RT-qPCR is a highly sensitive method for mRNA quantification in different samples. This technique requires an appropriate normalization of the data to minimize possible differences due to technical variations [21]. Reference genes are commonly used for transcript normalization because of their stable expression [25]. However, a large number of studies show that universal reference genes do not exist, and different experimental methods need to establish their own reference genes to ensure an accurate normalization strategy [23,26–28].

When damaged, the self-repair ability of cartilage tissue is very poor [29]. Cytokines or growth factors are indispensable for augmenting chondrocyte proliferation and cartilage regeneration [30–32]. By local administration, cytokines can increase thickness of the normal cartilage and can accelerate the repair of cartilage defects [33,34]. Cytokines exert their cartilage regeneration-enhancing abilities mostly by stimulating the expression of proliferative genes. In this study, we adopted IGF-1 as the therapeutic factor for repairing cartilage defects, and we applied it to samples at different stages of chondrocyte development to facilitate the determination of the most stably expressed reference genes in regenerating cartilage.

Artificial cartilage defects were induced in groups B and C by surgery, and the repair of defects was detected by histochemistry. H and E staining (Figure 1) confirmed that cartilage with this type of artificial defect had limited self-repair capacity; treatment with IGF-1 markedly stimulated chondrocyte proliferation and the formation of extracellular matrix, indicating that IGF-1 increases the rate of cartilage regeneration. Therefore, the regenerative status of the cartilage was different among the experimental groups.

We analyzed the RT-qPCR data using two different algorithms to determine the optimum number of, and the most stably expressed reference genes.

The geNorm and NormFinder algorithms are widely used and are based on divergent mathematical approaches. geNorm analyzes the expression stability of selected genes in all samples by calculating M values (M). However, NormFinder evaluates the expression stability of each reference gene independently and takes into account intra- and intergroup variations for normalization [35]. In this study, geNorm (Figure 3) and NormFinder (Figure 4) identified two genes, B2M and 18S rRNA, as the most stable single gene and the most stable reference gene pair to be used for normalizing RT-qPCR data from the cartilage regeneration model. 18S rRNA has been suggested to be suitable reference gene in different species (Homo sapiens, Rhodnius prolixus and Danio rerio) [20,36,37]. However, some studies have shown that 18S rRNA is not always a suitable reference gene [9,38]. These different results may be related to the species, tissue, organs or experimental factors. rRNA constitutes the largest fraction of RNA in the cell, and the high abundance of rRNA makes it suitable for high-abundant genes. B2M was among the least stable genes in rat MSCs, but was the most stable in rat chondrocytes [39]. Stephens et al.[40] reported that B2M stably expressed in osteoclasts and macrophages. In addition, we found that CYP, HPRT and GAPDH were unstable in rabbit cartilage tissue. From Figure 4A and NormFinder, it might be concluded that GAPDH is as reliable as B2M. Here, we show the detailed scores of each program calculation. geNorm analyzes the expression stability by calculating M values, the lower the value of M, the more stable of expression (GAPDH 0.344, B2M 0.249). With NormFinder, higher expression stability is indicated by a lower stability value (GAPDH 0.151, B2M 0.136). Together, geNorm and NormFinder showed that B2M was superior to GAPDH and that B2M and 18S rRNA is the most stable gene pair. Similar results have been shown in human cartilage. Pombo-Suarez M. et al.[15] reported that GAPDH and HPRT1 were less stably expressed in human osteoarthritic articular cartilage. Other studies also have shown that GAPDH has high expression variance in human mesenchymal stromal cells and hypoxia-cultured human chondrocytes [16,41]. Maccoux et al.[42] studied the expression stability of twelve newly identified housekeeping genes in canine osteoarthritic joint tissues. Their results showed that only a small number of the candidates are truly consistently expressed in a certain tissue type. In fact, when different candidate genes are studied, different genes might be identified as the most suitable references. Maccoux et al.[42] indicated that RPS4X and RPL13A are suitable reference genes for dog cartilage. However, their suitability for rabbit chondrocyte needs further investigation. Regardless, what their reports have in common with ours is that theirs also showed that GAPDH is less stably expressed than some other reference genes.

Our study shows that GAPDH is not suitable as a reference gene for rabbit cartilage even though it had been reported to be suitable for sheep [43]. This agrees with Dundas and Ling [44] and Too and Ling [45] that reference genes suitable for a specific organism or tissue may not be suitable for another organism or tissue. Thus, this paper compared the expression stability of the housekeeping genes in each tissue type under each pathological/physiological will undoubtedly be a good reference. So we recommend that researchers use B2M or 18S rRNA to calibrate the expression of their interest genes in studies of regenerating rabbit cartilage.

Trizol total RNA extraction reagent was purchased from Invitrogen (Invitrogen, Carisbad, CA, USA). Quantitative real-time PCR detection kit was purchased from Roche (Roche Ltd, Basel, Switzerland). Recombinant human IGF-1 was ordered from R&D (R&D Systems, Minneapolis, MN, USA). Experimental rabbits were housed in individual cages with standard food and water ad libitum under normal conditions. All procedures involving animals were approved by the Animal Care and Use Committee of China.

Three-month-old New Zealand white rabbits (2.8–3.0 kg) were obtained from VITAL RIVER (Beijing, China). Eighteen rabbits were randomly divided into three major groups (A, B and C). The two hind legs of each rabbit were studied; therefore, each group included 12 tissue samples. Rabbits in group A were not injured and served as controls. Artificial cartilage defects were induced in group B and C rabbits. Surgical procedures were performed under general anesthesia induced by intramuscular injection of ketamine hydrochloride (350 mg/kg body weight) and xylazine (5 mg/kg body weight); general anesthesia was maintained by additional injections at the same dosed every hour as required. After parapatellar arthrotomy, the patella was dislocated laterally, and a 5 mm diameter, 2 mm deep, partially articular cartilage defect was created in the patellar groove of the femur using a dentist drill. Defects were sufficiently lavaged with 0.9% saline, and all skin layers were sutured. Group B rabbits were not given IGF-1 treatment; group C rabbits were given IGF-1 after the operation. Postoperatively, the rabbits were returned to cages and allowed to move freely without splinting. One milliliter of rhIGF-1 (300 μg/mL in PBS) was injected into the articular capsules of the knee joints of group C animals every three days for 30 days, while group B rabbits received injections of PBS. The rabbits were sacrificed under general anesthesia. RNA was isolated from one half of each group and used for RT-qPCR. The remaining samples were subjected to histological evaluation.

For histological evaluation, the whole defect region, including surrounding cartilage and some hard bone tissue, was dissected and fixed in 10% buffered formalin for 1 week. Specimens were decalcified in 0.5% EDTA solution containing 0.1 M amino-n-caproic acid (Sigma Chemical Co., St. Louis, MO, USA) and 0.005 M benzamidine and embedded in paraffin. Four-micrometer sagittal sections of the defect area were collected and stained with hematoxylin and eosin (H and E) according to standard protocol.

Defect surrounded cartilage was carefully trimmed off and immediately put into Trizol reagent. Total RNA was extracted from 1mg cartilage homogenate. The concentration and quality of total RNA were determined using a biological spectrophotometer (Eppendorf, Germany). RNA samples had OD260/280 ratios between 1.7 and 2.0 and OD260/230 ratios below 2.0, reflecting high purity and the absence of protein. The integrity of each total RNA sample was assessed by determining the ratio of 28S/18S ribosomal RNAs after agarose gel electrophoresis. Total RNA (0.5 μg) was used as template in a 20 μL cDNA synthesis reaction (42 °C for 60 min) containing 0.2 μg random hexamer, 200 U of M-MuLV reverse transcriptase (Invitrogen, Carisbad, CA, USA), 20 U Ribonuclease inhibitor and dNTPs at a concentration of 1 mM. Each reverse transcript (0.5 μL) was used as template for PCR amplification. In addition, serial dilutions of the template cDNA were used to confirm reaction efficiency. Five reference genes, GAPDH, 18S rRNA, CYP, HPRT1, and B2M, were evaluated in this study. Primers for each gene are shown in Table 1. Optimum annealing temperatures were determined with gradient PCR prior to qPCR. Nontemplate controls were included for quality control purposes. To compare different RNA transcription levels, the Ct values were compared directly. Ct is the number of cycles required for the detection of the fluorescence signal reaches the set threshold level. Studies have shown that there is an inverse linear relationship between Ct and the amount of template nucleic acid. As the amount of template increases, the Ct value decreases [13]. Each 20 μL reaction contained 10 μL of 2× SYBR Green Mix, 0.5 μL of a 10 μM solution of each primer, 0.5 μL of cDNA and 8.5 μL of sterile distilled water. The RT-qPCR conditions were as follows: 95 °C for 5 min, followed by 35 cycles of 95 °C for 10 s, 58 °C for 20 s and 72 °C of 30 s. PCR was followed immediately by melting analysis, beginning at 72 °C for 1 min and ramping to 99 °C at a rate of 1 °C per 5 s. Each qPCR assay was run with triplicate technical samples. Relative quantification was performed using Light Cycler Software 1.5.0 (Roche Applied Science). To assess the expression stability of the selected reference genes, the raw Ct values were obtained by averaging the Ct values of the sample collection.

Data were analyzed after the raw Ct values were transformed into relative quantification data by the ΔCt method. In this study, the relative expression level equal was set to 1 for each gene measured with the minimum Ct, then, 2^−ΔCt was used to represent the relative expression level of the reference genes in the other samples(ΔCt = Ct value of each sample − minimum Ct value). These data were further analyzed with the geNorm [21] and NormFinder [22] algorithms. The expression levels of candidate reference genes in the three experimental groups are represented as boxplots. Boxes represent the lower and upper quartiles of the cycle thresholds range with medians. The graph was plotted with SPSS 15.0 software.