Troxerutin-mediated C9 inhibition is a disease-modifying treatment for inflammatory arthritis.


 Troxerutin (TXR) is a phytochemical reported to possess anti-inflammatory and hepatoprotective effects. In this study, we aimed to exploit anti-arthritic properties of TXR using an adjuvant induced arthritic (AIA) rat model. AIA induced rats showed highest arthritis score at disease onset and by oral administration of TXR (50, 100, 200 mg/kg body weight), reduced to basal level in a dose dependent manner. Isobaric tag for relative and absolute quantitative (iTRAQ) proteomics tool was employed to identify deregulated joint homogenate proteins in AIA and TXR treated rats to decipher probable mechanism of the TXR action in arthritis. iTRAQ analysis identified a set of 434 joint homogenate proteins with 65 deregulated proteins (log2 case/control ≥ 1.5) in AIA. Expressions of a set of important proteins (AAT, T-kininogen, vimentin, desmin, and nucleophosmin) that could classify AIA from healthy were validated using Western blot analysis. Western blot data corroborated proteomics findings. In silico protein-protein interaction study of joint homogenate proteome revealed that complement component 9 (C9), the major building blocks of the membrane attack complex (MAC) responsible for sterile inflammation, gets perturbed in AIA. Our dosimetry study suggests that a TXR dose of 200 mg/kg body weight for 15 days is sufficient to bring the arthritis score to basal levels in AIA rats. We have shown the importance of TXR as an anti-arthritis agent in AIA model and after additional investigation its arthritis ameliorating properties could be exploited for clinical usability.


Clustering coefficients ( ):
It is the ratio of number of triangular motifs created by a node 282 with its nearest neighbors and the total number of such motifs in the entire network. Therefore, 283 In the hierarchical network topology, ( ) shows power law against the degree , i.e., 295 ( )~ , here ~0.5 (26). The negativity or positivity of the exponent may be defined as 296 disassortivity or assortivity nature of a network topology, respectively(27). 297 Centrality measures. Betweenness centrality C B , closeness centrality C C , Eigenvector centrality 298 C E are the basic centrality measures and are the parameters for the estimation for a node's global 299 functional significance in a network regulation through information processing (28). 300 The total geodesic distance between a node and all of its connected nodes is given by . It also 301 determines how rapidly an information is spread within a network from one node to other 302 connected nodes (29). In a given network, of a node is calculated by dividing the total 303 14 number of nodes of network by the summation of geodesic path lengths between the nodes 304 and which is given by of equation 5. 305 The C B or betweenness centrality is the measure of a node which is share of all the shortest-path 307 traffic from all feasible routes through nodes to . So, it is the parameter of the ability of a node 308 to extract benefit from the flow of information throughout the network (30) and its ability to 309 control the signal processing over the other nodes within the network (31). If ( ) represents 310 the number of geodesic paths from one node to another node passing through the node , then 311 ( ) of node can be derived by the equation 6. 312

Joint radiographic and histopathology analyses show protective effects of TXR:
The 372 footpad thickness at disease onset was highest and resolved during the treatment (Fig 2D). The 373 limb roentgenograms of the experimental animals showed joint erosion with osteophyte 374 formation; edema and soft tissue with noticeable swelling (Fig 2E). AIA group showed 375 considerable damage with bone erosion and reduction in joint space. In TXR200 (p≤0.01) and 376 DS (p≤0.05) groups, a significant recuperation of the joint damage were observed. AIA group 377 exhibited drastic inflammation of the tibio-tarsal joints resulting in an increase in the thickness of 378 bones and cartilages (Fig 2E and 2F). All the treatment groups showed reduction in osteological 379 swelling in the tibiotarsal joints. The tibiofemoral joints were less affected as compared to 380 tibiotarsal joints in all the groups. However, AIA and DS groups exhibited signs of damage ( Fig  381   2E and 2F). AIA score at day 21 showed significant reduction in TXR200 group (Fig 2G). There 382 was a marked reduction in the radiographic score of TXR50 and TXR100 but were not 383 statistically significant (Fig 2H). In the plantar regions, reduced soft tissue swellings were 384 observed in the TXR treated groups. The histopathololgical data of all the groups was expressed 385 in terms of histological score (Fig 2I). 386

TXR treatment suppresses disease progression in arthritis:
The arthritic symptoms 387 appeared within 18-36 hours of adjuvant immunization and the inflammatory parameters were 388 edema, periarticular erythema, and functional decline in gait of the immunized rats (Fig 3). The 389 gait of the arthritic animals improved substantially in all the TXR treated groups as compared to 390 the AIA group, which was included as a parameter in arthritic score calculation throughout the 391 treatment period (Fig 3A) and on the day of sacrifice, i.e., day 21 ( Fig 2G). TXR (200 mg/kg) 392 treatment has brought down the arthritis score close to the basal level (healthy) (Fig. 3A) and 393 showed visible effect in the hind footpads of the rats (dorsal view, Fig 1C and ventral view, Fig  394   18 2D). Significant reduction of mean arthritic score was observed in a dose-dependent manner 395 from day 6 (dorsal view, Fig 1C and ventral view, Fig 2D). AIA score pattern was found to be 396 similar in TXR100 and DS groups whereas in TXR50 positive effect was observed by day 12. 397 The arthritis ameliorating potential of TXR at 50 and 100 mg/kg was found to be comparable to 398 DS treatment based on appearance of secondary lesions in the footpad and tail base. A marked 399 decline in AIA score in TXR200 from day 6 (p ≥ 0.05) to day 21 (p ≤ 0.0001) was observed (Fig  400   3A). Dose-dependent remissions of lesions in the subplantar region were observed in TXR 401 treated groups. TXR200 group showed rapid wound healing and erythema with respect to 402 TXR100, TXR50 and DS groups. AIA score, on the experiment termination day (day 21), 403 showed highly significant improvement (p<0.0001) in TXR200 followed by DS (p=0.0006) and 404 TXR100 and TXR50 groups (both p =0.002). 405

Reduction in footpad thickness upon TXR treatment: Significant reduction of mean 406
footpad thickness in TXR treated groups was observed in a dose-dependent manner ( Fig 3B). In 407 TXR200, a continuous reduction in swelling as compared to the AIA group from day 9 till day 408 21 was observed. TXR100 and DS exhibited similar trend throughout the experimental period 409 while TXR50 showed least inhibitory effect. gain till day 21. Body weight gain in the TXR50 group followed a different trend than the other 414 experimental groups (Fig 3C). TXR100 and TXR200 groups showed insignificant weight change 415 with respect to the AIA or healthy group. This can be attributed to the beneficial effects of TXR 416 overcoming its adverse effects due the lower dose. 417

Effect of TXR on liver and kidney histology:
We observed the absence of liver steatosis, 418 sinusoidal dilatation, Kuppfer cell hyperplasia, apoptosis, and necrosis in liver tissues from TXR 419 treated animals (Fig 3D, 3E). The presence of eosinophils in the portal tracts and sinusoids 420 suggested drug-induced liver injury (DILI). TXR treated groups showed insignificant difference 421 in the liver and kidney histological scores when compared to AIA group (Fig 3F and 3G). Based 422 on the liver histological score, AIA induction contributes to liver damage (Fig 3E and 3G). With 423 the administration of DS or TXR, the liver histological score remains similar to the AIA group. 424 At any given dose, TXR did not improve AIA-induced liver or kidney damage (p >0.05) when 425 compared to AIA. found up-regulated while 16 were down-regulated in AIA group with respect to healthy controls 432 (Table S2). When compared with AIA, in TXR treated groups, 11 (9↑ and 2↓) proteins with ≥1.5 433 fold change (Table S3), while 27 proteins were found to be significantly (p ≤ 0.05) deregulated 434 (Table S4). When DS group was compared with AIA, 19 (7↑ and 12↓) proteins were ≥1.5 fold 435 deregulated and 17 proteins were significantly deregulated. Three proteins (complement 436 component 9, C-reactive protein and α-1, β-glycoprotein) were found to be significantly 437 differentially expressed and are known inflammatory mediators playing critical roles in the 438 arthritis pathogenesis (Fig S3A). Two important inflammatory mediator proteins (C-reactive 439 protein and adenylate-kinase isoenzyme-1) showed deregulation in DS group (Fig S3A). A set of 440 20 5 proteins (AAT, T-kininogen, vimentin, nucleophosmin, and desmin) were sufficient to classify 441 the AIA diseased groups from the healthy group and were selected for further validation. We 442 have also compared the differential expression of proteins in TXR and DS treated groups with 443 the healthy animals to find that 28 (↑24 and ↓4) proteins in TXR (vs healthy group) (Table S5), 444 while 87 (↑76 and ↓11) proteins in DS (vs healthy group) were ≥1.5 fold changed. Communities at first hierarchical level exhibited the power law distribution for P(k) and C(k) 484 against degree distribution with negative exponents demonstrating further system-level 485 organization of the modules (Equation 9). CN (k) exhibits the power law against degree with a 486 22 positive exponent ( ~ 0.05, 0.13 and 0.14, respectively) ( Fig 5D). This specifies the assortivity 487 nature of the modules reflecting the possibility of the formation of rich-club and the hubs play a 488 very important role in the maintenance of network stability and properties(26). 489 Thus, there is an increase in the signal processing efficiency with higher degree nodes 501 emphasizing the significant roles of these nodes in the flow of information, regulation and 502 stabilization of the network. Therefore, the hub proteins must have played a significantly large 503 influence in the network regulation and pathogenesis of arthritis. The 122 modules of proteins 504 were taken into consideration to find out the most important proteins as they were present at each 505 topological level resulting in the identification of the most high-ranking key regulator proteins in 506 the arthritic network. After tracing hubs at every topological level, fifteen (15) proteins (C9, 507 Aldh2, Pdia3, Serpina6, Afm, Gyg1, Ppp1cc, Pfkp, Dhfr, Cat, Trhr, Vps29, Lta4h, Rac1, Lhpp) 508 (Table. S6) were established as the backbone of the entire network. The key regulators which 509 form the motifs with their partners (Fig 5D) might be instrumental in the network integrity, 510 optimization of signal processing, dynamics, maintaining the stability and most importantly 511 regulation of the network. Our community finding method confirmed that all the 15 key 512 regulators are interacting with each other (Fig 6A). This PPI was later validated using the 513 database of STRING 10.0 (Fig 6B). 514 3.9. C9 as the most common protein in different groups: We found four proteins (C9, protein 515 disulphide isomerase A3, thyrotropin releasing hormone receptor, and isoform gamma-2 of 516 serine / threonine -protein phosphatase 1) as common from list of in silico key regulators (n=15) 517 and the list of ≥1.5 fold changed proteins in AIA vs healthy (n=65) (Fig S4A). Whereas, 2 518 proteins (adenylate kinase isoenzyme -1 and C-reactive protein) were found to be common when 519 the list of ≥1.5 fold changed proteins in AIA vs DS (n=19) and the list of significantly (p ≤ 0.05) 520 deregulated proteins in AIA vs DS (n = 17) were matched (Fig S4B). The Venn diagram presents 521 the common list of proteins from in silico key regulator proteins (n=15), ≥1.5 fold changed 522 proteins in AIA vs healthy (n=65), ≥1.5 fold changed proteins in AIA vs TXR (n=11) and 523 significantly changed proteins between AIA vs TXR (n=27) (Fig S4C). C9 was identified as the 524 most common protein from these 4 lists. This signifies the importance of C9 as the molecular 525 target of TXR. No common protein was found in the Venn diagram of five lists together 526 (excluding n=65 list) (Fig S4D). 527

Molecular interaction of key regulators with TXR through in silico docking:
The 528 combination of these distinct methods viz. iTRAQ differential proteomics and in silico network 529 analysis studies helped to narrow down our search to the very specific protein with probability of 530 24 being the target of TXR, i.e., the complement component 9 (C9) (Fig 6C). The role of C9 is 531 critical for the formation of membrane attack complex (MAC) resulting in tissue injury which 532 further activates the entire complement pathway (Fig 6D). All these compounds were found to 533 position it in the deep cavity of the proteins which show several close interactions to their 534 catalytic residues (Fig S5) along with C9 ( Fig 6E). Here, several residues of binding pocket are 535 forming strong hydrogen bonds with the compound TXR in addition to several Van  Serpina6 (-5.9 kcal/mol), Afm (-7.9 kcal/mol), Gyg1 (-6.7 kcal/mol), Ppp1cc (-6.7 kcal/mol), 543 Pfkp (-7.3 kcal/mol), Dhfr (-7.5 kcal/mol), Cat (-7.1 kcal/mol), Trhr (-6.6 kcal/mol), Vps29 (-7.2 544 kcal/mol), Lta4h (-8.5 kcal/mol), Rac1 (-6.3 kcal/mol), Lhpp (-6.4 kcal/mol) (Table S6 and Fig  545   S5). Proteins associated with inflammatory processes such as acute phase plasma proteins or 546 others may indicate an intervention by TXR in the disease progression. Therefore, to investigate 547 the effect of therapeutic compound on the expression of plasma proteins in the arthritic rats was 548 of importance in our earlier study also(37). We used the network theoretical approach which has 549 considered the hubs, motifs and modules of the network with equal emphasis for the 550 identification of key regulators or the significant regulatory pathways preventing any bias 551 towards the overrepresented hubs or motifs. A relationship between hubs, motifs and modules 552 was established and the network used all proteins associated with the disease instead of merely 553 the manually curated datasets' usage. In conclusion, the hubs with highest degree were 554 identified, fifteen (15) were considered as the novel key regulators. 555

Discussion 556
The autoimmune disease, RA is considered to be an incurable and difficult to manage 557 disease, identification of pharmacological compounds with minimum side effects is a way 558 forward to for its management. The current pharmaceutical solutions involve disease modifying 559 arthritis drugs which work in ~60% of the cases and in certain populations can lead to adverse 560 reactions like pneumonia, tuberculosis, and interstitial pneumonitis. Identification of predictive 561 markers might facilitate to develop personalized therapy to gain optimum treatment benefit in 562 RA patients. In this study, we monitored the anti-arthritic potential of a phytochemical i.e., TXR 563 and investigated its probable mechanism of action by identifying its target proteins in the joint 564 homogenate. 565 In our in vitro studies, TXR has significantly reduced the nitrite levels with negligible 566 effects on cell viability up to a concentration of 674 μM as shown by MTT assay. It is widely 567 reported that TXR is a phytochemical with proven health benefits. It also has a very low toxicity 568 (LD50 of 27160 mg/kg body weight in rat) and safe for human use (38-41). 569 We used commonly adopted AIA animal model and observed manifestation of the RA disease 570 parameters. With the TXR treatment, there was a significant reduction of radiographic and 571 histological scores in all the AIA study groups receiving treatment in a dose-dependent manner. 572 Significant decrease in footpad thickness was observed and similar body weight gain was noticed 573 in the TXR treated AIA groups. Based on the histological analyses of liver and kidneys, it seems 574 that the damage caused by AIA was not corrected with TXR or DS treatment. We used DS as a conditions. We have found that TXR disrupts the MAC mediated complement pathway (Fig 6D), 610 however, further investigation at the biophysical level will be beneficial. Molecular docking study showed that TXR has high affinity towards C9, so it seems that 635 TXR will be even more effective to cause steric hindrance resulting in the disassembly of the 636 multimeric C9 in the MAC. This may inhibit the inadvertent cellular lysis and joint damage in 637 case of sterile inflammation or related pathogenesis (57). So, TXR may inhibit necrosis mediated 638 cell death by blocking C9 involvement in MAC formation. Since, TXR is well tolerated; 639 inhibition of its possible molecular targets may have minimal or no adverse effects. TXR has 640 been widely reported to possess hepatoprotective (58, 59) as well as renoprotective (60) 641 properties which can now be attributed to the MAC inhibition thus protecting the hepatic and 642 renal cells from xenobiotic mediated lysis. TXR is also a known antioxidant and inhibits 643 29 oxidative stress mediated cellular apoptosis (61) and may protect the synovial joints alleviating 644 severe damage in AIA and possibly in rheumatoid arthritis patients. 645 Natural products that could inhibit production of chemokines and cytokines and modulate 646 osteo-immune cross-talk could be useful in the treatment modalities for RA. These molecules 647 may vary through many other inflammatory mediators such as NF-κB, MAPK, and STAT3, etc. 648 It can also be inferred that the effect of TXR might have resulted in the inhibition of HMGB1 In conclusion, our quantitative proteomics approach demonstrated the anti-arthritic 660 properties of a phytochemical TXR and its probable interaction with synovial proteins. 661 Combination of quantitative proteomics study supported by the robust protein community 662 finding method provided a comprehensive tool to map probable targets of TXR and its 663 mechanism of action. The experimental approach adopted in this study will be useful at various 664 phases of drug discovery and validation in translational studies for various disease conditions. 665        Figure 1 Schematic representation of the experimental approach used in this study. A. Nitric oxide inhibition assay simultaneously with the cytotoxicity of TXR was evaluated using RAW264.7 cells. B. Induction of adjuvant induced arthritis (AIA) and the treatment plan. C. Joint homogenates from multiple study groups were used for proteomics experiment. D. Informatics analyses were used for identifying the key regulator proteins.  TXR treatment suppresses arthritis score in adjuvant-induced arthritis rats in a dose dependent manner. The arthritis score (mean ± SD) (A), change in foot pad thickness and (B) change in the body weight, (C) shows a trend with TXR administration. Photomicrographs of hematoxylin and eosin stained histological slides of kidney (D) and liver (E) of rats of different study groups. The histological scores of kidney (F) and liver (G) were derived from analysis of different parameters. n.s.: not signi cant at 95% con dence and p-value of less than 0.05 was considered as signi cant. *p <0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.