were stripped from the femoral condyles and tibial plateaux, cut into fine pieces and digested in growth media [Dulbecco’s modified eagles medium supplemented with 10% foetal calf serum (FCS), penicillin (100 units/ml), streptomycin (100 mg/ml), 2 mM L-glutamine] containing 1 mg/ml collagenase II (Worthington Biochemicals, CLS-2). After overnight incubation at 37 C, the sample was filtered to remove undigested cartilage and chondrocyte cells pelleted at 500g for 10 min. Cells were washed with growth medium and pelleted again. Cells were then grown in monolayer without passaging (5e7 days from isolation) before being utilised in experiments. A significant proportion of the cell preparations from OA cartilage contain crystals. Only the cells free of crystals were used in this study. The cells were dissociated from growing flasks by incubating in a small volume of 1 trypsin ethylenediaminetetraacetic acid (EDTA) for 15 min at 37 C after removing the growth medium and washing with phosphate buffered saline (PBS) (no Ca2þ, no Mg2þ). The cells were then harvested and plated at a density of 50,000 cells/well in growth medium in 96-well cell culture plates for experiment. For preparation of primary synoviocytes, the synovium was cut into small pieces and digested in growth media [Roswell Park Memorial Institute medium (RPMI) 1640 supplemented with 10% FCS, penicillin (100 units/ml), streptomycin (100 mg/ml), 2 mM L-glutamine] containing 1 mg/ml collagenase II. After the 2-h digestion at 37 C, the cell suspension was filtered to remove any undigested tissue. The filtered cell suspension was centrifuged at 500g for 10 min. The cells were resuspended into 10 ml of RPMI 1640 (with 2 mM L-glutamine) supplemented with 10% heat-inactivated FCS. Flow cytometry analysis of the OA synovial cells showed that the cell preparation was comprised mainly of fibroblast-like synoviocytes with 2e7% macrophages. Primary cells were then plated in a cell culture flask overnight before being utilised in experiments. The cells were dissociated from growing flasks by a gentle scraping using a cell scraper after removing the growth medium and washing with PBS (no Ca2þ, no Mg2þ). The cells were then harvested and plated at a density of 50,000 cells/well in growth medium in 96-well cell culture plates for experiment. Following plating in 96-well cell culture plates, the cells were incubated overnight at 37 C. The cell growth medium was then replaced with the fresh growth medium an hour before treatment with the crystals, interleukin (IL)-1b (CalBiochem), Pam3CSK (Invivogen) or anti-toll-like receptor 2 (TLR2) antibody (clone T2.5, Hycult Biotechnology). At the end of the experiment, the conditioned media from the cells were harvested for subsequent analyses and the cell viability tested.

DETERMINATION OF MMP-13 PROTEIN AND PGE2 BY ENZYME-LINKED IMMUNOSORBENT ASSAY (ELISA) Conditioned media from primary chondrocytes and synoviocytes were assayed for MMP-13 and PGE2 using commercially available ELISA kits (MMP-13 ELISA kit from Amersham Biosciences and PGE2 ELISA kit from R&D Systems).

DETERMINATION OF NITRITE LEVEL NO accumulation was measured by the Griess reaction. Fifty micro litres of conditioned media from primary chondrocytes and synoviocytes was mixed with 150 ml freshly prepared Griess reagent, comprising equal quantities of 0.1% (1-naphthyl)ethylenediamide dihydrochloride (Sigma) and 1% sulfanilamide in 5% orthophosphoric acid (Sigma). The nitrite concentrations were immediately determined by measuring absorbance at 550 nm.

CELL VIABILITY ASSAY Cell viability was assessed by the WST-1 assay (cell proliferation reagent WST from Roche). WST reagent was added to primary chondrocytes and synoviocytes at the end of experiments. The cells were incubated for 10 min at 37 C. Cell activity was determined by measuring absorbance at 440 nm.

STATISTICAL ANALYSIS All data are presented as mean  S.E.M. The n numbers refer to the sample number of OA patients tested. Where applicable, the t test for paired samples was performed for statistical analysis. P-values < 0.05 were considered significant.

Results CHARACTERISATION OF SYNTHETIC CALCIUM PHOSPHATE AND PYROPHOSPHATE CRYSTALS

Figure 1(a) and (b) shows the XRPD diffractogram of attempted synthesis of pure t-CPPD and m-CPPD, respectively.

1335

Osteoarthritis and Cartilage Vol. 17, No. 10

2000

a

*

*

*

1000

Counts

0 4000

b

*

*

*

2000 0 10000

c

*

5000

*

0 2000

d

*

*

* *

1000 0 40.0 20.0

8.0

6.0

4.0

d-spacing [Å] Fig. 1. X-ray diffractograms of (a) t-CPPD, (b) m-CPPD, (c) m-CPPTb and (d) DCPD.

The diffractogram for m-CPPD shows distinctive strong reflections at 7.29, 4.59 and 3.03 A˚ (asterisked) whilst the pattern for the triclinic form shows reflections at 8.02, 6.99 and 3.23 A˚ (asterisked). These reflections are consistent with those described by other workers and confirm successful synthesis of the pure CPPD forms28,38. Attempts to synthesise CPPT were also successful as shown by the X-ray diffraction pattern of the unbaked material [Fig. 1(c)]. The distinctive strong reflections at 11.33, 3.79 and 3.38 A˚ (asterisked) in the diffraction pattern are consistent with the b form of calcium pyrophosphate tetrahydrate16. However, baking the sample resulted in a phase change. Assessment of this new phase against potential phases did not provide a definitive confirmation of identity, though dehydration or reduction to calcium phosphate are probable mechanisms26,39. The specific identity of the crystals from the attempted BCP synthesis was confirmed by XRPD to be DCPD; the intense reflections at 7.58, 4.23 and 3.05 A˚ [asterisked in Fig. 1(d)] are consistent with literature values40. Baking the sample gave an X-ray diffractogram consistent with monatite, CaHPO441. Crystal samples are commonly baked to remove any LPS contamination that could lead to signalling through the TLR436. As described above, the process of baking crystals can change the physical form of the crystals, adding to the confusion as to the identity of the specific physical forms affecting cellular signalling. It became apparent that there was no significant level of LPS contamination in the unbaked crystal samples, as measured using a TLR4 reporter cell line. Optical microscopy reveals crystals were heterogeneous in size ranging from >1 mm (the optical limit of the microscope used) to 50 mm (Fig. 2). m-CPPD [Fig. 2(a)] and t-CPPD [Fig. 2(b)] were both ordered and needle like structures. DCPD [Fig. 2(c)] and m-CPPTb [Fig. 2(d)] were less well ordered. Crushing the crystals with a pestle and mortar effectively reduced the crystal size as exemplified by crushed m-CPPTb [Fig. 2(e)]. CHARACTERISATION OF CRYSTALS FROM THE OSTEOARTHRITIC KNEE JOINT

We observed that some OA articular cartilage samples contained visible crystals. Figure 3 shows the picture of

a representative crystalline cartilage sample from OA knee. The crystals appeared as white precipitates on the cartilage surface. We have also observed that intracellular crystals were present in OA synoviocytes, indicating that phagocytosis of crystals occurred in the OA joint (Fig. 4). The crystalline material extracted from OA cartilage [Fig. 2(f)] resembled the needle like t-CPPD and m-CPPD although the crystals are smaller (up to 10 mm) than the synthetic material. XRPD was performed to obtain more definitive proof of the identity of this crystalline material. Figure 5 shows the XRPD pattern of extracted material. Remarkably, the ratio of m-CCPD to t-CPPD was extremely consistent across all four donors. It should be noted that the extraction technique is reported not to affect the extraction of either hydroxyapatite or CPPD from cartilage37. The diffractograms of the extracted material were interrogated for any evidence of other crystalline phases other than CPPD, such as any BCP phases. All reflections in the diffractograms of the extracted materials could be accounted for by either m-CCPD or t-CPPD; no BCP was present. PROINFLAMMATORY AND CATABOLIC EFFECTS OF SYNTHETIC CALCIUM PHOSPHATE AND PYROPHOSPHATE CRYSTALS IN PRIMARY CHONDROCYTES AND SYNOVIOCYTES

Here, we showed that the catabolic activity of some synthetic crystal forms depends on the size; crushing m-CPPTb crystals increased the activity (Fig. 6). Our synthetic m-CPPD and t-CPPD crystals induced NO and MMP-13 release in human OA chondrocytes in a dose-dependent manner [Fig. 7(a and b)]. Synthetic DCPD crystals also showed a catabolic activity in human OA chondrocytes (Fig. 7). None of the crystals showed cytotoxic effects on the cell by measuring WST-1 activity. The observed crystal-induced response in the cells appeared specific as there were no induction of NO in the cells in response to other particles including iron, SiO2, ground glass and melanine (data not shown). A recent study reported that synthetic BCP induces PGE2 synthesis in human OA synovial fibroblasts42. In this study, DCPD crystals induced up to 70-fold upregulation of PGE2 release in human OA synoviocytes [Fig. 7(c)]. The synthetic pyrophosphate crystals also strongly increased PGE2 release in the cells.

1336

Y. Z. Liu et al.: Contribution of calcium-containing crystals in OA

Fig. 2. Micrographs of the calcium-containing crystals that have been generated using the methods described in the text: (a), m-CPPD; (b), t-CPPD; (c), DCPD; (d), m-CPPTb; (e), crushed m-CPPTb and (f), extract from OA cartilage. The scale bar represents 25 mm.

The extracted crystals from OA crystalline cartilage samples were tested in human OA chondrocytes and synoviocytes. The endogenous crystals induced NO, MMP-13 and PGE2 production in OA chondrocytes in a dose-dependent manner (Fig. 8). Similarly, the crystals increased PGE2 release in OA synoviocytes [Fig. 8(d)]. Interestingly, the response is similar to the well established potent catabolic cytokine IL-1b (Fig. 8). In order to exclude that the catabolic

response is mediated by other biological materials co-extracted with the crystals from OA cartilage, we processed OA cartilage with little crystal content and normal cartilage through crystal purification procedure and tested the end product. Neither of these samples showed similar catabolic response to the ones from crystalline OA cartilage. We have also confirmed that the physical interaction between the crystals and cells was required for the catabolic response by treating cells growing on the permeable membrane filters with crystals basolaterally without direct contact. Under those conditions, no catabolic response

Fig. 3. An image of human femoral condyles that has been used in this study highlighting the cartilage and the exposed bone. The arrow indicates visible crystal deposition on the surface of the cartilage.

Fig. 4. An image of primary human OA synovoial cells derived from the OA synovium. The inset is a view of the cell cytoplasm that contains crystals.

PROINFLAMMATORY AND CATABOLIC EFFECTS OF CRYSTALS FROM THE OSTEOARTHRITIC KNEE JOINT IN PRIMARY CHONDROCYTES AND SYNOVIOCYTES

1337

Counts

Osteoarthritis and Cartilage Vol. 17, No. 10

2000 1000 0 600 400 200 0 800 600 400 200 0

Donor 1

t

t t m

m

m

Donor 2

Donor 3

2000 Donor 4

1000 0

40.0 20.0

8.0

6.0

4.0

d-spacing [Å] Fig. 5. X-ray diffractograms of material extracted from the OA knee cartilage from four donors.

was observed. The crystals did not show cytotoxic effects (data not shown). THE ROLE OF TLR2 IN CALCIUM CRYSTAL-MEDIATED NO INDUCTION IN CHONDROCYTES

Synthetic m-CPPD and triclinic monosodium urate (t-MSU) crystals have been reported to use TLR2-mediate signalling in bovine chondrocytes to trigger NO production36. The same study also reported that TLR2 receptor is expressed in human primary articular chondrocytes; thus the crystals may induce catabolic responses via TLR2 in human OA chondrocytes. To explore the role of TLR2 in crystal-mediated response in OA chondrocytes, we used a TLR2 blocking antibody and examined the antibody mediated inhibition on both crystal-induced and TLR2 synthetic ligand Pam3CSK induced NO production (Fig. 9). Pam3CSK induced NO production, indicating the presence of functional TLR2 in human OA chondrocytes. As expected, the TLR2 blocking antibody significantly inhibited the response mediated by Pam3CSK. However, the blocking antibody failed to reduce the response mediated by either the endogenous crystals from OA knee cartilage or 25 Crushed

Nitrite (µM)

20

Uncrushed

15 10 5 0

V

0.125

0.25

0.5

1

m-CPPTbeta (mg/ml) Fig. 6. Crushing of synthetic m-CPPTb enhanced the catabolic effects in primary OA chondrocytes. Primary OA chondrocytes from three different donors were treated with crushed or uncrushed mCPPTb crystals for 24 h. The production of NO (n ¼ 3) over the 24 h treatment period was determined. Bars represent mean  S.E.M.

our synthetic crystals (Fig. 9). We also tested our synthetic t-CPPD, m-CPPD and DCPD crystals in our human TLR2 or TLR2/6 overexpressed recombinant HEK 293 cells. The crystals did not activate the human TLR2 receptors (data not shown). Our data suggest that CPPD crystals are unlikely to be ligands of human TLR2 receptors.

Discussion Synthetic calcium phosphate and pyrophosphate crystals have been assessed and compared with previous studies reported in the literature. We have sought to be very precise in our characterisation since we found the literature characterisation confusing. For example BCP is a general term and we have been unable to obtain a comprehensive list of phases that term includes, and characterisation by XRPD is variable. Furthermore due to the similarity of CPPD and CPPT nomenclature, some confusion has occurred; for example the incorrect description of o-CPPT as being the dihydrate38. The literature is further confused by the use of the term ‘‘dehydrate’’ for ‘‘dihydrate’’43,44. We believe this may be a consequence of automated spell checkers. It seems that we are still some way away from fully understanding the ‘‘complex crystal chemistry of the calcium pyrophosphates’’39. In this study we confirmed that we were successful in producing synthetic samples of m-CPPD, t-CPPD and mCPPTb. We were unsuccessful in confirming the identity of material formed by baking m-CPPTb. The formation of DCPD instead of one of the more common forms of BCP is potentially as a result of pH drift during the synthesis45. The synthetic crystals were larger than those isolated from OA knee tissue. Crushing of m-CPPTb crystals generated a sample more akin to the size of the native CPPD crystals. Biological activity could only be seen when the m-CPPTb crystals were reduced in size, hence the size of the crystals formed can be a determinant in their biological activity. This response to size is not simply a particle size effect; addition of inert material consisting of differing chemical and physical compositions but a similar particle size to the calcium crystals, did not generate a biological response. In contrast, the observed cellular response to CPPD was not dependent on particle size. We have frequently observed that some OA articular cartilage samples contain visible crystals. The extracted

1338

Y. Z. Liu et al.: Contribution of calcium-containing crystals in OA

25

Nitrite (µM)

20

a

m-CPPTbeta t-CPPD m-CPPD DCPD

25 Crystals IL-1

20

Nitrite (µM)

a

15 10

15 10 5

5

0

0

v

0.0625

0.125

0.25

0.5

V

1

0.01

Crystals (mg/ml) 120

m-CPPD 80

DCPD

60 40

1

80 Crystals IL-1

60 50 40 30 20

0

V

0.01

0 v

0.0625

0.125

0.25

0.5

0.03

0.1

0.3

1

Crystals (mg/ml)

1

Crystals (mg/ml)

c

35000 30000

100000 90000

m-CPPTbeta

80000

t-CPPD

70000

m-CPPD

60000

DCPD

PGE2 (pg/ml)

PGE2 (pg/ml)

0.3

10

20

c

0.1

70

t-CPPD

100

MMP-13 (ng/ml)

b

m-CPPTbeta

MMP-13 (ng/ml)

b

0.03

Crystals (mg/ml)

50000 40000

Crystals

25000

IL-1

20000 15000 10000 5000

30000

0

20000

V

0.01

10000

0.03

0.1

0.3

1

Crystals (mg/ml)

0 v

0.25

0.5

1

Crystals (mg/ml)

d

60000 Crstals

50000

IL-1

PGE2 (pg/ml)

Fig. 7. Induction of NO (a) and MMP-13 (b) in primary OA chondrocytes and PGE2 (c) release in primary OA synoviocytes by synthetic crushed m-CPPTb, t-CPPD, m-CPPD and DCPD crystals. The cells from three to five different donors were treated with crushed m-CPPTb, t-CPPD, m-CPPD or DCPD crystals for 24 h. The production of NO (n ¼ 5), MMP-13 (n ¼ 3) and PGE2 (n ¼ 5) over the 24 h treatment period was determined. Bars represent mean  S.E.M.

40000 30000 20000 10000 0

material from four donors characterised, gave very similar ratios of m-CPPD and t-CPPD. This suggests either that a biological mechanism is at work resulting in this, or that the extraction process gives rise to this ratio. Whilst not explicit in Mandel’s inference of a crystallographic relationship between the two forms38, the possibility of epitaxial growth as an explanation of the ratio is a reasonable one; epitaxy refers to the situation where a growth face of one crystal is sufficiently similar with respect to atomic positions to the growth face of a second phase, that the second phase grows concurrently. In the case of CPPD, as one hydrate grows, the other hydrate phase grows with it, thus resulting

V

0.01

0.1

1

Crystals (mg/ml)

Fig. 8. Induction of NO (a), MMP-13 (b) and PGE2 (c) release in primary OA chondrocytes and PGE2 (d) release in primary OA synoviocytes by purified endogenous OA knee joint crystals. The crystals were prepared from five different OA crystalline cartilage samples and tested. The cells from four to eight different donors were treated with the purified crystals or IL-1b (0.1 ng/ml) for 24 h. The production of NO (n ¼ 8), MMP-13 (n ¼ 5) and PGE 2 (n ¼ 4) from chondrocytes and PGE2 (n ¼ 6) from synoviocytes over the 24 h treatment period was determined. Bars represent mean  S.E.M.

1339

Osteoarthritis and Cartilage Vol. 17, No. 10

Nitrite (µM)

40 35

IgG1

30

Anti-TLR2

25 20

*

15 10 5

D 3

D 2

D 1

D 4 m D -C PP 5 Tb et a t-C PP D m -C PP D D C PD

Pa m

3C

SK

V

0

Fig. 9. Calcium phosphate and pyrophosphate crystal-induced NO release is not TLR2 dependent. Primary OA chondrocytes from three different donors were pre-treated with a TLR2 blocking antibody T2.5 at 100 mg/ml or IgG1 control for an hour before addition of synthetic and endogenous crystals (0.3 mg/ml) or the TLR2 ligand Pam3CSK (1 mg/ml). The endogenous crystals were prepared from five different OA crystalline cartilage samples and named as D1, D2, D3, D4 and D5. Following 24 h treatment the production of NO (n ¼ 3) over the treatment period was determined. Bars represent mean  S.E.M. *P < 0.05, significantly different from corresponding IgG1 value.

in a regularly observed ratio between the phases. No BCP crystals were detected in the purified OA cartilage crystal samples. This contrasts with reports that BCP crystals are frequently detected in OA synovial fluid46. It suggests that BCP crystals in synovial fluid are probably derived from bone remodelling in the OA joint. Synthetic calcium phosphate and pyrophosphate crystals have been reported to induce proinflammatory response in various cell types including human synovial fibroblasts, monocytes and bovine chondrocytes. However, limited studies have been carried out in primary human OA chondrocytes. Our studies further confirm the biological activity of calcium crystals in human primary OA chondrocytes and synoviacytes. Of interest is the activity of calcium-containing crystals depending on the size. For example, crushing of m-CPPTb crystals leads to the increase of activity. The underlying mechanism for the size-dependent effects is unclear. A previous study has reported the size-dependent calcium pyrophosphate crystal phagocytosis in polymorphonuclear leukocytes47. Since crystalecell interactions and inflammatory responses are influenced by crystal size and morphology, standardisation of methods of preparing and characterising calcium crystals is, therefore, important, particularly when analysing and comparing results from different research laboratories. A strong catabolic effect in human OA chondrocytes has been observed with the naturally occurring crystals. In our laboratory, we have tested a number of known catabolic factors (such as IL-1, IL-17, IL-18, tumour necrosis factor-alpha, oncostatin M, transforming growth factor-beta and various fibronectin fragments) in OA cells and have noted that IL-1a and IL-1b are the most potent and strongest stimuli. In this study, we have found that CPPD crystals can achieve similar efficacy to IL-1 in inducing NO, MMP-13 and PGE2 release. Some progress has been made in understanding the molecular mechanisms of calcium crystal-mediated cell activation48. Intracellular mitogen-activated protein kinase cascade appears to mediate the crystal-induced cellular

response13,49. Crystals have been shown to be phagocytosed by cultured synoviocyte and consequently stimulate the release of MMPs and prostaglandins50. Our finding that OA synovial cells from OA joints contain crystals supports that the crystal phagocytosis can occur in vivo. The direct effects of CPPD crystals on cells may also involve an interaction between crystals and cell surface receptors. It has been reported that CPPD crystals use TLR2 receptors in bovine chondrocytes to trigger NO production36. We have examined the role of TLR2 receptors in crystal-mediated cell activation in human primary cell system. In contrast to the previous finding, our data do not support a role of TLR2 in mediating the response induced by both synthetic and endogenous crystals. The reason for this discrepancy is not clear. It might be related to the difference of crystal preparation or species differences. The activation of bovine TLR2 receptors by CPPD crystals might not be directly translated to the human system. It is well established that the calcium-containing crystals are most commonly associated with joint diseases. The relationship between calcium crystal deposition and OA remains unclear, controversial and a difficult one to approach. There are two main hypotheses: one is that calcium crystals cause or worsen OA12e18; the opposite hypothesis is that OA causes or worsens calcium crystal deposition51e55. Here, we have shown that crystals from OA knee joints strongly induce NO, MMP-13 and PGE2 production from primary chondrocytes and synoviocyte, suggesting that the naturally occurring calcium crystals may contribute to cartilage degradation and synovitis in OA. It is unlikely that those biologically active calcium crystals present in the OA joints are just a benign bystander or a disease marker.

Conflict of interest There are no conflicts of interest from any of the authors of this manuscript that could have inappropriately influenced this work.

Acknowledgements We thank Dr Steve Delaney, Dr Ted Wells, Dr Clive Jackson, Dr Anne Osborne and Ms Kerry Moakes for discussion and support. We are grateful to all the patients and the orthopaedic surgeons in Edinburgh for providing clinical material. Special thanks to Dr George Nuki, Ms Christine Beadle and Ms Ying Zhou from University of Edinburgh for the supply of human tissue.

References 1. Goldring MB, Goldring SR. Osteoarthritis (Mini-review). J Cell Physiol 2007;213:626e34. 2. Burrage PS, Mix KS, Brinckerhoff CE. Matrix metalloproteinases: role in arthritis. Front Biosci 2006;11:529e43. 3. Scher JU, Pillinger MH, Abramson SB. Nitric oxide synthases and osteoarthritis. Curr Rheumatol Rep 2007;9:9e15. 4. Hill CL, Hunter DJ, Niu J, Clancy M, Guermazi A, Genant H, et al. Synovitis detected on magnetic resonance imaging and its relation to pain and cartilage loss in knee osteoarthritis. Ann Rheum Dis 2007;66: 1599e603. 5. Benito MJ, Veale DJ, FitzGerald O, Van den Berg WB, Bresnihan B. Synovial tissue inflammation in early and late osteoarthritis. Ann Rheum Dis 2005;64:1263e7. 6. Sutton S, Clutterbuck A, Harris P, Gent T, Freeman S, Foster N, et al. The contribution of the synovium, synovial derived inflammatory cytokines and neuropeptides to the pathogenesis of osteoarthritis. Vet J 2009;179:10e24.

1340

Y. Z. Liu et al.: Contribution of calcium-containing crystals in OA

7. Derfus BA, Kurian JB, Butler JF, Daft LJ, Carrera GF, Ryan LM, et al. The high prevalence of pathologic calcium crystals in pre-operatives knees. J Rheumatol 2002;29:570e4. 8. Halverson PB, McCarty DJ. Patterns of radiographic abnormalities associated with basic calcium phosphate and calcium pyrophosphate dihydrate crystal deposition in the knee. Ann Rheum Dis 1986;45: 603e5. 9. Pattrick M, Hamilton E, Wilson R, Austin S, Doherty M. Association of radiographic changes of osteoarthritis, symptoms, and synovial fluid particles in 300 knees. Ann Rheum Dis 1993;52:97e103. 10. Sanmarti R, Kanterewicz E, Pladevall M, Panella D, Tarradellas JB, Gomez JM. Analysis of the association between chondrocalcinosis and osteoarthritis: a community based study. Ann Rheum Dis 1996; 55:30e3. 11. Reuben PM, Wenger L, Cruz M, Cheung HS. Induction of matrix metalloproteinase-8 in human fibroblasts by basic calcium phosphate and calcium pyrophosphate dihydrate crystals: effect of phosphocitrate. Connect Tissue Res 2001;42:1e12. 12. Dayer JM, Evequoz V, Zavadil-Grob C, Grynpas MD, Cheng PT, Schnyder J, et al. Effect of synthetic calcium pyrophosphate and hydroxyapatite crystals on the interaction of human blood mononuclear cells with chondrocytes, synovial cells, and fibroblasts. Arthritis Rheum 1987;30:1372e81. 13. Ea HK, Uzan B, Rey C, Liote F. Octacalcium phosphate crystals directly stimulate expression of inducible nitric oxide synthase through p38 and JNK mitogen-activated protein kinases in articular chondrocytes. Arthritis Res Ther 2005;7:R915e26. 14. Fam AG, Morava-Protzner I, Purcell C, Young BD, Bunting PS, Lewis AJ. Acceleration of experimental lapine osteoarthritis by calcium pyrophosphate microcrystalline synovitis. Arthritis Rheum 1995;38:201e10. 15. Cheung HS. Role of calcium-containing crystals in osteoarthritis. Front Biosci 2005;10:1336e40. 16. Balic´-Zˇunic´ T, Christoffersen MR, Christoffersen J. Structure of the b form of calcium pyrophosphate tetrahydrate. Acta Crystallogr 2000;B56:953e8. 17. Gordon C, Swan A, Dieppe P. Detection of crystals in synovial fluids by light microscopy: sensitivity and reliability. Ann Rheum Dis 1989;48: 737e42. 18. Swan A, Chapman B, Heap P, Seward H, Dieppe P. Submicroscopic crystals in osteoarthritic synovial fluids. Ann Rheum Dis 1994;53: 467e70. 19. Mandel NS. The crystal structure of calcium pyrophosphate dihydrate. Acta Crystallogr 1975;B31:1730e4. 20. Davis NL, Mandel GS, Mandel NS, Dickerson RE. Structure of monoclinic calcium pyrophosphate tetrahydrate. J Crystallogr Spectrosc Res 1985;15:513e21. 21. Christoffersen MR, Balic´-Zˇunic´ T, Pehrson S, Christoffersen J. Growth and precipitation of a monoclinic calcium pyrophosphate tetrahydrate indicating auto-inhibition at pH7. J Cryst Growth 2000;212:500e6. 22. Christoffersen MR, Balic´-Zˇunic´ T, Pehrson S, Christoffersen J. Kinetics of growth columnar triclinic calcium pyrophosphate dihydrate crystals. Cryst Growth Des 2001;1:463e6. 23. Christoffersen MR, Christoffersen J. Effects of bisphosphonate (EHPD) on growth formation and dissolution of calcium pyrophosphate crystals. Cryst Growth Des 2003;3:79e82. 24. Christoffersen MR, Balic´-Zˇunic´ T, Christoffersen J. Kinetics and mechanisms of dissolution and growth of acicular triclinic calcium pyrophosphate dihydrate crystals. Cryst Growth Des 2002;2:567e71. 25. Christoffersen MR, Seierby N, Balic´-Zˇunic´ T, Christoffersen J. Kinetics of dissolution of triclinic calcium pyrophosphate dihydrate crystals. J Cryst Growth 1999;203:234e43. 26. Deepa M, Rajendra Babu K, Vaidyan VK. Growth and characterisation of calcium pyrophosphate dihydrate grown in silica gel. Bull Mater Sci 1994;17:105e10. 27. McCarty DJ, Hogan JM, Gatter RA, Grossman M. Studies on pathological calcifications in human cartilage. I. Prevalence and types of crystal deposits in the menisci of two hundred fifteen cadavera. J Bone Joint Surg 1966;48:309. 28. Swan A, Heywood B, Chapman B, Seward H, Dieppe P. Evidence for a causal relationship between the structure, size and load of calcium pyrophosphate dihydrate crystals and attacks of pseudogout. Ann Rheum Dis 1995;54:825e30. 29. Manjubala I, Sivakumar M. In-situ synthesis of biphasic calcium phosphate ceramics using microwave irradiation. Mater Chem Phys 2001;71:272e8. 30. McCarty DJ, Lehr JR, Halverson PB. Crystal populations in human synovial fluid. Identification of apatite, octacalcium phosphate, and tricalcium phosphate. Arthritis Rheum 1983;26:1220e4.

31. Hu J, Agrawal DK, Fang Y, Roy R. Synthesis of hydroxyapatite using phosphate-rich glasses in the system CaOeP2O5eH2O and acoustic waves. J Mater Sci 1993;28:5297e300. 32. Faure G, Netter P, Malaman B, Steinmetz J. Monocrystalline calcium hydrogen phosphate dihydrate in destructive arthropathies of chondrocalcinosis. Lancet 1977;2:142e3. 33. McCarty D. Crystals, joints, and consternation. Ann Rheum Dis 1983;42: 243e53. 34. McCarthy GM, Cheung HS, Abel SM, Ryan LM. Basic calcium phosphate crystal induced collagenase production: role of intracellular crystal dissolution. Osteoarthritis Cartilage 1998;6:205e13. 35. Cheng P, Pritzer KH. The effect of calcium and magnesium ions on calcium pyrophosphate crystal formation in aqueous solutions. J Rheumatol 1981;8:772e80. 36. Liu-Bryan R, Pritzker K, Firestein G, Terkeltaub R. TLR2 signalling in chondrocytes drives calcium pyrophosphate dihydrate and monosodium urate crystal-induced nitric oxide generation. J Immunol 2005; 174:5016e23. 37. Swan AJ, Heywood BR, Dieppe PA. Extraction of calcium containing crystals from synovial fluid and articular cartilage. J Rheumatol 1992;19:1764e73. 38. Mandel GS, Renne KM, Kolbach AM, Kaplan WD, Miller JD, Mandel NS. Calcium pyrophosphate crystal deposition disease: preparation and characterisation of crystals. J Cryst Growth 1988;87:453e62. 39. Brown WE, Gregory TM. Calcium pyrophosphate chemistry. Arthritis Rheum 1976;19(3):446e62. 40. Record 72-0713, McClune WF, Eds. Powder Diffraction File. Swarthmore, Pennsylvania: International Centre for Diffraction Data; 1993. 41. Record 09-0080, McClune WF, Eds. Powder Diffraction File. Swarthmore, Pennsylvania: International Centre for Diffraction Data; 1993. 42. Molloy ES, Morgan MP, Doherty GA, McDonnell B, Hillard M, O’Byrne J, et al. Mechanism of basic calcium phosphate crystal-stimulated cyclooxygenase-1 up-regulation in osteoarthritic synovial fibroblasts. Rheumatology 2008;47:965e71. 43. Jibiki M, Shimoda S, Nakagawa K, Kawasaki K, Asada K, Ishibashi K. Calcifications of the disc of the temporomandibular joint. J Oral Pathol Med 1999;28:413e9. 44. Lin SH, Hsieh ET, Wu TU, Chang CW. Cervical myelopathy induced by pseudogout in ligamentum flavum and retro-odontoid mass: a case report. Spinal Cord 2006;44:692e4. 45. Hafez IT, Paraskeva CA, Toliza A, Klepetsanis PG, Koutsoukos PG, Gustavsen O, et al. Calcium phosphate overgrowth on silicon sand. Cryst Growth Des 2006;6:675e83. 46. Rosenthal AK, Fahey M, Gohr C, Burner T, Konon I, Daft L, et al. Feasibility of a tetracycline-binding method for detecting synovial fluid basic calcium phosphate crystals. Arthritis Rheum 2008;58:3270e4. 47. Schumacher HR, Fishbein P, Phelps P, Tse R, Krauser R. Comparison of sodium urate and calcium pyrophosphate crystal phagocytosis by polymorphonuclear leukocytes. Arthritis Rheum 1975;18:783e92. 48. Liu-Bryan R, Liote F. Monosodium urate and calcium pyrophosphate dihydrate (CPPD) crystals, inflammation and cellular signalling. Joint Bone Spine 2005;72:295e302. 49. Nair D, Misra RP, Sallis JD, Cheung HS. Phosphocitrate inhibits a basic calcium phosphate and calcium pyrophosphate dihydrate crystal-induced mitogen-activated protein kinase cascade signal transduction pathway. J Biol Chem 1997;272:18920e5. 50. Cheung HS, Halverson PB, McCarty DJ. Release of collagenase, neutral protease, and prostaglandins from cultured mammalian synovial cells by hydroxyapatite and calcium pyrophosphate dihydrate crystals. Arthritis Rheum 1981;24:1338e44. 51. Neogi T, Nevitt M, Niu J, LaValley MP, Hunter DJ, Terkeltaub R, et al. Lack of association between chondrocalcinosis and increased risk of cartilage loss in knees with osteoarthritis: results of two prospective longitudinal magnetic resonance imaging studies. Arthritis Rheum 2006;54:1822e8. 52. Ishikawa K, Masuda I, Ohira T, Yokoyama M. A histological study of calcium pyrophosphate dihydrate crystal-deposition disease. J Bone Joint Surg Am 1989;71:875e86. 53. Kirsch T, Swoboda B, Nah H. Activation of annexin II and V expression, terminal differentiation, mineralization and apoptosis in human osteoarthritic cartilage. Osteoarthritis Cartilage 2000;8:294e302. 54. Johnson KA, van Etten D, Nanda N, Graham RM, Terkeltaub RA. Distinct transglutaminase 2-independent and transglutaminase 2-dependent pathways mediate articular chondrocyte hypertrophy. J Biol Chem 2003;278:18824e32. 55. Viriyavejkul P, Wilairatana V, Tanavalee A, Jaovisidha K. Comparison of characteristics of patients with and without calcium pyrophosphate dihydrate crystal deposition disease who underwent total knee replacement surgery for osteoarthritis. Osteoarthritis Cartilage 2007;15:232e5.