Histochemical analysis of lymphatic endothelial cells in the pancreas of non-obese diabetic mice


  • P. Qu,

    1. Division of Morphological Analysis, Department of Anatomy, Biology and Medicine, Oita Medical University, Japan
    Search for more papers by this author
  • R. C. Ji,

    Corresponding author
    1. Division of Morphological Analysis, Department of Anatomy, Biology and Medicine, Oita Medical University, Japan

      Dr R. C. Ji, Division of Morphological Analysis, Department of Anatomy, Biology and Medicine, Oita Medical University, 1–1 Idaigaoka, Hasama-machi, Oita-gun, Oita 879–5593, Japan. Tel.: 08197 5865623; fax: 08197 5865623; e-mail: JI@oita-med.ac.jp
    Search for more papers by this author
  • S. Kato

    1. Division of Morphological Analysis, Department of Anatomy, Biology and Medicine, Oita Medical University, Japan
    Search for more papers by this author

Dr R. C. Ji, Division of Morphological Analysis, Department of Anatomy, Biology and Medicine, Oita Medical University, 1–1 Idaigaoka, Hasama-machi, Oita-gun, Oita 879–5593, Japan. Tel.: 08197 5865623; fax: 08197 5865623; e-mail: JI@oita-med.ac.jp


We studied the relationship between insulitic development and function–structural changes of pancreatic lymphatics in non-obese diabetic (NOD) mice using combined 5′-nucleotidase (5′-Nase) enzyme histochemical and secondary lymphoid tissue chemokine (SLC/CCL21) immunohistochemical methods. Interlobular lymphatic vessels were positive for 5′-Nase throughout the pancreas, and dependent on both blood vessels and pancreatic ducts. Intralobular initial lymphatics were rare and occasionally ran in the neighbourhood of islets. During the non-insulitic stage, the 5′-Nase-reactive product was evenly distributed on the surface of lymphatic endothelial cells (LECs) with weak expression of CCL21. The activity of 5′-Nase on lymphatic vessels became slightly reduced as insulitis developed. The increasing blood glucose values appeared to be consistent with an increasing CCL21 expression by the endothelial lining, especially on the surface of LECs adjacent to the infiltrated islets and tissues. Lymphocytes and dendritic cells (DCs) were frequently located in the connective tissue, surrounding the lymphatic wall with deposition of 5′-Nase precipitates. As the infiltration became severe, lymphocytes and DCs accumulated within lymphatic vessels and expressed high levels of CCL21. The most significant finding was that many DCs adhered to lymphatic vessels, transmigrating via the thin and indented endothelial walls. The activity of 5′-Nase was increased on the adhesion surface between DCs (or lymphocytes) and LECs. The latter were characterized by open intercellular junctions and obvious cytoplasmic protrusions. These results suggest that LECs closely interact with DCs and lymphocytes, and play a key role in the migration of DCs and lymphocytes via lymphatic vessels during the pathological processes of insulitis in NOD mice.


The autoimmune response has been analysed since the 1980s in the non-obese diabetic (NOD) mouse, a model of human insulin-dependent (type I) diabetes mellitus (IDDM). In prediabetic NOD mice, dendritic cells (DCs) and macrophages are usually located around vascular vessels adjacent to the islets of Langerhans. These cells absorb relevant antigens and then migrate via afferent lymphatics towards lymph nodes, where the antigens are presented to T lymphocytes. Activated or self-reactive T lymphocytes move through blood vessels into the pancreatic parenchyma and eventually lead to insulin-producing β-cell destruction (Jansen et al. 1994). During autoimmune pancreatitis, hyperglyceamia affects several important metabolic pathways in vascular endothelial cells, through which functional changes occur and cause diabetic complications (Tooke, 1995). Abnormal structures of blood vessels have been encountered in the NOD thymus (Savino et al. 1991). Lymphatic vessels in normal mammalian tissues (liver and skin) might participate in the migration of DCs and lymphocytes, a process that could be mediated by adhesion molecules and chemokines (Gunn et al. 1998; Saeki et al. 1999). The most likely candidate for the link of lymphatics with DCs and lymphocytes is secondary lymphoid tissue chemokine (SLC/CCL21), which was recently identified in the secondary lymphoid tissues and intestinal lymphatics of normal mice (Gunn et al. 1998) and in the pancreatic lymphatics of wild-type RIP-BLC1 transgenic mice (Luther et al. 2000). CCL21 expression on lymphatic vessels was thought to be involved in recruiting mature DCs from afferent lymphatics to the T-cell area of lymph nodes (Cyster, 1999). However, difficulties in distinguishing initial lymphatics from blood capillaries have hampered studies of intra-organic lymphatic vessels. Recent investigations of lymphatics and parenchymal components of the pancreas have focused on the normal tissues of several species (Ji & Kato, 1997; Regoli et al. 2001). To the best of our knowledge, it remains unknown whether any effects of the morphological and functional alterations in lymphatic endothelial cells (LECs) are exerted during the pathological processes of autoimmune insulitis.

We used 5′-Nase enzyme-histochemistry to distinguish lymphatics from blood vessels, and immunohistochemistry with CCL21 antibody to identify LEC functional changes in chronic inflammatory tissues (Kato et al. 1991; Ji & Kato, 2003). We investigated the morphological properties of pancreatic lymphatics, and explored the dynamic migration of immune cells (DCs and lymphocytes) via lymphatic walls to constrain better the development of diabetes in NOD mice.

Materials and methods

Seventy-five female NOD/shi jic mice were deeply anaesthetized with ether. The pancreas was excised at 4, 7, 10, 13 and 17 weeks (15 or more animals per interval), embedded in Tissue-Tek OCT compound, and then rapidly frozen in dry ice/acetone. Tissue samples were stored at −80 °C. Normal female BALB/c mice served as controls. All experiments were performed in compliance with the Guidelines for the Care and Use of Laboratory Animals at Oita Medical University.

5′-Nase histochemical staining

For light microscopy, consecutive 6–8 µm cryosections were fixed in cacodylate buffer (pH 7.2) containing 4% paraformaldehyde for 10 min, washed and then stained for 50 min at 37 °C with 5′-Nase–lead medium. The reaction medium consisted of 0.2 m Tris-maleate buffer (pH 7.2), 20 mL; adenosine 5′-monophosphate (AMP, sodium salt for substrate; Sigma, St Louis, USA), 25 mg; 0.1 m MgSO4 (an activator of enzyme), 5 mL; sucrose, 3 g; distilled water, 22 mL; 2% Pb(NO3)2, 3 mL and l-tetramisole, 20 mg (specific inhibitors of endogenous alkaline enzyme). The samples were developed with 1% ammonium sulphide for 1–2 min at room temperature. To distinguish initial lymphatics from blood capillaries further, the sections after 5′-Nase staining were incubated for 20–25 min at 4 °C with ALPase reaction medium, which contained 0.1 m Tris-HCl (pH 8.5), 40 mL; fast blue BB, 40 mg; NN′-dimethyl formamide, 2 mL; and naphthol AS-MX phosphate, 40 mg.

For transmission electron microscopy (TEM), tissue samples of about 2–3 mm3 from the tail of each pancreas were immersed in 4% paraformaldehyde (1% CaCl2 and 7% sucrose) for 30 min at 4 °C and chopped into small pieces using a razor blade. The tissue blocks were reacted with 5′-Nase–cerium medium for 40–60 min at 37 °C (Kato et al. 1991; Ji & Kato, 1997, 2001). The reaction medium contained 0.2 m Tris-maleate buffer (pH 7.2), 20 mL; AMP, 16 mg; 20 mm MgCl2, 4 mL; sucrose, 2 g; distilled water, 12 mL; 20 mm CeCl3 (capture agent), 4 mL; and l-tetramisole, 16 mg (Kato & Miyauchi, 1989). The specimens were post-fixed in 2% osmium for 2 h at 4 °C, dehydrated through graded ethanol concentrations, and then embedded in Epok 812. Sections (1 µm) stained with 0.1% toluidine blue and nabox were analysed to localize fine distribution and to identify the general structure of the vessels in the endocrine and exocrine tissues. Ultrathin sections (90–95 nm) were cut with a diamond knife on Reichert-Nissei Ultracuts. After staining with uranyl acetate and lead citrate, sections were examined under a JEM-1200 EX II electron microscope (JEOL, Tokyo, Japan). Control experiments for 5′-Nase staining were established by inactivating the enzyme for 60 min at 60 °C before incubating samples, by omitting the substrate and by adding 5′-Nase standard medium with 50 mm NiCl2 or 5 mm l-tetramisole.


Cryosections were fixed with acetone for 10 min at 4 °C and then washed in three changes of phosphate-buffered saline (PBS, pH 7.4) with 0.1% Triton X-100. The sections were blocked for 15 min with 0.3% H2O2 in PBS to inactivate endogenous peroxidase and an incubation with 10% normal rabbit serum for 20 min prevented background staining and non-specific antibody binding. The sections were incubated overnight at 4 °C with goat antimouse CCL21 (Santa Cruz Biotech, Santa Cruz, USA) at a dilution of 1 : 50–1 : 100 with PBS, and then with biotinylated rabbit anti-goat IgG (Nichirei, Tokyo, Japan) for 60 min at room temperature. After three washes in PBS, the sections were immersed in streptavidin–biotinylated peroxidase complexes for 30 min. Immunoreactivity was visualized using 0.02% 3,3′-diaminobenzidine tetrahydrochloride (DAB) for 2–4 min (Luther et al. 2000). The sections were counterstained with haematoxylin and examined by a light microscope (Olympus, Tokyo, Japan). Staining with CCL21 was read in a blinded manner by counting lymphatic and blood vessels around and within infiltrated areas or islets. To identify lymphatic and blood vessels, serial sections stained with CCL21 were further processed with 5′-Nase and ALPase enzyme reaction. Staining was semi-quantitatively scored as follows: –, negative; +, weak; ++, moderate and +++, strong. Negative controls with secondary antibody and normal rabbit serum but without CCL21 were not positively stained (data not shown).

Serological analysis of diabetes

Diabetes was characterized by weight loss, hyperglycosuria as assessed by urine chemostrips, dark tail vein (increased HbA1) and persistent hyperglycaemia (> 250 mg dL−1) as measured using an ANTSENSE II monitor (Daikin, Osaka, Japan). Curves of blood glucose values were compared among groups using Student's t-test.


Insulitis developed in 80% of the female NOD mice from 7 weeks of age and IDDM occurred in 20–30% of the mice at the age of 17 weeks. Blood glucose values increased with age in NOD mice (Fig. 1).

Figure 1.

Relationship between blood glucose and age (weeks) in NOD mice. Blood glucose values are shown as means ± SEM for 10 NOD mice per group. Values at 4 weeks of age significantly differ from those of 13-week-old (*) and 17-week-old (**) NOD mice ( P  < 0.05 and 0.01, respectively).

Interlobular lymphatic vessels that were 5′-Nase-positive were extensive in the connective tissue between the lobules and preferentially associated with blood vessels that had ALPase activity and pancreatic ducts (Fig. 2a), among which infiltrating cells accumulated in the NOD pancreas (Fig. 2b). 5′-Nase-positive intralobular lymphatic vessels that began among the acini (Fig. 2c), as absorbing initial lymphatics, were adjacent to the infiltrated islets of Langerhans during the insulitic period (Fig. 2d). The amount of 5′-Nase activity between normal and NOD mice significantly differed in both intralobular and interlobular lymphatics (Fig. 2a–d). 5′-Nase activity was unevenly distributed on the lymphatic wall in the NOD pancreas (Fig. 2e,g), especially at the serious insulitic stage.

Figure 2.

Photomicrographs of pancreatic cryosections with 5′-Nase-ALPase enzyme staining in normal (a,c) and NOD mice (b,d,e,g), and CCL21 immunochemical staining in NOD mice (f,h,I,j). (a) 5′-Nase-positive interlobular lymphatic vessel (dark brown) runs parallel to ALPase-positive blood vessels (blue). (b) Infiltrating cells are distributed along 5′-Nase-positive interlobular lymphatic vessels and around blood vessels and pancreatic ducts in 7-week-old NOD mice. (c) Typical 5′-Nase-positve intralobular lymphatic vessels with thin and indented wall among acini with few connective tissues. (d) At 10 weeks of age, an intralobular lymphatic vessel with weak 5′-Nase activity is adjacent to infiltrated area of the islets of Langerhans. (e,f) In serial sections of 7-week-old NOD mice, CCL21 staining (f) is uneven in 5′- Nase-positive lymphatic vessel and in surrounding connective tissue, but absent in the islet of Langerhans, pancreatic ducts and ALPase-positive blood vessels. (g,h) In serial sections of 10-week-old NOD mice, staining of CCL21 appears more intense in the infiltrated islet and surrounding matrix, and in 5′-Nase-positive interlobular lymphatics adjacent to the infiltrated islet than in ALPase-positive blood vessels. (i) CCL21 immunoreactivity in lymphatic endothelium and infiltrated areas (asterisks) of 13-week-old NOD mice. (j) At 17 weeks of age, expression of CCL21 is high on intralobular lymphatic vessel, infiltrating cells in the islets and surrounding areas. L, lymphatic vessel; B, blood vessel; D, pancreatic duct; I, the islet of Langerhans. Scale bar = 50 µm.

Immunohistochemical analysis from several age-groups of the NOD mice detected CCL21 protein expression on pancreatic lymphatics. CCL21 expression on LECs rarely appeared in the pancreas of healthy wild-type mice. The expression of CCL21 on intralobular and interlobular pancreatic lymphatics was up-regulated with age in NOD mice (Table 1). In 7-week-old NOD mice, immunoreactivity for CCL21 was identified in 5′-Nase-positive LECs and in surrounding connective tissues, but not in blood endothelial cells, epithelial cells of the pancreatic ducts or other cell types (Fig. 2e,f). In 10-week-old mice, CCL21 was stained in the interlobular lymphatic vessels, infiltrating cells in islets and surrounding matrixes. CCL21 activity was restricted to the surface of LECs adjacent to the infiltrated islets of Langerhans (Fig. 2g,h). The immunoreactivity for CCL21 was more intense on the surface of LECs and infiltrated islets in 13-week-old (Fig. 2i) than in 7-week-old (Fig. 2f) mice. At the age of 17 weeks, intralobular lymphatics and infiltrated islets were stained more intensely than blood vascular cells (Fig. 2j).

Table 1.  CCL21 expression on pancreatic lymphatic and blood vessels during the development of insulitis in NOD mice
MouseAge (weeks)Lymphatic vesselBlood vessel
  1. Scoring intensity: –, negative staining; +, weak; ++, moderate; +++, strong.


At the non-insulitic stage, 5′-Nase-reactive precipitates were more evenly distributed on the luminal than on the abluminal surface of LECs in TEM. Lymphatic vessels possessed rich valves, by which lymphatic capillaries were differentiated from collecting lymphatic vessels. These valves consisted of double layers of endothelial cells separated by a connective tissue core, to which 5′-Nase-reactive product was also attached (Fig. 3a). The basal laminae of lymphatic vessels were irregular or sometimes entirely absent, being surrounded by 5′-Nase-negative nerve fibres (Fig. 3b). 5′-Nase granules were found on the walls of initial lymphatics, especially on intercellular junctions (end-to-end, overlapping, interdigitating), and many pinocytotic vesicles were found in the endothelial cytoplasm (Fig. 3c).

Figure 3.

Electron micrographs of pancreatic lymphatics in NOD mice at 4–7 weeks of age. (a) 5′-Nase-reactive product is more evenly distributed on the luminal than the abluminal surface of lymphatic endothelial cells. Endothelial surface of the lymphatic valve (arrows) is decorated with 5′- Nase-reactive precipitates. Inset: higher magnification of the area indicated by an asterisk. (b) The interlobular lymphatic vessel with 5′-Nase-reactive product is surrounded by 5′-Nase-negative nerve fibres (N) and blood vessels (B). Inset: higher magnification of the area indicated by an asterisk. (c) Fine 5′-Nase-reactive granules extend into initial lymphatic intercellular junctions, which are characterized by overlapping (arrowhead), end-to-end (arrow) and interdigitating (double arrow). Many pinocytotic vesicles are seen in the endothelial cytoplasm. L, lymphatic vessel. Scale bar = a,b, 2 µm; a inset, b inset, c, 0.5 µm.

As insulitis progressed, a few infiltrating cells (DCs and lymphocytes) were detected in close opposition to the wall of 5′-Nase-positive lymphatic vessels at 10 weeks of age (Fig. 4a). Initial lymphatic vessels with 5′-Nase activity usually contained DCs. Mature DCs were characterized by an irregularly shaped nucleus with perinuclear condensed heterochromatin, electron-lucent veils and multivesicular bodies in the cytoplasm. The cell membrane of DCs retained weak 5′-Nase activity. The lymphatic endothelium represented obvious protrusions into the luminal and abluminal aspects (Fig. 4b). In 13-week-old NOD mice, many lymphocytes were positioned around and within lymphatic vessels with weak 5′-Nase activity. Lymphatic endothelia adjacent to the accumulation area of lymphocytes occasionally protruded into surrounding connective tissues to enclose the lymphocytes. The overlapping junction appeared in LECs entrapping lymphocytes (Fig. 4c). The lymphatic wall was frequently interrupted by a wide gap, clearly in contrast to the tightly structured wall of blood vessels. Sometimes, DCs and lymphocytes were closely adjacent to the open junction of lymphatic vessels (Fig. 4d). The most significant finding was that DCs with various cellular components transmigrated through the intercellular space of neighbouring endothelial cells by the age of 17 weeks. Cytoplasmic processes of DCs extended from the connective tissue to the lumen of lymphatic vessels (Fig. 5a). Lymphocytes frequently adhered to the luminal surface of LECs during this period. The adhesion surface between LECs and lymphocytes was obviously covered with 5′-Nase-reactive product (Fig. 5b).

Figure 4.

TEM micrographs of 10–13-week-old NOD pancreas. (a) Dendritic cells are located near the lymphatic vessel with attached 5′-Nase-reactive precipitates. (b) 5′-Nase-positive initial lymphatic vessel contains a dendritic cell with lobulated nucleus, obvious processes and multivesicular cytoplasmic bodies. Lymphatic endothelium also shows significant cytoplasmic protrusions into lumen and adjacent matrix. Inset: higher magnification of the DC. (c) Higher magnification of the inset area (asterisk). Lymphatic endothelia with weak 5′-Nase activity (arrows) protrude into surrounding connective tissues to enclose a lymphocyte. (d) Lymphatic endothelial wall is frequently interrupted by a wide gap and lymphocyte is close to open intercellular junction (arrow). L, lymphatic vessel; B, blood vessel; LC, lymphocyte; DC, dendritic cell. Scale bar = a–d, b inset, 2 µm; c inset, 10 µm.

Figure 5.

TEM views of lymphatics and blood vessels in pancreas of 17-week-old NOD mice. (a) Many lymphocytes and DCs appear in lymphatic vessels, on which 5′-Nase-reactive product is unevenly distributed. A dendritic-like cell (DC) is transmigrating through intercellular junction from matrix into lumen of lymphatic vessels. (b) 5′- Nase-positive lymphocytes adhere to luminal surface of 5′-Nase-positive LEC. Adhesion surface between LECs and lymphocytes has intense 5′-Nase product. Insets: higher magnification views of asterisk areas of a and b. L, lymphatic vessel; B, blood vessel; EAC, exocrine acinar cell; LC, lymphocyte. Scale bar = a,b, 5 µm; a inset, b inset, 1 µm.


5′-Nase-positive intralobular lymphatics began among the acini as absorbing lymphatic vessels. Lymph formation might depend chiefly on the drainage of lymph scattered through these intralobular lymphatics adjacent to the parenchyma (Ji & Kato, 1997). Interlobular lymphatic vessels were located throughout the pancreas around blood vessels and excretory ducts, and had obvious valves to prevent lymph back-flow. They may play important roles not only in the communication between intralobular and extra-organ lymphatics, but also in the regulation of draining tissue fluids in the interstitium, especially to prevent oedema during chronic inflammation in NOD mice.

The present study demonstrated that the transendothelial migration of DCs or lymphocytes through lymphatic vessels might be facilitated by structural changes of LECs, such as open junctions between adjacent endothelial cells, as well as increased cytoplasmic protrusions and vesicles in the NOD pancreatic lymphatics. Open junctions were frequently present on LECs in the NOD mouse pancreas, but were rare in normal controls. In this respect, pancreatic lymphatics may be similar to those of the lung and liver (Niiro & O’Morchoe, 1985; Roberge et al. 1985), but different from those of the dermis, in which the open junction is a feature that constitutes a major pathway for transport (Leak & Burke, 1968). The open junctions of LECs are important not only for the absorption of tissue fluids and removal of destroyed cell components and large amounts of enzyme released into the interstitium as occurs in severe pancreatitis, but also for the transport of immune cells, such as lymphocytes and DCs. The fold of endothelial cytoplasm, which abluminally entrapped immune cells, might be involved in providing a pathway across the endothelial walls of lymphatic vessels. Many pinocytotic vesicles appeared in the lymphatic endothelium in the NOD pancreas, although a few were present in normal LECs (O’Morchoe, 1997). As insulitis developed, the vesicles, as key molecule transporters, might be considerably activated in the lymphatic endothelium and fused to form intralymphatic channels.

Generally, mobilization of immune cells into lymphatics, whether in a steady or Ag-triggered state, is a passive phenomenon owing to the absorbing effect of the negative pressure in lymphatic vessels. In chronic inflammation, however, adhesion molecules and chemokines expressed on LECs become the initial guidance for the migration of lymphocytes and DCs in lymphatic vessels (Chan et al. 1999; Saeki et al. 1999). The close apposition of DCs and LECs identified in the present study might be a necessary signal for cell transmigration through lymphatic vessels. We found that DCs often entered and moved within lymphatics as insulitis progressed, whereas this seldom occurred in normal mice. Moreover, the adenylate and guanylate cyclase, 5′-Nase, has been used in various mammalian tissues to study the distribution and fine structure of lymphatic vessels with special reference to their relationship with ALPase-positive blood vessels (Kato, 2000). During severe insulitis, 5′-Nase activity of lymphatics was much weaker in the NOD pancreas than in control groups, indicating that altered endothelial cell components directly induce functional expression on cell membranes to influence the exchange of interstitial tissue fluids, permeability modifications and the absorptive capacity of lymphatic vessels. At areas of chronic inflammation, a few activated 5′-Nase-positive lymphocytes migrated into 5′-Nase- and CCL21-positive lymphatic vessels and travelled along with antigen-bearing DCs into the T-cell areas of regional lymph nodes, ensuring a rapid start to the secondary immune response.

Both intralobular and interlobular pancreatic lymphatics in NOD, rather than in normal mice, unevenly expressed CCL21, and initially appeared close to DCs and lymphocytes. This suggests that endothelial cells secrete or contain CCL21, being induced by inflammatory stimuli to a greater or less extent (Hjelmstrom et al. 2000). These LECs up-regulate adhesion molecules such as CD31 and CCL21 chemokine. The latter is widely expressed in lymphatic vessels and in surrounding tissues, and it interacts with a G protein coupled receptor (CCR7) on lymphocytes and DCs (Chan et al. 1999; Saeki et al. 1999). Firmly attached DCs and lymphocytes on endothelial cells may easily follow a chemotactic gradient into lymphatic vessels. In a previous study, the mice homozygous for the paucity of a lymph node T-cell (plt) mutation do not express CCL21 and they have defects in T lymphocyte homing and DC localization in the spleen and lymph nodes (Gunn et al. 1999). Once DCs and lymphocytes enter lymphatic vessels, 5′-Nase and CCL21 are considered to participate in the migration towards lymph nodes. The present study has demonstrated that CCL21 immunoreactivity localized weakly on infiltrating cells in or adjacent to islets in the NOD pancreas, which is consistent with its chemotactic activity. The expression profile was similar to that observed in the spleen and pancreatic lymph nodes of wild-type animals, where CCL21 activity was detected on stromal and vascular endothelial cells (Luther et al. 2002).

The distribution characteristics of 5′-Nase and CCL21 on lymphatic vessels appear to reflect their roles in chronically inflamed tissues. LECs that produce CCL21 attract immune cells (DCs and lymphocytes), and 5′-Nase is involved in immune cell binding to the endothelium, which might promote further understanding of the mechanism in autoimmune diseases.


We thank Mr T. Kajiwara (Department of Anatomy, Biology and Medicine), Mr H. Kawazato and Miss A. Yasuda (Electron Microscopy Unit, Research Laboratory Center) for their excellent technical assistance.