Extracellular Distribution of Collagen II and Perifibrillar Adapter Proteins in Healthy and Osteoarthritic Human Knee Joint Cartilage

Abstract

Perifibrillar adapter proteins, interconnecting collagen fibrils, and linking the collagen network with the aggrecan matrix seem to play a crucial role in the pathogenesis of osteoarthritis (OA). Therefore, we examined immunohistochemically the extracellular distribution of collagen II and the main perifibrillar adapter proteins—collagen IX, decorin, cartilage oligomeric matrix protein (COMP), and matrilin-3—in human samples of healthy (n=4) and OA (n=42) knee joint cartilage. Histopathology assessment was performed using an OA score. Staining patterns were evaluated in relation to the disease stage. The perifibrillar adapter proteins were uniformly distributed in the upper zones of healthy cartilage. In moderate OA (n=8; score 14.3 ± 4.7), all proteins analyzed were locally absent in the fibrillated area or the superficial and upper mid zone. In advanced OA (n=20; score 18.9 ± 5.3), they were uniformly distributed in these zones and accumulated pericellularly. Perifibrillar adapter proteins are important for the stabilization of the collagen network in the upper zones of healthy cartilage. Their degradation might be a critical event in early OA. In advanced OA, there are indications for an increased synthesis in an attempt to regenerate the lost tissue and to protect the remaining cartilage from further destruction.

Introduction

The extracellular matrix (ECM) of articular cartilage consists of two supramolecular compartments, the fibrillar collagen network and the extrafibrillar matrix, mainly composed of the proteoglycan aggrecan.¹ The collagen network provides tensile strength for cartilage tissue, and it further restricts the swelling of the tissue caused by the high water binding capacity of aggrecan. The resulting swelling pressure makes the cartilage resistant against compression.² The interface between the two compartments is constituted by a highly complex fibrillar periphery.¹ Perifibrillar adapter proteins, including fibril associated collagens with interrupted triple helices (FACIT), small leucine-rich proteoglycans (SLRP), and other noncollagenous matrix proteins, interconnect the collagen fibrils and mediate interactions between the collagen network and the aggrecan matrix (Fig. 1).

Figure 1.

Schematic illustration of the supramolecular network formation (not to scale). Interactions between the collagenous fibrillar structures and the aggrecan matrix are mediated by the perifibrillar adapter proteins matrilin-1, -3, and -4,¹⁷ decorin,¹³ COMP,⁹ and collagen IX.⁸ The adapter molecules interconnect collagen II/XI fibrils and link them to collagen VI microfibrils⁷⁰ or aggrecan.^13,16 Modified from Klatt et al.¹⁷ Abbreviation: COMP, cartilage oligomeric matrix protein.

Collagen fibrils, consisting of fibril-forming collagens II (>90%) and XI (~3%), are decorated with collagen IX (~1%), a member of the FACIT collagens.³ The N-terminal short arm of collagen IX projects into the perifibrillar space and, thus, provides binding sites for other matrix proteins while its long arm, oriented parallel to the fibril, allows covalent interactions with collagen II as well as other collagen IX molecules.^4–8 Matrilins and the cartilage oligomeric matrix protein (COMP) are members of noncollagenous matrix protein families. In immature cartilage, the pentameric glycoprotein COMP^9,10 catalyzes collagen fibril formation¹¹ while in adult cartilage with little collagen fibrillogenesis, it primarily seems to cross-link and stabilize the collagen network^12,13 through binding to the mature collagen fibril via collagen IX,¹⁴ matrilin-3, and -4.^4,15 In addition, COMP interconnects the collagen network with the extrafibrillar matrix by binding to aggrecan.¹⁶ Matrilin-3, a homotetramer, forms complexes or even heterooligomers with matrilin-1 and -4 that are linked via the SLRPs biglycan and decorin to collagen VI microfibrils.¹⁷ While the core protein of decorin is associated with collagen VI, its glycosaminoglycan side chain can interact with other extrafibrillar molecules.^18–20

Perifibrillar adapter proteins are not only fundamental for the mechanical stabilization and interconnection of ECM components, but also for the regulation of the collagen fibril diameter during fibrillogenesis, thus playing a key role in proper collagen network formation.^18,21–24 Therefore, it is not surprising that mutations or a lack of distinct perifibrillar adapter proteins cause a broad spectrum of skeletal disorders.¹ A severe form of chondrodysplasia, the pseudoachondroplasia, is caused exclusively by mutations in COMP,^25,26 whereas a milder form, the multiple epiphyseal dysplasia, has, in addition, been linked to mutations in collagen IX^27,28 and matrilin-3.²⁹ Patients with mutations in these genes often develop premature OA. In knockout models, collagen IX and matrilin-3 deficient mice have shown OA-like alterations with proteoglycan depletion and a loss of intact collagen II as well as a higher OA incidence and severity.^30–32

In summary, perifibrillar adapter proteins are essential for healthy articular cartilage because they play a crucial role in proper collagen fibril formation, collagen network stabilization, and interconnection of the fibrillar and extrafibrillar matrix.

Concerning the pathogenesis of OA, several authors state that the characteristic swelling of cartilage appearing in early OA may rather result from degradation of molecules in the fibrillar periphery than the collagen network itself.^13,33,34 However, current knowledge is limited to just a few studies that analyzed only a few perifibrillar adapter proteins in human articular cartilage. These studies observed an increased expression and altered extracellular distribution of adapter molecules in OA cartilage.^35–41 To extend the understanding of OA, an important next step is to make coordinated observations of the distribution patterns of all participating perifibrillar adapter proteins according to stages of disease progression. To our knowledge, we are the first that analyzed simultaneously the extracellular distribution of four major perifibrillar adapter proteins, namely, collagen IX, decorin, matrilin-3, and COMP, and of collagen II, as primary component of the fibrillar matrix, in healthy and osteoarthritic human knee joint cartilage. For the first time, distribution patterns of collagen II and the four perifibrillar adapter proteins were described in OA cartilage and related to the disease stage by using an OA histopathology score.

Materials and Methods

Tissue Samples

A total of 42 cartilage samples from the tibial plateau and the underlying subchondral bone were collected from 11 OA patients undergoing total knee replacement (Table 1). Due to different surgical techniques, the number of suitable samples ranged from two to six per donor (mean: 3.8 ± 1.3).

Table 1.

Demographic and Anthropometric Variables of All OA Patients.

	N	Age (Years)	Body Mass (kg)	Body Height (m)	BMI (kg/m²)
Females	7	65.0 ± 7.6 (55–76)	87.6 ± 17.6 (70–114)	1.60 ± 0.05 (1.51–1.69)	34.0 ± 5.8 (27.0–44.0)
Males	4	73.0 ± 4.7 (69–78)	88.8 ± 16.7 (64–100)	1.77 ± 0.05 (1.70–1.80)	28.2 ± 4.2 (22.1–31.2)
Total	11	67.9 ± 7.6 (55–78)	88.0 ± 16.4 (64–114)	1.66 ± 0.10 (1.51–1.80)	31.9 ± 5.8 (22.1–44.0)

Values are presented as mean ± SD (min. – max.). Abbreviations: OA, osteoarthritis; BMI, body mass index; SD = standard deviation.

Four macroscopically normal tibial cartilage/bone samples were gained from a 24-year-old male accident victim (86.4 kg, 1.86 m, body mass index 24.9 kg/m²). The study was approved by the local ethics committee, and informed written consent was obtained from tissue donors. Samples were fixed in 4% paraformaldehyde (24 h), decalcified with 20% EDTA (6 weeks), embedded in paraffin, and cut in 10-µm-thick sections.

IHC

Endogenous peroxidase was blocked with methanol and 1% H₂O₂. For the detection of collagen II, the sections were digested with pepsin (0.5% in 0.01N Hydrochloric Acid; pH 1.85) for 1 h at RT while all other sections were incubated with chondroitinase (40 mU/ml in TBS containing 0.01% BSA) for 1 h at 37C. Afterward, the sections were blocked for 30 min with TBS containing 0.01% BSA. Sections were incubated with primary antibodies overnight at 4C. A mouse monoclonal antibody against collagen II (1:500; Calbiochem, Darmstadt, Germany) and rabbit polyclonal antibodies against collagen IX (α1/NC4; 1:1000),⁴ decorin (1:500),⁴² matrilin-3 (1:500),⁴³ and COMP (1:1000)⁴⁴ were used. Horseradish peroxidase conjugated secondary antibodies, a swine polyclonal antibody against rabbit IgG, and a rabbit polyclonal antibody against mouse IgG (both 1:100; Dako, Glostrup, Denmark) were applied for 1 h at RT. Finally, the sections were stained using 3-amino-9-ethylcarbazole (Sigma Aldrich, St. Louis, MI). To exclude unspecific binding of the secondary antibodies, control stainings were carried out without primary antibody (data not shown). In addition, we performed stainings of the not commercially primary antibodies using isotype controls and preimmunesera, respectively. All these controls were negative (Supplemental Fig. 1A). Furthermore, the specificity of not commercially available primary antibodies was validated on respective knockout extracts and/or tissue sections in earlier studies (supplemental Table 1A).

Osteoarthritis Research Society International (OARSI)-Score

The OARSI OA histopathology grading system was applied for all samples.⁴⁵ Scoring was performed by two independent observers on Safranin-O stained sections that were made according to standard histochemical protocols. The score (0–24) is the product of the OA severity (six grades) and the horizontal extent of the involved cartilage surface (four stages). Values are presented as mean ± standard deviation (SD).

Histological Analysis

Sections were divided into a superficial, a mid, and a deep zone.⁴⁶ In addition, an area that corresponded to grade 2 (surface discontinuity), grade 3 (vertical fissures), and grade 4 (erosion) of the OARSI histopathology assessment⁴⁵ was determined as fibrillated area. The ECM was further divided into pericellular, territorial, and interterritorial regions.⁴⁶ All analyses were conducted with a light microscope (Nikon ECLIPSE 80i, Nikon Instruments Inc., New York, NY).

Results

OA cartilage samples had a mean OA score of 17.5 ± 5.4. As expected, the score of the control samples (the “healthy samples”) was zero. If the distribution pattern of a protein appeared in more than 10 OA cartilage samples, it was designated as a staining pattern. In OA cartilage, three different staining patterns (I, II, and III) were classified for each of the five ECM proteins analyzed. These patterns are described below. The frequency of occurrence together with the mean OA score of the corresponding cartilage samples are summarized in Table 2. Beyond that, two general distribution patterns (I + II), simultaneously describing the staining patterns of all five ECM proteins, were detected. Please note that the Roman numerals of the staining patterns are just for their differentiation and that there is neither a correlation between the numerals and the OA score, nor between the numerals of the single and the general staining patterns. Furthermore, it has to be taken into account that dependent on the location, several samples showed more than one staining pattern.

Table 2.

Distribution Pattern I to III of Collagen II, Collagen IX, Decorin, COMP, and Matrilin-3 and the Frequency of Occurrence (N) Together With the OA Score (mean ± SD) of the Corresponding OA Cartilage Samples.

	OA Staining Patterns
	I		II		III
	N	OA Score	N	OA Score	N	OA Score
Collagen II	28	16.6 ± 4.9	35	17.6 ± 4.9	32	17.9 ± 5.4
Collagen IX	18	16.4 ± 5.4	10	14.6 ± 4.2	38	17.8 ± 5.4
Decorin	17	17.1 ± 4.9	21	16.8 ± 5.7	19	18.3 ± 5.4
COMP	17	16.6 ± 5.0	20	15.8 ± 6.4	18	20.0 ± 3.6
Matrilin-3	14	15.0 ± 3.6	20	15.2 ± 5.5	16	22.0 ± 2.5

Please note that several samples showed more than one staining pattern. Abbreviations: COMP, cartilage oligomeric matrix protein; OA, osteoarthritis; SD = standard deviation.

Collagen II

Healthy cartilage showed an even distribution of collagen II (Fig. 2A and E). In the mid and deep zone, the pericellular staining was locally pronounced. However, all zones also had areas with a lack of pericellular staining.

Figure 2.

Immunohistological staining patterns of collagen II in healthy (A, E) and OA cartilage (B–D, F–H). Long bar = 1000 µm, short bar = 100 µm. Abbreviation: OA, osteoarthritis.

In OA pattern I, collagen II was locally absent in the superficial zone and fibrillated area, respectively (Fig. 2B and F). OA pattern II showed a uniform collagen II staining in the fibrillated area, superficial, and upper mid zone with a lack of pericellular staining (Fig. 2C and G). In OA pattern III, collagen II was detectable in the fibrillated area or superficial and upper mid zone while in the lower mid and deep zone, there was only a pericellular staining, and almost no staining in the interterritorial matrix (Fig. 2D and H).

Collagen IX

Healthy cartilage showed a uniform staining in the superficial and partly in the upper mid zone with a locally increased pericellular staining (Fig. 3A and E). In the mid zone, collagen IX was located in the pericellular matrix.

Figure 3.

Immunohistological staining patterns of collagen IX in healthy (A, E) and OA cartilage (B–D, F–H). Long bar = 1000 µm, short bar = 100 µm. Abbreviation: OA, osteoarthritis.

OA pattern I was characterized by a lack of collagen IX in the superficial zone and fibrillated area, respectively (Fig. 3B and F). Characteristic for OA pattern II was a collagen IX staining that was exclusively detectable in the pericellular matrix in the superficial and upper mid zone or fibrillated area (Fig. 3C and G). OA pattern III showed a collagen IX staining in the superficial and upper mid zone or fibrillated area that was in part increased in the pericellular matrix. In the lower zones, the staining was limited to the pericellular matrix (Fig. 3D and H).

Decorin

In healthy cartilage, decorin was localized in the superficial and upper mid zone (Fig. 4A and E). In the upper mid zone, the staining in the pericellular matrix was locally increased while it was missing in the interterritorial matrix. Two of four samples also showed a slight staining in the deep zone.

Figure 4.

Immunohistological staining patterns of decorin in healthy (A, E) and OA cartilage (B–D, F–H). Long bar = 1000 µm, short bar = 100 µm. Abbreviation: OA, osteoarthritis.

In OA pattern I, decorin was absent in the superficial zone and fibrillated area, respectively (Fig. 4B and F). OA pattern II was characterized by a uniform decorin staining in the fibrillated area, superficial and upper mid zone, and a pericellular staining in the lower zones (Fig. 4C and G). OA pattern III was similar to pattern II, but there was no staining of decorin in the lower mid and deep zone (Fig. 4D and H).

COMP

In healthy cartilage, COMP was detectable in the superficial and upper mid zone (Fig. 5A and E). Compared with the interterritorial and territorial matrix, the staining in the pericellular matrix was sometimes slightly increased.

Figure 5.

Immunohistological staining patterns of COMP in healthy (A, E) and OA cartilage (B–D, F–H). Long bar = 1000 µm, short bar = 100 µm. Abbreviations: COMP, cartilage oligomeric matrix protein; OA, osteoarthritis.

In OA cartilage, COMP could be detected in the fibrillated area, superficial, and upper mid zone. In OA pattern I, there was a lack of COMP in the superficial zone and fibrillated area, respectively (Fig. 5B and F). Contrary to healthy cartilage, OA pattern II was characterized by a reduced or no staining of COMP in the interterritorial matrix and a pericellular accumulation of COMP (Fig. 5C and G). OA pattern III was similar to pattern II but the pericellular staining was much more pronounced (Fig. 5D and H). In all samples, there was no staining in the deeper zones.

Matrilin-3

In healthy cartilage, a uniform staining in the superficial and upper mid zone and, in part, a pericellular staining that extended to the mid zone could be observed (Fig. 6A and E). In two of four samples, the staining in the superficial zone was locally reduced.

Figure 6.

Immunohistological staining patterns of matrilin-3 in healthy (A, E) and OA cartilage (B–D, F–H). Long bar = 1000 µm, short bar = 100 µm. Abbreviation: OA, osteoarthritis.

OA pattern I was characterized by a loss of staining in the superficial zone and fibrillated area, respectively (Fig. 6B and F). In the lower zones, a matrilin-3 staining was found in the interterritorial, territorial, and pericellular matrix with the latter showing the most intense staining. OA pattern II showed a uniform staining of matrilin-3 in the superficial zone and fibrillated area as well as an increased pericellular staining in the lower zones (Fig. 6C and G). In OA pattern III, matrilin-3 was located in the fibrillated area and the staining in the pericellular matrix was increased (Fig. 6D and H).

General Distribution Patterns

To clarify the changes in the extracellular distribution of the analyzed proteins that occur in parallel in OA cartilage, the staining patterns of all proteins were compared, and two general distribution patterns could be described in 28 OA cartilage samples. In the remaining 14 OA samples, the staining patterns of the five proteins analyzed could not be assigned consistently to a distinct distribution pattern.

Distribution pattern I was found in eight OA cartilage samples (OA score 14.3 ± 4.7; Fig. 7F–J). In contrast with healthy cartilage (Fig. 7A–E), this pattern was characterized by a local distinct lack of collagen II and IX, decorin, matrilin-3, and COMP in the fibrillated area, superficial, and upper mid zone.

Figure 7.

Distribution of collagen II, collagen IX, decorin, COMP, and matrilin-3 in healthy (A–E) and OA cartilage according to pattern I (F–J) and II (K–O). The microscopic images (magnification × 20, bar = 1000 µm) for each of the three distribution patterns originate from the same samples and, thus, same location on the tibia. Abbreviations: COMP, cartilage oligomeric matrix protein; OA, osteoarthritis.

Distribution pattern II was found in 20 OA cartilage samples (OA score 18.9 ± 5.3; Fig. 7K–O). It was characterized by a uniform collagen II staining in the fibrillated area, superficial, and upper mid zone with a lack of pericellular staining. In the deep and lower mid zone, collagen II was localized in the pericellular matrix, whereas it was absent interterritorially. In the fibrillated area, superficial, and upper mid zone, all four perifibrillar adapter proteins were distributed uniformly. In addition, collagen IX, matrilin-3, and COMP showed a partly increased pericellular staining in these zones. In the lower zones, the staining of collagen IX, matrilin-3, and decorin was limited to the pericellular matrix.

Discussion

Perifibrillar adapter proteins are of great importance for proper collagen fibril formation, collagen network stabilization, and interconnection of the fibrillar and extrafibrillar matrix in articular cartilage. Although they seem to play a crucial role in the pathogenesis of OA, studies analyzing their localization in human articular cartilage during the progression of OA are rare. For the first time, we analyzed in the same sample the extracellular distribution of collagen II as main representative of the fibrillar matrix and of collagen IX, decorin, COMP, and matrilin-3 as main perifibrillar adapter proteins in healthy and osteoarthritic human knee joint cartilage. We identified specific distribution patterns for each of these proteins, and show a correlation with defined disease stages using an OA score.

In healthy cartilage, the perifibrillar adapter proteins were localized exclusively in the upper zones, whereas collagen II could be detected across all zones and in every matrix compartment. This is in accordance with previous studies that described the localization of COMP,³⁶ matrilin-3,³⁷ and collagen II⁴⁷ in healthy human articular cartilage. However, Koelling et al.³⁵ reported that collagen IX was found throughout all cartilage layers in healthy articular cartilage. The authors applied a monoclonal antibody D1-9 against the α1 chain of collagen IX whereas in our study a rabbit polyclonal antibody against collagen IX (α1/NC4) was used. This apparent discrepancy might be explained by the fact that the NC4 domain of collagen IX could be released from cartilage by matrix metalloproteinase-13.⁴⁸ The tangential collagen fibrils in the superficial zone have to resist tensile and shear stresses. They transfer compressive loads from directly loaded areas to adjacent tissue and, thus, recruit a larger area of cartilage in the deep zone to carry compressive loads.^49–51 Perifibrillar adapter proteins are not only vital for proper formation and stabilization of the collagen network in the upper zones of articular cartilage, but also seem to be of high importance for the mechanical functionality of cartilage as a whole.^31,52

Compared with healthy cartilage, all proteins analyzed showed distinct alterations in their extracellular distribution in OA cartilage, especially collagen II and COMP. A pericellular loss of collagen II in the fibrillated area and upper mid zone could be identified as the most common feature confirming previous studies.^47,53 Chondrocytes can release collagenases and, therefore, might be responsible for the destruction of collagen II in the pericellular matrix themselves.^54,55 As indicated by COMP pattern II and III, showing a pericellular accumulation of the protein, it is also possible that too high COMP concentrations relative to collagen II caused the pericellular lack of collagen. COMP catalyzes the fibrillogenesis by binding five collagen molecules at the same time and bringing them in close proximity.¹¹ However, in case of too high concentrations, COMP can act as an inhibitor by saturation of sites with single molecules precluding cross-bridging and, thus, may lead to a defective fibrillogenesis and impaired repair.¹² The distribution patterns of collagen II and COMP provided first indications of a mutual interference of the individual ECM proteins. In addition, we could identify three main staining patterns for each of the perifibrillar adapter proteins supporting the notion that OA is not a uniform disease entity. OA is more an end-stage description of various joint degeneration phenotypes with most likely slightly different pathomechanisms. As the disease is multifactorial, the complex and diverse pathomechanisms might strongly depend on the individual patient and their genetic, mechanical, and/or injury history. However, the detailed description of these varying pathomechanisms is essential for the development of new diagnostic and treatment options. Future experiments should, for example, analyze if the staining patterns correlate with biomarkers of cartilage metabolism in blood, urine, or synovial fluid. This would help to validate biomarkers for early diagnosis and progress of OA in personalized therapeutic approaches.

However, coordinated observations of the distribution patterns of all participating perifibrillar adapter proteins according to stages of disease progression have still not been carried out. In our study, we were able to describe two general distribution patterns of all analyzed perifibrillar adapter proteins that provide indications of important events in the course of disease.

With a mean OA score of 14.3 ± 4.7, lying in the middle third of the total score range, distribution pattern I can be seen as typical for moderate OA. The key characteristic of this pattern was a local lack of all proteins analyzed in the fibrillated area, superficial, and upper mid zone. Studies analyzing collagen II revealed that degradation and cleavage is initiated around chondrocytes near the articular surface, extending with progressive disease to the territorial and interterritorial matrix and the lower zones.^53–55 Distribution pattern I confirms the observation that tissue degradation starts near the articular surface and, in addition, shows that besides collagen II, also the molecules in the fibrillar periphery are depleted. There are indications that the depletion of one adapter molecule has effects on its binding partners. The association of matrilin-3 with collagen fibrils depends on the presence of collagen IX and to some extent of COMP.⁴ Thus, a lack of collagen IX leads to a dramatic reduction in COMP and matrilin-3.⁵⁶ Although we cannot draw any conclusions about the exact time course of ECM degradation, it is very likely that in our OA cartilage samples, the perifibrillar adapter proteins were depleted first. Studies inducing tissue breakdown by factors triggering chondrocytes to secrete proteolytic enzymes have shown an initial degradation of aggrecan, followed by fragmentation of molecules such as COMP and later collagen IX, and a final depletion of the collagen network.^12,34,48,57 A loss of perifibrillar adapter proteins might lead to a reduced mechanical stabilization of the collagen network in the upper zones and, thus, to an impaired functionality of the cartilage as a whole. This, in turn, could alter the loading of the chondrocytes, increase chondrocyte death, and diminish the capacity of cartilage to maintain and repair the ECM.^13,58–60 In addition, distribution pattern I provides clear indications that in early or moderate OA, a marked degradation of the proteins analyzed took place in the upper zones of human knee joint cartilage. In this way, our study supports the approach to use molecules such as COMP,⁶¹ matrilin-3,⁶² collagen IX,⁴⁸ and collagen II⁶³ that are degraded in OA and whose fragments can be assayed in body fluids such as synovial fluid, blood, or urine, as biomarkers of cartilage metabolism.¹² In conclusion, the loss of perifibrillar adapter proteins in the upper zones of articular cartilage seems to be a critical event in early OA.

With a mean OA score of 18.9 ± 5.3, distribution pattern II can be seen as a characteristic pattern for advanced OA. The main feature was a uniform distribution of all proteins analyzed in the fibrillated area. This is an interesting observation because distribution pattern II includes samples with advanced OA that are characterized by erosion of cartilage,⁴⁵ and, thus, perifibrillar adapter proteins were detectable in zones in which their evidence in healthy cartilage was missing. This might be explained by the mechanosensitivity of chondrocytes, enabling them to remodel the ECM in response to altered mechanical loading.^60,64,65 A loss of perifibrillar adapter proteins in the upper zones of cartilage, as shown in distribution pattern I, and, thus, a change in the mechanical properties of the cartilage in these zones could lead to such an altered loading environment and consequently changed biosynthetic activity of the chondrocytes, explaining the unusual occurrence of perifibrillar adapter proteins in the lower zones of cartilage. The pericellular accumulation of all proteins analyzed in distribution pattern II provides further evidence for an increased anabolic activity. Several authors could show up to five times higher mRNA levels of decorin, collagen IX, and COMP in areas adjacent to the main defect compared with macroscopically intact areas^35,36,38 and significantly increased matrilin-3 mRNA levels in cartilage with severe compared with minor OA.³⁷ Increased collagen II mRNA levels^66–68 and an elevated content of type II procollagen in the mid and deep zone of OA cartilage⁶⁹ were found. In conclusion, distribution pattern II indicates an increased anabolic activity of the chondrocytes in the remaining cartilage in advanced OA. This might be seen as an attempt to regenerate the lost tissue and to protect the remaining cartilage from further destruction.

Nevertheless, the current study also has limitations. We did not analyze the protein amounts (e.g., western blotting) or expression levels (e.g., PCR) of the perifibrillar adapter proteins, which could give further important insight into the relationship between catabolic and anabolic reactions or reexpression of cartilage proteins. The immunohistological analysis can only provide information about the status quo regarding the localization of distinct proteins so that we can merely speculate on prior pathological processes. In addition, we cannot draw any conclusions if the immunohistochemically stained proteins are intact, full length, and/or partially degraded. Due to the difficulty to get healthy human articular cartilage, we had a limited number of control samples.

In summary, we detected a uniform distribution of perifibrillar adapter proteins in the superficial and upper mid zone of healthy, mature human knee joint cartilage. This confirms their importance in the mechanical stabilization of the collagen network in the upper zones. In moderate OA, this integrity and, thus, functionality seem to be impaired due to a local lack of the collagen network and the molecules in the fibrillar periphery in the fibrillated area, superficial, and upper mid zone. In cartilage with advanced OA, there are indications for an increased anabolic activity. This could result from an altered loading of the chondrocytes due to the loss of integrity in the upper zones, and it might be seen as an attempt to restore the lost tissue or to stabilize and protect the remaining cartilage from further destruction.