Effect of Porphyromonas gingivalis infection on gut dysbiosis and arthritis exacerbation in 1 mouse model

49 Background : Porphyromonas gingivalis (Pg) infection causes periodontal disease and is one 50 of the causative bacteria of rheumatoid arthritis (RA) exacerbation. Gut microbiota dysbiosis 51 has shown strong associations with systemic diseases, including RA, diabetes mellitus, and 52 inflammatory bowel disease, and inoculation of periodontopathogenic bacteria (i.e. Pg) can 53 alter gut microbiota composition. Therefore, this study investigated dysbiosis-mediated 54 arthritis exacerbation by Pg oral inoculation in an experimental arthritis model mouse. 55 Methods : Pg inoculation in the oral cavity twice a week for 6 weeks was performed to induce 56 periodontitis in SKG mice. Concomitantly, a single intraperitoneal (i.p.) injection of 57 laminarin (LA) was administered to induce experimental arthritis (Pg-LA mouse). 58 Citrullinated protein (CP) and IL-6 levels in periodontal, intestinal, and joint tissues, and 59 serum were measured by ELISA. Gut microbiota composition was determined by Miseq after 60 DNA purification of mouse feces. Fecal microbiota transplantation (FMT) was performed by 61 transferring Pg-RA-derived feces to normal mice. The effect of the Pg peptidylarginine 62 deaminase (PgPAD) in arthritis progression was determined using PgPAD knockout mutant.


Evaluation of alveolar bone level in mouse 171
The alveolar bone level (ABL) of SKG mice was evaluated by the Kawai's methods 172 described previously [23]. Briefly, after methylene blue staining (Sigma-Aldrich) for 10 min, 173 the upper molar jaw was washed with PBS three times. The length of the blue-stained root 174 surface of all molar teeth from the enamel-cement junction to the top of the alveolar bone was 12 measured. Differences between treated mice and the control group ( After blocking each well with 1% BSA in PBS supplemented with 0.05% Tween 20 (PBST), 196 the sample or standard (diluted in PBST from 1 ng/mL to zero) was applied to each well. Technologies, Santa Clara, CA). Real-time PCR was performed on the pooled library using a 254 KAPA Library Quantification Kit for Illumina, following the manufacturer's protocols, and a 255 sample library with a 20% denatured PhiX spike-in was sequenced by Miseq using a 600 256 cycles kit to obtain 2 × 250 bp paired-end reads. Taxonomic assignments and estimation of 257 relative abundance of sequencing data were performed using the analysis pipeline of the 258 QIIME software package [28]. An operational taxonomic unit (OTU) was defined at 97% 259 similarity. OTUs indicating relative abundance under 0.05% were filtered to eliminate noise. 260 The OTU was assigned a taxonomy based on a comparison with the Silva database using 261 UCLUST [29]. 262 263

Evaluation of periodontal tissue in experimental arthritis model mice with Pg infection 286
Oral infection of Pg induced alveolar bone loss in the Pg and Pg/LA groups (Fig. 1a  287 and 1b). However, the Ctrl and LA groups did not show any differences in ABL compared to 288 Pg or Pg/LA. Additionally, IL-6 production in mouse gingival tissue was measured, which 289 was increased in the Pg and Pg/LA groups compared to the Ctrl group (Ctrl: 92.2 pg/10 mg 290 tissue, Pg: 318.3 pg/10 mg tissue, Pg/LA: 319.4 pg/10 mg tissue). IL-6 production in the LA 291 group was similar to the Ctrl group (Fig. 1c). 292 293

Assessment of experimental arthritis in mice 294
To assess experimental arthritis in mice with Pg oral inoculation, serum and joint 295 tissues were examined. AS was determined, which was increased in the LA group after 5 296 weeks of LA injection. The Pg/LA group was also significantly elevated from 4 weeks of LA 297 injection and Pg inoculation, same as our previous report [16]. However, AS of Ctrl and Pg 298 groups did not show the induction of joint swelling (Fig. 1d). HE-stained sections of joint 299 tissues in the LA group showed mild infiltration of immune cells and growth of granulation 300 tissue. Furthermore, severe inflammation and pannus formation in the Pg/LA group were 301 observed ( Fig. 1e-h). Serum ACPA levels and IL-6 levels in serum and joint tissues were 302 20 elevated in the LA, Pg, and Pg/LA groups compared to the Ctrl group, with the highest levels 303 in the Pg/LA group (Fig. 1i-k). 304 305

Effect of Pg oral administration on gut microbiota composition 306
The composition and abundance of gut microbiota in mouse fecal samples were 307 assessed by next-generation sequencing of 16s rRNA. The relative abundance in phylum 308 level in the LA, Pg, and Pg/LA groups was changed compared to the Ctrl group (Fig. 2a). 309 Further, the LA, Pg, and Pg/LA groups showed increased relative abundance at the order 310 level of Bacteroides compared to controls (Fig. 2b). Conversely, the relative abundance of 311 Firmicutes, Deferribacteres, and Clostridiales was decreased in LA, Pg, and Pg/LA groups 312 compared to Ctrl (Fig. 2c-e). There was a significant decrease in the relative abundance of 313 Deferribacteres in the Pg and Pg/LA groups compared to the LA group (Fig. 2d). At the 314 family level, the relative abundance of S24-7 in the LA, Pg, and Pg/LA groups was 315 significantly increased compared to Ctrl, but Pg and Pg/LA groups were lower than LA (Fig.  316   2f). 317 318

Effect of Pg oral administration on gut inflammation 319
To assess gut inflammation, IL-6 production in small and large intestinal tissues was 320 measured ( Fig. 3a and b). IL-6 production in the small intestine of the Pg/LA group was 321 remarkably elevated compared to other groups (Pg/LA group: 184.8 pg/50 mg tissue). In the 322 Pg group, IL-6 production in the small intestine was slightly increased compared to the Ctrl was performed. Six weeks after LA injection, the composition and abundance of gut 341 microbiota in mouse fecal samples were assessed by next-generation sequencing (Fig. 4a-d). 342 Composition at the phylum level of LA-FMT and Pg/LA-FMT groups were similar to the LA 343 and Pg/LA groups , respectively. Further, there were no differences in the relative abundance 344 of Bacteroides, Firmicutes, and Deferribacteres. 345 To further assess the effect of FMT on the onset of experimental arthritis, mouse 346 periodontal and joint tissues were analyzed. FMT did not show any effect in periodontal 347 tissues, as alveolar bone resorption was not observed (Fig. 5a and b). However, severe joint 348 swelling was observed in the LA-FMT and Pg/LA-FMT groups compared to the LA and 349 Pg/LA groups ( Fig. 5c-g, Fig. 2f and h). The maximum AS of the LA-FMT and Pg/LA-FMT 350 23 groups was much higher than that of LA and Pg/LA, respectively. Pannus formation in the 351 Pg/LA-FMT group was found in joint tissues, similar to the Pg/LA group. Further, pannus 352 formation and erosion of joint bone tissues were strongly observed in the LA-FMT group 353 ( Fig. 5f and g). Additionally, FMT of feces from the LA and Pg/LA groups increased IL-6 in 354 joint, small intestinal, and large intestinal tissues (Fig. 5h-j). CP levels in joint, small 355 intestinal, and large intestinal tissues were also elevated in the FMT-LA and FMT-Pg/LA 356 groups, with the Pg/LA group showing the highest levels ( Fig. 5k-m). 357 358

Effect of PgPAD-deficient mutant on exacerbation of experimental arthritis 359
To determine the effect of PgPAD on CP generation and exacerbation of experimental 360 arthritis, the PgPAD-deficient Pg strain was inoculated into the oral cavity of SKG mice. 361 Bone resorption around the upper molar teeth of Pg 33277 wild type strain and PgPAD 362 deficient Pg strain was measured (Fig. 6a). Oral inoculation of the PgPAD-deficient Pg strain 363 showed slight bone resorption compared to the Ctrl group, which was much lower than that 364 the wild type strain (35.2% lower, Fig. 6b). IL-6 production in gingival tissue of the wild 365 type-inoculated group was elevated. However, this increase was suppressed in the deficient 366 24 strain (Fig. 6c). Additionally, AS of the PgPAD-deficient Pg strain was significantly lower 367 than the wild type strain (Fig. 6d). CP generation in serum and joint, small intestinal, and 368 large intestinal tissues by PgPAD-deficient Pg strain inoculation was also lower than the wild 369 type strain (55.6%, 38.2%, 35.9%, and 20.2% decrease, respectively, Fig. 6e-h). Serum 370 ACPA levels in the PgPAD-deficient Pg strain was significantly decreased compared to \wild 371 type (45.8% decrease, Fig. 6i). In this study, Pg inoculation affected gut microbiota composition. Pg inoculation 386 resulted in decreased relative abundance of Deferribacteres and S24-7 in Pg/LA compared to 387 the LA group ( Fig. 2d and f). By contrast, the abundance of Bacteroides, Firmicutes, and 388 Clostridiales did not show significant differences between groups, although their relative 389 abundance changed dramatically compared to the control group (Fig. 2)  ACPAs target proteins/peptides with citrullinated epitopes and serve as informative 417 RA biomarkers, which are useful for RA diagnosis [34]. ACPA are generated within 418 synovium and possibly at extra-articular sites prior to disease onset. Recent investigations 419 have begun to elucidate the different mechanisms by which ACPAs may be directly 420 pathogenic in RA. CP is a specific target of ACPA and involved in ACPA generation. This 421 study found increased CP in gingival, small intestinal, large intestinal, and joint tissues 422 following Pg inoculation in the experimental arthritis model mouse (Fig. 3 and 6). In humans, 423 four citrullinated autoantigens, fibrinogen/fibrin, vimentin, α-enolase, and type II collagen, 424 are now well accepted as ACPA targets [35][36][37][38]. In our study, the specific origin of CP is 425 unclear. However, Pg can rapidly generate CP from α-enolase or fibrinogen by proteolytic 426 cleavage at Arg-X peptide bonds using arginine gingipains, followed by citrullination of 427 carboxyterminal arginines by bacterial PAD [13]. Taken together, one of the mechanisms 428 may lead to ACPA generation by PgPAD citrullination. To support this hypothesis, the Pg 429 knockout mutant of PgPAD resulted in less CP generation in gingival tissue, serum, small 430 28 intestine, large intestine, and joint tissue (Fig. 6e-h). Furthermore, the Pg 33277 wild type-431 stimulated serum ACPA was diminished by stimulation with the PgPAD knockout mutant 432 (Fig. 6i). However, endogenous PAD, PADI4, is also involved in CP and ACPA generation, tissue, and serum (Fig. 1i, j and 3a, b), resulting in PADI4 induction and subsequent CP and 441 ACPA generation. 442 Surprisingly, PgPAD knockout showed both decreased CP and ACPA production and 443 suppressed inflammation of periodontal tissues, including alveolar bone resorption (Fig. 6a,  444 b), which depends on osteoclast activation via RANKL signaling [44]. A previous report 445 showed that citrullinated vimentin induced endogenous PADI4 and RANKL in fibroblast-like 446 29 synoviocytes derived from RA patients [45]. However, no previous study has shown a 447 correlation between PgPAD and inflammation. Yet, the combination of gingipain and PgPAD 448 generated citrullinated -enolase, and its CP might be an initiator of inflammation. Therefore, 449 elimination of Pg as the causative bacteria of CP production is very important. To determine the effect of Pg infection in periodontal tissue, Pg W83 (10⁸ CFU/50 654 mL/mouse in 2% CMC/PBS solution) was administered into the oral cavity of SKG mice (6-655