|Home | About | Journals | Submit | Contact Us | Français|
Subchondral bone and articular cartilage play complementary roles in load bearing of the joints. Although the biomechanical coupling between subchondral bone and articular cartilage is well established, it remains unclear whether direct biochemical communication exists between them. The calcified cartilage between these two compartments was generally believed to be impermeable to transport of solutes and gases, previously. However, recent studies found that small molecules could penetrate into the calcified cartilage from the subchondral bone. To quantify the real-time solute transport across the calcified cartilage, we developed a novel imaging method based on fluorescence loss induced by photobleaching (FLIP). Diffusivity of sodium fluorescein (376 Da) was quantified to be 0.07±0.03 and 0.26±0.22 μm2/sec between subchondral bone and calcified cartilage and within the calcified cartilage in the murine distal femur, respectively. Electron microscopy revealed that calcified cartilage matrix contained non-mineralized regions (~22% volume fraction) that are either large patches (53±18 nm) among the mineral deposits or numerous small regions (4.5±0.8 nm) within the mineral deposits, which may serve as transport pathways. These results suggest that there exists a possible direct signaling between subchondral bone and articular cartilage, and they form a functional unit with both mechanical and biochemical interactions, which may play a role in the maintenance and degeneration of the joint.
The subchondral bone layer supports the softer articular cartilage, as both distribute the mechanical loads across joint surfaces with a gradual transition in the stress and strain1. Despite this well-established mechanical coupling between the subchondral bone and articular cartilage, direct biochemical communication between these two compartments through the calcified cartilage is poorly understood. It has been established that when the loading pattern of a joint is altered, as in osteoarthritis (OA), the subchondral bone exhibits increased bone turnover and morphological changes such as sclerosis, formation of cysts, and development of osteophytes 2-6. These changes occur with a release of cytokines and growth factors during subchondral remodeling, which may move to the articular cartilage. It is proposed that these signals initiate OA and result in a vicious cycle of positive and negative feedbacks that eventually favor OA progression7; 8. The nature of such agents and their transport pathways are not well defined.
The calcified cartilage that connects the subchondral bone with articular cartilage is generally believed to be impermeable to transport of solutes and gases9-12, but this has been recently challenged from several lines of research. Firstly, direct transport pathways have been found. Anatomical studies identified connecting “vascular channels” between subchondral bone and articular cartilage that invade calcified cartilage of normal joints and appear more frequently in aged and OA joints13-15. Also, microcracks and structural defects have been shown to penetrate into calcified cartilage in mature joints, and especially in those that have been overused, traumatized, or degenerated from OA16-18. A recent study demonstrated convincingly the existence of these pathways by measuring perfusion of small molecules (rhodamine and fluorescein) into equine calcified cartilage19. Secondly, biological interactions have been observed among the various cells present in these tissues indicating the presence of the biochemical mechanisms for communication. Co-culture studies revealed that the function and phenotype of superficial chondrocytes are modulated by the cells in the deeper tissues, including chondrocytes in calcified cartilage20 and subchondral osteoblasts7; 21; 22. The exact nature of the communicating signals is being actively pursued7; 23.
The objectives of this study are to identify the transport pathway between bone and articular cartilage, and to quantify the diffusivity of a small test molecule. We will provide results for perfusion across the calcified cartilage layer but will also utilize a novel technique, fluorescence loss induced by photobleaching (FLIP), to measure transport in real-time and calculate the diffusivity between chondrocytes and osteocytes. The methodology will increase the spatial and temporal accuracy and may be used in vivo. The results will determine if small nutrients and signaling molecules can diffuse between subchondral bone and articular cartilage in vivo and if they form one functional unit with both mechanical and biochemical interactions.
Distal femurs from adult C57BL/6J mice (3-5-month-old, The Jackson Laboratory, Bar Harbor, ME) were used. Twelve isolated femurs were used for ex vivo perfusion of sodium fluorescein (0.12 mg/mL, Sigma-Aldrich, St. Louis, MO) from either the articular surface (by vertically immersing the distal 2mm ends in the tracer solution) or subchondral bone (by injecting 5 μL of tracer solution into the epiphyses using a 30G needle inserting into the marrow cavity and passing through the growth plate) for 2, 19 or 69 h at 4°C. Care was taken not to damage the subchondral bone and articular cartilage during perfusion. For in vivo perfusion studies, eight live animals were tail injected with a bolus of sodium fluorescein solution (10 mg/mL, 0.5 mL), and sacrificed 20 min later. Distal femurs were harvested. The procedures were approved by the Institutional Animal Care and Use Committee.
The femoral distal ends (~ 2-3 mm) were sagittally split between the two condyles and trimmed at -40°C using a cryomicrotome (Leica, Wetzlar, Germany) equipped with a diamond knife (Diatome, Switzerland). Samples were mounted on a cover glass and imaged using an inverted confocal laser scanning microscope (Zeiss LSM 510, Standort Göttingen, Germany) equipped with a 40× (1.2NA) water immersion lens (Zeiss Korr UV-Vis-IR). Tracer distribution was examined 5-20 μm below the cutting surface using 488/520 nm excitation/emission and 1024 × 1024 pixel images captured at a speed of 15.73 sec per frame.
Two murine distal femurs were harvested, fixed in 4% (w/v) paraformalderhyde and treated with osmium tetroxide (EM Sciences, Hatfield, PA). The dehydrated samples were embedded in Spurr resin (EM Sciences). Thin sections of 60-80 nm thickness were collected and stained with alcoholic uranyl acetate, followed by Reynolds lead citrate. Images were taken using a Philips CM-12 transmission electron microscope (FEI, Hillsboro, Oregon) at an 80 kv accelerating voltage. EM photographs of five random extracellular matrix areas (~1.4 μm2/location) were converted to binary images using the mean gray levels as the thresholds. The mineralized pixels were assigned “0” and non-mineralized pixels assigned “1.” The volume fraction of the non-mineralized space was reported as the percentage of non-mineralized pixels relative to the total pixels. The linear dimensions of the ellipse-shaped non-mineralized spaces were obtained by measuring and averaging the diameters in the major and minor directions, and the spacing of the regularly distributed non-mineralized channels was obtained by measuring a group of channels using custom Matlab codes.
The anterior regions of the eight in vivo perfused femurs were studied since there were fewer soft tissue attachments. The focal plane was set 5-20 micron below the cutting surface to avoid artifacts and to limit the transport in the depth (z) direction. A single spherical chondron was readily identified in the well-perfused calcified cartilage and outlined using the region of interest (ROI) tool in the microscopy software. By continuously photobleaching this chondron, a transport “sink” was created where the unbleached tracer concentration remained low and thus drove the unbleached tracer moving out from the neighboring chondrons or osteocyte lacunae (“source”). These dynamic out-fluxes were used to calculate the tracer diffusivities between chondrons and between osteocyte lacuna and chondron, as detailed in the next section. The scan settings were 488/520 nm for excitation/emission wavelength, 512 × 512 pixel image size and a scan speed of 3.93 sec per frame. The FLIP procedure consisted of two pre-bleach scans of the entire field using low-intensity laser (0.5-5% transmission), followed by 50-200 repetitions of photobleaching the selected ROI (100% transmission) and one post-bleach scan (0.5-5% transmission). It usually took ~500-2000 sec until a steady state of intensity was achieved in sources. Less than six FLIP experiments were performed per sample.
A bio-spherical model has been developed to describe the movement of unbleached tracer from the source into the sink through the ECM of the calcified cartilage and subchondral bone during the FLIP experiments (details in Appendix).24 Assuming the fluorescence intensity is proportional to the tracer concentration, the normalized intensity in the source was predicted to follow an exponential drop:
where I(t) is the fluorescence intensity for the source at time point t; I0 is the intensity at the source immediately after the concentration of the sink is decreased and remains at a low level, Istd is the steady state intensity of the source; α is the correction factor (2.5-4) accounting for the unsteady diffusion as a function of the size and spacing of the source and sink (detailed the appendix and also ref24); D is the tracer diffusion coefficient. The term in the brackets at the right side of the equation indicates the influence of the anatomical features, η1 and η2 are the biospherical coordinates on the surfaces of the source and sink [ ], d is the distance between the source and sink centers, R1 and R2 are the source and sink radii 24.
The size and spacing of the source and sink (R1, R2, d) were first obtained from the pre-bleach images using the overlay function in the microscopy software. The time courses of the fluorescence intensities I(t) and I0, Istd were calculated for each source, and corrected for the autofading during the recording period using a reference (e.g., an osteocyte lacuna far away from the photobleached region). The diffusion coefficient of the tracer D for each source-sink pair was calculated from the slope of the fitting line (ln vs. time). From the total 8 animals, a total of 40 FLIP source-sink pairs were analyzed to calculate the diffusion coefficients between chondron and chondron (C-C) in calcified cartilage (n = 26) and between osteocyte lacuna in the subchondral bone and chondron (O-C) (n = 14). The difference between C-C and O-C diffusions was tested using a two-tailed Student's t-test with a significance level of p = 0.05.
A spatially inhomogeneous staining was found through the distal femur in all samples, as seen after 2 h in vitro perfusion (Fig. 1). Directly beneath the synovial surface, a band (20-50 μm) of strong staining was shown in the superficial uncalcified cartilage (UC) and abruptly discontinued at the wavy tidemark (TM). Punctuated staining of hypertrophic chondrocytes (C) was seen in the calcified cartilage (CC). The osteochondral interface (OI, white dashed line, Fig. 1) between CC and subchondral bone plate (SB) could be inferred from the signature lacunar-canalicular system of osteocytes (OC). The contour of the OI was very complicated and some times in direct contact with UC (OI-C, big solid arrows, Fig. 1). The overall staining pattern remained the same for various perfusion times (2, 19, 69 h), perfusion directions, and for both ex vivo and in vivo perfusion (data not shown).
The calcified cartilage contained non-mineralized regions with two distinct patterns: i) discrete, larger non-mineralized patches (NM) among denser mineral deposits (MD), and ii) numerous smaller linear spaces within the whisker-like mineral deposits (Fig. 2). Dispersed among the randomly distributed MDs (Fig. 2A) were patches of NM varying from 20 to 75 nm in linear dimensions (~53 ± 18 nm, n = 42, Fig. 2B). Inside individual MD, a quasi-periodic mineralized fibrous mesh (MFM) and non-mineralized spots () and channels (→) in size of ~3-6 nm (4.5 ± 0.8, n = 52) were shown (Fig. 2C). The total volume fraction of the non-mineralized space was ~ 22%. Transport can occur through these non-mineralized spaces, which may contain organic materials (e.g., proteoglycans) that are invisible in the EM and that may reduce the effective pore sizes for solute and fluid transport.
Dynamic transport between chondrons themselves and between chondrons and osteocyte lacunae was recorded during FLIP experiments, and a good agreement was achieved between the model prediction and experimental data (Fig. 3). A representative FLIP experiment is illustrated in Fig. 3, with the repetitively photobleached chondron (1, sink) and its surrounding chondron (2, source) and osteocyte lacuna (3, source) as well as a reference osteocyte for autofading correction (4, Fig. 3A). Within 20-50 sec after FLIP was initiated, the intensity of the sink chondron was rapidly decreased and remained at a low level. This lower sink concentration (I0) caused the intensity drop in the surrounding source cells (2 and 3) until a steady state (Istd) is reached. Tracer fluxes between source chondron and sink chondron (C-C: 2→1,) and between subchondral osteocyte lacuna and sink chondron (O-C: 3→1) are shown in Fig. 3B. As predicted from our model, the experimental data fit a logarithmic function [(ln[(I(t)-Istd)/(I0-Istd)] vs. time with good agreement (R2 > 0.85) and the apparent diffusion coefficient (D) was obtained for transport between C-C and O-C (Fig. 3C). The Ds for C-C pairs (n = 26) were 0.05-0.9 μm2/s with a mean ± std of 0.26±0.22 μm2/s, while those for O-C (n = 14) were significantly smaller (p = 0.0003), in a range of 0.02 to 0.12 μm2/s with a mean ± std of 0.07±0.03 μm2/s, (Fig. 4).
The present study demonstrated that solute transport between subchondral bone and articular cartilage did occur through a permeable calcified cartilage utilizing a novel FLIP method. Quantitative measurements of the diffusivity of a small testing molecule (sodium fluorescein, 376 Da) suggest that similar sized small nutrients and signaling molecules may be able to perfuse over tissues in mature joints in vivo. These results support the concept that subchondral bone and articular cartilage are one functional unit with both mechanical and biochemical interactions.
This approach was designed to directly visualize and measure intrinsically slow dynamic transport processes with a high temporal-spatial resolution. The method overcomes the limited accuracy of determining the spatial tracer profile in early studies19, and has the obvious advantage of introducing the tracer in live animals, eliminating the time-consuming tracer permeation process10; 19. FLIP has been successfully used in studying anisotropic diffusion in articular cartilage25. Compared with the more commonly used fluorescent recovery after photobleaching (FRAP),26 FLIP utilizes repetitive photobleaching to create a greater driving force for solute movement and thus improve the detection sensitivity. Because of the small dimensions of the photobleached chondrons (~ 7 μm radius) and the short distance among the transport sink and source (~ 25 μm), the dynamic responses, tracer out flux, can be observed within a reasonable time frame (~20 min), even when the permeability is on the order of 0.1 μm2/s. Since sink is maintained in our FLIP approach, transport can be forced to cross the osteochondral interface, which is difficult to achieve using FRAP26.
The apparent diffusion coefficient (D) of sodium fluorescein was 0.26 μm2/s in calcified cartilage, comparable with the 0.9 μm2/s obtained previously in equine calcified cartilage samples19. Although diffusion in the deeper layers of articular cartilage and subchondral bone is one hundred-fold slower than in the superficial layer of the cartilage10; 27; 28, perfusion of these small biological molecules over the entire joint is, in fact, surprisingly fast. Molecules similar in size to sodium fluorescein, such as prostaglandin E2, nitric oxide, or glucose, could readily perfuse through the mineralized joint tissues over a distance of 40 μm within half an hour (i.e., √ (4Dt) = 43 μm, for D = 0.26 μm2/s and t = 0.5 h) (Fig. 1). Our preliminary tests showed that parvalbumin and lysozyme (12-14 kDa) perfused over 2 hours (data not shown). Both big and small non-mineralized regions (50 vs. 5 nm) were identified in calcified cartilage (Fig. 2). We believe that solute transport occurs through the pores within these spaces and the level of tissue mineralization may influence the sizes and porosity of the pores embedded in the calcified cartilage. The spatial distribution of the non-mineralized spaces may be quite heterogeneous and results in the large diffusivity variation observed in calcified cartilage. This needs further investigation. Most importantly, our results suggest the osteochondral interface is permeable to solute transport, although solute diffusivity over the interface is reduced 3-4 fold compared to that of the calcified cartilage. Using FRAP, the diffusivity of sodium fluorescein within the discrete bone lacunar-canalicular system was found to be ~330 μm2/s. 26 Assuming the porosity of the LCS in the subchondral bone is 5%-10%, the averaged diffusivity in the bone tissue is expected to be ~16-33 μm2/s, which is approximately two orders of magnitude higher than the diffusivity measured in calcified cartilage. Therefore, the reduced diffusivity across the osteochondral interface is unlikely due to calcified cartilage and subchondral bone and is very likely due to the materials deposited along the interface.12
A permeable calcified cartilage layer implies that communication is possible between chondrocytes in different zones as well as among chondrocytes and bone cells or stromal cells in the bone marrow. This communication is involved in the modulation of cartilage mineralization and metabolism in vitro7; 20; 22. The cytokines and growth factors released during subchondral bone remodeling could reach the overlying articular cartilage and be involved in the initiation and progression of OA 8. How these transport characteristics change in OA joints needs to be established. We expect an overall increase of transport in OA because of the larger variations of mineralization29, which may result in increased permeability in some locations. In advanced stages OA, vessel invasion14; 15 shortens the distance between bone and uncalcified cartilage, bringing bone and the marrow much closer to articular cartilage and increasing hydraulic conductance29. Normal joints may only occasionally contain such close contact locations (OI-C, Fig. 1). The present study suggests that both mechanical and biochemical coupling between bone and cartilage may be involved in OA progression and may be a possible pathway for pharmaceutical treatment 8; 15; 23; 30.
The relative opaqueness of the joint tissues and limited laser penetration depth required that the knee joint be surgically opened in our studies. The use of a multi-photon microscope could offer deeper penetration and allow imaging study to be repeated in live animals. Other concerns were the effects of cryosectioning and photobleaching on the cells and tissues during sample preparation and FLIP experiments. Since the measurements focused on transport through the interstitial mineralized matrix that contained limited water, this possible damage was not expected to influence our final results. Finally, the mathematical model had several assumptions including macroscopically homogeneous matrix and neglecting the other surrounding cells 24. Although a correction factor was introduced to account for the unsteady diffusion in FLIP, the apparent diffusion obtained here was an averaged value from the initial faster diffusion to the later slower diffusion. 24
We thank Dr. C. Ni for help in cryosectioning, Ms. O. O'Shea and Mr. T. Labassiere for electron microscopy, Drs. R.L. Duncan, M.C. Farach-Carson, L.P. Wang, and S. Advani for insightful discussion. This study was supported by grants from NIH (P20RR016458; R01AR054385), and University of Delaware Research Foundation, Project 111(B0602, China).