Female Wistar rats of 6-8 weeks (weight 130-150 gm) were taken for the induction of arthritis. All these rats were issued and maintained at the Central animal facility of Jawaharlal Nehru University, New Delhi (India). These experimental rats were housed separately in cages, fed with standard rodent chow (Hindustan Lever Limited, India) and water ad libitum. Standard parameters were taken into care during their housing and whole experimental phase. These were photoperiods, darkness, environmental temperature, and humidity. Photoperiods equalled 12 h of light and 12 h of darkness daily, with the environmental temperature and humidity maintained at 25 °C + 2 and 42 % + 5, respectively. All experimental protocols were approved by the Committee for the Purpose of Control and Supervision of Experimental Animals (CPCSEA) and the Institutional Animal Ethical Committee (IAEC) of Jawaharlal Nehru University, New Delhi, India.

All the chemicals were purchased from Sigma Aldrich Chemical Co. (St. Louis, MO). Antibodies of TNF-α, IL-1β and IL-6 were purchased from Peprotech Inc., USA. All other chemicals and solvents used were of analytical grade.

These experimental female Wistar rats were randomly allocated into the two groups of five animals each (a) Group I: Rats served as healthy 'control' with no collagen immunization, (b) Group II: Rats immunized with collagen designated as "CIA".

The CIA was induced as previously described by Srivastava et al. (2014[51]). Briefly, porcine immunization-grade native collagen type II (Chondrex) was dissolved in 0.05 mol/l acetic acid at a concentration of 2 mg/ml by stirring overnight at 4 °C. The emulsion was prepared by homogenizing collagen type II with Freund's complete adjuvant (1:1) at 4 °C. Each rat was immunized with 200 μg of emulsified collagen type II, intradermally at multiple sites on the back and the base of the tail followed by a booster dose with the same antigen preparation on day 7. Arthritis was allowed to develop for 28 days.

In all the experimental rats, the clinical scoring of CIA rats was performed. Clinical score was based on the composite index of disease severity and the number of affected limbs. The clinical score was assessed every 4 days to monitor disease progression as described previously (Cuzzocrea et al., 2000[9]). Since the CIA is more pronounced in the hind-limbs, so, the scoring of rat hind-limbs only has been taken into account to assess the severity. Two independent observers, blinded to the experimental groups, performed the scoring.

In all the experimental rats, the hind paw edema was measured every 4^th day throughout the experiment using Vernier callipers. The arthritic indices (AI) were calculated as described in the previous report (Coelho et al., 2004[7]) using the formula:

The body weight was measured every 4^th day in all the experimental rats. The measurement of body weight was performed according to our previous study (Srivastava et al., 2014[51]). Variation in body weight (%) of each individual animal after the onset of arthritis on day 12 was calculated using the formula below.

On day 28, all the experimental rats were sacrificed. Hind limbs were removed and right hind leg was fixed at 10 % buffered formalin for radiological evaluation. All radiographs were taken with X-ray film (Kodak Diagnostic Film) using MBR-1505R (Hitachi Medical Corporation, Tokyo, Japan). Settings for radiography were 5 mA, 40 kV and 1 s exposure. Films were placed 60 cm below the X-ray source.

Arthritic damage was also examined through histology. The knee joint was excised from the fixed hind leg, its skin and muscular parts were trimmed off and subjected to decalcification in 5 % nitric acid for 7-10 days, processed for paraffin embedding, sectioned at 8 μ followed by staining with hematoxylin-eosin and studied using light microscopy.

The activity of antioxidant enzymes SOD (superoxide dismutase), CAT (catalase), GPx (glutathione peroxidase) and LP (lipid peroxidation) were measured in joint tissue of CIA rats and control rats by applying the procedure of Singh et al. (2000[48]). The LP was estimated in joint tissue of CIA rats and control rats by the method of Ohkawa et al. (1979[35]).

For the preparation of joint tissue homogenate, the joint tissues were removed, rinsed quickly in cold saline solution and homogenized in 0.05 M Tris-HCl buffer (pH =7.4) to yield a 5 % (w/v) homogenate. The homogenate was centrifuged at 1000 g for 10 min. The resulting supernatant was transferred into pre-cooled centrifugation tubes and centrifuged at 12,000 g for 30 min. The supernatant (containing the cytosolic fraction) was diluted appropriately followed by the removal of residual lipid layers. It was then used to estimate the total cytosolic SOD, CAT, GPx and LP. The homogenate was stored at -80°C till biochemical analysis.

The activity of total cytosolic SOD, GPx and CAT enzymes in joint tissue homogenate were determined and expressed as units/mg protein (Singh et al., 2000[48]).

Thiobarbituric acid-reactive substance (TBARS), an index of lipid peroxidation in joint tissue homogenate was estimated by the method of Ohkawa et al. (1979[35]). The amount of TBARS was determined spectrophotometrically at 532 nm and expressed as nanomoles of malondialdehyde (MDA)/mg protein. Tetramethoxypropane (TMP) was used as a standard.

Cytokine measurements were also performed on joints of all the experimental rats. For this purpose, whole joints were powdered with liquid nitrogen by using a pestle-mortar. Powdered tissue was weighed and homogenized (1:5 w/v) in 0.02 M Tris-HCl buffer ( pH =7.4) containing 20 mM MgCl₂, 1 mM PMSF, 10 μg leupeptin and aprotinin, 1 mM ATP, 5 mM EDTA, 50 mM DTT and 3 μl protease inhibitor cocktail per 10 ml buffer. Homogenate was first centrifuged at 4000 rpm, 4°C for 15 min to remove the debris and bony portions. The collected supernatant was further centrifuged at 16,000 rpm for 15 min and the resultant clear supernatant was aliquoted and stored at -80 °C for estimation of cytokines. Concentrations of plasma and joint cytokines, TNF-α, IL1- β and IL-6 were measured by indirect ELISA.

Plates (Nunc) were coated with 10 μg proteins/well diluted in carbonate buffer at 37 °C for 1-2 h. The plates were washed thrice with PBST (0.2 % Tween 20 in 1X PBS) and blocked using blocking buffer (1 % skimmed milk in 1X PBS) for 1 h at 37°C. After washing, 100 μl/well anti-rat polyclonal antibodies against TNF-α, IL1-β and IL-6 were added and incubated at 37 °C for 1 h. This was followed by washing the wells and then 100 μl/well (1:1000 diluted) anti-rabbit HRP labelled IgG secondary antibody was added. After incubation followed by washing, developing buffer (10 mg/ml orthophenylene diamine dissolved in 150 mM citrate-phosphate buffer, pH 5.6 with 30 % H₂O₂) was added. The plates were incubated for 5-10 min at RT (room temperature), the reaction was stopped with the stop solution (2 N H₂SO₄) and the plate was read at 490 nm using Spectra Max Plus from Molecular Devices using the SoftMax-Pro software. Bovine serum albumin, at a concentration of 5 μg/ml (100 μl/well) was included as a negative control in each assay performed and the wells incubated with coating buffer instead of plasma were used as the blank.

The concentration of cytokines was expressed in pg/ml. Each sample was measured in triplicates and values were derived from the calibration curve.

Aqueous metabolites were extracted from joint tissue of the CIA and control rats by the method of perchloric acid extraction, as described previously (Srivastava et al., 2008[50]). After completing the extraction, the sample was lyophilized. All samples were re-dissolved in 0.5 ml deuterated water (pH = 7.0) and taken in 5 mm NMR tube (Wilmad no. 528, USA). The ¹H-NMR experiments were performed on a Bruker Avance 400 MHz spectrometer (Bruker Biospin, Zurich, Switzerland) at 25 °C temperature.

One-dimensional proton NMR experiments, using single pulse sequence were performed. Spectral width used was 8000 Hz with time domain data points of 32 K. Flip angle of the radio frequency pulse was 90° with a complete relaxation delay of 7 seconds (relaxation delay plus acquisition time) to ensure complete recovery of magnetization to the equilibrium between the scans. Typically, 128 scans were accumulated for each sample and the resultant data were Fourier transformed after multiplying by exponential window function using a line broadening of 0.34 Hz. Sealed and reusable glass capillary with a known concentration of TSP [3-(trimethylsilyl) propionic-2, 2, 3, 3-d4 acid, sodium salt] in D₂O was used as a standard reference. The NMR spectra of all the samples were recorded under uniform conditions.

Assignments of the metabolites in the perchloric acid extract of joint tissue of CIA rats as well as control rats were completed with the help of two-dimensional (2D) double quantum filtered correlation spectroscopy (DQF-COSY) experiment. The assignments were confirmed by literature (Zhong et al., 2012[64]; Shao et al., 2014[43]; Atherton et al., 2006[2]; Hwang et al., 2010[21]; Sharma et al, 2003[45][46]). Spectra of standard components were recorded individually as well as in a mixture with an appropriate concentration. Protons of amino acids are pH sensitive, the resonances from glutamate, glutamine and alanine were confirmed by recording the spectra of these compounds in the pH range 7-9. The N⁺(CH₃)₃ resonances of GPC (Glycerophosphocholine) and carnitine were identified by carrying out the experiment on the mixture of these two compounds.

Quantification of metabolites was performed by using a computer program, which is devised for such purposes and also used in the previous studies (Srivastava et al., 2008[50], 2014[51]). Metabolite concentrations were calculated in the mmol/mg weight of the joint tissue.

The mean level of selected metabolites in the perchloric acid extraction of the joint tissue of the CIA rats and control rats was compared by student “t” test for two independent groups. The p-value less than 0.05 was considered to be significant. Correlation and DFA (discriminant function analysis) were also performed. The data management and analysis were carried out using statistical software SPSS version 15.0.

The outcome of characteristic parameters of CIA rats as compared to control rats are described in the following way:

Collagen induced arthritis was developed in the experimental rats immunized with type II collagen as shown in Figure 1(A)(Fig. 1). Clinical presentation of arthritis (mild periarticular erythema and edema) exposed by the redness and swelling of the joints first appeared in the hind paws approximately 10 days post challenge. Progression of disease was enhanced in frequency and severity in a time-dependent manner and maintained a plateau of the peak of CIA response from day 15-30 (mean AI of 125.01) [Figure 1(B)(Fig. 1)]. There was no enlargement appearing in hind paw diameter over time. Hind paw swelling and erythema also presented a time-dependent enhancement in severity with maximum arthritis scores of ~8 observed between days 12 and 30 in rats immunized with CII [Figure 1(C)(Fig. 1)]. There was no macroscopic confirmation of the hind paw swelling and erythema in the control rats.

The body weight of all the experimental rats was recorded for every 4^th day during the whole duration of the experiment. The onset stage of arthritis was connected with severe body weight loss that continued till the end of experimental duration, i.e., day 30 in the CIA rats [Figure 1(D)(Fig. 1)]. Control rats showed a gradual increase in their body weight with respect to time.

Radiographic analysis of the hind paws from CIA rats presented soft tissue swelling, bone matrix resorption and joint space narrowing on day 30 of the experiment [Figure 2(A)(Fig. 2)]. There was no radiographical evidence of arthritic features in the control rats. After finishing the experiment on day 30, histological evaluation of the joints of arthritic animals presented a distinct characteristic feature of severe arthritis like synovial hyperplasia, massive mixed infiltration (neutrophils, macrophages and lymphocytes) [Figure 2(B)(Fig. 2)] along with articular cartilage and bone erosion. There was no proof of arthritic features of inflammation and tissue destruction in the joints of control rats.

Level of lipid peroxidation was found to be higher in CIA rats as compared to control rats [Figure 3(A)(Fig. 3)]. The lipid peroxide levels were increased by 2.84 fold (p < 0.0001) in the joints of CIA rats as compared to control rats. The results showed that the joints were much affected and are vulnerable to peroxidation due to increased oxidative stress in the CIA.

SOD, GPx and CAT activities were determined in the joints of CIA rats and control rats [Figure 3(B-D)(Fig. 3)]. SOD, GPx and CAT activities were found to be higher (p < 0.0001) in the CIA rats as compared to the control rats. CAT showed the maximum increase of 2.87 fold in the arthritic rat joints followed by rising of 2.22 and 1.65 fold in the activity of SOD and GPx, respectively.

Levels of TNF-α, IL-1β and IL-6 were found to be higher (p < 0.0001) in joint tissue of CIA rats as compared to control rats (Figure 4(Fig. 4)).

Assignments of various metabolites in the perchloric acid extract of the joint tissue of the CIA rats as well as control rats were represented in the Figure 5(Fig. 5).

Quantification of BCA (Branched chain amino acids: valine, leucine and isoleucine), lactate, alanine, acetate, carnitine, choline, GPC and creatinine was performed in the perchloric acid extract of joint tissue of CIA rats as well as control rats. Level of acetate was reduced in CIA rats as compared to control rats, but independent sample “t” test showed no significant difference (p = 0.78). Similarly, carnitine level was also reduced in CIA rats as compared to control rats, but the statistically significant difference is not observed (p = 0.54). Lactate, alanine, BCA and creatinine levels were higher in CIA rats as compared to control rats. There is a significant difference observed between lactate (p = 0.004), alanine (p = 0.025), BCA (p = 0.006) and creatinine levels (p = 0.023). Level of choline and GPC were reduced in CIA rats as compared to control rats. Choline (p = 0.038) and GPC (p = 0.009) were statistically significant in CIA rats as compared to control rats. Statistical analysis represented the quantitative difference for metabolites in CIA rats as compared to control rats (Figure 6(Fig. 6)).

Correlation in between alanine and lactate was poor (Pearson correlation = 0.34) for CIA rats, but no correlation was observed in control rats. The coefficient of correlation in between carnitine and BCA was somehow better (Pearson correlation = 0.44) than alanine to lactate correlation for CIA rats. Correlation in between choline and GPC was good and negative (Pearson correlation = -0.63) for CIA rats as well as for control rats (Pearson correlation = -0.79).

DFA analysis was performed for the discrimination or classification (100 %) of CIA rats and control rats on the basis of two parameters (choline and GPC). The weightage for choline and GPC was 142.18 and 25.92, respectively. Constant value for the DFA equation was -4.755 and cut off value was zero.