Since the efficiency of reverse transcription reactions depends on the total amount of RNA, the input RNA for the reverse transcription step should be identical if various groups of samples are compared 26. This requires accurate quantification of the isolated RNA as well as assessment of the RNA integrity. In the present study, RiboGreen assays of RNA isolated from differently treated C-28/I2 chondrocytes revealed RNA amounts ranging from 6.96 μg to 36.3 μg. No correlation between the amount of total RNA and the different treatments was found. The integrity of the RNA preparations was verified by the inspection of the 28S and 18S ribosomal RNA bands using agarose gel electrophoresis. Efficiency of the DNase digest was additionally assessed by implementing minus RT controls for each RNA preparation in the qPCR experiments. Some of the minus RT control samples recorded C_t values of around 32 indicating minor genomic DNA contamination. However, the ΔC_t – separating the minus RT controls from the unknown samples – of more than 13 C_t was regarded as high enough to be negligible 37. In summary, all RNA samples were found suitable for qPCR reactions.

Optimization of primer concentrations is critical for increasing specificity and sensitivity of the qPCR assay 33. The results of the described primer optimization experiments are presented in Table 1. For all qPCR reactions of the study, these respective primer concentrations were used. Afterwards, standard curves were prepared to examine the quality of the overall qPCR assay. Standard curves of all examined genes revealed high amplification efficiencies over 4 to 5 orders of magnitude (Table 1) covering the C_t range of all samples. The R² values of the standard curves exceeded 0.996. Gene-specific amplification was confirmed by a single peak in the melting-curve analyses (data not shown). Only ACTB amplification resulted in the detection of a second peak, which is known to originate from the presence of regions with highly different GC content rather than from unspecific reactions 38. Moreover, no primer-dimer was detectable in any of the melting-curves indicating optimal performance of the primers. The no-template controls (NTC), included in all qPCR runs, did not record any positive C_t value throughout the study.

A real-time qPCR assay, based on SYBR Green detection, was designed for the transcriptional profiling of six frequently used reference genes (ACTB, GAPDH, B2M, HPRT1, SDHA, YWHAZ) in the cDNA samples. Figures 1A, B and 1C show the distribution of the C_t values of all samples under all culture conditions (Table 2) for glucosamine, curcumin and diacerein experiments, respectively. This 'whole error approach' is a simplified way to visualize the overall error introduced during the whole assay and gives an overview of the abundance of the genes in the samples. Since lower C_t values indicate higher numbers of starting cDNA molecules, the most abundant reference genes in C-28/I2 cells were ACTB, GAPDH and B2M followed by YWHAZ. The amplification of HPRT1 and SDHA occurred about 5 C_t values later. However, this 'whole error approach' does neither take into account the influence of qPCR efficiencies, nor does it provide satisfactory resolution for statistical evaluation of small differences between study groups. In a further step, the C_t values of each sample were related to the average C_t value of the respective untreated control group by the equation E^-ΔCt. Tables 3, 4 and 5 show the relative reference gene expression levels of IL-1β treated chondrocytes under the influence of glucosamine, curcumin and diacerein, respectively. IL-1β did not provoke significant differences of any reference gene except for ACTB which was significantly up-regulated in the glucosamine (Table 3) and diacerein (Table 5) experiments (p < 0.05). Regarding the impact of the CPA used, a trend for dose-dependent down-regulation of all reference genes was observed for both the pre-treated and the simultaneously treated study groups. In contrast, time-dependent down-regulation seems to occur at the high concentrations of CPA used. We found that both glucosamine (Table 3) and curcumin (Table 4) treatment affected the expression level of all reference genes significantly (p < 0.05). Comparing two samples, a twofold change in gene expression level is commonly regarded as the benchmark of biological meaningfulness 2935. Therefore, a less than twofold change is considered as a requirement for suitability of a reference gene. In our study, a 48 h incubation of chondrocytes with high concentrations of glucosamine (Table 3) and curcumin (Table 4) (5 mM and 50 μM, respectively) induced most reference genes to exceed this benchmark of biological meaningfulness. The only exceptions were B2M in the case of glucosamine treatment (Table 3) and YWHAZ in the case of curcumin treatment (Table 4). The data further indicate differences in the reference gene regulation caused by different CPA. For diacerein-treated chondrocytes, GAPDH, B2M, HPRT1 and SDHA were not significantly affected by any culture condition used.

In order to determine the most stable reference genes in our study, the data were analyzed using the geNorm software. Table 6 shows the ranking of the 6 candidate reference genes according to their expression stability under the influence of glucosamine, curcumin and diacerein. The optimal number of reference genes necessary for the calculation of reliable normalization factors was assessed calculating the pairwise variations V_n/V_n+1 according to Vandesompele et al. 32. RT-PCR normalization factors were calculated on the basis of the 3 most stable genes and stepwise inclusion of the other genes in the order of their expression stability. The pairwise variation V_n/V_n+1 was calculated between the 2 sequential normalization factors (NF_n and NF_n+1). A large variation means that the added gene has a significant effect on the newly calculated normalization factor and should preferably be included for calculation. In all 3 experiments, the inclusion of the 4^th reference gene did not contribute significantly to the variation of the normalization factor (V_3/4 value < 0.15). Based on the proposed cut-off value of 0.15, below which the inclusion of an additional control gene is not required, the use of 3 control genes would therefore be sufficient for normalization in our study. The 3 most stable reference genes for glucosamine-treated samples were GAPDH, B2M and SDHA. In contrast, for curcumin and diacerein-treated samples, HPRT1, GAPDH and B2M provided the most stable expression levels with SDHA ranking the 6^th and 5^th positions, respectively. It is remarkable that ACTB, one of the most prominent and frequently used reference genes, proved to be one of the most variably expressed genes under the influence of selected CPA.

Glucosamine hydrochloride, curcumin, diacerein and 0.25 % Trypsin-EDTA were purchased from Sigma (St. Louis, MO, USA). Recombinant human IL-1β was obtained from Strathmann Biotec (Hamburg, D). Primers for real-time PCR experiments and the mi-Gel Extraction Kit were from Metabion (Martinsried, D). Dulbecco's modified Eagle's medium and 100x ITS solution were purchased from Gibco (Lofer, A). The StrataScript First-Strand Synthesis System and the Brilliant SYBR Green QPCR Master Mix were from Stratagene (La Jolla, CA, USA). All other chemicals were of analytical grade and obtained from Merck (Darmstadt, D) unless otherwise specified.

Immortalized C-28/I2 chondrocytes were used for all experiments in this study. In order to determine the influence of glucosamine, curcumin and diacerein on reference gene expression levels in chondrocytes, three sets of individual experiments – one for each compound – were performed.

2.3 × 10⁵ chondrocytes were seeded into 25 cm² culture flasks (Corning, NY, USA) and cultured for 5 days in DMEM supplemented with 10 % fetal calf serum (Biochrom AG, Berlin, D), 2 μl/ml gentamycin and 50 μg/ml ascorbate. On days 5 and 6 the culture medium was changed to serum-free DMEM supplemented with 1 % (v/v) 100× ITS (10 μg/ml insulin + 5.5 μg/ml transferrin + 6.7 ng/ml sodium selenite) and 2 μl/ml gentamycin. Additionally, the particular CPA and/or IL-1β (10 ng/ml) were added as indicated in Table 2. Thereby, 7 different conditions for each CPA experiment were created: one control group without additional supplements, one IL-1β treated group, two groups simultaneously treated with both IL-1β and CPA at two different concentrations, two groups pre-incubated with CPA at two different concentrations prior to IL-1β addition, and one group treated with the respective CPA at the higher concentration without the addition of IL-1β. Each of these study groups consisted of three identically treated 'biological' replicates. Throughout cell culturing, cells were maintained in a humidified 5 % CO₂/95 % air atmosphere at 37°C.

On day 7 of the culture period, chondrocytes were harvested by trypsination using 0.05 % Trypsin-EDTA. The cells were washed with PBS and total RNA was extracted immediately using the NucleoSpin RNA II Kit (Macherey-Nagel, Dueren, D) according to the manufacturer's instructions. On-column DNase digestion was performed for all samples during the isolation process. To control for the integrity of the isolated RNA preparations, all samples were qualitatively assessed according to standard methods prior to use. Briefly, a 1.5% denaturing agarose gel was loaded with 2 μg of total RNA samples and stained with ethidium bromide after electrophoresis. The quantity of extracted RNA was determined using the Quant-It RiboGreen reagent (Molecular Probes, Eugene, USA) and the quantitative plate-read function of the Stratagene MxPro QPCR software. An R² > 0.990 of the standard curve plots was considered acceptable for accurate RNA quantification. Reverse transcription (RT) was performed using the StrataScript First-Strand Synthesis System, following the manufacturer's instructions. Briefly, 2 μg of total RNA were reverse transcribed in a 20 μl reaction volume using the manufacturer-supplied oligo(dT) primers. Minus RT controls were prepared for each sample using the identical procedure except for the omission of reverse transcriptase. All cDNA preparations and minus RT controls were diluted at a ratio of 1:10 with RNase free water prior to real-time PCR.

Six reference genes (ACTB, GAPDH, B2M, HPRT1, SDHA, YWHAZ) belonging to different functional classes were selected to reduce the chance that these genes might be co-regulated (Table 1). Primers for GAPDH were designed using AlleleID 2.01 software (Premier Biosoft, CA, USA) and the Genbank sequence NM_002046. In order to avoid non-specific product formation, the primers were designed across exon junctions and with minimized self and cross dimer ΔG values. Additionally, BLAST analyses were performed to verify specificity. Primer sequences for all other reference genes were used as described by Vandesompele et al. 32. Amplicon sizes were 140 bp for ACTB, 165 bp for GAPDH, 86 bp for B2M, 94 bp for HPRT1, 86 bp for SDHA and 94 bp for YWHAZ. Gene specific primers were used to amplify respective PCR products in order to provide sufficient template material for primer concentration optimization. PCR amplifications were performed using GoTaq DNA Polymerase (Promega, Wisconsin, USA) in a 50 μl reaction volume (10 μl of 5x Colourless Go Taq Reaction Buffer, 1 μl of PCR Nucleotide Mix, 5 μl of gene specific forward and reverse primer, 0.25 μl of GoTaq DNA polymerase, 2 μl of template DNA, and nuclease-free water to 50 μl). The PCR program consisted of an initial denaturation step at 95°C for 2 min, followed by 35 cycles at 95°C for 45 s, at 55°C for 45 s and at 72°C for 1 min, and a final extension step at 72°C for 5 min. Amplified PCR products were diluted with bromophenol blue solution at a ratio of 1:4 and purified by gel electrophoresis using a 1% agarose gel. After staining with ethidium bromide, the designated bands were cut out of the agarose gel and the products were extracted from the gel using the mi-Gel Extraction Kit according to the manufacturer's instructions. Optimization of primer concentrations for the qPCR experiments was performed by qPCR assays as described below using different primer concentrations and a 1:10³ dilution of purified PCR products as templates. Optimal primer pairs for each reference gene were identified selecting the combination that gave the lowest C_t value and the highest ΔR_n value.

All qPCR reactions were performed in 25 μl reaction mixtures containing 1 μl cDNA, 12.5 μl Brilliant SYBR Green QPCR Master Mix, primer pairs as listed in Table 1, and nuclease-free water to 25 μl. Each biological replicate was run in triplicate on a Mx3000P QPCR system (Stratagene). Thermocycling conditions consisted of an initial polymerase activation step at 95°C for 10 min, followed by 35 cycles at 95°C for 30 s, at 55°C for 1 min, and at 72°C for 1 min. Afterwards, melting curves were generated to confirm a single gene-specific peak and to detect primer-dimer formation by heating the samples stepwise from 55°C to 95°C while continuously monitoring the fluorescence. NTC were included in each run to control for contaminations. Minus RT controls were run to verify the efficiency of the on-column DNase digestion. The results were analyzed using the MxPro real-time QPCR software (Stratagene). Baselines and thresholds were automatically set by the software and used after manual inspection. The crossing point of the amplification curve with the threshold represented the cycle threshold (C_t). PCR efficiencies for each primer pair were derived from standard curves using five-fold dilution series covering a 4 to 5 log dynamic range starting from one randomly selected undiluted cDNA sample of the untreated control group. The reactions were run in duplicate and no-template controls were included.

Results were exported to Microsoft Excel and GraphPad Prism for further analyses. For each test compound, the distribution of the expression levels (C_t values) for each reference gene under the 7 experimental conditions was displayed as Box and Whiskers plots. In a next step, data within each experiment were normalized to the average level of mRNA expression (C_t value) of the respective control group using the relationship E^-ΔCt (E, efficiency; ΔC_t, difference between C_t values). Statistics were performed using one-way ANOVA with post-hoc Dunnett tests comparing each study group with the untreated control group (significance p < 0.05). The geNorm application software for Microsoft Excel was additionally used as described by Vandesompele et al. 32 to identify the most stable reference gene under the described conditions, and to determine the optimal number of reference genes required for reliable normalization of qPCR data.