Age, but not short-term intensive swimming, affects chondrocyte turnover in zebrafish vertebral cartilage

Zebrafish strain and maintenance

The AB wild-type line of zebrafish purchased for exercise experiments (GenDanio Aquaculture System, New Taipei City, Taiwan) were randomly segregated into two groups: control (Ctrl) vs. exercise (Exe), and all the comparisons between before (+0 days) vs. after (+14 days) exercise training program were done with these zebrafish. The AB wild-type line of zebrafish at 3–4 months of age and 11–13 months of age were purchased (Azoo Co., Taipei, Taiwan) for the age comparisons (4- vs. 12-month-old). All zebrafish were kept individually (in 200 mL water) at 28.5 °C with a light cycle of 14-h light/10-h dark and were fed twice daily for this study (Chang et al., 2016; Westerfield, 2000). All experimental procedures in this study were reviewed and approved by the Institutional Animal Care and Use Committee of National Taiwan University (NTU105-EL-00037) and were performed in accordance with the approved guidelines.

Zebrafish intensive exercise training

To force zebrafish to go through intensive exercise training, a simple training system was designed and assembled with an aquatic powerhead (Rio+1400; Technological Aquatic Association Manufacturing, Thousand Oaks, CA, USA) connected to a polyisoprene tube (52 cm in length, 2.5 cm in diameter) and a mesh covering the end opening (Fig. 1A). Each system housed an individual zebrafish in the tube during the training session and a 60-L polyethylene tank housed three training systems placed in a stack fashion.

Each zebrafish in the exercise group was assessed for their maximal resisting speed. Briefly, the flow of the aqua pump was increased every minute until the zebrafish fail to resist and reside the mesh (fail speed). The maximal speed (i.e., the flow speed immediately before the fail speed) was then calculated according to the milliliter/minute of the aqua pump and the cross-sectional area of the tube (Table 1). Each zebrafish in the exercise group was transferred into the system 30 min after the morning feed and rested for 30 min before the 8-h training session at maximal speed began (Fig. 1B). After the training session, the zebrafish was transferred back to the housing system, received an excessive night feed. The training session lasted for 14 days, and the control group in this experiment was managed in the same way without turning the aqua pump on.

Body measurements

To assess the effects of age and exercise training on overall physiological condition, body weight and body length of each zebrafish were measured and recorded. Briefly, each zebrafish was anesthetized in 0.016% ethyl 3-aminobenzoate methanesulfonate (MS-222; Sigma-Aldrich, St. Louis, MO, USA) before the morning feed. The body weight was measured on a precision balance (PJ3600; Mettler-Toledo, Columbus, OH, USA) after excessive water was removed and a photograph was taken to measure the body length from the mouth tip to the end of the tail-fin using ImageJ (Schneider, Rasband & Eliceiri, 2012).

Micro-computed tomography and bone mineral density

To assess the effects of age and exercise training on the skeletal system, bone mineral density (BMD) of each zebrafish was estimated using images from micro-computed tomography (microCT). Briefly, zebrafish were anesthetized using a mixture of 100 ppm MS-222 and 100 ppm isoflurane (Abbott Laboratories, Queenborough, UK) in water (Huang et al., 2010), restrained between two wet sponges, and scanned (SkyScan-1076; Bruker microCT, Kontich, Belgium) with nine μm resolution, 80 kV, 124 μA, 0.5° rotational step, 1,700 ms exposure and a 0.5 mm aluminium filter (Table S1). To standardize and calibrate the intensity, two scanning phantoms with 0.25 and 0.75 g/cm³ of densities were used. The scale was designed according to a pre-defined parameter (air, HU = −1,000, color index = 0) and scanning results (including water and phantoms) to generate the mapping reference between color index (0–255) and Hounsfield units (HUs; in our case, −1,000 to 3,184). The 3D rendering was done by using CTvox software (Bruker microCT, Kontich, Belgium) and the BMD was analyzed by using CTAn software (Bruker microCT, Kontich, Belgium) with the fourth (a Weberian vertebra) or all vertebrae as the region of interest (Bird & Mabee, 2003; Hur et al., 2017).

Labeling and tracing of 5-bromo-2’-deoxyuridine (BrdU)

To understand the effect of age and short-term intensive exercise training on chondrocyte proliferation, 5-bromo-2’-deoxyuridine (BrdU; Sigma-Aldrich, St. Louis, MO, USA) was used to label the proliferative cells. For age comparisons, the zebrafish were anesthetized in 0.016% MS-222 3 h after the night feed with five μL of BrdU (2.5 mg/mL in distilled water) were administered via oral gavage once a day for consecutive 15 days (Reimer et al., 2008). To reduce stress for the exercise comparisons, zebrafish were immersed in 200 mL of BrdU (150 μg/mL) each day during the dark period of the 2-week training session (Rowlerson et al., 1997). To determine whether quiescent cells exist in the cartilage, the BrdU labeled zebrafish were chased for an additional 30 days.

Histology preparation

To observe the effect of age and short-term intensive exercise training on the morphology and composition of cartilage, qualitative and quantitative histology were analyzed. Briefly, the zebrafish were sacrificed in 20 mL 0.4% MS-222, and fixed in 20 mL 4% paraformaldehyde (Merck, Darmstadt, Germany) at 4 °C for 5 days after the abdomen was opened with a scalpel for better penetration of the fixative. The zebrafish were then immersed in 20 mL of 10% ethylenediaminetetraacetic acid (EDTA, Amresco, Solon, OH, USA) for 3 days for de-calcification. After the fixatives and EDTA were washed away with water, dehydration was with 20 mL each of a 30–100% ethanol gradient. The sample was then immersed two times in 20 mL of xylene (J.T. Baker, Center Valley, PA, USA) for 1 h each and embedded in paraffin (Surgipath Medical Industries, Richmond, IL, USA). The tissue blocks were sectioned at five μm to produce consecutive sagittal sections using a rotary microtome (HM315; Microm, Walldorf, Germany). The tissue slides were then rehydrated using xylene followed by a 100–30% ethanol gradient, and finally immersed two times in phosphate buffered saline (PBS; Amresco, Solon, OH, USA) for 5 min. From the most lateral edge of the vertebral column to the midline, about 30 (in 4-month-old) or 40 (in 12-month-old) consecutive sections could be obtained. For all quantitative image analysis, five sections from the same subject with consistent 20 μm (in 4-month-old) or 25 μm (in 12-month-old) interval between sections were used for the same staining analysis and the sum of the results from five slides represented the result for the subject.

Histochemistry

To observe the GAG content in the cartilage, the tissue slides were immersed in hematoxylin (Surgipath Medical Industries, Richmond, IL, USA) for 5 min, washed with 1% HCl (Merck, Darmstadt, Germany) and distilled water, immersed in 0.02% fast green (Merck, Darmstadt, Germany) for 1 min, washed with 1% acetate (Merck, Darmstadt, Germany) and distilled water, immersed in safranin O (Merck, Darmstadt, Germany) for 10 min, washed with 95% and 100% ethanol, and finally sealed with mounting medium (Muto Pure Chemicals, Tokyo, Japan).

Immunohistochemistry

To observe the distribution of type II collagen and BrdU labeling/retention in the cartilage, immunohistochemistry was performed. Briefly, after the tissue slides were rehydrated, epitope retrieval was achieved by proteinase K (20 μg/mL; GeneMark, Taichung, Taiwan) incubation at room temperature for 30 min (for type II collagen). For BrdU, incubation was with sodium citrate 2.94 mg/mL, pH 6.5; Sigma-Aldrich, St. Louis, MO, USA) at step-up temperatures from 65 to 95 °C for 10 min. After washing with 0.1% Phosphate Buffered Saline (PBS) with Tween-20 (PBST) (Amresco, Solon, OH, USA) twice, the tissue slides were blocked with blocking buffer (3% bovine serum albumin (Sigma-Aldrich, St. Louis, MO, USA) in PBS) at room temperature for 30 min and incubated with the primary antibodies against type II collagen (1:10 in blocking buffer; Developmental Studies Hybridoma Bank, Iowa City, IA, USA) for 2.5 h or against BrdU (1:100 in blocking buffer; AbD Serotec, Kidlington, UK) for 30 min at room temperature. After washing with 0.1% Tween-20 in PBS, the primary antibodies were detected by goat anti-mouse IgG conjugated with Alex Fluor 555 (1:300 in blocking buffer; Abcam, Cambridge, UK) for 1 h at room temperature. The nuclear counter-staining was done by 4′,6-diamidino-2-phenylindole (DAPI, 10 mg/mL; Biotium, Fremont, CA, USA) and the tissue slides were sealed with Fluoroshield mounting medium (Abcam). A primary-free, secondary-only antibody staining was used as a negative control.

Terminal deoxynucleotidyl transferase dUTP nick end labeling (TUNEL)

To assess the effect of age and exercise training on chondrocyte apoptosis, the terminal deoxynucleotidyl transferase (TdT) dUTP nick end labeling (TUNEL) assay was performed using a commercial kit (In Situ Cell Death Detection Kit, TMR red; Roche, Basel, Switzerland). Briefly, after rehydration, the tissue slides were perforated with proteinase K (20 μg/mL) for 15 min at room temperature and washed with PBST. The tissue slides were then incubated with the TdT and dUTP mixture (1:10) at 37 °C for 1 h, counter stained with DAPI, and sealed. A DNase I (two U/μL; Geneaid Biotech, Taipei, Taiwan) treatment at room temperature for 10 min after perforation was used as a positive control, while a TdT-free reaction was used as a negative control for this staining.

Micrographs and image analysis

All the tissue slides were documented by a tissue scanner (TissueFAXS;, TissueGnostics, Vienna, Austria) or by a confocal microscope equipped with differential interference contrast (TCS SP5 II; Leica Microsystems, Wetzlar, Germany). For the quantitative image analysis, three (for type II collagen) or five slides from the same subject with a consistent interval between sections were used for the same staining analysis. The sum of the results represented the subject. The software HistoQuest (TissueGnostics, Vienna, Austria) was used to quantitatively analyze the contents of GAGs and type II collagen in the cartilage, while ImageJ was used to count the nuclei with DAPI, BrdU or TUNEL staining (Schneider, Rasband & Eliceiri, 2012).

Scanning electron microscope

For the scanning electron microscope (SEM) imaging, the tissue sections were collected onto the cover slips (32 × 24 mm). The sections were then de-waxed twice in xylene for 10 min, then three times in 100% ethanol for 10 min each followed by two immersions in acetone for 10 min. Each slide was then critical point dried in liquid CO₂ in a critical point dryer (Hitachi, Tokyo, Japan) and ion coated (IB-2, Eiko, Tokyo, Japan) before documentation in the SEM (Inspect S, FEI, Hillsboro, OR, USA).

Statistical analysis

To minimize the potential bias brought about by small sample sizes, non-parametric statistical approaches were used in this study. To minimize the influence due to individual variations, the Wilcoxon matched-pairs signed rank test was used for the comparisons between before and after exercise training sessions. For the comparisons between age groups (4-month-old vs. 12-month-old) and exercise groups (exercise vs. control), the Mann–Whitney U-test was used. For the comparisons among 0-, 15- and 30-day chase, Kruskal–Wallis test was used. Statistical significance was considered when P ≤ 0.05.

Zebrafish continues to grow after sexual maturity while intensive exercise hinders this growth

In mammals, hormones such as estrogen fluctuate dramatically during sexual maturity and trigger the halt in skeletal growth including the closure of epiphyseal plates in bones (Zhong et al., 2011). The body measurements of zebrafish indicate that zebrafish continues to grow after sexual maturity as the body length increased significantly from 2.65 cm at 4 months of age to 3.12 cm (Fig. 1C; Table S2). Intriguingly, although the body length of the control zebrafish was not significantly changed in 2 weeks, the body length of zebrafish was significantly shortened after the 2-week intensive exercise-training program (Fig. 1D; Fig. S1 and Table S2).

Similarly, the body weight was significantly increased from 0.16 g in 4-month-old zebrafish to 0.23 g at 12 months of age (Fig. 1E; Table S3). The body weight continued to increase during the 2-week experimental period in the 12-month-old control group, but the body weight was not altered in the zebrafish experiencing intensive exercise training (Fig. 1F; Table S3). These results indicate that the zebrafish body continues to grow between 4 and 12 months of age, especially the body weight, whereas short-term intensive exercise halts the growth.

The BMD continues to increase after sexual maturity while intensive exercise negatively affects this trend

Previous studies in the human skeletal system indicate that the BMD peaks between 30 and 40 years of age. Appropriate nutrition and exercise enhance the BMD level, but aging especially in females after 50 years of age increases the risk of osteoporosis and OA (Lee et al., 2013; Warming, Hassager & Christiansen, 2002). To assess the potential impact of short-term intensive exercise on the vertebrae column, microCT scan with the fourth vertebrae selected as region of interested was used to estimate the BMD (Figs. 2A and 2B). The result shows that the BMD in the fourth vertebrae continued to grow significantly in the control group during the 2-week experimental period (Fig. 2C; Table S4). The BMD level remained at comparable levels before and after the 2-week intensive exercise training in the exercise group (Fig. 2D; Table S4). These results indicate that zebrafish BMD level in the fourth vertebrae continues to increase even at 12 months of age, but the short-term intensive exercise hinders this increase in BMD.

Zebrafish continue to accumulate cartilage ECM after sexual maturity

The GAGs and collagens, especially type II collagen, are the predominant constituents of articular cartilage, and the networking of these macromolecules buffers and disperses the mechanical pressures applied to the joints (Hedlund et al., 1999; Van Der Rest & Mayne, 1988). One of the signatures of OA is the loss of these ECM components (Pritzker et al., 2006). Previous study shows that, as the zebrafish ages, the fourth and fifth vertebrae develop bone and cartilage deformities similar to OA symptoms in humans (Hayes et al., 2013). To assess the effects of age and mechanical loading on ECM accumulation in cartilage, histochemistry, immunohistochemistry and the quantitative analysis of micrographs were performed with the fourth Weberian vertebra (Figs. 3A and 3B) as the region of interest since it includes the largest cartilage ECM, including type II collagen (Fig. 3A) and GAGs (Fig. 3B), compared to other vertebrae. Furthermore, the cartilage at the fourth Weberian vertebra resembles the histological features of hyaline cartilage as it contains no prominent collagen fibers (bracket in Fig. 3C) and is surrounded by fibrous perichondrial organization (arrow and bracket in Fig. 3D).

Interestingly, the distribution of type II collagen was more prominent at the ventral and dorsal end of the cartilage in 4-month-old zebrafish (Fig. 4A), but much more prominent at cartilage margins in 12-month-old zebrafish with no discernible difference between exercise and control groups (Figs. 4B and 4C). Quantitative analysis of immunohistochemical micrographs of type II collagen show significant lower levels in both distribution areas (Fig. 4D; Table S5) and signal density (Fig. 4E; Table S6) in 4-month-old zebrafish compared to 12-month-old zebrafish. On the other hand, both the distribution area (Fig. 4F; Table S5) and signal density (Fig. 4G; Table S6) of type II collagen were at comparable levels in exercise and control groups.

The histochemistry of safranin O, fast green and hematoxylin can provide well-discerned depictions of GAGs, collagens and cell nuclei in a tissue (Figs. 5A–5C). Compared to type II collagen (Figs. 4A–4C), the GAGs were more prominently distributed in the core of the cartilage, especially in 12-month-old zebrafish, while occupying a larger area in the vertebra (Figs. 5A–5C). Interestingly, although the GAG area size was significantly smaller in 4-month-old zebrafish than in 12-month-old zebrafish (Fig. 5D; Table S7), the signal densities were at a comparable level (Fig. 5E; Table S8). Similar to type II collagen, both the distribution areas (Fig. 5D; Table S7) and signal densities (Fig. 5E; Table S8) of GAGs were at comparable levels between exercise and control groups. Taken together, the short-term intensive exercise training does not result in discernible change in the zebrafish cartilage, while the cartilage continues to grow after the sexual maturity of zebrafish at 4 months of age. Interestingly, the accumulation of type II collagen seems less mature than GAGs in 4-month-old zebrafish, since the signal density of GAGs but not type II collagen remained at comparable levels between 4 and 12 months of age.

Cellular dynamics decreased with age

Since the homeostasis of cartilage ECM, especially the GAGs, depend on the balance of catabolism and anabolism of chondrocytes, it is essential to evaluate the chondrocytes in the cartilage. As hematoxylin staining and the existence of lacunae clearly depicted the distribution of chondrocytes and their nuclei (Figs. 5A–5C), the cell counts and cell densities were also evaluated using the same tissue slides. The zebrafish vertebral chondrocytes did not distribute with apparent orientations, but the cells lacking surrounding lacunae were predominately located at marginal regions (Figs. 5A–5C). The 4-month-old zebrafish had significantly greater cell density than the 12-month-old zebrafish (Fig. 5F; Table S9), but the cartilage in 12-month-old zebrafish contained more cells (Fig. 5G; Table S9). Again, both the cell densities and cell counts were at comparable levels between the exercise and control groups. These results indicate that, between 4 and 12 months of age, the continuous growth of cartilage is contributed both by the accumulation of ECM and by the increase in chondrocytes.

To further elucidate whether the increase in chondrocytes was a static accumulation or a result of a dynamic equilibrium, TUNEL staining and BrdU labeling were performed. The TUNEL staining indicates that the apoptotic cells were predominantly located at the outer regions of the vertebral cartilage (most anterior/posterior and dorsal/ventral tips) (Fig. 6A; Table S10). Quantitative analysis shows that cartilage in 12-month-old zebrafish contained a significantly greater percentage of apoptotic cells than cartilage in 4-month-old zebrafish (Fig. 6B), whereas the exercise group was not significantly different from the control group (Fig. 6B; Table S10). On the other hand, BrdU labeling (Fig. 6C; Table S11) indicates that cartilage (within GAG-positive as well as type II collagen-positive regions) in 4-month-old zebrafish contains far greater percentages of proliferating cells than cartilage in 12-month-old zebrafish, and intensive exercise training does not alter the proliferative potential of chondrocytes (Fig. 6D; Table S11). Interestingly, the average percentage of BrdU-positive cells in 12-month-old zebrafish was merely 0.117% (12-month-old), 0.055% (control group) and 0.045% (exercise group) in contrast to the 3.24% in 4-month-old zebrafish, while there were no BrdU-positive cells in many of the 12-month-old cartilage. These results indicate that the cellular renewal is gradually lost as zebrafish aged.

Although the BrdU-positive cells were sporadically distributed in the cartilage with no specific localizations, some of the BrdU-positive cells were located in the peripheral regions where the GAGs and type II collagen were not accumulated (Fig. 6C; Table S11). These cells resided at a location resembling perichondrium with elongated nuclei similar to superficial cells in mammalian articular cartilage. Recent studies imply that these perichondrial superficial cells could serve as stem cells or progenitor cells to provide new cells for chondrocyte turnover (Candela et al., 2014; Karlsson et al., 2009; Li et al., 2017). To assess whether stem cell-like quiescent cells are present in zebrafish cartilage, BrdU pulse-chase was performed in attempting to look for cells retaining the label (Fig. 6E; Table S12). However, we found no BrdU-positive cells in the entire fourth Weberian vertebra in any 12-month-old zebrafish samples (data not shown). In the 4-month-old zebrafish cartilage, the BrdU-positive percentages decreased from 3.24% to 0.58% after a 15-day chase and 1.69% after a 30-day chase (Fig. 6E; Table S12) with no statistical significance. Furthermore, as HMGB2 was previously reported a potential molecular marker for mesenchymal stem cell-like chondrocytes in mouse articular cartilage (Taniguchi et al., 2009), immunohistochemistry using anti-HMGB2 antibody also found no cells being labeled in the entire fourth Weberian vertebra in both 4- and 12-month-old zebrafish (data not shown). Taken together, age indeed affects chondrocyte dynamics and we found no evidence to suggest the existence of cartilage stem cells in mature zebrafish at 12 months of age.