Revisiting spatial distribution and biochemical composition of calcium-containing crystals in human osteoarthritic articular cartilage

Osteoarthritis (OA) is a whole-joint disease characterized by cartilage destruction with cartilage mineralization, subchondral bone modifications and mild synovial inflammation. The two main types of calcium-containing (CaC) crystals encountered in hyaline and fibrous cartilage mineralization are calcium pyrophosphate (CPP) and basic calcium phosphate (BCP) crystals. Cartilage and meniscus calcification evidenced on conventional radiographs is named chondrocalcinosis [1]. BCP crystals are a heterogeneous group of CaC crystals that encompass hydroxyapatite, carbonated apatite (CA), octacalcium phosphate, amorphous carbonated calcium phosphate (ACCP), tricalcium phosphate and whitlockite (WH) (magnesium-substituted) crystals. Both monoclinic and triclinic CPP crystals and the different members of BCP crystals have been found in synovial fluid [2]. The pathogenic role of CPP and BCP crystal deposition in cartilage is still unclear and controversial [3, 4]. However, growing clinical and experimental evidence indicates that CaC crystals may induce real microcrystal stress on synoviocytes and chondrocytes, leading to OA lesion worsening and/or development [5].

Articular calcification is a well-known phenomenon observed in late-stage OA [6‐8]. However, recent findings suggest that cartilage mineralization is a dynamic process occurring during or prior to OA development. Scotchford and colleagues detected BCP crystals in 12 cartilage samples from normal femoral heads of patients undergoing prosthetic replacement due to fracture of the femoral neck or distal femoral tumor. Transmission electron microscopy revealed calcifications in juvenile femoral head cartilage from patients as young as 10 years old and in five patients younger than 25 years old [9]. Similarly, Mitsuyama and colleagues, in a cadaveric study of 56 individual donors (mean age 50.3 years, range 12 to 74 years), showed calcifications in knee cartilage of young donors (23 donors <40 years old) and in normal-to-mildly affected cartilage, with 40% of tibial cartilage samples and 58% of femoral cartilage samples showing only grade 1 (macroscopically normal hyaline appearance) or grade 2 (minimal fibrillation) cartilage [10]. Moreover, the authors showed that calcification was diffused in the joint and comparable between femoral condyles and tibial plateaux. However, they did not distinguish superficial areas from deep zones, and the type of CaC crystals was not assessed [10]. Fuerst and colleagues reported that calcification was constantly identified in OA cartilage harvested during total knee replacement and that 40% of the mineralized cartilages were from patients with only mild OA lesions [6]. Moreover, chondrocytes isolated from OA cartilage showed differentiation toward a pro-mineralizing hypertrophic phenotype, whereas those isolated from normal cartilage did not [6]. Unfortunately, the authors examined only the most loaded and destroyed compartment of the knee joint, namely the medial femoral condyle cartilage. Sun and colleagues also showed that chondrocytes from OA meniscus could differentiate toward a pro-mineralized phenotype as compared with normal meniscus chondrocytes [11]. All of these findings suggest that hyaline and fibrous cartilage mineralization is a dynamic process involving chondrocyte metabolism dysfunction.

Extracellular inorganic phosphate (Pi) and extracellular inorganic pyrophosphate (PPi) concentrations are critical determinants of mineralization. Extracellular Pi and extracellular PPi are regulated by several proteins. The extracellular PPi concentration is controlled by three ectoenzymes: plasma-cell membrane glycoprotein 1 (PC-1)/ectonucleotide pyrophosphatase phosphodiesterase 1, which hydrolyzes nucleotide triphosphate into PPi; and AMP, the multipass transmembrane ankylosis (ANK) transporter; as well as the tissue-nonspecific alkaline phosphatases (TNAPs), which hydrolyze PPi into 2-Pi [12]. PPi is a potent inhibitor of apatite crystal formation, but its excess leads to CPP crystal deposition. The mRNA expression of these three molecules is higher in OA than in normal chondrocytes [11, 13, 14]. Sun and colleagues also showed increased expression of PC-1 and human ANK (ANKH) genes in chondrocytes isolated from OA meniscus [11].

We speculated that cartilage mineralization could affect the whole joint. We assessed the distribution of these mineral phases in human knee OA cartilage and their biochemical composition and morphological aspects, as well as changes in the expression of genes involved in the mineralization process. We found that cartilage calcification was a widespread phenomenon during OA, involving all joint compartments including nonweight-bearing zones and the whole depth of cartilage. Several types of CaC crystals were identified and the crystals displayed various morphological aspects. Finally, we observed that ANKH, PC-1 and TNAP expression was upregulated in chondrocytes isolated from CaC crystal-containing cartilages. Our results suggest that cartilage mineralization in OA may be associated with generalized chondrocyte dysfunction and changes toward a promineralizing phenotype.

Mineralization of hyaline cartilage and fibrocartilage is a common phenomenon of end-stage OA and has been associated with OA progression and cartilage destruction [6, 11]. We confirmed the high frequency of cartilage mineralization in knee articular cartilage harvested during total knee replacement. Furthermore, cartilage mineralization was widespread in OA and occurred with similar frequency in all articular cartilage compartments, including nonweight-bearing zones and those free of lesions (such as intercondyle cartilage). In addition, the presence of CaC crystals was associated with a significant upregulation of the expression of genes, including ENPP1 (PC-1) and ANKH, which are involved in Pi and PPi homeostasis.

In a cadaveric study of 106 knees, Mitsuyama and colleagues showed that calcification was widespread in knee cartilages and involved both femoral condyles and tibial plateaus [10]. Similarly, in an anatomic, radiographic and magnetic resonance imaging study of 10 cadaveric knees of older subjects, Abreu and colleagues found diffuse calcifications involving menisci and hyaline cartilage but also other joint tissues such as cruciate ligaments, popliteus tendon and joint capsule in 40% of the joints [22]. Finally, in a cadaveric study of 68 knees, computed tomography frequently detected calcifications in menisci, femoro-tibial cartilages as well as proximal tibio-fibular cartilages [23]. However, all of these studies only assessed the presence of calcifications without characterizing the type of CaC crystals. Here, we used FT-IR spectroscopy and scanning electron microscopy to describe for the first time the distribution and type of CaC crystals in the whole knee joint, systematically analyzing eight to 10 cartilage areas for each knee, according to a standardized sample-harvesting protocol.

Although CA crystals were the main mineral detected, cartilage mineralization consisted of several types of CaC crystals with different physico-chemical properties and different shapes. As in Fuerst and colleagues' study, BCP crystals were detected in all samples whereas CPP crystals were found in only 40% [6]. However, the authors only analyzed the medial femoral condyle and did not differentiate superficial from deep zone cartilage. We rarely found CPP and BCP crystals in the same sample or zone, which suggests a tight balance between Pi and PPi metabolism. Indeed, excess PPi can lead to CPP crystal deposition, whereas low levels favor BCP crystal formation [24]. However, the observation of both CaC crystal types in some samples suggests that the Pi/PPi balance could change during OA disease.

The mechanisms of articular cartilage calcification in normal and OA cartilages are unknown. Several factors identified include genetics, aging, extracellular matrix modifications, imbalance between inhibitors and promineralizing factors, dysregulation of PPi and Pi metabolism, and chondrocyte phenotype alterations such as hypertrophic differentiation and apoptosis [25, 26]. Several studies have shown that chondrocytes isolated from OA cartilage were prone to induce calcifications in vitro. The phenotype of these chondrocytes differed from that of chondrocytes isolated from normal cartilage. They were hypertrophic, expressed type X collagen and/or had upregulated expression of several genes involved in the mineralization process including ENPP-1 and ANKH [6, 11, 27]. We also observed an increase in mRNA expression of ENPP-1 (PC-1) and ANKH in chondrocytes from tibial cartilage containing CPP crystals as compared with cartilage without CPP crystals. However, IHC revealed no differences at the protein level. This result could be explained by the mild degradation of the cartilage samples used for IHC assessment. Further studies assessing more severe OA cartilage are required. Hirose and colleagues found a similar expression of PC-1 and ANK in OA cartilage with and without CPP crystal deposition [28]. However, extracellular PPi production was greater in chondrocytes from CPP crystal-containing cartilage than OA cartilage without crystals. These proteins may have both post-transcriptional and post-translational regulation mechanisms, but other proteins may be involved in the mineralization process. Recently, CD73 protein was found to have a potent inhibitor role in vascular and peri-articular mineralization [29]. St Hilaire and colleagues showed nonfunctional mutations in NT5E in members of three families with symptomatic arterial and joint calcifications. This gene encodes CD73, which participates in the extracellular pathway that converts AMP to adenosine. PC-1 hydrolyzes ATP into PPi and AMP, which is then converted to adenosine and Pi by CD73. Adenosine inhibits TNAP activity. CD73 deficiency therefore leads to increased TNAP activity, thus enhancing PPi degradation and Pi accumulation, both of which favor mineralization [29]. Whether the same phenomenon exists in OA cartilage needs to be further assessed. We observed that the NT5E mRNA expression was similar in chondrocytes isolated from cartilage with and without calcification. The protein level was not assessed because no efficient antibody is available.

In this study we did not show any association of radiographic OA grade or age and cartilage calcification, probably because our study was underpowered with its small patient sample and because our samples were mostly end-stage OA. However, Fuerst and colleagues showed that a radiographic Kellgren-Lawrence score and histological lesions correlated with cartilage calcification [6]. Moreover, Nalbant and colleagues reported that the presence of CaC crystals in OA synovial fluid was a predictor of OA progression [30]. Recently, we showed that intra-articular injection of BCP crystals into mouse knee induced cartilage degradation along with synovial inflammation [31]. Altogether, these data indicate that cartilage calcification may represent a significant pathogenic process during OA. Several issues remain to be addressed in CaC crystal-associated OA management, including the development of diagnostic tools to conveniently detect BCP crystals in synovial fluid and imaging techniques to detect CaC crystals in OA joints in vivo such as high-resolution computed tomography or ultrasonography, which could be used on a routine basis.

Our results suggest that cartilage mineralization is a widespread and constant process in the OA joint, occurring even in nonweight-bearing intercondyle cartilage. Mineralization involved several compartments and several types of CaC crystals. This process is associated with an increase in the expression of genes involved in Pi and PPi homeostasis, which suggests a switch toward a promineralizing chondrocyte phenotype in OA. Molecular and biomechanical mechanisms controlling the articular chondrocyte fate toward a promineralizing phenotype and driving cartilage mineralization in OA remain largely undetermined. Further investigation with larger samples is required to address these questions and could lead to new specific therapies in OA targeting the associated mineralization process.