To address NO as an initiating factor of inflammatory synovitis, we simply administered mice with SNP by intra-articular injection. Fascinatingly, a single injection with SNP can lead to the significant edema on the paw after only one day. As expectation, coinjection with a mixture of SNP and CII-CFA induces a severe inflammatory phenotype, confirming that NO legitimates a causative effector leading to inflammatory lesions. The exogenous NO-induced acute synovitis, therefore, represents a rapidly constructed early phase model of RA in mice.

As a critical hypoxic parameter, SpO₂ is very low (55–57 %) in mice after injection with SNP regardless of treatment with or without CII-CFA. At the same time, we also found that CII-CFA induces the moderate SpO₂ (62–66 %), which are lower than control mice (82 %). It is worthy of noting that SpO₂ seems not to correlate the severity of inflammation. For example, we saw that the inflammatory signature is more severe in CII-CFA-injected mice with high SpO₂ (62 %) than SNP-injected mice with lower SpO₂ (55–57 %). These results implied that CII-CFA-triggered immune activation, in addition to NO, should be implicated in the acceleration of inflammatory progression.

To validate bacterial infection as the origin of NO generation, we determined the serum NO level in BIA mice after subcutaneous injection of CEF, ART, or ART + CEF for three days (twice a day). As results, either treatment can considerably decline the NO level lower than the control. Conceivably, ART declines the NO level because it inhibits iNOS even though infection still exists. CEF allows an equal NO level to the control because it kills bacteria and blocks infection. The combined treatment leads to a lower NO level owing to the dual effects of CEF-suppressed infection and ART-inhibited NO production. These results verified that persistent gastrointestinal infection and potent NO burst can be abrogated by antibacteria and/or iNOS inhibition.

We can conveniently evaluate the therapeutic efficacy of CEF, phytol, alcohol, or a drug combination on BIA by measuring the hypoxic parameters, including NO, LA, and SpO₂. From these measurements, we noticed that CEF injection or diluted alcohol drinking can eradicate the hypoxia-derived consequences, but phytol only partially normalizes the hypoxic parameters (Table 5.1).

In CIA mice established by intradermal or intra-articular injection of CII-CFA, on the other hand, administration of ART or RAP enables a dramatic decrease of the NO level, even lower than the control. A combination of ART with RAP, or RAP with alcohol in BIA-CIA mice leads to the considerable repression of NO production. In particular, an extremely lower NO level (0.448 mM) was detected in an ART-injected CIA mouse. In contrast, untreated BIA-CIA and CIA mice exhibit higher NO levels, up to 10 mM in the maximum.

Intriguingly, CIA mice without any treatment can give rise to a low NO level almost equal to that of the control, which might be attributed to the decay of CII-CFA-triggered NO after a longer duration of postimmunization. Actually, three weeks have been passed from CII-CFA boosting to NO detection.

According to the morphological and histochemical identifications after treatments, it can be confirmed that articular synovitis would be ameliorated in antiarthritic drug-treated mice, although their merits are not on an average level. In mice with synchronous intra-articular CII-CFA injection and ART or RAP administration, no intimal hyperplasia and subintimal fibrosis occur in the synovial tissue of mice. However, dispersed synovial infiltration by inflammatory lymphocytes was observed in ART-treated mice, but not in RAP-treated mice. These results indicated that ART as an iNOS inhibitor only suppresses hyperplasia, while RAP as an immunosuppressant that blocks immune activation-dependent NO production can ameliorate synovial hyperplasia and lymphocytic infiltration.

Among CIA mice, postmodeling treatment by ART and RAP aborts both intimal hyperplasia and subintimal fibrosis, but fails to block the mild and dispersed inflammatory infiltration into synovial tissues by mononucleated cells. Coadministration of CIA mice with RAP and alcohol cannot hamper the synovial progression to local hyperplasia and lymphocytic infiltration. Postmodeling injection of BIA-CIA mice with ART and RAP abolishes the subintimal fibrosis and inflammatory infiltration, but intimal hyperplasia cannot be blocked completely. In contrast, RAP and alcohol can remit synovial hyperplasia, but do not inhibit the eventual progression to the mild and dispersed inflammatory infiltration.

All above results from different regimens demonstrated that pretreatment prior to NO generation is more effective than posttreatment after NO generation. In other words, synovial damage made prior to drug administration cannot be ameliorated by any regimens of posttreatment. These results conclusively indicated that NO should represent one of the most important initiators leading to RA-like synovitis and arthritis. NO is mainly responsible for synovial hyperplasia, so NO-initiated synovial lesions may be irreversible by iNOS inhibitors.

Through pathogenic infection-triggered NO production, iNOS is involved in an immune attack against active invaders. For example, bacterial infection of human colon epithelial cells has been reported to allow upregulated iNOS expression and enhanced NO production (Witthoft et al. 1998). Apart from the gastrointestinal infection following live bacterial feeding, immunization of mice with CII-CFA also provokes potent NO burst, hence suggesting that NO production is dependent on the immune activation regardless of the pathogenic infection. This is the reason why live bacterial feeding and CII-CFA injection can equally induce the synovial inflammation.

The modeling of RA in mice generally requires the heat-killed M. tuberculosis-containing CFA in addition to CII, implying that anti-CII responses are insufficient to induce the classic arthritic lesions in mice. Some authors thought that anti-CII reactivity might be a consequence of inflammation rather than the cause (Courtenay et al. 1980). This is the reason why so many kinds of autobodies can be detected in the blood of RA patients or experimental arthritic rodents, such as those against citrullinated proteins, glucose-6-phosphate isomerase, integrin, and fibrin, etc. (McDevitt 2000; Wilder 2002; Humby et al. 2009). So we assumed that CII is likely dispensable for modeling RA in mice. Indeed, we successfully induced acute arthritis in mice by intra-articular injection with CFA alone instead of CII-CFA. Therefore, any immune activators including autoantigens and bacteria, either live or dead, can evoke the immune responses and make the inflammatory lesions by provoking the NO-driven hypoxia.

The mucosal response to an enteric infection includes the production of chemoattractant cytokines (chemokines), anti-inflammatory cytokines, and proinflammatory cytokines, in which TNFα is an important proinflammatory cytokine that amplifies the epithelial immune response to the bacterial infection (Eckmann and Kagnoff 2005). TNFα, IL-1β, and other proinflammatory cytokines can upregulate iNOS in chondrocytes and synovial cells of osteoarthritis (Abramson 2004). In the present study, we observed a global activation of chemokines, cytokines, and corresponding receptors in BIA and CIA mice, which drives a complicated and chronic inflammatory process (Cejka et al. 2010). The unexpected outcome that TNFα is downregulated in CIA mice seems puzzling, but it is most likely because the attenuation of immune responses to CII-CFA after several weeks of immunization. In fact, sampling for cytokine chip profiling was carried out after 28 days of modeling.

Although the implication of NO in the pathogenesis of experimental arthritis was poorly understood, CII-CFA-triggered NO burst has been noticed in CIA mice (Cannon et al. 1996). We have observed the formation of double NO peaks after primary challenging and boosting with CII-CFA, and also noticed a stable NO level higher than the control during daily live bacterial feeding. Albeit in distinct manners, either CII-CFA or bacteria can provoke NO production, permanently or transiently, in BIA and CIA mice. Surprisingly, we also found that cotreatment by CII-CFA with bacteria allows a global downregulation of proinflammatory cytokines, suggesting the bacteria-conferred suppression to CII-CFA-activated immune responses. In consistence with our findings, other authors have previously reported that Schistosoma japonicum infection can significantly attenuate the clinical signs, reduce the histological damages, alter the humoral immune responses, and inhibit the splenocyte proliferation in CIA mice (Song et al. 2011).

Additionally, immunosuppression by bacteria has been addressed in a recently published review (Kelly et al. 2012). Because arthritic development is mostly dependent on the systemic immune activation, immunosuppression can, of course, alleviate the inflammatory arthritis. RAP is a well-known immunosuppressant that reduces pannus formation, cartilage erosion, and joint damage in rats with adjuvant-induced arthritis (Teachey et al. 2009). ART also plays an antiarthritic role in CIA mice (Wang et al. 2008). Even though the pharmacological mechanisms of RAP and ART as arthritic therapeutics are thought to be multifaceted, it is unambiguous that both drugs can suppress NO production, and administration of RAP and/or ART can effectively block the onset of synovitis. So there should be an association between the increase of arthritic inflammation and the decrease of NO production.

Antibacteria by CEF, anti-inflammation by diluted alcohol, proapoptosis by phytol, or a combination of multidrugs in live bacterial feeding mice improves the inflammatory feature of synovial tissues, which could be reflected by the normalization of hypoxic parameters. CEF can dramatically decrease those hypoxic parameters to the values comparative to the control. Phytol, alcohol, or their combinations also normalize, more or less, the hypoxic parameters. Support evidence of alcohol beneficial to antiarthritis is from a clinical cohort demonstrating that alcohol consumption is inversely associated with the risk and severity of RA (Maxwell et al. 2010). On the other hand, phytol as an oxidative burst inducer was used for treatment of experimental arthritis in rats, by which autoimmune responses are suppressed and both acute and chronic arthritis ameliorated (Hultqvist et al. 2006).

We observed the remarkable synovial hyperplasia and angiogenesis from the histochemical analysis of articular sections. Angiogenesis is an early event in the inflammatory joint, which enables the activated monocytes entering the synovium and expanding them throughout a pannus via the recruitment of endothelial cells, eventually resulting in cartilage degradation and bone destruction (Kennedy et al. 2010). Hypoxia can induce the expression of angiogenesis-related genes including HIF-1α and VEGF (Kasuno et al. 2004). NO can also activate HIF-1α under the normoxic conditions (Natarajan et al. 2003). So we concluded that NO is eligible as a hypoxic inducer capable of initiating angiogenesis and hyperplasia.

The synovium itself is a relatively hypoxic tissue, in which O₂ tension in cartilage ranges from 7 % (53 mmHg) in the superficial layer to less than 1 % (7.6 mmHg) in the deep zone (Fermor et al. 2007). NO can also accelerate its own consumption by increasing its entry into red blood cells (Han et al. 2003). NO inhibits the mitochondrial enzyme COX in competition with O₂, leading to so-called a “metabolic hypoxia” situation, in which cells cannot use O₂ although it is available (Xu et al. 2005). High levels of NO inhibit cell respiration by binding to COX, whereas slow and small-scaled NO release can stimulate mitochondrial biogenesis in diverse cell types (Nisoli and Carruba 2006). Our results also indicated that high NO levels are correlated with low SpO₂, hence validating NO conveying signals for angiogenesis and hyperplasia.

Due to the hypoxic induction, blood sugars are anaerobically catabolized and necessarily converted to LA by glycolysis, which can be accumulated in the bloodstream unless O₂ supply is rehabilitated. By monitoring the dynamic changes of NO and LA, we observed a proportional fluctuation of NO with LA in arthritic modeling mice. The high LA level is a new and simple parameter for quantifying hypoxia and indicating transversion from normoxia to hypoxia. Furthermore, it is known that hypoxia can activate HIF-1α, which in turn binds to the promoter of downstream hypoxia-inducible genes such as VEGF for starting transcription and translation (Olson and van der Vliet 2011). We detected the overexpression of HIF-1α and VEGF in the inflamed synovial tissue of CIA mice. When SNP was injected into the hypoderm of mice, HIF-1α and VEGF are expressed in higher levels, thereby confirming a relevance of NO-driven overexpression of HIF-1α and VEGF with synovial angiogenesis during hyperplasic induction.

In view of different roles playing by NO and proinflammatory cytokines, we believe both of which are likely important in the initiation and progression of inflammatory arthritis in mice. But it is possible that NO is mainly responsible for synovial hyperplasia, whereas proinflammatory cytokines are apparently relevant to inflammatory infiltration. NO may induce synovial angiogenesis and hyperplasia by hypoxia, and can subsequently guide proinflammatory cytokines penetrating deeply into the synovium along with the newborn blood vessel (Ng et al. 2010). Our results indicated that NO promotes synovial angiogenesis by activating HIF-1α and VEGF, but then enhanced angiogenesis mitigates glycolysis. These results would become a solid basis for future arthritic treatment by inhibiting NO-driven angiogenesis. Currently, published data have demonstrated that anti-VEGF treatment by bevacizumab reduces blood supply, increases glycolytic metabolites, and promotes tumor metastasis in glioblastoma (Keunen et al. 2011), underlining that no amelioration would be reached if inflammation-originated hypoxia is not yet alleviated.

Our study has answered a long-term unanswered question about the association of distal or systemic infection with inflammatory arthritis: gastrointestinal infection can serve as an etiological initiator of inflammatory arthritis by dually upregulating proinflammatory cytokines that allow lymphocytic infiltration and triggering NO to drive synovial hypoxia and hyperplasia. These achievements should shed light on the prophylactic and therapeutic interventions of RA and other human autoimmune diseases in the future.

BIA that simulates CIA was developed in mice upon daily live bacterial feeding. The morphological lesions of paw erythema and edema together with the histological alterations such as synovial hyperplasia and lymphocytic infiltration emerge as the early phase manifestation of RA. Bacteria and collagen induce the global upregulation of proinflammatory cytokines, accompanying with the elevation of serum NO levels and decline of SpO₂. NO-driven hypoxia is evident from the accumulation of LA, an end product from glycolysis. Upregulation of HIF-1α and VEGF validates hypoxia-induced angiogenesis. The administration of SNP also causes articular inflammation by inducing synovial hypoxia. Antibacteria by CEF and/or immunosuppression by RAP or inhibition by ART can abrogate NO production, mitigate hypoxia, and considerably ameliorate or even completely abort synovitis, hence highlighting NO may serve as an initiator of inflammatory arthritis. Taken together, bacteria can mimic collagen to enable synovial lesions via upregulating proinflammatory cytokines, triggering NO production, driving hypoxic responses, and inducing synovial angiogenesis and hyperplasia, suggesting sustained infection might be, in part, responsible for the onset of synovitis and arthritis in mice.