Histological and morphological studies have demonstrated that embryonic motion in ovo is required for normal diarthrodial joint development. Chicks that have been paralyzed during embryogenesis develop arthrogrypotic joints (1). However, controversy exists regarding the extent to which joint morphogenesis is determined by genetic programming and modulated by mechanical factors. Specifically, it remains to be determined whether motion is critical to joint formation throughout morphogenesis or only at specific time points.
Articular cartilage enables low-friction movement of diarthrodial joints. When compressed, articular cartilage deforms to increase the joint contact surface area, thereby lowering the stress applied to the cartilage and the subchondral bone. Cartilage increases its compressive stiffness by binding water to its matrix phase (2). When articular cartilage is compressed, water is extruded from the articular cartilage and forms a thin film that helps lubricate the joint surface. When compression is removed, the cartilage returns to its original volume by reabsorbing water into the matrix. Glycosaminoglycan (GAG) is the major component of the matrix phase of articular cartilage that contributes to its unique biomechanical properties and enables it to support large compressive loads. GAG has a high, fixed negative-charge density that creates electrostatic repulsive forces. When GAG is hydrated, it binds water and swells until it is restricted by the collagen network, the other major constituent of cartilage. Negatively charged chondroitin sulfate is the predominant GAG in cartilage at birth and accounts for 85% of the total GAG content in adult cartilage (3). Together with highly anionic tyrosine-sulfate sites, GAG stabilizes the collagen network by providing multiple linkage sites between adjacent collagen fibers (2, 4). One of the earliest changes observed in cartilage degeneration is a decrease in the GAG content of the matrix phase.
Since GAG content contributes to the mechanical stiffness of cartilage, as shown in animal studies (5) and suggested by in vivo results (6, 7), the measurement of GAG can also provide a measure of cartilage health and performance. However, the histological evaluation of cartilage using stains for GAG, and the measurement of GAG content after cartilage proteolysis are destructive and cannot be used to assess GAG content in vivo. Allen et al. (8) demonstrated that one can assess local GAG content nondestructively by using MRI to measure the distribution of an administered contrast agent, gadolinium (Gd(DTPA)2−), in articular cartilage. Gd(DTPA)2− shortens the tissue-specific longitudinal T1 relaxation time measured by MRI in a concentration-dependent manner. Negatively-charged Gd(DTPA)2− distributes in inverse proportion to the concentration of the negatively-charged GAG molecules in the cartilage matrix. Thus as GAG is depleted from the cartilage, a higher concentration of Gd(DTPA)2− occurs in the cartilage and the measured T1 relaxation time is shortened. The T1 relaxation times can be mapped to reflect the GAG contents distributed throughout the articular cartilage (9). This technique, termed delayed gadolinium-enhanced MRI of cartilage (dGEMRIC), has been validated using excised human cartilage (10) and cultured chondrocytes (11), and applied in vivo to evaluate cartilage degeneration in patients following an anterior cruciate ligament tear or meniscal injury, and in individuals with different levels of physical activity (6, 12–15).
The effect of immobilization on the concentration of GAG in articular cartilage is of current interest. However, to the best of our knowledge, this has not been evaluated during embryonic limb development. In a canine study of adult animals, Palmoski and Brandt (16) showed that GAG was depleted after prolonged immobilization. In contrast, in other studies exercise increased the GAG content in the knees of healthy individuals and osteoarthritic patients (6, 7).
The chick embryo is a well-established animal model for developmental research and has been widely used to assess the effects of immobilization on the development of diarthrodial joints (1, 17–19). The staging of the embryonic development of the chick was well documented by Hamburger and Hamilton (20). In ovo paralysis of chick embryos by decamethonium bromide (DMB) interferes with diarthrodial joint development and has been used as a model for the study of arthrogryposis (1). Immobilization results in a decline in the growth rate of the bone (17) and in the disappearance of the menisci (18). Mitrovic (19) reported that DMB paralysis prevented cavitation of the cartilage anlage, which is necessary for subsequent formation of articular surfaces and intra-articular structures. Our objective was to investigate the effect of in ovo paralysis on joint morphogenesis, and to measure GAG content in the developing articular cartilage of the embryonic chick knee using the dGEMRIC technique. Conventional histologic analysis was compared with dGEMRIC for measuring GAG in the chondral structures of developing embryonic chick knees paralyzed in ovo with DMB at two different time points after gestation.
MATERIALS AND METHODS
The Beth Israel Deaconess Medical Center Institutional Animal Care and Use Committee approved all of the animal and embryo protocols used in this study. Fertilized Leghorn eggs (SPAFAS, CT, USA) were incubated in a desktop incubator (Lyon Electric Co., CA, USA) under atmospheric pressure, at 38°C and 54–59% humidity. Incubation times were scheduled based on estimates of the time of fertilization provided by the commercial supplier.
Two groups of chick embryos were evaluated: a nonparalyzed control group and an in ovo paralyzed group (Table 1). Since independent embryonic movement does not occur until 8–10 days after fertilization, and embryonic movement is reduced by day 17 after fertilization, we evaluated two time points during embryonic development of the chick knee—days 13 and 16 after fertilization—to evaluate what effect paralysis had on the progressive development of the embryonic knee joint. A postsynaptic depolarizing agent, decamethonium bromide (DMB; Sigma, St. Louis, MO, USA), was used to induce paralysis (18, 19, 21). A volume of 0.25 ml of DMB solution (0.2 mg/ml) was injected into the chorioallantoic membrane via a 2-mm-diameter hole drilled through the shell every other day starting on day 10 of incubation (Fig. 1). The access hole was covered with sterile tape after each application of DMB. A successfully paralyzed embryo at the time of harvest was defined as an embryo with a beating heart, significantly decreased muscle mass, and no observed motion for 60 s after removal of the embryo from the shell. At harvest the chick embryos were killed and their hind limbs were disarticulated at the hip joint. The hind limbs were gently rinsed in physiologic saline to avoid artifacts due to residual blood and organic debris during MRI.
Table 1. Numbers of Samples for T1 Measurement
Day 13 control
Day 13 paralyzed
Day 16 control
Day 16 paralyzed
Since the T1 relaxation time in a tissue with a Gd-complex as the contrast agent is determined by the intrinsic relaxation time and relaxivity, we compared the diffusion of uncharged Gd(HPDO3A) (ProHance®, Bracco Diagnostic Inc., Princeton, NJ, USA) to negatively-charged Gd(DTPA)2− (Magnevist®, Berlex, Wayne, NJ, USA) in the cartilage anlage of paralyzed and nonparalyzed embryonic chick hind limbs on days 13 and 16 after incubation (Table 1). The differences between the T1 relaxation times measured when the hind limbs were equilibrated in Gd(HPDO3A) vs. Gd(DTPA)2− reflects the influence of the negatively-charged GAGs on the diffusion of Gd through the cartilage. The hind limbs were equilibrated in either 1 mmol Gd/l Gd(DTPA)2− or 1 mmol Gd/l Gd(HPDO3A) for 48 hr prior to MRI. We also measured the intrinsic T1 relaxation time of the day-16 embryonic hind limbs after equilibration for 48 hr in saline.
Prior to imaging, all of the knees were rinsed with freshly prepared 1-mM Gd solution or saline. The limbs were mounted on plates and placed in 10-mm plastic tubes filled with freshly prepared Gd solution or saline. MRI was conducted using a 8.45-Tesla micro-imaging system (DRX-350™, Bruker Bio-spin, Karlsruhe, Germany). Multiple T2-weighted fast spin-echo (FSE) images were acquired in the sagittal plane of the knee joint to identify appropriate imaging planes that delineated both distal femoral and proximal tibial cartilage canals and avoided partial-volume effects. One imaging plane was selected for each sample for quantitative analysis. T1 measurements were conducted using a saturation recovery method with a set of six different repetition times (TRs) ranging from 0.05 s to 3 s (0.05, 0.1, 0.2, 0.3, 0.6, and 1.0 s with Gd, and 0.1, 0.3, 0.6, 0.9, 1.5, and 3 s with saline), 6.6-ms echo time (TE), 0.5-mm slice thickness, 10-mm field of view (FOV), 128 × 128 matrix size, and four (for day 13 samples) or two (for day 16 samples) excitations, yielding a (78 μm)2 in-plane resolution.
MRI Data Analysis
On each MR image the signal intensity in the femoral and tibial epiphyses was measured in five different regions of interest (ROIs) (Fig. 2). Magnetization was then plotted for each TR to determine the T1 relaxation times in each ROI. After the highest and lowest values were eliminated, the remaining three T1 relaxation times for each ROI were averaged to obtain a representative value of GAG content for each epiphysis. To assess the consistency of our technique, we also measured the T1 of the surrounding solution, which should be constant. Color-scaled T1 maps for each imaging plane were generated using Matlab software (The MathWorks, Natick, MA, USA).
Four specimens in each group were fixed in 4% paraformaldehyde in phosphate-buffered saline (PBS), decalcified with 14% EDTA (pH 7.4), embedded in paraffin, and cut into 5-μm-thick sections. Toluidine and Safranin O were used to stain for GAG in the cartilage anlage. Three undecalcified sections were also prepared from each group and sectioned as above. Alkaline phosphatase (ALP) activity was assessed using an enzymatic stain to indicate osteoblast activity. Silver nitrate (von Kossa) and Alizarin red stain were used to evaluate calcium deposition and the appearance of mineralized tissue.
To measure GAG content by MRI, the chick hind limbs must be freshly excised, since with time the hind-limb tissues will begin to break down. It was not possible to subject each specimen to imaging after equilibration in Gd(HPDO3A), Gd(DTPA)2−, and saline to perform repeated-measures analyses. Instead, separate groups of chick hind limbs were prepared and equilibrated in Gd(HPDO3A), Gd(DTPA)2−, and saline. All measured T1 relaxation times were expressed as the mean ± standard deviations (SDs). An analysis of variance (ANOVA) was used to compare T1 relaxation times between the control and paralyzed groups at each time point was performed for Gd(HPDO3A), Gd(DTPA)2−, and saline. Comparisons were made after 13 and 16 days of incubation. All calculations were performed using JMP IN (SAS Institute, Cary, NC, USA).
Control and paralyzed groups with Gd(DTPA)2− were compared to determine the GAG content of cartilage using dGEMRIC. T1 maps were generated for control and paralyzed hind limbs at 13 and 16 days of incubation (Fig. 2). These maps delineated the epiphyses of the knee joint adjacent to the chondral surfaces of both the distal femur and proximal tibia. At 13 days of incubation with Gd(DTPA)2−, the mean T1 relaxation times for femurs were longer in control (387 ± 18 ms) than paralyzed (357 ± 16 ms, P = 0.02) hind limbs. At 16 days of incubation, the mean T1 relaxation times for femurs were shorter in control (373 ± 20 ms) than paralyzed (417 ± 17 ms, P < 0.01) hind limbs. The mean tibial T1 relaxation times were the same on day 13 (395 ± 20 ms and 372 ± 37); however, on day 16 the T1 relaxation times were longer in the paralyzed group (395 ± 34 ms and 447 ± 42, P = 0.04; Fig. 3a).
Next, Gd(HPDO3A) was used to determine the relaxivity of the contrast agent in cartilage. Unlike Gd(DTPA)2−, Gd(HPDO3A) is an electrically neutral molecule, and thus it diffuses into the epiphyses independently of the GAG tissue concentration. At 13 days of incubation with Gd(HPDO3A), the mean T1 relaxation times of the control and paralyzed hind limbs were statistically the same in both the femur and tibia (femur: 325 ± 30 ms and 353 ± 18 ms; tibia: 341 ± 54 ms and 370 ± 70 ms). At 16 days the mean T1 relaxation times for femurs were longer in control than paralyzed hind limbs (382 ± 23 ms and 341 ± 7 ms, respectively; P < 0.01). The tibial T1 relaxation times were not statistically different (354 ± 29 ms and 353 ± 21 ms, respectively; Fig. 3b).
Since we found a difference in the femur with Gd(HPDO3A) on day 16, we measured the T1 relaxation time without contrast agent to determine the intrinsic T1 relaxation time. Hind limbs with saline showed no difference in T1 relaxation times between the paralyzed and control groups. In all cases within an experimental group, T1 relaxation times for the surrounding solutions were consistent for each measurement.
Histology with Toluidine blue and Safranin O staining demonstrated that the distal femur and proximal tibia were cartilage anlage. Cavitation and formation of the knee was more pronounced for control limbs than paralyzed limbs at 13 and 16 days of incubation (Fig. 4). The cartilage canals, the avian equivalent of secondary ossification centers in mammals, were larger in the control limbs than in the paralyzed limbs. However, staining differences were not sufficient to quantitatively assess any difference in GAG content between control and paralyzed limbs on either day 13 or day 16 of incubation. ALP activity was present on the surface of the diaphysis on days 13 and 16 of incubation in both control and paralyzed limbs. ALP was present only in the cartilage canals at the epiphyses in the 16-day incubated controls (Fig. 5). Mineralized tissue stained by silver nitrate (von Kossa) and Alizarin red was present on the surfaces of the diaphysis of both control and paralyzed limbs equally, but was not present in the cartilage canals (data not shown).
Since the diffusion of Gd(DTPA)2− into cartilage is inversely proportional to its GAG content and proportionately affects the T1 relaxation time measured by MRI, we used this technique to study nondestructively the effect of in ovo paralysis on GAG content in the developing embryonic chick knee using micro-MRI. Independent hind-limb movement does not occur until 8–10 days after incubation in normal chick embryos, and is reduced after 17 days of incubation (22). Therefore, in this study we induced paralysis from day 10 to day 16 after incubation to investigate the importance of movement on the developing chick hind limb. In agreement with other studies involving different species (23), we demonstrated that 13 days after incubation the GAG content was lower in the articular cartilage of paralyzed chick hind limbs compared to the nonparalyzed control chicks. In the adult rabbit knee, Eronen et al. (24) noted a decline in GAG synthesis in tibial and femoral condyle cartilage after 2 and 6 days of immobilization.
In the embryonic chick hind limb, 16 days after incubation, GAG content was increased in the femoral and tibial cartilage canal of the paralyzed chicks compared to control chicks. One explanation for this unexpected result is that paralysis actually caused an increase in GAG synthesis. In a previous study of adult rabbits, after 17 days of immobilization GAG content at the proximal tibia and distal femur was increased (25). Although it is possible that a similar mechanism was operational 16 days after incubation in the paralyzed embryonic chick hind limb, this is improbable. The response of chondrocytes to mechanical stress is likely quite different in a developing embryo compared to an adult; however, to our knowledge, no study has evaluated GAG synthesis during embryogenesis. A much more likely explanation is that limb movement is necessary for proper embryonic knee-joint development, and that paralysis delays proper joint development. Formation of the cartilage anlage occurs in four stages: cell migration, aggregation, condensation, and chondrification. The development of long bones from the cartilage anlage occurs by endochondral ossification, which involves chondrocyte hypertrophy cartilage matrix calcification, vascular invasion, and ossification. This process is initiated when the cells in the central region of the anlage begin hypertrophy, which increases the cellular fluid volume. Angiogenesis in the hypertrophic zone is required for the replacement of cartilage by bone. Vascular endothelial growth factor (VEGF) is released from hypertrophic chondrocytes and promotes vascular invasion. These events of cartilage matrix remodeling and vascular invasion are prerequisites for the migration of osteoclasts and osteoblasts, which begin to remove mineralized cartilage matrix and replace it with bone (4). The result that GAG increased on day 16 of incubation may also reflect delayed maturation of the cartilage canal of the paralyzed chicks, as suggested by assessment of ALP activity (which is indicative of osteoblast activity). ALP was active in the cartilage canal of the control chicks and was absent in the paralyzed chicks. The delay in the appearance of the cartilage canal of the paralyzed chick knee joint would result in a relatively larger volume of cartilage in the cartilage canal compared to the nonparalyzed control chicks on day 16 of incubation.
MR contrast agents shorten tissue-specific T1 relaxation times in a concentration-dependent manner. For tissues in which a Gd-complex is used as a contrast agent, the T1 relaxation time is determined by
where 1/T1(observed) is the observed relaxation time, 1/T1(intrinsic) is the intrinsic relaxation time, R1 is the relaxivity, and c is the concentration of the contrast agent in the tissue (26). In our study the intrinsic T1 relaxation time was identical when the hind limbs of both the control and paralyzed groups were equilibrated in saline. R1 is specific to each contrast agent, but it also depends on the surrounding tissue environment. If the tissue content of the cartilage were to change (e.g., as a result of increased protein content or the formation of mineralized cartilage), R1 might also change. This would affect the observed T1 relaxation time, provided that the tissue concentration of the contrast agent remained the same. Unlike Gd(DTPA)2−, Gd(HPDO3A) is an electrically neutral molecule, and thus it diffuses into the cartilage independently of the GAG tissue concentration. Since the R1's of Gd(DTPA)2− and Gd(HPDO3A) are otherwise similar (27, 28), if the T1 relaxation times in the control and paralyzed limbs are similar when Gd(HPDO3A) is used as the contrast agent, then any difference in the observed T1 relaxation time using Gd(DTPA)2− must be secondary to differences in the GAG content of the cartilage.
T1 relaxation times were shorter in the femoral cartilage of paralyzed chicks 13 days after incubation compared to the control chicks using Gd(DTPA)2−, while both femoral and tibial T1 relaxation times showed no differences between the two groups using Gd(HPDO3A). This suggests that initially paralysis decreased the production of GAG in the cartilage of the developing chick hind limb. However, after 16 days of incubation, when Gd(DTPA)2− was used the T1 relaxation times were longer in the femoral and tibial cartilage of the paralyzed chicks compared to the nonparalyzed control chicks. This suggests that the GAG content was higher in the paralyzed chick hind limbs compared to the nonparalyzed control chick hind limbs after 16 days of incubation. Before coming to any conclusion, we may need to consider why the T1 relaxation times were shorter in the femoral (but not the tibial) cartilage in the paralyzed chicks when Gd(HPDO3A) was used as the contrast agent on day 16. There are two possible explanations: First, the relaxivity is higher in the femoral cartilage of paralyzed chicks compared to nonparalyzed chicks, presumably because of differences in tissue composition. If this were the case, the increased relaxivity in the cartilage of the femur would lead to an underestimation of GAG content. While the T1(observed) was shorter when Gd(HPDO3A) was used, the T1(observed) was longer when Gd(DTPA)2−was used. These observations, along with Eq. , support our contention that GAG was increased in the cartilage of the paralyzed hind limbs on day 16 of incubation, since the relaxivity must change in the same direction for both contrast agents.
Using Eq. , with Gd(DTPA)2− or Gd(HPDO3A) as the contrast agents, T1 is expressed as:
where T1MV(observed) is the T1 with Gd(DTPA)2−, T1PH(observed) is the T1 with Gd(HPDO3A), R1MV is the relaxivity of Gd(DTPA)2−,. R1PH is the relaxivity of Gd(HPDO3A), cMV is the concentration of Gd(DTPA)2−, and cPH is the concentration of Gd(HPDO3A) in tissue. Therefore, assuming that R1MV and R1PH are the same (R), subtracting Eq.  from Eq.  results in
where T1PH(observed) reflects the diffusion of the uncharged Gd molecule, and T1MV(observed) reflects the additional effect of the charge on the distribution of Gd(DTPA)2− throughout the tissue (27). Therefore, Eq.  directly reflects the GAG content throughout the tissue and supports our contention that GAG content was decreased in the cartilage of the paralyzed hind limbs on day 13, but was increased in the cartilage of the paralyzed hind limbs on day 16 of incubation (Fig. 6).
The second possible explanation for the longer T1's in the control group with Gd(HPDO3A) on day 16 is that the percentage of tissue fraction in the cartilage for the contrast agent to distribute was reduced. If this were the case, a lower content of Gd(HPDO3A) in the femoral cartilage of the control limbs compared to the paralyzed limbs would be explained by a reduction in the relative volume of cartilage present on day 16 of incubation for control chicks as hind-limb maturation proceeded. Osteoblasts appeared at the cartilage canal and started to produce mineralized tissue. Since the appearance of osteoblasts was not observed, and presumably was delayed in the paralyzed chicks on day 16 of incubation, there would be a relatively higher tissue fraction of cartilage in this group.
We questioned whether our observed result was purely a consequence of immobilization of the developing hind limb or a side effect of DMB, which was used as the paralyzing agent. DMB is a postsynaptic depolarizing agent that is used extensively by other researchers to achieve paralysis in ovo (18, 19, 21). It induces paralysis in the embryonic hind limb by depolarizing the postsynaptic membrane. We validated that residual DMB in the tissues did not affect the T1 relaxation times by comparing DMB with saline imaged by MRI. Drachman and Sokoloff (1) compared DMB, botulinum toxin, and spinal cord extirpation in terms of their ability to induce limb paralysis. All three methods produced identical joint abnormalities. Therefore, the change in the GAG content of the cartilaginous tissues is likely a consequence of limb paralysis and joint immobilization rather than DMB.
In the present study, Toluidine blue and Safranin O staining, which depend on the diffusion of cations through the cartilage, did not prove sensitive enough to show any histological differences in GAG content between the paralyzed and nonparalyzed chick hind limbs. dGEMRIC allows the nondestructive quantitative measure of GAG in small, intact embryonic hind limbs. MRI clearly showed different T1 relaxation times between the control and paralyzed groups, indicating that dGEMRIC is more sensitive than conventional histologic cation stains used for this application. Quantitative staining is affected by the time duration for diffusion, pH differences, and electrolyte concentration (29). Other methods used to measure GAG content in cartilage, such as spectrophotometry with dimethylmethylene blue, may be as sensitive as MRI but are destructive (30). The limitation of this study is that we did not perform a dimethylmethylene blue assay, which is standard for GAG measurement, because the volume of chick cartilage was not sufficient for this procedure.
In summary, micro-MRI analysis of the embryonic chick hind limb allows for nondestructive imaging of the chondral surface morphology and mapping of the distribution of Gd(DTPA)2− throughout the cartilage canal of the developing chick embryonic knee joint. This technique provides a functional, nondestructive assessment of the cartilage matrix composition of the developing knee joint. An increased T1 measurement indicated that a higher GAG content existed after 16 days of incubation in the paralyzed chick hind limbs compared to the nonparalyzed controls, whereas after 13 days of incubation the femurs from the paralyzed chicks had a lower GAG content than the nonparalzyed controls. We believe that hind-limb paralysis and joint immobilization delayed the maturation of limb development, as evidenced by the absence of osteoblasts and delay in the formation of secondary centers of ossification in the cartilage canals on day 16 of incubation for the paralyzed chicks. The importance of limb movement and the effect of immobilization on proper embryonic joint development were indicated by the differences in the GAG content of the cartilage in the developing knee joint of paralyzed and nonparalyzed embryonic chicks measured nondestructively by dGEMRIC and micro-MRI.
We thank Takeshi Muneta, M.D., Ph.D., Tokyo Medical and Dental University, for helpful discussion.