ADDENDA AND ERRATA
A correction has been published for this article. To view the correction, click here.

Calcifications in human osteoarthritic articular cartilage: ex vivo assessment of calcium compounds using XANES spectroscopy

Calcium (Ca²⁺)-containing crystals (CCs), including basic Ca²⁺ phosphate (BCP) and Ca²⁺ pyrophosphate dihydrate (CPPD) crystals, are associated with severe forms of osteoarthritis (OA). Growing evidence supports a role for abnormal articular cartilage mineralization in the pathogenesis of OA. However, the role of Ca²⁺ compounds in this mineralization process remains poorly understood. Six patients, who underwent total knee joint replacement for primary OA, have been considered in this study. Cartilage from femoral condyles and tibial plateaus in the medial and lateral compartments was collected as 1 mm-thick slices cut tangentially to the articular surface. First, CCs presence and biochemical composition were assessed using Fourier transform infrared spectroscopy (FT-IR). Next, Ca²⁺ compound biochemical form was further assessed using X-ray absorption spectroscopy (XAS) performed at the Ca²⁺ K-absorption edge. Overall, 12 cartilage samples were assessed. Using FT-IR, BCP and CPPD crystals were detected in four and three out of 12 samples, respectively. Ca²⁺ compound biochemical forms differed between areas with versus without CCs, when compared using XAS. The complete set of data shows that XANES spectroscopy can be used to accurately characterize sparse CCs in human OA cartilage. It is found that Ca²⁺ compounds differ between calcified and non-calcified cartilage areas. In calcified areas they appear to be mainly involved in calcifications, namely Ca²⁺ crystals.

1. Introduction

Osteoarthritis (OA) affects the entire joint. Thus, in addition to articular cartilage destruction, OA induces changes in other joint components including the bone, menisci, synovium, ligaments, capsule and muscles (Hunter & Felson, 2006). Multiple factors including the genetic background, mechanical overload, ageing and intra-articular calcium (Ca²⁺)-containing microcrystal (CC) deposition contribute to the pathogenesis and progression of OA. CCs encountered in human crystal deposition-associated joint diseases include Ca²⁺ pyrophosphate dihydrate (CPPD) and basic Ca²⁺ phosphate (BCP) crystals. CPPD crystals (Ca₂P₂O₇·2H₂O) are found as triclinic and/or monoclinic forms (t-CPPD and m-CPPD, respectively) (Mandel & Mandel, 1988). BCP crystals encompass octacalcium phosphate [OCP; Ca₈(HPO₄)₂(PO₄)₄·5H₂O], carbonated apatite [CA; Ca₁₀(PO₄)₆(OH)_x(CO₃)_y], tricalcium phosphate [Ca₃(PO₄)₂] (McCarty et al., 1983) and magnesium whitlockite (Scotchford & Ali, 1995) crystals. A pathogenic role for CC-associated microcrystalline stress in OA is supported by recent clinical and experimental findings (Ea et al., 2011; Nguyen et al., 2010). CCs have been identified in 60% of OA synovial fluids and in 100% of OA cartilages obtained during total knee or hip joint replacement (Derfus et al., 2002; Fuerst, Bertrand et al., 2009; Fuerst, Niggemeyer et al., 2009). CCs are associated with destructive forms of OA, as well as with chondrocyte differentiation toward a pro-mineralizing phenotype (Fuerst, Bertrand et al., 2009). In vitro, CCs can activate various cell types by several mechanisms and display mitogenic, catabolic, inflammatory and apoptotic properties (Ea & Lioté, 2009; McCarthy & Cheung, 2009). In vivo, their presence can precede cartilage breakdown (Bendele & Hulman, 1988; Evans et al., 1994), and intra-articular injection of CPPD crystals worsens rabbit OA lesions induced by anterior cruciate ligament section (Fam et al., 1995).

The mechanisms underlying articular cartilage mineralization during OA are incompletely understood. Briefly, calcification can occur when the balance between inhibitors and pro-mineralizing factors is disrupted by any of multiple influences such as genetics, ageing, extracellular matrix modifications, dysregulation of inorganic pyrophosphate (PPi) and phosphate (Pi) metabolisms, changes in extracellular Ca²⁺ levels, and chondrocyte phenotype alterations (Ea & Lioté, 2009). Cartilage calcifications are formed at least partly in chondrocyte-derived apoptotic bodies (Johnson et al., 2001) and matrix vesicles (MVs), which are membrane-enclosed particles released from the plasma membrane of mineralization-competent cells (Anderson, 1995). Apatite crystal formation within MVs in OA cartilage has been documented by electron microscopy (Kirsch et al., 2000). Concentrations of Ca²⁺ (30–50 mmol l⁻¹) and inorganic phosphate (Pi) (15–30 mmol l⁻¹) in MVs are well above the minimum level of [Ca]·[Pi] supersaturation, at which amorphous Ca²⁺ phosphate, considered a precursor of biological apatite, spontaneously precipitates at physiological pH (Termine & Eanes, 1974). Ca²⁺ binding to proteins and lipids within the MV lumen, together with an imbalance between Pi and PPi (Camerlain et al., 1975; Terkeltaub, 2001), is among the driving forces of early mineral ion uptake (Sauer & Wuthier, 1988) and ultimately leads to tissue mineralization. Compared with normal cartilage, OA cartilage consistently showed increased Ca²⁺ and Pi contents, of 128.9 mmol kg⁻¹ versus 52.6 mmol kg⁻¹ for Ca²⁺ and 66.9 mmol kg⁻¹ versus 23.1 mmol kg⁻¹ for Pi (Pritzker et al., 1987). Increased extracellular Ca²⁺ drives chondrocyte differentiation toward a pro-mineralizing phenotype (Bonen & Schmid, 1991), whereas extracellular Ca²⁺ depletion maintains a stable cellular phenotype (Gigout et al., 2005). However, little is known regarding the state of Ca²⁺ compounds in human OA articular cartilage.

Accurate characterization of calcifications is difficult to achieve using conventional techniques available in clinical laboratories. Among the techniques that are often used to characterize calcifications, diffraction techniques (using neutrons, X-rays or electrons as probes) (Guinier, 1956) and vibrational spectroscopy [including Raman and Fourier transform infrared (FT-IR) spectroscopy] (Kneipp et al., 1999; McKelvy et al., 1998) are very effective tools for studying mineralized tissues. Recent experimental developments permit ex vivo and in vivo experiments at the micrometre scale (Bertsch & Hunter, 2001; Dumas et al., 2006). Physico-chemical techniques available for the structural characterization of cartilage and other biological samples include techniques specific to large instruments (Bazin et al., 2006), such as synchrotron radiation X-ray fluorescence (Bazin et al., 2007; Zoeger et al., 2006), X-ray diffraction (Yarker et al., 1984) and X-ray absorption spectroscopy (XAS) (Eichert et al., 2005). XAS is a sensitive technique, specific to synchrotron radiation, which has been used to characterize the local environment of specific elements such as Ca²⁺ and Zn²⁺ in biological samples. Measurements can be performed directly on the sample with minimal preparation (Bazin et al., 2009a,b; Carpentier et al., 2010). XAS techniques include X-ray absorption near-edge structure spectroscopy (XANES) and extended X-ray absorption fine-structure spectroscopy (Eichert et al., 2005; Sowrey et al., 2004) and provide an accurate description of the electronic state, geometry of very first neighbours, and first coordination spheres of the selected element.

The objective of this study was to use XAS performed at the Ca²⁺ K-absorption edge (XANES) to compare the state of Ca²⁺ compounds between human OA articular cartilage areas with CCs versus areas without CCs.

2. Experimental design

2.1. Patients

Specimens of knee joint cartilage were obtained from six patients undergoing total knee replacement surgery for OA at a single orthopaedic department. Patients were selected at random and included prospectively. One rheumatologist (CN) reviewed the medical file of each patient. Clinical data [age; sex; body mass index (BMI); and presence of genu varum or genu valgum] and preoperative plain knee radiographs were obtained and analysed. Kellgren and Lawrence staging was used to characterize OA severity. Patients with secondary OA (inflammatory disease or trauma) were not included.

2.2. Cartilage sample preparation

Specimens included femoral condyle and tibial plateau cartilages from both the medial and the lateral compartments (Fig. 1a), yielding a theoretical total of eight cartilage zones per knee. Specimens were washed three times with saline under aseptic conditions to avoid contamination by bone mineral crystals released at the time of surgery. For each femoro-tibial compartment, samples were collected as 1 mm-thick slices cut with a scalpel tangentially to the articular cartilage surface. Cartilage areas were labelled as follows: 1, medial condyle; 2, lateral condyle; 3, medial tibial plateau; 4, lateral tibial plateau; S, superficial layer; and D, deep layer (Fig. 1b). Cartilage samples were either frozen directly and stored at 253 K for FT-IR or fixed in 70° alcohol at 277 K for XAS. This protocol was used in order to preserve the physico-chemical integrity of the mineral part of the samples. Of note, preservation of the balance between intra- and extra-cellular ions such as Ca²⁺ requires a different and specific preparation protocol (Boxer et al., 2009).

Figure 1 Knee joint specimen obtained during arthroplasty and schematic representation of the sample collection protocol. The specimen included femoral condyle and tibial plateau cartilage from both the medial and the lateral compartments (a). Cartilage areas were labelled as follows: 1, medial condyle; 2, lateral condyle; 3, medial tibial plateau; 4, lateral tibial plateau; S, superficial layer; D, deep layer (b).

2.3. Fourier transform infrared (FT-IR) spectroscopy

Cartilage samples were first characterized using FT-IR spectrometry (Vector 22; Brucker Spectrospin, Wissembourg, France) as previously described (Estepa & Daudon, 1997). Data were collected in the absorption mode between 4000 and 400 cm⁻¹, with a resolution of 4 cm⁻¹. The absorption bands of Ca²⁺ biological reference compounds, namely CA, amorphous Ca²⁺ carbonated phosphate (ACCP) and CPPD, are well documented. Regarding CA, the ν₁ and ν₃ P—O stretching vibration modes are measured at 960–962 cm⁻¹ and 1035–1045 cm⁻¹, respectively, while the O—P—O ν₄ bending mode corresponds to the doublet at 602–563 cm⁻¹. The bands at 3570 and 633 cm⁻¹, corresponding to the stretching and vibrational modes of the OH— groups characteristic of hydroxyapatite, are absent from CA. Disappearance of the CA shoulder in the ν₃ absorption band is a fingerprint for ACCP. Regarding m-CPPD, O—P—O bending is recorded at 535 cm⁻¹ and 508 cm⁻¹. P—O stretching vibrations generate absorptions at 923 cm⁻¹ and 991 cm⁻¹, whereas asymmetric stretching vibrations give rise to absorptions at 1037 cm⁻¹ and 1089 cm⁻¹.

2.4. X-ray absorption spectroscopy

Experiments were performed at room temperature and atmospheric pressure, on the DIFFABS beamline located at the D13-1 bending magnet at the SOLEIL synchrotron facility (Saint Aubain, France). The typical energy of the storage ring was 2.75 GeV with an average current of 300 mA in the new TOP/UP configuration. Monochromatization and horizontal focus were achieved using a double Si (111) crystal monochromator that provided an energy resolution of ΔE/E = 10⁻⁴, which was necessary for XANES resolution. The monochromator was positioned between two cylindrical vertically focusing mirrors (50 nm Rh deposited on Si substrates). At the position of the sample, spot size was defined using slits to reach 100 µm × 250 µm (H × V, FWHM). Spectra at the Ca K-absorption edge (4086 eV) were collected in the 4020–4140 eV energy range, with energy steps of 0.5 eV and a 3 s dwell time per point. At least four different areas of each sample were analysed. Additional details on the experimental set-up have been reported previously (Carpentier et al., 2010).

2.5. Ethical considerations

All human sample collection procedures complied with the Helsinki Declaration. According to French Law on human research (Law 2007-1110, article 1211-2), cartilage sample collection (surgical waste) was authorized unless opposition from patients; all patients were informed and gave oral consent. As quoted above, formal approval from an ethical committee and a written consent from the patients are not required for studies using surgical waste materials.

3. Results

3.1. Clinical, demographic and OA characteristics

Overall, six patients (five females) were included. The mean (SD) age and body mass index at surgery were 74 (9) years (58 to 81 years) and 28 (9) kg m⁻², respectively. Preoperative plain radiographs showed uni-, bi- and tri-compartmental knee OA in 1, 3 and 2 patients, respectively. In all cases, OA was most severe in the medial femoro-tibial compartment, where the mean (SD) Kellgren and Lawrence score was 4 (0) (Table 1).

Table 1
Demographic, clinical and osteoarthritis features in the six patients with knee osteoarthritis

Data are number (%), unless otherwise indicated. SD, standard deviation; OA, osteoarthritis.

Age in years, mean (SD)	74 (9)
Male gender	1
Body mass index, kg m^−2, mean (SD)	28 (9)
Genu varum	5 (83)
Kellgren, mean (SD)	4 (0)
Unicompartmental OA	1 (17)
Bicompartmental OA	3 (50)
Tricompartmental OA	2 (33)
OA mainly affecting the medial femoro-tibial compartment	6 (100)
Preoperative radiographic calcifications	1 (17)
Calcifications by macroscopic examination	6 (100)
Calcifications by microscopic examination	6 (100)

3.2. Identification and quantification of calcium-containing crystals in human OA articular cartilage using FT-IR

Although eight specimens per knee would have been obtained in theory, we only examined 12 specimens selected at random from the six patients and free of calcifications by macroscopic examination (Table 2) owing to limited XAS beam time. FT-IR showed CCs in at least one sample from four of the six patients. The main compounds identified were mixtures of proteins, triglycerides and carbohydrates. The total mineral phase accounted for 10% of the total sample on average. Two types of mineral phase were detected, CA and m-CPPD (in four and three samples, respectively) (Fig. 2). In five samples no mineral phase was detected (Table 2).

Table 2
Identification and quantification of Ca²⁺-containing crystals in 12 samples of osteoarthritis human articular cartilage using FT-IR spectroscopy

Data are percentages. CA, carbonated apatite; CPPD, Ca²⁺-pyrophosphate dihydrate; ACCP, amorphous carbonated Ca²⁺ phosphate; prot, proteins; TG, triglycerides; CH, carbohydrate. See §2 for explanations of sample labelling.

CA	CPPD	ACCP	Other (prot, TG, CH)	Total mineral phase
0	0	0	100	0	XANES indicated Ca^2+ in tissue
0	15	0	85	15	XANES indicated Ca^2+ in CPPD
0	15	0	85	15
10	0	0	90	10	XANES indicated Ca^2+ in tissue
0	0	0	100	0
0	0	0	100	0	XANES indicated Ca^2+ in tissue
0	0	0	100	0
5	0	0	95	5	XANES indicated Ca^2+ in tissue
0	10	0	90	10
0	0	0	100	0	XANES indicated Ca^2+ in tissue
5	0	0	95	5
5	0	0	95	5

Figure 2 FT-IR absorption spectra representative of cartilage samples exhibiting calcium (Ca^2+)-containing crystals (CC), namely, carbapatite (a) and Ca^2+ pyrophosphate dehydrate crystals (b). The absorption bands of carbapatite (CA) and Ca^2+ pyrophosphate dehydrate (CPPD) are well documented. Regarding CA, the ν1 and ν3 P—O stretching vibration modes are measured at 960–962 cm^−1 and 1035–1045 cm^−1, respectively, while the O—P—O ν4 bending mode corresponds to the doublet at 602–563 cm^−1. The bands at 3570 and 633 cm^−1, corresponding to the stretching and vibrational modes of the OH— groups characteristic of hydroxyapatite, are absent from CA. Regarding m-CPPD, O—P—O bending is recorded at 535 cm^−1 and 508 cm^−1. P—O stretching vibrations correspond to absorptions at 923 cm^−1 and 991 cm^−1, whereas asymmetric stretching vibrations give rise to absorptions at 1037 cm^−1 and 1089 cm^−1.

3.3. XANES spectra of biological reference compounds

The three reference compounds used in this study and previously characterized by FT-IR analysis were one biological m-CPPD sample and two biological BCP samples, namely CA and ACCP. To obtain a signal corresponding to Ca²⁺ from the tissue, thus including both intra- and extra-cellular Ca²⁺, we collected the XAS spectrum from cartilage areas with no calcifications by FT-IR (Fig. 3). After pre-peak A ascribable to a 1s → 3d transition, the most intense transition resonance of the spectrum was observed. This structure included a shoulder-like feature (feature B, 1s → 4s transition). For CA there was a double peak (features C1 and C2, 1s → 4p transition), whose relative intensities depended on the type of Ca²⁺ involved [Ca(I) or Ca(II)]. For m-CPPD and ACCP a single peak was present. Of note, the peak corresponding to Ca²⁺ from the tissue displayed a higher intensity. In addition, the shape of the edge of the sample was very different from the shapes of the other reference compounds, suggesting that the XANES spectrum reflected Ca²⁺ atoms present in the tissue. For m-CPPD, an additional feature (D) was present at 4057 eV.

Figure 3 XANES spectra of reference compounds. Absorption spectra were obtained for biological cartilage samples previously characterized by FT-IR. Red solid line, carbapatite; red dotted line, amorphous carbonated Ca^2+ apatite; black solid line, Ca^2+ pyrophosphate dehydrate; blue solid line, non-diffusible tissue Ca^2+. Note the pre-peak A (1s → 3d transition), the shoulder B (transition 1s → 4s), the double peaks C1–C2 [1s → 4p transition, whose relative intensities depend on the type of Ca^2+, namely, Ca(I) or Ca(II) involved in apatite] and the single C peak for non-diffusible tissue Ca^2+.

3.4. XANES spectra of human osteoarthritic articular cartilage compounds

Part of the sample was first analysed using FT-IR and the other part was investigated using XANES. In all five cartilage samples without CCs by FT-IR the XANES spectra were similar and displayed a large and very intense C peak corresponding to Ca²⁺ in the tissue. In patient 2, whose sample contained 15% of m-CPPD by FT-IR, the shape and position of the XANES spectrum were similar to those of the reference m-CPPD compound, whereas no tissue Ca²⁺ spectrum was detected (Fig. 4).

Figure 4 XANES human osteoarthritic articular cartilage compounds. Absorption spectra were obtained for biological cartilage samples previously characterized by FT-IR. Red solid line, carbapatite; red dotted line, amorphous carbonated Ca^2+ apatite; black solid line, Ca^2+ pyrophosphate dehydrate; blue solid line, non-diffusible tissue Ca^2+. (a) Patient 1. (b) Patient 3. (c) Patient 4. (d) Patient 5. (e) Patient 6. (f) Patient 2.

4. Discussion

Although the presence of CCs in osteoarthritic cartilage was recognized several decades ago, the relationship between CCs and OA remains controversial. Several recent studies support a pathogenic role for CCs in OA (Fuerst, Bertrand et al., 2009; Fuerst, Niggemeyer et al., 2009), thus confirming many earlier clinical and experimental studies, and suggest that articular cartilage mineralization may be a consistent feature of end-stage OA. However, the role for Ca²⁺ compounds in the mineralization process remains poorly understood. In the present study using XANES spectroscopy at the Ca²⁺ K-edge, a powerful technique specific to synchrotron radiation, we found differences in Ca²⁺ compound biochemical composition between cartilage areas with and without CCs from patients with end-stage OA. Thus, in areas without CCs, a specific XANES spectrum was found for tissue Ca²⁺ compounds, whereas in CCs-containing areas a distinct and specific XAS fingerprint was identified for CPPD crystals.

We first used FT-IR spectroscopy to distinguish cartilage areas with CCs from those without in samples that were initially selected based on the absence of calcifications by macroscopic examination. CCs were present in seven of 12 specimens, from four of six patients, suggesting that FT-IR spectroscopy was sensitive for detecting calcifications in our biological samples. However, FT-IR spectroscopy may have failed to detect calcifications in some of the samples, owing to heterogeneous location of CCs within the OA cartilage.

FT-IR reflects only the overall chemical composition of the species contained in the sample at the molecular scale, including lipids, proteins and PO₄ groups associated with Ca²⁺ (Estepa & Daudon, 1997). Therefore, we also used XANES spectroscopy at the Ca²⁺ K-edge to specifically characterize Ca²⁺-compound-containing species. We found at least two distinct types of Ca²⁺ compounds in OA articular cartilage. One type was found in OA cartilage areas without CCs and seemed to be composed of tissue Ca²⁺ compounds, i.e. not linked to mineral compounds, as the XANES spectrum was similar to that found in hydrated tissues. In these areas, tissue Ca²⁺ was the main Ca²⁺ compound, whereas Ca²⁺ involved in CCs was absent or less abundant. The other type of Ca²⁺ compound was identified in CCs-containing OA articular cartilage areas and had an XANES spectrum specific of CPPD crystals. As mentioned above, our sample preparation protocol was suitable for preserving the physico-chemical integrity of the mineral part of the sample. Preservation of tissue ions such as Ca²⁺ may require specific preparation protocols (Boxer et al., 2009).

Despite the technical difficulty in preserving diffusible Ca²⁺ ions, XANES appears suitable for accurately characterizing Ca²⁺ compounds in articular cartilage and provides data that complement those supplied by FT-IR spectroscopy. Since XANES is insensitive to polydispersity (Moonen et al., 1995), it allows an assessment of the relative quantity of Ca²⁺ embedded in the calcifications compared with Ca²⁺ in tissues. That XANES is successful in characterizing Ca²⁺ compounds is closely linked to the ability of XANES to characterize poorly crystallized compounds. Numerous studies based on XAS at the Ca²⁺ K-edge have been performed on other biological samples (Carpentier et al., 2010). Thus, XAS has been used to evaluate Ca²⁺ in physiologically calcified tissues such as bones and teeth or apatitic and non-apatitic Ca²⁺ phosphates of biological interest.

5. Conclusion

In summary, in the present study we show that XANES spectroscopy can be used to accurately characterize sparse CCs in human OA cartilage. We find that Ca²⁺ compounds differ between calcified and non-calcified cartilage areas. In calcified areas they appear to be mainly involved in calcifications, namely Ca²⁺ crystals. Further application of synchrotron-radiation-related technologies should bring additional information about other important chemical elements involved in these calcifications.