Immunohistochemical study of lymphoid tissues in adjuvant arthritis (AA) by image analysis; relationship with synovial lesions

Abstract

The aim of this study was to examine leucocyte populations in lymphoid organs during AA and to ascertain the relationship with lesions in synovial joints. Popliteal lymph nodes, spleen and knee synovial membranes were removed from both healthy and AA rats at intervals of 3–4 days over a 3-week period. Cryostat sections were stained with MoAbs directed against lymphocyte and macrophage subpopulations, and studied by image analysis. Throughout the arthritic period, high numbers of ED1⁺ and ED3⁺ macrophages were seen in both lymphoid compartments and intercellular adhesion molecule-1 (ICAM-1) expression also increased in some zones of lymph nodes and spleen. The percentages of CD4⁺ and CD8⁺ cells rose in the splenic zones studied but fell in the lymph node cortex. Very few natural killer (NK) cells were found in lymphoid tissues, but the number rose after AA induction. In synovia from AA rats, ED2⁺ macrophages proliferated but α/β T cell infiltration was only occasionally observed, accompanied by ED1⁺ cells and ICAM-1 expression. In conclusion, synovitis developing after AA induction seems to be caused directly by macrophages and indirectly by lymphocytes placed both in popliteal lymph nodes and spleen.

Introduction

The immune response requires perfect co-ordination among diverse subsets of lymphoid and accessory cells. Lymphoid organs provide microenvironments where interactions among T, B, and antigen-presenting cells (APC) result in the effective induction of the immune response. Moreover, the traffic of these immunocompetent cells between organs is vital for these intricate cellular interactions.

AA is a model of chronic inflammation induced in rats by a mycobacterial suspension intradermally injected. Ten to 15 days after induction, inflammation of the hind paws, or even of all four paws appears, though the mechanisms involved are unknown. Although histological analysis of inflamed joints in AA was first performed many years ago [1–3], today MoAbs are used to establish cell subpopulations, which should help in understanding the pathogenic mechanisms of this disease. Earlier, we detected lymphocytosis in arthritic animals accompanied by a decrease in the blood CD4/CD8 T ratio [4]. Moreover, an immunohistochemical study of the arthritic knee synovial membrane showed that only a few T cells were present [5]. Nevertheless, it has been established that AA is T cell-mediated. Thus, the disease can be transferred by T cells [6,7] and treatment with antibodies directed against α/β T cells [8] or CD4⁺ T cells [9,10] ameliorates or inhibits the disease completely. On the other hand, macrophages and their cytokines seem to have a major bearing on arthritis [11]. To determine the involvement of several cell subpopulations in AA and to elucidate the relationship between immunocompetent cells in lymphoid organs and lesions in synovial joints, several MoAbs directed against the main and some minor lymphocyte and macrophage subpopulations are used here for the immunohistological analysis of spleen, popliteal lymph nodes and knee synovial membrane during the time course of this disease. In analysing and quantifying cell populations, image analysis techniques are applied.

MATERIALS and METHODS

Animals

Outbred female Wistar rats (Charles River, Barcelona, Spain), weighing 160–200 g, were housed five per cage with food and water ad libitum. Temperature (20 ± 2°C), relative humidity (55%) and light cycle (12 h on) were constantly monitored.

Induction and assessment of arthritis

Under ether anaesthesia, rats were intradermally injected at the base of the tail with 0·1 ml of a liquid vaseline suspension of heat-killed Mycobacterium butyricum (0·5%) (Difco Labs, Detroit, MI). Arthritis severity was quantified by scoring each paw from 0 to 4 (0 = normal, 4 = maximum) based on levels of swelling and periarticular erythema. The sum of the scores for all four paws was used to calculate the arthritic index.

Synovial damage was quantified by scoring each knee synovial membrane from 1 to 4 (1 = normal, 4 = maximum) depending on the size and aspect of harvested synovial membrane. The sum of the scores for the two knee synovial membranes was used to calculate the synovium index. Both arthritic and synovium indices were established in a blind manner.

Samples

At intervals of 3–4 days over a 3-week period, animals were decapitated under ether anaesthesia. Healthy age- and sex-matched rats were killed on day 0. Spleen, popliteal lymph nodes and both knee synovial membranes were harvested. Tissues from control and arthritic rats were embedded in OCT, frozen in liquid nitrogen and stored at −80°C until cryostat sectioning. Lymphoid tissues were cut at −20°C, and synovia at −30°C. Six micrometre-thick sections were obtained. After fixing in acetone for 10 min at 4°C, sections were stored at −20°C until immunohistochemical assay.

Immunohistochemical assay

The peroxidase–anti-peroxidase (PAP) method was used as described elsewhere [5]. Briefly, frozen sections were rehydrated with PBS pH 7·2. Endogenous peroxidase activity was blocked and, after washing in PBS, the tissue was incubated with rabbit serum to avoid background. Sections were incubated with each primary MoAb (Table 1) and after washing they were also incubated with the secondary antibody (rabbit anti-mouse immunoglobulin; Dako, Glostrup, Denmark; containing 5% rat serum to avoid background). After further washing, slides were incubated with mouse PAP (Dako). Peroxidase activity was demonstrated by adding 3,3′-diaminobenzidine tetrahydrochloride (DAB; Sigma, Madrid, Spain) as a chromogen. The reaction was stopped with PBS. Nuclear staining was performed with Delafield haematoxylin (Merck, Barcelona, Spain). Negative controls were performed both without the primary antibody and with an isotype-matched control antibody. All incubations were performed at room temperature.

Table 1.

Names, specificity and references of the mouse anti-rat MoAbs used

Name	Specificity	References
R73*	TCRα/β+ cells	[12]
W3/25†	CD4 in T-helper cells, some macrophages, some dendritic cells	[13]
OX8†	CD8 in T suppressor/cytotoxic cells, NK cells	[14]
OX6†	MHC class II (RT1B) in B lymphocytes, interdigitating dendritic cells	[15]
3.2.3*	NKR‐P1 in NK cells	[16]
V65*	TCRγ/Δ+ cells	[17]
1A29§	ICAM-1 (CD54) in vascular endothelial cells, stimulated APC	[18]
ED1‡	Monocytes, macrophages, dendritic cells	[19,20]
ED2‡	Resident macrophages	[20]
ED3‡	Lymphoid tissue macrophages	[20]

MoAbs were kindly provided by

^*Dr T. Hünig (Würzburg, Germany)

^†the late Dr A. F. Williams (Oxford, UK)

^‡Dr C. Dijkstra (Amsterdam, The Netherlands)

^§purchased from Seikagaku Corp. (Tokyo, Japan).

Image analysis

Spleen and lymph node sections were studied by means of a computerized image analysis system. After immunohistochemical staining, tissue sections were examined under a microscope (Nikon Apophat, Tokyo, Japan) at a ×500 magnification. The microscope was equipped with a TK-870E camera (JVC, Tokyo, Japan) with one CCD which delivers a 24 bit/pixel signal to the image analysis system (Microm UP, Barcelona, Spain). Images were transferred by means of the RGB (red, green, blue) system. The areas of the tissue sections that were captured were randomly chosen, although areas with debris or insufficient histological quality were excluded interactively.

To perform spleen image analysis, a strategy described elsewhere was applied [21]. Briefly, a central arteriole was considered as a geometric centre from which three equidistant segments were drawn. Each segment began in the vascular epithelium of the central arteriole and ended in the red pulp. White pulp was considered to be the zone that extended from the beginning of the segment up to a point optically defined by the observer. A second zone, considered as being a marginal zone, was determined optically and the remainder of the segment was defined as red pulp. In studying lymph node sections, the image analysis strategy consisted in drawing three parallel and equidistant segments which included cortical and medullar zones. The point of separation between both zones was established by the observer. Results from each zone for every tissue corresponded to the mean of the three segments. Computer software enabled us to calculate, as a percentage, that portion of the entire length of each zone which was positively stained, i.e. longitudinal spread of brown colouring as a proportion of the total length of each zone (percentage of positive staining).

Image analysis was carried out by the same investigator and in a blind manner. Preliminary studies were performed to establish tissue homogeneity: frozen pieces were completely cut and six sections belonging to different parts were analysed for each primary antibody. No statistically significant differences were observed between these sections.

Image analysis was not applied to synovial membrane sections because, following immunohistochemical staining, a highly heterogeneous positive staining was obtained.

Statistical analysis

Arthritic and synovium indices from each post-induction day were compared with basal levels (day 0), before the induction of the disease, and were analysed by means of the Mann–Whitney U-test. Immunohistochemical results obtained by image analyses from arthritic rats were compared with values from healthy animals by anova (Snedecor’s F, α < 0·05).

Results

Articular inflammation

Synovium index (Fig. 1a), which is a measure of macroscopic knee synovial changes, and arthritic index (Fig. 1b), which is a measure of hind paw articular swelling and erythema, were significantly increased from day 11 (P < 0·05).

Fig. 1.

(a) Time course of synovium index, established by grading each knee synovial membrane from 1 to 4 depending on size and aspect. Each point represents the mean value +s.e.m. of five rats. (b) Time course of arthritic index, established by scoring each paw from 0 to 4 based on levels of swelling and periarticular erythema. Each point represents the mean value +s.e.m. of five rats.

Immunohistochemical characterization of synovial membrane

TCRα/β⁺ cells were absent in all healthy tissues and in almost all synovia from arthritic animals. Interestingly, a few tissues obtained on day 14 after AA induction showed some positive cell clusters near the synovial lining (Fig. 2A). No TCRγ/Δ⁺ lymphocytes and no natural killer (NK) cells were observed on healthy sections or following AA induction.

Fig. 2.

Knee synovial membrane sections from arthritic rats stained by the PAP method. (A) TCRα/β⁺ cells, AA day 14. (B) MHC class II⁺ cells, AA day 14. (C) ED1⁺ cells, AA day 14. (D) ICAM-1⁺ cells, AA day 14. Mag. × 200.

By means of the anti-MHC class II MoAb, few positive cells were found in healthy synovial tissue sections, whereas an increase in the number of positive cells occurred in the synovial lining of all arthritic animals after day 14 post-induction until the end of the study (Fig. 2B). Some ED2⁺ macrophages were also seen in all healthy synovial membranes distributed in lining and subintimal areas. A higher proportion of these cells was seen after AA induction (data not shown).

Healthy synovial membranes did not show any ED1⁺ or intercellular adhesion molecule-1 (ICAM-1)⁺ cells. After AA induction, however, some clusters of ED1⁺ (Fig. 2C) or ICAM-1⁺ (Fig. 2D) cells were observed in those sections that had also shown a positive staining for the anti-TCRα/β MoAb. Moreover, consecutive sections showed that ED1⁺ and ICAM-1⁺ cells were in the same location as clusters of TCRα/β⁺ lymphocytes.

Immunohistochemical characterization of popliteal lymph nodes

Figure 3 shows the time course of positive staining obtained after applying image analysis to sections labelled with anti-TCRα/β, anti-CD4, anti-CD8 and anti-MHC class II MoAb. For the four antibodies, the time course of the disease affected the positive staining both in lymph node cortex and in medulla (P < 0·01).

Fig. 3.

Time course of percentage of positive staining from different MoAb staining in cortical (•) and medullar zones (○) from popliteal lymph node sections. (a) Anti-TCRα/β. (b) Anti-CD4. (c) Anti-CD8. (d) Anti-MHC class II MoAb. Results were obtained applying the image analysis strategy for lymph nodes described in Materials and Methods. Each point represents the mean value +s.e.m. of five rats.

TCRα/β⁺ lymphocytes in healthy tissues were mainly found in the cortex without occupying the germinal centres (Fig. 4A). After AA induction, significantly higher percentages of positive staining of TCRα/β in arthritic rats than in healthy ones were only obtained on day 14 in cortex (P < 0·05) and on days 14 (P < 0·05) and 21 (P < 0·001) post-induction in medulla (Fig. 3a). CD4⁺ staining showed significantly lower values in lymph node cortex from day 11 onwards than in healthy animals on day 0 (P < 0·001). Values of CD4⁺ staining in medulla were significantly higher on days 14 and 18 post-induction than those in healthy rats (P < 0·001) (Fig. 3b). A significant decrease in CD8⁺ cells was observed in cortex (P < 0·01) on days 7 and 11 post-induction. In medulla however positive staining stood at around 10% throughout the period studied (Fig. 3c).

Fig. 4.

Popliteal lymph node sections from rats stained by the PAP method. (A) TCRα/β⁺ cells, healthy. (B) MHC class II⁺ cells, AA day 14. (C) NKR-P1⁺ cells, AA day 4. (D) ED1⁺ cells, AA day 14. (E) ED3⁺ cells, AA day 11. (F) ICAM-1⁺ cells, AA day 14. Mag. × 200 (A–D,F); × 400 (E).

The staining with anti-MHC class II MoAb revealed that, after AA induction, positive cells were more frequent in both the cortical and medullar regions than in healthy lymph nodes, where hardly any positive cell was found. Increases in both cortex and medulla were already statistically significant on day 4 post-induction and remained so in the cortex until the end of the study (P < 0·001) and in the medulla until day 18 (P < 0·01) (Fig. 3d). A representative section stained with anti-MHC class II MoAb from a rat killed on day 14 post-induction is shown in Fig. 4B.

T cells bearing γ/Δ receptor were not found in any healthy or arthritic lymph nodes (data not shown). On the other hand, NK cells in healthy tissue were faintly scattered in the cortex, with the exception of germinal centres. After AA induction, an increase in cortical positive staining was observed in some tissues from days 4, 18 and 21. Figure 4C shows a lymph node section from day 4 post-induction stained with the anti-NKR-P1 MoAb.

As for macrophages, ED1⁺ cells were seen in the cortex and principally in the medulla of healthy lymph nodes. After AA induction and with the progression of the disease, the ED1⁺ population increased in both compartments. Figure 4D shows a representative section from day 14 of AA stained with the ED1 MoAb. ED3⁺ macrophages from healthy tissues were mainly found in the medullar area and some positive cells were also seen in the cortex and in the subcapsular zone surrounding the cortex. When AA was induced, ED3⁺ cells increased in the subcapsular zone from day 11 (Fig. 4E) and continued increasing until the end of the study.

By means of the anti-ICAM-1 MoAb, positive staining was observed in healthy lymph nodes all along venule sinus and cortex, with the exception of germinal centres. After AA induction, positive staining increased progressively from day 4 until day 14 and remained until the end of the study. Figure 4F shows cortical ICAM-1⁺ staining in a tissue from day 14 post-induction.

Immunohistochemical characterization of spleen

Results obtained after staining with anti-TCRα/β, anti-CD4, anti-CD8, anti-MHC class II, ED3 and anti-ICAM-1 MoAbs are summarized in Fig. 5. The time course of AA affected the percentages of positive staining of all the MoAbs analysed in the three zones studied (P < 0·05) with the exception of white pulp stained with ED3 MoAb. For all antibodies but ED3 the percentages of positive staining were higher in white pulp than in marginal zone and red pulp.

Fig. 5.

Time course of percentage of positive staining from different MoAb staining in white pulp (○), marginal zone (•) and red pulp (Δ) from spleen sections. (a) Anti-TCRα/β. (b) Anti-CD4. (c) Anti-CD8. (d) Anti-MHC class II. (e) ED3. (f) Anti-ICAM-1. Results were obtained applying the image analysis strategy for spleen described in Materials and Methods. Each point represents the mean value +s.e.m. of five rats.

The TCRα/β⁺ lymphocyte proportion was maintained in the three splenic zones with the exception of a peak on day 4 post-induction (Fig. 5a). On the other hand, lymphocytes bearing TCRγ/Δ were not found in any healthy or arthritic spleen (data not shown). CD4⁺ cells were more abundant in each zone of arthritic sections than in those of healthy sections, from day 7 to the last day studied (P < 0·001) (Fig. 5b). A representative section from day 14 is shown in Fig. 6A. Otherwise, the proportion of CD8⁺ cells began to increase in the three splenic zones with the establishment of AA, showing significantly higher percentages from day 14 until the end of the period studied (P < 0·001) (Fig. 5c). A representative section from day 14 post-induction stained with anti-CD8 is shown in Fig. 6B.

Fig. 6.

Spleen sections from arthritic rats stained by the PAP method. (A) CD4⁺ cells, AA day 14. (B) CD8⁺ cells, AA day 14. (C) MHC class II⁺ cells, AA day 14. (D) ED3⁺ cells, AA day 11. (E) ICAM-1⁺ cells, AA day 14. (F) NKR-P1⁺ cells, AA day 14. Mag. × 200.

Results obtained with anti-MHC class II MoAb showed that positive staining increased on days 11–14 (P < 0·01 on day 11) (Fig. 5d). On days 18 and 21, however, these values dropped significantly below healthy values (P < 0·01). An increase in positive staining was observed in the marginal zone on day 14 (P < 0·05). Later, positive staining in the marginal zone and red pulp decreased (P < 0·05 on day 18). A tissue stained with anti-MHC class II MoAb is shown in Fig. 6C.

When statistical analysis (anova) was applied to ED3 staining results, the test revealed that AA time course affected positive staining in the marginal zone and red pulp (P < 0·05), but it had no effect on white pulp positive staining, where values were low (< 6%) throughout the period studied (Fig. 5e). In the marginal zone a higher presence of positive cells than that of white pulp and red pulp was observed throughout the period studied. On day 11, i.e. when inflammatory signs were first identifiable, significantly higher values were obtained with respect to healthy values on day 0 (P < 0·05) (Figs 5e and 6D). In red pulp, the presence of cells stained with the ED3 MoAb increased significantly on days 11–21 with respect to healthy values on day 0 (P < 0·05) (Fig. 5e).

Figure 5f shows the results obtained after analysing ICAM-1 expression. A significant decrease in positive staining was observed on day 4 in the three zones studied (P < 0·05). However, on days 11 and 14 increased values were found in both white pulp and the marginal zone (P < 0·05) and on day 14 in the marginal zone (P < 0·001). Figure 6E shows the ICAM-1 distribution in a splenic section from day 11 post-induction.

NK cells were scarce in spleen and therefore image analysis was not applied to all the sections. Percentage of positive staining in healthy rats was about 0·23 ± 0·14% (mean ±s.e.m.) in white pulp, 8·21 ± 5·12% in the marginal zone and 3·28 ± 1·3% on red pulp. After AA induction, a significant increase in NKR-P1⁺ cells in red pulp on days 14 (12·02 ± 2·08%) and 18 (6·96 ± 1·87%) (P < 0·05) was recorded, while percentages in white pulp and the marginal zone remained invariable. Figure 6F shows a representative section of NKR-P1⁺ cells in a spleen from day 14 post-induction.

Discussion

Rat lymphoid tissues and knee synovial membrane were immunohistochemically analysed during the development and establishment of AA in order to determine the involvement of several cell subpopulations in this pathology and to elucidate the relationship between lymphoid organ cells and articular lesions. Cell populations in lymphoid tissue sections were objectively quantified by means of image analysis.

Alteration of the knee synovial membrane and the outer inflammation were first detected on day 11, but T lymphocyte infiltration was only detected in a few tissues from day 14. The lack of significant lymphocyte infiltration agrees with earlier results obtained in established AA [5] and with those of Meacock et al. [22] when studying the ankle joint in the CP20961-induced arthritis. The T cell clusters found in these few tissues were always accompanied by ED1⁺ macrophage infiltration and by a positive expression of ICAM-1 in the same zone of the section. These results corroborate the role of adhesion molecules in inflammatory cell infiltration, and agree with findings that indicate that an ICAM-1-dependent pathway is involved in the pathogenesis of arthritis [23,24]. On the other hand, synovial membrane from arthritic animals presented a high number of both ED2⁺ macrophages and cells expressing MHC class II antigen. These increases are consistent with those reported by Dijkstra et al. [19], and corroborate the essential role of synovial macrophages in arthritis development [11,25]. However, T CD4⁺ cells are also essential in AA development, as evidenced by transference studies [6,7] or by treatment with anti-TCRα/β[8] or anti-CD4 MoAb [9,10]. In order to study changes in T cell subpopulations in lymphoid tissues after AA induction, popliteal lymph nodes and spleen were immunohistochemically analysed.

An enlargement of popliteal lymph nodes was observed from day 4 post-induction, but a significant increase in T α/β cell percentages by image analysis was not detected until the establishment of inflammation. Before any evidence of articular inflammation, lymph nodes showed a higher frequency of cortical and medullar cells bearing MHC class II molecules in arthritic rats than in healthy tissues. Cortical MHC class II⁺ cells may be B lymphocytes, dendritic cells or activated T lymphocytes. It is known that after an antigenic priming, activated B cells proliferate and remain in lymph nodes for a long period [26], while activated T cells proliferate and leave lymph nodes towards peripheral sites where most of their activity occurs [27]. This would explain why a great increase in the lymph nodes of T cell proportions was not seen. However, we did not find a large number of these T cells in knee synovial membrane.

After AA induction, CD4⁺ and CD8⁺ cell numbers significantly increased in each splenic compartment. These results agree with findings reported by Sugawara et al. [28], who described an increase in spleen CD4⁺ lymphocyte percentage by FACS. The increase in spleen T cells might be the result of in situ lymphocyte proliferation or T cell migration from lymph nodes. The high expression of ICAM-1 found in the present study in the splenic marginal zone and red pulp in arthritic rats, and the fact that T lymphocytes seem to reach the spleen mainly in the marginal zone and also in the red pulp [29], suggest that splenic CD4⁺ T cell increase is due, at least in part, to lymphocyte migration. On the other hand, the high ICAM-1 expression in lymph nodes from arthritic animals might reflect a rise in the expression of this molecule in high endothelial venules of the inner cortex that allow lymphocyte migration. All these results suggest the migration of lymphocytes between lymph nodes and spleen and indicate that few of these cells reach the affected knee synovial membrane. This suggestion agrees with a study about AA transference to naive recipients in which transferred T cells were not found in synovia, but in lymph nodes, spleen and liver [30].

ED3⁺ cells form a particular subpopulation of macrophages consisting of marginal zone macrophages and marginal metallophils [20]. Metallophilic macrophages are good candidates for transferring of antigens to B cells in the inner splenic white pulp, and they are believed to be primarily involved with the processing and presentation of particulate antigens [31]. The fact that we found an increase in ED3⁺ cells in spleen and lymph nodes suggests that these macrophages might be involved in initiating cell activation and proliferation.

In the synovial tissue from patients with early rheumatoid arthritis, granzyme-positive cytotoxic cells, mainly NK cells, were found to have increased and their presence correlates with disease activity [32,33]. In the present study we found a larger amount of NK cells in the cortex of lymph nodes and in the red pulp of spleen of AA rats than in those of healthy tissues. Therefore, although no NK cells were found at the local site of inflammation, it cannot be ruled out that these cells play a role in AA.

TCRγ/Δ⁺ lymphocytes preferentially recognize mycobacterial antigens both in humans [34] and mice [35], and it has been reported that patients with rheumatoid arthritis have high numbers of TCRγ/Δ⁺ cells in peripheral blood and synovial fluid [36], although this is controversial [37]. On the other hand, TCRγ/Δ⁺ cells do not seem to be essential in promoting or perpetuating AA [38]. In the present study, the absence of TCRγ/Δ⁺ cells confirms the non-important role of this minority subpopulation in developing AA.

In conclusion, our results suggest that AA induction promotes the mobilization of macrophages and lymphocytes both from lymph nodes and spleen. Consequently, TCRα/β⁺ lymphocytes and B cells, some NK but not TCRγ/Δ⁺ lymphocytes, proliferated. B cells mainly remain in lymph nodes and T cells move from lymph nodes to blood and spleen and some of them reach synovia. Therefore, synovitis would be produced directly by macrophages and indirectly by lymphocytes placed both in lymph nodes and spleen.