Bacterial lipopolysaccharide acts as an adjuvant to induce autoimmune arthritisin mice

Abstract

We investigated the ability of lipopolysaccharide (LPS) as an adjuvant to induce autoimmune arthritis. LPS from Escherichia coli was intraperitoneally injected into DBA/1J mice together with the joint cartilage component type II collagen (CII) on day 0. Thereafter, the injection of CII and LPS was continued every 2 weeks up to day 56. The results showed that mice injected with CII plus LPS had signs of arthritis on day 55 and the joint inflammation reached a peak on day 75. Injection of CII or LPS alone induced no arthritis. Histologically, marked oedema of synovium and intense infiltration of inflammatory cells, including neutrophils, were observed 3 days after the onset of joint inflammation. Twenty‐one days later, there were marked proliferation of synovial tissues with many mononulear cells and destruction of cartilage. Anti‐CII immunoglobulin G (IgG) and IgG2a antibodies were markedly produced in mice injected with CII plus LPS. Pronounced secretion of cytokines, including interleukins‐12 and ‐1β, interferon‐γ and tumour necrosis factor‐α, was also observed in these animals. Arthritis was passively transferred into naive syngeneic mice with sera but not with lymphoid cells from mice given CII with LPS. Other types of LPS from Salmonella enteritidis, Salmonella typhimurium and Klebsiella pneumoniae as well as lipid A from E. coli, induced inflammation in joints when administered with CII. Polymixin B sulphate mixed with LPS or lipid A blocked the induction of joint inflammation. These results indicate that LPS appears to play an important role as an adjuvant in the induction of arthritis in which autoimmunity to CII is involved.

Introduction

Lipopolysaccharide (LPS) is a component of the Gram‐negative bacterial cell wall that activates B cells, resulting in marked production of polyclonal antibodies.^1,2 LPS is also a potent substance that secretes various kinds of mediators, including interleukin‐12 (IL‐12) and interferon‐γ (IFN‐γ), involved in cellular immunity.^3,4 Therefore, a number of studies have demonstrated that LPS plays a role in some diseases in which autoantibodies or self antigen‐specific T cells are involved. For instance, LPS enhances MRL/lpr nephritis,⁵ experimental autoimmune uveitis,⁶ experimental autoimmune myocarditis⁷ and experimental autoimmune enterocolitis.⁸ However, few studies have clearly demonstrated a role for LPS in the induction of autoimmune arthritis.

Experimental models of autoimmune arthritis include collagen‐induced arthritis in mice, which is caused by immunization with type II collagen (CII) emulsified with complete Freund’s adjuvant consisting of mineral oil (incomplete Freund’s adjuvant) and heat‐killed mycobacteria, followed by a booster injection.^9,10 In this model of arthritis, which resembles rheumatoid arthritis in humans in certain clinical and histological aspects,^11,12 autoimmunity to CII is critically involved since the disease can be passively transferred to naive recipients with anti‐CII immunoglobulin G (IgG) and IgG2a antibodies.^13,14 However, it has not been clearly shown whether LPS has ability as an adjuvant to induce this type of autoimmune disease.

In the present study we show that administration of CII in combination with LPS, but not of CII or LPS alone, was followed by induction of arthritis which was associated with marked production of anti‐CII IgG and IgG2a antibodies as well as secretion of cytokines including IL‐12, IFN‐γ, IL‐1β and tumour necrosis factor‐α (TNF‐α), suggesting that LPS plays an important role as an adjuvant in the onset of autoimmune arthritis.

Materials and methods

Animals

Male DBA/1J mice, 8–9 weeks of age, were used in all experiments. The mice were bred in the animal breeding unit of Saga Medical School, Saga, Japan. They were maintained in a temperature‐ and light‐controlled environment with free access to standard rodent chow and water.

Induction of arthritis by co‐administration of type II collagen (CII) with lipopolysaccharide (LPS)

One hundred micrograms of CII extracted from native calf articular cartilage (Funakoshi Co., Tokyo, Japan) was dissolved in 100 µl of 0·005 m acetic acid and injected intraperitoneally (i.p.) into mice (day 0). Thereafter, the CII injection was repeated i.p. on days 14, 28, 42 and 56. As a control, 100 μl of 0·005 m acetic acid alone was administered i.p. on the same days. Varying doses of LPS from E. coli 011:B4 (Difco Laboratories, Detroit, MI) dissolved in 0·1 ml phosphate‐buffered saline (PBS) were also given i.p. immediately after each injection of CII. PBS was similarly administered as a control. In some experiments, LPS from Salmonella enteritidis, Salmonella typhimurium (Difco), and Klebsiella pneumoniae (Sigma Chemical Co., St. Louis, MO) and lipid A from E. coli K12D31m4 (Funakoshi Co., Tokyo, Japan) were administered i.p.

Evaluation of arthritis

To evaluate the severity of arthritis, the lesions of the four paws were each graded from 0 to 3 according to the increasing extent of erythema and oedema of the periarticular tissue as described elsewhere.¹⁵ The maximum possible score is 12.

Histology

Mice were killed 3 and 21 days after onset of arthritis and their hindpaws were amputated, fixed in 4% formalin and decalcified. The tissues were embedded in paraffin, sectioned at 4 µm, and stained with haematoxylin and eosin.

Measurement of antibodies to CII

Mice were killed on day 80 and their sera were heat inactivated at 56° for 30 min. Anti‐CII IgG and IgG2a antibodies were measured using an enzyme‐linked immunosorbent assay (ELISA).¹⁶ In brief, 96‐well flat‐bottomed microtitre plates were incubated with 100 µl/well of CII (100 µg/ml) at 37° for 1 hr and washed three times with PBS containing 0·05% Tween‐20. The wells were then blocked by incubation with 100 µl of PBS containing 1% ovalbumin (Sigma) at 37° for 1 hr. After washing, the plates were incubated with 100 µl of a 1 : 600 dilution of each serum sample at 37° for 30 min. The plates were washed, and 100 µl/well of a 1 : 1000 dilution of rat anti‐mouse IgG and IgG2a labelled with alkaline phosphatase (Pharmingen, San Diego, CA) was added and incubated at 37° for 1 hr. After washing, 100 µl of 3 mm of p‐nitrophenylphosphate (Bio‐Rad laboratories, Hercules, CA) was addedper well and the plates were incubated in the dark at room temperature for 15 min. The absorbance was then measured at 405 nm in a Titertec Multiscan spectrophotometer (EFLAB, Helsinki, Finland). The results were expressed as absorbance units for an optical density at a wavelength of 405 nm (OD₄₀₅) ± SEM

Measurement of cytokines

Spleens were removed on day 70 and cell suspensions were prepared. Erythrocytes in the cells were lysed with Tris–NH₄Cl. A total of 5 × 10⁶ cells in 1 ml RPMI‐1640 (Flow Laboratories, Inc., Mclean, VA) containing 1 mm glutamine, 100 U/ml penicillin, 100 µg/ml streptomycin, 5 × 10⁻⁵ m 2‐mercaptoethanol and 1% heat‐inactivated autologous mouse serum were cultured in 24‐well tissue culture plates, either medium alone or with 50 µg/ml CII.¹⁷ Forty‐eight hours later, supernatants were harvested and stored at −70° until assayed. Cytokine production was quantified by ELISA. The ELISA kits for IL‐12, IFN‐γ, IL‐1β and TNF‐α were purchased from Funakoshi Co. (Tokyo, Japan).

Transfer of arthritis with sera and lymphoid cells

For passive transfer of arthritis with sera, mice were killed on day 60 and 1 ml of pooled sera were injected intravenously (i.v.) into naive syngeneic mice. Passive transfer of arthritis with lymphoid cells was attempted by using the method of Taurog et al.¹⁸ Briefly, lymph node and spleen cells were cultured in RPMI‐1640 medium (Flow Laboratories Inc.) with concanavalin A (Con A; 3 µg/2 × 10⁶ cells/ml). RPMI‐1640 contained 1 mm glutamine, 100 U/ml penicillin, 100 µg/ml streptomycin, 5 × 10⁻⁵ m 2‐mercaptoethanol and 1% heat‐inactivated autologous mouse serum. After 2 days culture, 2 × 10⁸ cells in 0·5 ml of PBS were i.v. injected into the tail of normal recipient mice.

Induction of antigen‐induced arthritis

Antigen‐induced arthritis in mice was induced according to the method described previously.¹⁹ Briefly, mice were immunized with 100 µg of bovine serum albumin (BSA) emulsified with complete Freund’s adjuvant (Difco) and 21 days later 2 µg of BSA dissolved in 20 µl of PBS was intra‐articularly injected into the right and left ankle joints.

Administration of polymyxin B sulphate (PMB)

Fifty micrograms of PMB (Sigma) dissolved in 100 µl of PBS was mixed with 5 µg of LPS or 2 µg of lipid A and administered i.p. immediately after each injection of CII. To test the effect of PMB on antigen‐induced arthritis, 10 µg of the antibiotic dissolved in 20 µl of PBS containing 2 µg of BSA was intra‐articularly injected 21 days after immunization with BSA.

Results

Ability of LPS as an adjuvant to induce arthritis

To examine the ability of LPS as an adjuvant to induce arthritis, CII plus PBS (CII/PBS), CII plus 0·2, 1, or 5 µg of LPS (CII/0·2 µg LPS, CII/1 µg LPS, or CII/5 µg LPS) and acetic acid plus 5 µg of LPS (AA/5 µg LPS) were injected i.p. into mice on days 0, 14, 28, 42 and 56. The incidence of arthritis is shown in Table 1. Injection of CII/1 µg LPS induced arthritis in two of 10 mice on day 70 and four of 10 mice on days 80 and 100. When mice were treated with CII/5 µg LPS, two of 10 animals had joint inflammation on day 55. All the mice developed arthritis by day 70. In contrast, CII/PBS, CII/0·2 µg LPS and AA/5 µg LPS did not induce inflammation in joints at any of the times examined. The time–course of arthritis is shown in Fig. 1. Treatment with CII/5 µg LPS induced severe joint inflammation that reached a peak on day75 followed by a gradual decrease in its severity by day 100. Mice treated with CII/1 µg LPS had moderate arthritis.

Table 1.

Incidence of arthritis induced by co‐administration of CII with LPS

	Treatment
Days after injection	AA/PBS	CII/PBS	CII/0·2 µg LPS	CII/1 µg LPS	CII/5 µg LPS	AA/5 µg LPS
50	0/10 (0)*	0/10 (0)	0/10 (0)	0/10 (0)	0/10 (0)	0/10 (0)
55	0/10 (0)	0/10 (0)	0/10 (0)	0/10 (0)	2/10 (20)	0/10 (0)
60	0/10 (0)	0/10 (0)	0/10 (0)	0/10 (0)	7/10 (70)	0/10 (0)
70	0/10 (0)	0/10 (0)	0/10 (0)	2/10 (20)	10/10 (100)	0/10 (0)
80	0/10 (0)	0/10 (0)	0/10 (0)	4/10 (40)	10/10 (100)	0/10 (0)
100	0/10 (0)	0/10 (0)	0/10 (0)	4/10 (40)	10/10 (100)	0/10 (0)

^*Data are given as number of mice with arthritis/total number of mice used (% of incidence) and are representative of three experiments. Mice were i.p. injected with 100 µg of CII dissolved in 100 µl of 0·005m acetic acid (AA) on days 0, 14, 28, 42 and 56; 100 μl of 0·005m AA alone was given as a control. The indicated doses of LPS dissolved in 100 µl of PBS were also given i.p. immediately after each CII injection. As a control, 100 µl of PBS was administered. The incidence of arthritis was examined on the indicated days.

Figure 1.

Severity of arthritis induced by co‐administration of CII with LPS. Mice were injected i.p. with 100 µg of CII dissolved in 100 µl of 0·005 m acetic acid (AA) on days 0, 14, 28, 42 and 56. One hundred microlitres of 0·005 m AA alone was given as a control. LPS (0·2, 1 and 5 µg) dissolved in 100 µl of PBS was also given i.p. immediately after each CII injection. As a control, 100 µl of PBS was administered. The severity of arthritis was determined every 5 days after CII and LPS injection. Bars show the mean ± SEM of two to ten mice with arthritis. *P < 0·05 versus AA/PBS, CII/PBS and AA/5 µg LPS. Data are representative of three experiments.

Effect of administration of CII and LPS on histological changes in joints

Histological changes in the tarsal joints of mice treated with CII and 5 µg LPS were examined after onset of arthritis. Animals treated with AA/PBS, CII/PBS, or AA/LPS showed no histological changes in joints seen in normal mice (Fig. 2a). In contrast, treatment with CII/5 µg LPS resulted in marked oedema of synovial tissues, with massive infiltration of polymorphonuclear cells on day 3 (Fig. 2b). Twenty‐one days later, proliferated synovium, intense cell infiltration in which mononuclear cells predominated, and marked destruction of cartilage were observed (Fig. 2c). These histological results are summarised in Table 2.

Figure 2.

Histological changes in tarsal joints of mice with arthritis. Mice were injected i.p. with 100 µg of CII dissolved in 100 µl of 0·005 m acetic acid on days 0, 14, 28, 42 and 56. Five micrograms of LPS dissolved in 100 µl of PBS was also given i.p. immediately after each CII injection: (a) normal joint; (b) 3 days after onset of arthritis; (c) 21 days after onset of arthritis. Haematoxylin and eosin stained, × 200 (original magnification).

Table 2.

Summary of the histological data

Days after onset	Synovial oedema	PMN infilt.	MC infilt.	Cartilage destruction
0	–	–	–	–
3	+ +	+ +	+	±
21	±	±	+ +	+ +

PMN, polymorphonuclear cells; MC, mononuclear cells; infilt., infiltration; –, none; ±, weak; +, moderate; + +, strong.

Effect of administration of CII and LPS on anti‐CII antibody production

The effect of administration of CII plus LPS on the production of anti‐CII IgG and IgG2a antibodies was investigated. Mice treated with CII/0·2 µg LPS showed significantly greater production of anti‐CII IgG antibody compared with those given CII/PBS (Fig. 3). Treatment with CII/1 µg LPS and CII/5 µg LPS further facilitated the IgG antibody production. AA/5 µg LPS treatment was also effective in enhancing the level of anti‐CII IgG antibodies, although the enhanced antibody production by AA/5 µg LPS was much weaker than that by CII/5 µg LPS. Anti‐CII IgG2a antibody production was also augmented by CII plus LPS in a dose‐related manner. Mice treated with AA/5 µg LPS had greater amounts of anti‐CII IgG2a antibodies than those administered with CII/PBS, but lesser amounts than those given CII/5 µg LPS.

Figure 3.

Anti‐CII antibody production by co‐administration of CII with LPS. Mice were injected i.p. with 100 µg of CII dissolved in 100 µl of 0·005 m acetic acid (AA) on days 0, 14, 28, 42 and 56. One hundred microlitres of 0·005 m AA alone was given as a control. LPS (0·2, 1 and 5 µg) dissolved in 100 µl of PBS was also given i.p. immediately after each CII injection. As a control, 100 µl of PBS was administered. The serum levels of anti‐CII IgG and IgG2a antibodies were determined on day 80. Bars show the mean ± SEM of five mice. *P < 0·05 versus AA/PBS; **P < 0·05 versus CII/PBS; #P < 0·05 versus CII/5 µg LPS, Mann–Whitney analysis. Data are representative of three experiments.

Effect of administration of CII and LPS on the secretion of cytokines

To examine the effect of administration of CII plus LPS on the secretion of cytokines, IL‐12, IFN‐γ, IL‐1β and TNF‐α were measured. As shown in Table 3, treatment with CII/0·2 µg LPS, CII/1 µg LPS and CII/5 µg LPS was followed by increases in all these cytokines in a dose‐related manner. There were also significantly high levels of IL‐12, IFN‐γ, IL‐1β and TNF‐α in animals treated with AA/5 µg LPS, although their secretion by this combined administration was much weaker than that by CII/5 µg LPS.

Table 3.

Secretion of cytokines in mice treated with CII and LPS

	Cytokine (pg/ml)
Treatment	IL‐12	IFN‐γ	IL‐1β	TNF‐α
AA/PBS	84 ± 6	108 ± 12	64 ± 4	95 ± 10
CII/PBS	136 ± 12*	180 ± 14*	154 ± 10*	204 ± 23*
CII/0·2 µg LPS	347 ± 22*,**	275 ± 16*,**	483 ± 47*,**	558 ± 36*,**
CII/1 µg LPS	2196 ± 253*,**	1657 ± 202*,**	2378 ± 342*,**	4378 ± 375*,**
CII/5 µg LPS	7392 ± 530*,**	3352 ± 283*,**	5842 ± 478*,**	8326 ± 639*,**
AA/5 µg LPS	1648 ± 153*,**,***	927 ± 105*,**,***	1288 ± 110*,**,***	2196 ± 184*,**,***

Mice were i.p. injected with 100 µg of CII dissolved in 100 µl of 0·005 m acetic acid (AA) on days 0, 14, 28, 42 and 56; 100 μl 0·005 m AA was given as a control. The indicated doses of LPS dissoved in 100 µl PBS were also given i.p. immediately after each CII injection. As a control, 100 µl PBS was administered. The secretion of cytokines from spleen cells was examined on day 70. Data are representative of three experiments.

^*P < 0·05 versus AA/PBS

^**P < 0·05 versus CII/PBS

^***P < 0·05 versus CII/5 µg LPS, Mann–Whitney analysis.

Passive transfer of arthritis with sera and lymphoid cells

To learn whether arthritis induced by a combination of CII and LPS is mediated by immune responses to CII, sera and lymphoid cells from mice given CII/5 µg LPS were injected i.v. into normal recipient mice. Signs of arthritis were observed 3 days after injection of the sera and the joint inflammation reached a peak on day 5 (Fig. 4). In contrast, there was no joint inflammation observed in the recipient mice injected with the lymphoid cells.

Figure 4.

Passive transfer of arthritis with sera from mice treated with CII plus LPS. Mice were injected i.p. with 100 µg of CII dissolved in 100 µl of 0·005 m acetic acid (AA) on days 0, 14, 28, 42 and 56. Five micrograms of LPS dissolved in 100 µl of PBS was also given i.p. immediately after each CII injection. Mice were killed on day 60 and 1 ml of the sera or 2 × 10⁸ of the lymphoid cells were injected i.v. into the tail of naive recipient mice: •, sera injected; ○, lymphoid cells injected. Data are representative of two experiments.

Effect of varying types of LPS and lipid A on the induction of arthritis

LPS from other Gram‐negative bacteria and lipid A from E. coli were also used to test their ability as an adjuvant to induce arthritis. As shown in Fig. 5, i.p. injection of all types of LPS from S. enteritidis, S. typhimurium and K. pneumoniae together with CII induced joint inflammation comparable to that caused by the endotoxin from E. coli. Lipid A from E. coli was also active in inducing arthritis.

Figure 5.

Induction of arthritis by administration of CII with varying types of LPS or lipid A. Mice were injected i.p. with 100 µg of CII dissolved in 100 µl of 0·005 m acetic acid (AA) on days 0, 14, 28, 42 and 56. One hundred microlitres of PBS containing 5 µg of LPS from E. coli, S. enteritikdis, S. typhimurium and K. pneumoniae and 2 µg of lipid A from E. coli were also administered i.p. immediately after each CII injection. The severity of arthritis was determined on day 75. Bars show the mean ± SEM of eight mice. *P < 0·05 versus PBS, Mann–Whitney U‐test. Data are representative of two experiments.

Effect of PMB on arthritis induced by CII plus LPS or CII plus lipid A

The effect of PMB which neutralizes LPS^20,21 on the induction of arthritis by CII plus LPS was investigated. For this study, the antibiotic was mixed with LPS or lipid A before its injection. In order to examine the specificity of the effect of PMB, the effect of the antibiotic was also tested on antigen‐induced arthritis which was induced by immunization with BSA and then by intra‐articular injection of the same antigen.¹⁹ The results are shown in Fig. 6. Admininstration of PMB completely blocked the induction of joint inflammation by CII/LPS or CII/lipid A. Conversely, PMB failed to affect the antigen‐induced arthritis.

Figure 6.

Prevention of CII plus LPS‐induced arthritis by PMB. Mice were injected i.p. with 100 µg of CII dissolved in 100 µl of 0·005 m acetic acid (AA) on days 0, 14, 28, 42 and 56, or subcutaneously immunized with 100 µg of BSA dissolved in 50 µl of PBS emulsified with an equal volume of complete Freund’s adjuvant on day 0. One hundred microlitres of PBS containing 5 µg of LPS or 2 µg of lipid A alone, or 100 µl of PBS containing 5 µg of LPS or 2 µg of lipid A mixed with 50 µg of PMB was given i.p. immediately after each CII injection. Twenty microlitres of PBS containing 2 µg of BSA alone or 20 µl of PBS containing 2 µg of BSA mixed with 10 µg of PMB was intra‐articularly injected 21 days after immunization with BSA. *P < 0·05 versus LPS/PBS, **P < 0·05 versus Lipid A/PBS, ***P < 0·05 versus BSA/PMB, Mann–Whitney U‐test. Data are representative of two experiments.

Relationship between the frequency of CII plus LPS injections, anti‐CII antibody production and induction of arthritis

A relationship between the frequency of co‐administration of CII with LPS, production of anti‐CII IgG and IgG2a antibodies, and the incidence and severity of arthritis was examined. As shown in Table 4, levels of anti‐CII IgG and IgG2a antibodies were increased by more frequent injections of CII plus LPS. There were no signs of arthritis in mice given two injections of CII plus LPS. Three, four and five injections of both substances induced arthritis in two, six and seven of seven mice, respectively. There was no difference in the disease severity between arthritic mice administered with three, four and five injections of CII plus LPS.

Table 4.

Correlation between the frequency of administration of CII plus LPS, production of anti‐CII antibodies, and the incidence and severity of CIA

Days of administration	Anti‐CII IgG (A405)	Anti‐CII IgG2a (A405)	Incidence (%)	Mean joint score
0	0·14 ± 0·02†	0·09 ± 0·01†	0/7 (0)‡	0
0, 14	0·41 ± 0·03*	0·34 ± 0·02*	0/7 (0)	0
0, 14, 28	0·69 ± 0·04*,**	0·62 ± 0·04*,**	2/7 (29)	6·5 ± 2·27§,*,**
0, 14, 28, 42	0·89 ± 0·06*,**,***	0·76 ± 0·05*,**,***	6/7 (86)	7·1 ± 1·03*,**
0, 14, 28, 42, 56	0·93 ± 0·06*,**,***	0·80 ± 0·06*,**,***	7/7 (100)	7·8 ± 0·94*,**

Mice were i.p. injected with 100 µg of CII dissolved in 100 µl of 0·005 m acetic acid (AA) on the indicated days. Five micrograms of LPS dissolved in 100 µl of PBS were given i.p. immediately after each CII injection. The serum levels of anti‐CII IgG and IgG2a antibodies and the incidence and severity of arthritis were determined on day 80.

^†Mean ± SEM of seven mice

^‡Number of mice with arthritis/total number of mice used

^§Mean ± SEM of two to seven mice with arthritis.

^*P < 0·05 versus day 0

^**P < 0·05 versus days 0, 14

^***P < 0·05 versus days 0, 14, 28.

Discussion

The present study demonstrates that bacterial LPS may play a role as an adjuvant in the induction of autoimmune arthritis since mice injected with CII and LPS from E. coli as well as from S. enteritidis, S. typhimurium and K. pneumoniae developed joint inflammation in which autoimmunity to CII was involved. Administration of CII in combination with lipid A from E. coli LPS was also effective in inducing the autoimmune disease. Neither CII nor LPS alone induced arthritis. The specific role of LPS and lipid A in the induction of arthritis was confirmed by the blockade of the joint inflammation by PMB that neutralized LPS and lipid A.^20,21 It was previously demonstrated that LPS or its lipid A portion enhanced MRL/lpr nephritis in mice,⁵ experimental autoimmune uveitis,⁶ experimental autoimmune myocarditis⁷ and experimental autoimmune enterocolitis.⁸ However, few studies clearly showed a role for LPS in autoimmune arthritis.

Marked increases in anti‐CII IgG and IgG2a antibodies were observed in mice treated with CII plus LPS. There was also a good correlation between the frequency of administration of CII plus LPS, the serum level of anti‐CII IgG and IgG2a antibodies, and the incidence and severity of arthritis. In addition, arthritis induced by administration of CII with LPS was passively transferred to naive recipient mice by sera from animals given CII plus LPS, suggesting that humoral responses to CII were critically involved in the induction of the joint inflammation. Attempts to transfer the disease passively with lymphoid cells from the animals failed, indicating that cellular responses to CII might not be directly involved in the onset of the joint inflammation. Histologically, there were marked oedema of synovium and infiltration of many neutrophils in the early phase of arthritis and this was followed by the chronic destructive phase in which pronounced proliferation of synovium containing mononuclear cells and erosion of cartilage were observed. Because these histological and immunological features are also seen in the previously reported model collagen‐induced arthritis in mice which was caused by immunization with CII emulsified with complete Freund’s adjuvant,^9,10 the pathogenic mechanism of our model of arthritis appears to be similar to that of collagen‐induced arthritis. In other words, LPS or lipid A appears to act as an adjuvant in the same manner as complete Freund’s adjuvant.

The pronounced production of anti‐CII IgG2a antibodies seen in our experiments may be explained in part by the enhanced secretion of IFN‐γ in mice treated with CII plus LPS because this T helper type 1 (Th1) cytokine plays a critical role in the production of IgG2a.²² Furthermore, the enhanced secretion of IFN‐γ may be due to the augmentation by LPS of the secretion of IL‐12 that stimulates production of the Th1 cytokine²³ since the endotoxin does not normally recognize T cells, but activates non‐lymphocytes, including macrophages that produce IL‐12.³

The induction of joint inflammation by CII plus LPS may be also in part explained by augmented secretion of IL‐1β and TNF‐α by LPS since these cytokines are known to produce various inflammatory mediators, including nitric oxide and arachidonic acid metabolites.^24,25 Stimpson et al.²⁶ demonstrated that arthritis induced by the toxic effect of peptidoglycan‐polysaccharide polymers injected intra‐articularly in rats was reactivated by a systemic injection of LPS. It is also likely that anti‐CII antibodies and these inflammatory mediators acted synergistically to induce arthritis. Terato et al.²⁷ showed that injection of LPS reduced the dose and the number of anti‐CII monoclonal antibodies required for induction of joint inflammation, suggesting the synergistic effect of LPS on arthritis.

Injection of LPS or CII alone was also followed by enhanced production of anti‐CII IgG and IgG2a antibodies as well as IL‐12, IFN‐γ, IL‐1β and TNF‐α, although the extent of the antibody and cytokine production was much weaker than that in mice administered with CII plus LPS. This is probably mainly due to the non‐specific activation of macrophages and B cells by the endotoxin. We treated mice with LPS alone every 2 weeks over a period of 140 days to test whether repeated treatment of LPS alone was effective in inducing arthritis. However, no joint inflammation developed in these animals in which levels of anti‐CII‐specific IgG and IgG2a antibodies, IL‐12, IFN‐γ, IL‐1β and TNF‐α did not reach those seen in mice administered with CII plus LPS (data not shown). Therefore, both CII and LPS appear to be required for the marked production of anti‐CII antibodies and those cytokines necessary for the development of arthritis. In other words, the enhanced levels of the antibodies as well as the cytokines caused by LPS or CII alone appear to be not sufficient to induce arthritis. There is one report showing that oral administration of CII or LPS alone, or both, induced arthritis in mice.²⁸ However, our repeated attempts to reproduce the results failed. This appeared to be due to the weak production by feeding CII or/and LPS of anti‐CII IgG and IgG2a antibodies in serum that was not sufficient for the induction of arthritis. Thus, in our experiments, arthritis was readily induced by systemic but not oral administration of both CII and LPS.

Higher levels of antibodies to E. coli²⁹ as well as LPS‐binding protein³⁰ are observed in sera and synovial fluids from patients with rheumatoid arthritis than in those from control healthy subjects. These findings imply a role for LPS in the autoimmune disease which may be supplied following infection with Gram‐negative bacteria or by normal flora in the gut, including E. coli.^31,32 As shown in the present study, LPS not only from E. coli but also from S. enteritidis, S. typhimurium and K. pneumoniae and the LPS active site lipid A from E. coli induced arthritis when co‐administered with CII. We also recently found that oral administration of LPS caused exacerbation of CIA that was associated with increased amounts of the endotoxin in blood (unpublished data). CII is a major component of cartilage and IgG as well as IgM antibodies specific for CII are seen in patients with rheumatoid arthritis.^33,34 Furthermore, CII and anti‐CII antibody complexes are present in sera from the patients.³⁵ These findings suggest that LPS could contribute to the production of anti‐CII antibodies in rheumatoid arthritis, although the precise role of the antibodies in the human disease remains obscure.

In summary, systemic injection of both CII and LPS, but not CII or LPS alone, induced arthritis that was associated with marked production of anti‐CII IgG and IgG2a antibodies as well as pronounced secretion of IL‐12, IFN‐γ, IL‐1β and TNF‐α, suggesting that the endotoxin may act as an adjuvant in the induction of autoimmune joint inflammation such as rheumatoid arthritis.

Glossary

Abbreviations

AA

acetic acid

CII

type II collagen

PMB

polymyxin B sulphate