The nonobese diabetic (NOD) mouse is considered to be an animal model suitable for studying the genetic and pathologic features of human autoimmune insulin-dependent (type I) diabetes mellitus. In the initial phase of the disease, antigen-presenting cells (APCs), such as dendritic cells (DCs) and macrophages, infiltrate periductal and perivascular spaces adjacent to the islets of Langerhans (Rosmalen et al., 2000; Shinomiya et al., 2000). These cells capture self-antigens, migrate via afferent lymphatics into T-cell zones of lymph nodes and provide stimulation to naive T-lymphocytes originating from the thymus, thus eliciting continuous generation of memory T-lymphocytes against autoantigens (Savino et al., 1991; Yoon et al., 1999). Activated or self-reactive T-lymphocytes have been implicated as effector cells for progressive destruction of insulin-producing β-cells. The migratory property of lymphocytes and DCs is regulated by various adhesion molecules and chemokines expressed on lymphatic and blood vessels (Newman, 1997; Martin-Fontecha et al., 2003). Recently, we have revealed the expressing features of 5′-nucleotidase (5′-Nase) and chemokine CCL21 on pancreatic lymphatics and their relationships with the dynamic migration of immune cells (DCs and lymphocytes) via lymphatic walls during the pathological processes of autoimmune insulitis in wild-type NOD mice (Qu et al., 2003). These observations proposed that lymphatic vessels not only contributed to the removal of excess fluid, but also played a pathogenic role in maintaining and possibly worsening the diabetes.
The diabetic therapeutics has been concentrated on the suppression of autoimmune responses in the pathological process especially with immunoadjuvants. Complete Freund's adjuvant (CFA), one of the most commonly used immunomodulators, has well-documented therapeutic effects on several autoimmune diseases in animals and humans, such as experimental autoimmune encephalomyelitis (Kahn et al., 2001), pristane-induced arthritis (Zheng et al., 2002), and psoriasis (Balagon et al., 2000). Administration of CFA between the age of 4 and 10 weeks prevented the diabetes and reduced the insulitis in NOD mice (Sadelain et al., 1990; Mcinerney et al., 1991; Lee et al., 2004). However, it remains unknown how CFA treatment affects the infiltration by different subsets of mononuclear cells and what accounts for CFA-induced reduction in the insulitis. We previously examined the structural organization of pancreatic lymphatics in NOD mice with special reference to histochemical characteristics of the endothelial cells (Qu et al., 2003). In the present study, we thus analyzed CD11c-, CD4-, and F4/80-positive infiltrating cells in the islets and compared the expression of adhesion molecule CD31 and chemokine CCL21 on the pancreatic lymphatics and infiltrating cells among 7- to 30-week-old CFA-treated or untreated NOD mice. Furthermore, substantial information as to the relationships among pancreatic lymphatics, infiltrating cells, and insulitic development was provided to constrain how LECs can actively participate in pathological processes of the diabetes.
MATERIALS AND METHODS
Female NOD/shi jic and BALB/c mice, purchased from Clea Japan (Tokyo, Japan), were maintained in the Division of Laboratory Animal Science, Institute of Scientific Research, Oita University.
Administration of CFA
Five-week-old NOD mice were injected into the pad of hind foot with 0.05 ml complete Freund's adjuvant (CFA, containing mycobacterium strain H37 Ra; Sigma, St. Louis, MO) subcutaneously. After the mice were deeply anesthetized with ether, the pancreases were removed respectively at 7, 10, 13, 17, and 30 weeks of age with approval of the Animal Research Ethics Committee in Oita University Faculty of Medicine. Each group consisted of 10 mice. CFA-untreated NOD and CFA-treated BALB/c mice served as control groups.
Serological Analysis of Diabetes
Diabetes was assessed by hyperglycosuria and persistent hyperglycemia (> 250 mg/dl) measured with an ANTSENSE II monitor (Daikin, Osaka, Japan). Percentage of diabetes was calculated as 100% × the number of diabetic NOD mice × (the initial number of experimental NOD mice)−1.
For light microscopy, samples were embedded in OCT compound and frozen until use. Serial 6–8 μm cryosections were cut, air-dried, and fixed by 4% paraformaldehyde in cacodylate buffer (pH 7.2) for 10 min. 5′-Nase activity for identifying lymphatics was conducted with 5′-Nase-lead medium (Kato et al., 1997). The sections were immersed with 1% ammonium sulfide for 1–2 min at room temperature (RT). To distinguish initial lymphatics from blood capillaries, the specimens after 5′-Nase treatment were incubated for 20–25 min at 4°C with ALPase reaction medium (Kato, 2000; Ji and Kato, 2001).
For transmission electron microscopy (TEM), pancreatic tissues were fixed in 4% paraformaldehyde for 30 min at 4°C and chopped into small pieces. The specimens were allowed to react with 5′-Nase-cerium medium for 40 min at 37°C (Kato et al., 1997). After postfixed in 2% osmium for 2 hr at 4°C, the blocks were dehydrated and embedded in Epok 812. Ultrathin sections (90–95 nm) were cut and stained with uranyl acetate and lead citrate, then examined under a JEM-1200 EX II electron microscope (Jeol, Tokyo, Japan). Control experiments for 5′-Nase staining were undertaken, respectively by inactivating the enzyme for 60 min at 60°C before samples were incubated or by omitting the substrate or adding 5′-Nase standard medium with 50 mM NiCl2 or 5 mM L-tetramisole.
Serial 6–8 μm cryosections were air-dried and fixed in acetone, then washed in 0.1 M phosphate-buffered saline (PBS; pH 7.4). After the sections were treated with 0.3% H2O2 in PBS for 15 min, they were incubated with 10% blocking serum for 20 min and respectively with rat antimouse CD31 (1:300; platelet endothelial cell adhesion molecule-1, PECAM-1; Cymbus Biotechnology, Hants, U.K.), goat antimouse CCL21 (1:100; R&D Systems, Minneapolis, MN), hamster antimouse N418 for CD11c (1:200; Endogen, MA), rat antimouse CD4 (1:200; Southern Biotechnology Associates, AL), rat antimouse F4/80 (1:400; Cosmo Bio, Tokyo, Japan) for 60 min at RT or overnight at 4°C. After rinsing in PBS, the sections were reacted respectively with biotinylated goat antihamster IgG for CD11c, rabbit antigoat IgG for CCL21, or goat antirat IgG for CD4, F4/80, and CD31 at a 1:200 dilution for 60 min at RT. The sections were subjected to reagents of streptavidin-biotinylated peroxidase complexes (SBPCs) for 60 min. Immunoreactivity was visualized with a solution of 0.02% 3,3′-diaminobenzidine tetrahydrochloride (DAB) in 0.05 M Tris-HCl (Sigma) buffer (pH 7.4) for 2–4 min. The sections were counterstained with Mayer's hematoxylin and examined by a light microscope (Olympus, Tokyo, Japan).
For preembedding immunoelectron microscopy (Ji and Kato, 2001), pancreatic tissues were fixed in 4% paraformaldehyde for 4 hr at 4°C. The samples were mounted in OCT compound and rapidly frozen in dry ice and acetone; 15 μm consecutive cryosections were cut and processed for the identification of CCL21 and CD31 activities by means of the above-mentioned indirect immunoperoxidase technique. After SBPC treatment, the sections were fixed in 1% glutaraldehyde for 10 min at 4°C. The samples were further incubated with DAB solution for 5–10 min, postfixed with 2% OsO4 for 1 hr before dehydration through a graded ethanol series. Finally, the sections embedded in Epok 812 were cut into ultrathin sections and observed without additional staining. In all cases, negative controls for immunohistochemical staining were provided by replacement of primary antibodies with PBS or nonimmune mouse serum, or omitting DAB.
By using a postembedding immunogold technique (Ji et al., 2003), fixed pancreas was dehydrated in graded ethanol series at 4°C. They were immersed with a 1:1 LR White:100% ethanol mixture for 1 hr before being transferred to 100% LR White for 2 hr. The blocks were then embedded in LR White (London Resin, Basingstoke, U.K.) and polymerized at 60°C for 24 hr. Ultrathin sections were cut and treated with 10% blocking serum for 15 min and incubated with goat antimouse CCL21 (1:100) at 4°C overnight. The sections were further incubated with rabbit antigoat IgG-coated 5 nm colloidal gold (1:80) for 1 hr at RT. After silver enhancement was performed in some samples, the sections were counterstained for 2–5 min with diluted aqueous uranyl acetate and lead citrate and examined under an electron microscope.
Quantification of Immunocytochemistry
The staining with CCL21 and CD31 was identified in a blind manner by counting lymphatic and blood vessels around and within infiltrated areas or islets, as in our previous study (Qu et al., 2003). The staining was semiquantitatively scored as follows: −, negative; +, weak; ++, moderate; and +++, strong.
The number of CD11c-positive DCs, CD4-positive T-lymphocytes, F4/80-positive macrophages, CCL21- and CD31-positive cells in each group was quantified by direct counts of cells per islet on immunoperoxidase-labeled cryosections. Densely labeled cells were analyzed using a grid micrometer. The number of grid intersections overlying the islets and positive cells was counted. Counts were made from at least 60 islets (at a magnification of 100) from six to eight random NOD mice in each group. An islet is defined as infiltration with cells displaying cytoplasmic staining with CCL21 or membranous staining with CD11c, CD4, F4/80, and CD31, respectively, and there is at least one cell around or within the islet. Values were presented as the mean number of cells per islet (mean ± SD) at various stages of insulitis and counts were made blindly.
The comparison among groups was analyzed with a factorial analysis of variance. Differences between CFA-treated and untreated groups were evaluated using the Student's t-test. A P value of less than 0.05 was considered to be statistically significant.
Analysis of Blood Glucose Values
The incidence of diabetes increased with age in CFA-untreated NOD mice, reaching 70–80% in 30-week-old. In CFA-treated groups, NOD mice did not develop diabetes, and blood glucose values were in normal condition as those of BALB/c mice, but obviously lower compared with those in age-matched untreated groups. The differences of blood glucose values among treated groups were not statistically significant (Fig. 1).
Kinetics of Lymphatics and Infiltrating Cells
The expression of CCL21 and CD31 on pancreatic lymphatics reduced distinctly in the treated groups compared with that in age-matched untreated groups, although their activities were upregulated with age in the untreated groups (Table 1). The tendency of decreased number of DCs, T-lymphocytes, CCL21- or CD31-positive cells in the islets was observed, but macrophage infiltration did not alter in the treated groups (Fig. 2, Table 2).
Table 1. CCL21 and CD31 expression on pancreatic lymphatics in CFA-untreated and -treated NOD mice*
Table 2. Kinetics of CCL21-, CD31-, CD11c-, F4/80-, and CD4-positive cells in the islets of Langerhans between CFA-untreated and -treated NOD mice*
Mean number of cells per islet ± SD.
5.13 ± 0.89
5.17 ± 0.54
3.10 ± 0.42
3.20 ± 0.62
0.90 ± 0.39
2.00 ± 0.76
2.77 ± 0.46
1.17 ± 0.34
2.47 ± 1.56
0.63 ± 0.37
9.83 ± 1.68
9.83 ± 0.98
13.53 ± 1.96
4.27 ± 0.78
16.20 ± 3.28
2.69 ± 0.67
3.20 ± 0.61
4.10 ± 0.72
4.40 ± 0.95
1.57 ± 0.90
21.37 ± 5.06
15.77 ± 1.46
25.96 ± 3.21
5.80 ± 1.04
28.93 ± 3.31
7.03 ± 1.85
5.83 ± 0.97
5.53 ± 1.86
5.40 ± 1.48
2.03 ± 0.84
32.20 ± 3.91
23.73 ± 2.24
59.50 ± 6.12
8.00 ± 2.21
48.63 ± 5.54
5.43 ± 1.10
8.07 ± 0.99
13.73 ± 2.78
7.33 ± 2.49
7.63 ± 1.39
7.27 ± 1.33
8.50 ± 1.26
13.00 ± 1.81
8.20 ± 2.12
5.50 ± 1.69
Distribution of Lymphatics, T-Lymphocytes, and APCs
In 7-week-old NOD mice, no significant difference between CFA-treated and untreated groups was shown in the number of macrophages and T-lymphocytes (Fig. 2, Table 2). CCL21 immunoreactivity was weak in 5′-Nase-positive pancreatic lymphatics and surrounding matrix in the untreated group, but was not detectable in most cell types of the age-matched treated group (Table 1). In the 10-week-old untreated group, CCL21 expression was more intense in the infiltrated islets and surrounding matrix, and in 5′-Nase-positive interlobular lymphatics adjacent to infiltrated islet than in ALPase-positive blood vessels (Fig. 3a and b). Strong CD31 staining (Fig. 3c) was visualized in lymphatic and blood vessels. CD11c-positive DCs and CD4-positive T-lymphocytes were present in peri- and intrainsular areas (Fig. 2, Table 2). CCL21 expression was weaker in lymphatic endothelium with 5′-Nase activity and in surrounding matrix in the 10-week-old treated group than in the untreated group (Fig. 3d and e), but did not show obvious differences in comparison with that in the 13-week-old treated group. Weak CD31 staining was identified in lymphatic endothelial walls and infiltrated islets (Fig. 3f). The number of DCs and T-lymphocytes decreased in the perivascular and parainsular areas in 10-week-old treated NOD pancreas (Fig. 2, Table 2). As the diabetes developed, high density of CCL21 was shown on the infiltrated islet and intralobular lymphatics in the 17-week-old untreated group (Fig. 4a). CD4-positive T-lymphocytes were predominantly found in most peri- and intrainsular areas, surrounding intralobular lymphatics with 5′-Nase activity (Fig. 4b and c). Numerous CD11c-positive DCs accumulated in peri- and intrainsular areas (Fig. 4d), but a small number of macrophages were located on peri-insular areas (Fig. 4e). In age-matched treated groups, the infiltrated islet and lymphatic vessel displayed weak expression of CCL21 (Fig. 4f). The islet adjacent to 5′-Nase-positive lymphatic vessel (Fig. 4g) was infiltrated by a few T-lymphocytes (Fig. 4h) and DCs; the latter were at the border of T-cell infiltration (Fig. 4i). CD31 immunoreactivity was weak on the lymphatic vessel (Fig. 4j).
Immunocytochemical Localization of CCL21 and CD31 Expressing Lymphatics
A postembedding immunogold technique displayed CCL21 reaction product on the luminal and abluminal surfaces and perinuclear areas of LECs in CFA-untreated groups (Fig. 5a). CCL21 conjugated with gold particles produced weak label mainly on the luminal surface and cytoplasm of LECs in CFA-treated groups (Fig. 5b). A preembedding technique revealed that the electron-dense product of CD31 was deposited on endothelial cell surfaces and cytoplasm (Fig. 5c), and sometimes on intercellular junctions of LECs in untreated groups (Fig. 5d). CD31 product was distinctly weaker on the abluminal than on luminal surfaces of LECs in treated groups (Fig. 5e). Immunoreactivities of CCL21 and CD31 on pancreatic lymphatics were relatively higher in the untreated than in age-matched treated groups.
Ultrastructural Features of Pancreatic Lymphatics
In CFA-untreated groups, many vesicles and cytoplasmic processes were found in LECs with weak 5′-Nase activity (Fig. 6a), and lymphatic endothelial wall was also frequently interrupted by intercellular junctions, to which the dendritic cell was close (Fig. 6b). In CFA-treated groups (Fig. 6c), few cytoplasmic processes were observed in the lymphatics. 5′-Nase reaction product was evenly distributed on the surfaces of lymphatic vessels, but not in blood vessels. In contrast to untreated groups, open junctions were seldom detected in lymphatic vessels, around which few immune cells were located (Fig. 6d). As insulitis progressed, many lymphocytes and DCs recruited around lymphatic vessels, on which 5′-Nase-cerium granules were not discernible. It was interesting that DCs and lymphocytes migrated across the intercellular junction simultaneously at the age of 17 weeks in the untreated group (Fig. 6e). The cytoplasmic process of dendritic cells was extending into the intercellular junction where LECs were closely connected with surrounding tissues by fine elastic fibers (Fig. 6f). T-lymphocyte was also transmigrating through the intercellular junction, probably from matrix into the lumen of lymphatic vessels, in which overlapping junctions were found (Fig. 6g). In all treated groups, the migration of immune cells was not detected, and occasionally a few immune cells were present around the lymphatics.
Inhibitory Effects of CFA on Autoimmune Response
Complete Freund's adjuvant, as an immunomodulator, appeared to influence the maintenance of autoimmune response. Two factors were involved in the autoimmune process, the sufficient autoantigens captured by APCs and the islet-associated lymphoid-like tissues in the pancreas (Ludewig et al., 1998; Yoon et al., 1999).
In the present study, we demonstrated that the insulitic development was significantly inhibited in CFA-treated NOD mice. In inflammation tissues, lymphatic vessels produced the homing chemokine CCL21, whose CC chemokine receptor (CCR7) was expressed on activated DCs and some lymphocytes (Ploix et al., 2001; Martin-Fontecha et al., 2003). The increased expression of CCR7 and CCL21 promoted the molecular interaction between DCs and LECs, directing the migration of DCs to lymph nodes through lymphatic vessels. Chronic overexpression of CCL21 in pancreatic lymphatics and islets in CFA-untreated NOD mice suggested that they should recruit DCs and lymphocytes to support the development of chronic inflammation tissues that increased the efficiency of islet antigen-specific T-cell responses (Qu et al., 2003). After CFA treatment, pancreatic lymphatics expressed weak immunoreactivity for CCL21 or not. Chemotactic affinity between lymphatics and infiltrating cells became decreased, inhibiting activated DC entry into lymphatic vessels with a chemotactic gradient. Simultaneously, CD31 (PECAM-1) was expressed not only on DC subsets and T-lymphocytes, but also on the intercellular junctions and surfaces of LECs, indicating a homophilic CD31 interaction between DCs and LECs. Some experiments also showed that CD31 participated in the process of DC and lymphocyte through the endothelial cell lining after antigenic stimulation (Bogen et al., 1992; Newman, 1997). In contrast to adhesion molecule CD45/ICAM-1, CD31 supported the migration of activated T-cells in the absence of chemokines (Zocchi et al., 1996). CD31/PECAM-1 adhesion may play an important role in the chemokine-independent transmigration of leukocytes through the lymphatic vessels. Following CFA treatment, reduced expression of CD31 on lymphatic vessels may affect the migration of DCs. The hypothesis was supported by the fact that anti-CD31 mAbs could not block the migration of leukocytes toward intercellular junctions, but block transmigration of these leukocytes through vessels (Muller et al., 1993). CFA treatment suppressed the infiltration of T-lymphocytes to the islet by influencing migratory processes of DCs in the insulitis lesion into lymphatics. Recent studies have shown that CFA immunization stimulated NK cells to prevent the diabetes in NOD mice (Lee et al., 2004).
Alternatively, DCs and lymphocytes accumulated gradually in perivascular areas and in peri- and parainsular areas during the development of diabetes. Some mature DCs in these areas expressed CCL21 to attract more activated CCR7-positive DCs, resulting in the arrest of DCs, which did not leave the pancreas. Instead, DCs were trapped in these areas and initiated an aberrant T-cell response to maintain the chronic inflammation for several weeks without tissue destruction before the onset of diabetes (Martin-Fontecha et al., 2003). The chronic inflammation had characteristics of lymphoid neogenesis. The pancreatic expression of CCL21 was sufficient to cause the progression toward autoimmunity (Fan et al., 2000; Ploix et al., 2001). The overexpression of chemokine CCL21 and adhesion molecule CD31 on pancreatic tissues definitely participated in the establishment of insulitic lesion in NOD mice, although the roles of other chemokines and adhesion molecules could not be ruled out. In CFA-treated groups, infiltrating cells in the absence of immunoreactivities of CCL21 and CD31 might be no longer able to attract a large number of DCs and T-lymphocytes to the peri- and parainsular areas. Accumulation of effector cells was thus inhibited in the pancreas. Moreover, no obvious changes in the number of macrophages were detected in the experiment, which suggested that the cell surfaces might lack the expression of some molecules (such as CCL21 and CD31). In fact, CFA treatment appeared to reduce rather than completely prevent the expression of CCL21 and CD31 on the pancreatic tissues (LECs and infiltrating cells). Under the circumstances, peri-insulitis may occur without intraislet infiltration and newly formed lymphoid-like tissues adjacent to the islets. Therefore, the present investigation provided the first proof that CFA treatment promoted the functional modulation of pancreatic lymphatics and infiltrating cells in the autoimmune pathological processes, thus inhibiting the insulitic development.
Effects of CFA on Lymphatic Structures
After CFA treatment, the transendothelial migration of DCs and T-lymphocytes through pancreatic lymphatics might not be facilitated by the lack of CCL21 and CD31 on LECs and altered lymphatic structures; the latter were characterized by the decreased number of cytoplasmic protrusions and vesicles. Open junctions were frequently present on the initial lymphatics in the CFA-untreated NOD pancreas, but seldom in those of BALB/c control and CFA-treated NOD groups. Open junction and endothelial cytoplasm played important roles in transmigration of DCs and T-lymphocytes through lymphatic vessels (Qu et al., 2003). Obviously, the lack of open junctions in the treated groups reduced the transport of these cells significantly. Pinocytotic vesicles, as the key transporter of molecules and tissue fluids, might predominate in lymph formation (O'Morchoe et al., 1997), although it needs to be further confirmed whether the number of vesicles in the treated groups can return to normal state as those in BALB/c control mice. With lymphatic modification, LECs showed a relatively high 5′-Nase activity after CFA treatment. For 5′-Nase, enzyme histochemical staining has been widely used to investigate organ specificity and structural organization of lymphatic capillary networks in several mammals during the pathophysiological processes (Kato, 2000; Ji and Kato, 2001, 2003). 5′-Nase activity and functional capability indicated that altered endothelial cell components directly induced the expression on cell membranes to influence permeability modifications and absorptive capacity of lymphatics (Ji et al., 2003), even though it remains to be determined whether 5′-Nase activity on lymphatics reflect the diabetic progression.
Different Regulatory Mechanisms of CFA
CFA, which is composed of mineral oil and killed mycobacteria, induces a polarized Th1 response and activates the innate immune system. Furthermore, heat-shock protein (HSP)-specific regulatory T-cells induced by mycobacteria in CFA might contribute to the suppressive effect of CFA on diabetes (Kaufmann et al., 1987; Qin et al., 1993). When NOD mice were treated with CFA at an early age, the mycobacterial antigens in CFA might be presented to T-cells by different APCs (Elias et al., 1990). It was possible that the immune response to the antigens led to some decreases in the availability of the autoantigens, which were captured and presented by DCs in NOD mice. Interestingly, CFA might allow induction of clonal anergy in the effector cells that cause β-cell destruction (Ulaeto et al., 1992). In the state of T-cell dormancy, the islets were no longer subject to an immune attack.
In the present study, the changes in ultrastructures of lymphatics and in the expression of CD31 and CCL21 on pancreatic lymphatics were observed, which may signal a shift of infiltration of DCs and T-lymphocytes to the islets. The relationship between pancreatic lymphatics, infiltrating cells, and insulitis indicates the novel role of pancreatic lymphatics during the insulitic development of nonobese diabetes.
The authors thank Mr. T. Kajiwara and Mrs. M. Maki (Department of Anatomy, Biology and Medicine) as well as Mr. H. Kawazato and Ms. A. Yasuda (Division of Biomolecular Medicine and Medical Imaging, Department of Life Science, Institute of Scientific Research) for their assistance.