CD26 expression determines lung metastasis in mutant F344 rats: involvement of NK cell function and soluble CD26

Animals, injection of tumor cells and processing of lungs

F344USA substrain (F344/Crl from a Charles River colony in Portage, Mich.) and F344JAP substrain (DuCrj from a Charles River colony in Atsugi, Japan) were obtained from the different countries via Charles River, Sulzfeld, Germany. F344HAN (F344/Ztm) rats, initially derived from the F344USA substrain, were obtained from a breeding colony at the Central Animal Laboratory at Hannover Medical School, Germany. All substrains were bred for one generation and maintained in a specific pathogen-free facility at 25°C under a 12-h light/12-h dark cycle (light on at 0700 hours) with ad libitum access to food and water. For the experiments, age-matched 8-week-old male F2 offspring of all three substrains were used. All research and animal care procedures were approved by the Lower Saxony District Government (Hannover, Germany) and followed the principles described in the European Community's Council Directive of 24 November 1986 (86/609/EEC) and the Principles of Laboratory Animal Care (NIH publication no. 85-23, revised 1985). Every attempt was made to reduce animal suffering and the number of animals needed to obtain reliable results. The exact number of animals used per experiment is indicated by the F values with at least four animals per condition and time-point.

Cell culture, injection of tumor cells, dissection of the animals and immunohistochemistry were conducted as previously described [22, 28]. In brief, 1×10⁶ MADB106 tumor cells (clone kindly provided by Prof. S. Ben-Eliyahu, Tel Aviv, Israel) derived from the log phase of tumor growth were injected via the lateral tail vein. Lungs were removed at different time-points thereafter. For in situ quantification of tumor cells at early time-points after injection (15 min, 30 min and 1 h), cells were stained with vital dye using the fluorescein derivative 5-(and 6-)carboxyfluorescein diacetate succinimidyl ester (CFSE; Molecular Probes, Eugene, Ore.) before injection. For quantification of lung surface colonies at later time-points (2 weeks after tumor cell inoculation), en-bloc dissected lungs and heart were injected with 8 ml Bouin's solution (72% saturated picric acid solution, 23% formaldehyde and 5% glacial acetic acid) and fixed in the same solution until lung surface nodules were counted. Three areas per lung were examined using multiple square centimeters of lung surface as a standard measurement. This was achieved by using a counting frame (1 cm²) as a gauge and lung surface colony numbers are expressed as mean number per centimeter squared, according to the method of Wexler [29].

Immunohistochemistry of lung CD26, NK cells and CFSE-labeled tumor cells

Immunostaining of CD26, NK cells and CFSE-labeled MADB106 tumor cells was achieved using the following mAbs: (1) characterizing rat CD26/DPPIV: MRC OX61 (Serotec, Oxford, UK), mouse, 1:100; and 6A3 [16] (kindly provided by Prof. B.U. Pauli, Cancer Biology Laboratory, Department of Pathology, Cornell University, College of Veterinary Medicine, Ithaca, N.Y.); (2) characterizing the NK-RP1 receptor (CD 161) on the NK cell surface (anti-NKRP1): Chambers 3.2.3 (Serotec, Raleigh, N.C.), mouse anti-rat, 1:10,000; and (3) characterizing the intracellular CFSE antigen (anti-CFSE): DE1 (Boehringer, Mannheim, Germany), mouse, 1:100.

In the rat, mAb 3.2.3 strongly reacts with NK cells while subsets of T lymphocytes and monocytes are only weakly 3.2.3-positive. Two different anti-CD26 mAbs (OX61 and 6A3) were chosen. MAb 6A3 inhibits adhesion of lung-derived endothelial membrane vesicles to lung metastatic carcinoma cells [16] and pilot studies using immunohistochemistry have indicated that mAb OX61 recognizes an epitope of CD26 absent in F344JAP rats. For immunohistochemistry, one or two consecutive alkaline phosphatase anti-alkaline phosphatase (APAAP) stainings were performed as previously described [22, 28]. Control sections were included in which one or both primary antibodies were omitted.

Quantification of tumor targets: in vivo/in situ cell adhesion assay

Vital dye (CFSE) labeling of MADB106 tumor cells allows the quantification of tumor cells and NK cells in thick sections of lung tissue by stereology in situ [28]. In the present study, we produced thin sections (8 μm) of the same lungs (n=10) and performed additional microscopic counting by image analysis of CFSE-labeled tumor cells, which appeared as mAb DE1-positive MADB106 cells in histological sections. This was done to further simplify the previously validated stereological quantification technique and also to obtain factors for the estimation of absolute numbers on the basis of relative counts obtained by image analysis. Simple regression analysis between stereological measurements (absolute tumor cell numbers) in thick sections and cell numbers per grid counted by two independent observers revealed a highly significant correlation of R=0.893 and R ²=0.797 with a regression plot of y=0.01+1.6⁻⁵×x between the two measures (F(1,8)=31.4; P=0.0005). Therefore, in the present study, the assessment of DE1-positive tumor cells in lung tissue from different substrains 15 min, 30 min and 1 h after tumor inoculation was carried out using image analysis. All CFSE-labeled MADB106 tumor cells and leukocyte subsets within a grid on the ocular lens were counted (Zeiss Kpl-W 12.5×; grid 0.75×0.75 mm=0.5625 mm²/grid, using a Zeiss Neofluar objective ×10, NA=0.3). Each right upper lobe of the lungs was sectioned at six randomly chosen non-adjacent levels. From each level, three sections were evaluated. On average, 30 grids per section were examined (i.e. 0.5625 mm²/grid × 30 grids × three sections × six levels) resulting in an area per animal of 3.04 cm².

Quantification of CD26 activity in lungs of F344 substrains

Lung tissue samples were kept frozen at −80°C until use. Tissue (100 mg) was homogenized in 0.14 M NaCl, 20 mM Hepes buffer, pH 7.4 (1 ml), with CD26 enzyme activity determined by incubating an aliquot with the substrate glycylproline-p-nitroanilide (Gly-Pro-pNA, 0.5 mM in the above buffer; Bachem, Germany), and measuring the color development after 30–60 min incubation at 37°C at 405 nm versus controls [20]. One unit was defined as the enzymatic release of 1 μmol nitroaniline per minute under these conditions. A further aliquot of the homogenate was used to measure the protein concentration by the Coomassie brilliant blue binding method using bovine serum albumin as standard.

Quantification of NK cell cytotoxicity in spleens of F344 substrains

For detection of splenic NK cell-mediated lysis in F344 substrains, a single cell suspension of splenocytes was prepared for each subject by gently pressing spleen tissue using the end of the stamp from sterile plastic syringes (n=3 per F344 substrain in three independent experiments). After centrifugation, erythrolysis was performed for 10 min at room temperature using 8 ml ACK erythrolysis buffer. After two washes in PBS, cells were resuspended in exactly 10 ml RPMI-1640. Leukocyte numbers were determined using a Coulter cell counter and the splenocyte concentrations adjusted to 6×10⁶ cells/ml with supplemented RPMI-1640 containing 10% fetal bovine serum. NK cytotoxicity was measured in classical ⁵¹Cr-release assays using MADB106 target cells. Splenocytes, prepared through a Ficoll-Hypaque gradient [15], were used as effectors and effector-to-target (E:T) ratios of 12.5:1, 25:1, 50:1 and 100:1 were obtained. Effector and target cells were coincubated either for 4 h with the addition of 1000U/ml IL-2 (EuroCetus, Amsterdam, The Netherlands) or for 18 h [27]. Control wells containing only labeled targets were also plated to determine spontaneous release. For determination of maximum possible release, the targets in one set of control wells were lysed with Triton X-100. The spontaneous release was always less than 10% of the maximum release. Plates were centrifuged for 4 min prior to incubation at 37°C in an atmosphere containing 5% CO₂ and again prior to harvesting 75 μl of the supernatant for determining ⁵¹Cr release in a gamma counter. The specific cytotoxicity was calculated from the following formula: [(experimental release)−(spontaneous release)]/[(maximum release)−(spontaneous release)]×100.

Immunocytochemistry of cytospin-centrifuged MADB106 cells and peripheral blood mononuclear cells after preincubation with mutant F344 rat serum

Peripheral blood mononuclear cells (PBMC) were obtained from age-matched untreated F344 rat substrains (n=3 per F344 substrain). MADB106 cells and PBMC were centrifuged in a cytospot centrifuge onto gelatin-coated microscope slides [24]. In order to investigate the expression of plasma FN (pFN) and adhesion of soluble CD26 to MADB106 tumor cells, cells were preincubated overnight at 4°C with serum from each of the F344 substrains. Anti-CD26 mAb (6A3, see above) and anti-FN polyclonal Ab (Cedarlane, Ontario, Canada; rabbit, 1:100) were used as primary antibodies. PBMC of these three substrains incubated with each serum were also stained for FN and CD26 to clarify the influence of pFN and soluble CD26 on these cells. Cytospin-centrifuged cells were stained as described above except for the reacting time to the above antibodies, which was 24 h at 4°C.

Western blotting of mutant CD26

Serum samples obtained from F344 substrains were reduced and denatured in sample buffer, separated by SDS-PAGE (10% gel) and transferred by semidry blotting (90 min, 65 mA) to a polyvinylidene difluoride membrane (PVDF; Millipore, Bedford, Mass.) as previously described [10]. The blot was blocked with 5% casein, 0.1% Tween and 1 mM EDTA in PBS for 60 min, incubated overnight at 4°C in 0.1% Tween and 1 mM EDTA in PBS with anti-CD26 (6A3), washed (15 min three times, 1 mM EDTA in PBS) incubated (30 min) with peroxidase-labeled goat anti-rabbit IgG (Sigma, Deisenhofen, Germany; 1:40,000), washed (15 min three times, 1 mM EDTA in PBS), incubated with enhanced chemiluminescence (ECL) reagent (1 min) and exposed to Hyperfilm as recommended by the manufacturer (Amersham, Little Chalfont, UK). Molecular masses were determined with rainbow colored proteins (Amersham).

Statistical analysis

Data from the in vivo adhesion assay and in vitro NK cytotoxicity assay were analyzed by two-way analyses of variance (ANOVA) (factors: "substrain" and "time", i.e. 15 min, 30 min and 1 h after inoculation, in the former; and "substrain" and "E:T ratio", i.e. 100:1, 50:1, 25:1 and 12.5:1, in the latter). These data were further analyzed by one-way ANOVA and Fisher's PLSD post hoc tests, if appropriate. Data obtained from simple measures such as CD26 activity or colony numbers were analyzed by one-way (factor: "substrain") ANOVA. An asterisk indicates significant post hoc effects versus F344USA substrain (control) obtained by Fisher's PLSD. All data are presented as means±SEM.

CD26 activity in lungs of F344 substrain: lack of activity in F344JAP rats

The endogenous enzymatic CD26-like activity in lungs of F344 substrains is shown in Fig. 1A. CD26-like activity was lacking in F344JAP rats (F(2,14)=44.6; P<0.0001).

Tumor cell adhesion in vivo: reduction in F344JAP rats

To study the adhesion of MADB106 tumor cells, CFSE-stained cells were injected i.v. The mean number of CFSE-positive cells in lung tissue at 15 min, 30 min, and 1 h after inoculation of MADB106 tumor cells is illustrated in Fig. 1B. Two-way ANOVA showed a significant effect of the factors "substrain" (F(2,30)=12.2; P<0.0001) and "time" (F(2,30)=7.2; P=0.0027), but no significant interaction due to the fact that in F344JAP rats at all time-points the number of tumor cells was reduced. However, one-factor ANOVA revealed that the strain factor was significant at 30 min (F(2,8)=5.5; P=0.03) and 1 h (F(2,12)=6.3; P=0.008). The results of this newly established in vivo assay suggest that the reduced tumor adhesion found in F344JAP starts as early as 15–30 min after entry of metastatic cells into the blood.

Metastasis formation: inhibition in F344JAP rats

The number of lung surface tumor colonies (see Fig. 2M–O) of different substrain F344 rats 2 weeks after injection of MADB106 tumor cells is illustrated in Fig. 1C. One-factor ANOVA revealed a significant effect of substrain (F(2,24)=9.5; P=0.0003) on colony numbers due to a decreased number of colonies in F344JAP rats.

Differential NK cytotoxicity in mutant F344 substrains: CD26 modulates NK cell function

NK specific lysis of MADB106 tumor cells using a classical ex vivo NK cell functional assay is shown in Fig. 3. In the mutant F344JAP substrain lacking CD26-like enzymatic activity, NK cell-mediated lysis against MADB106 tumor targets was significantly decreased compared to that in wild-type F344USA rats. This effect was observed in both assays (4 h coincubation in the presence of IL-2, Fig. 3A; and 18 h incubation, Fig. 3B). Two-way ANOVA showed a significant effect of "substrain" (F(1,24)=10.3; P=0.0006) and "E:T ratio" (F(3,24)=5.2; P=0.0065) in the 4-h assay. Similarly, two-way ANOVA of the 18-h assay data revealed a significant effect of "substrain" (F(1,24)=11.5; P=0.0003) and "E:T ratio" (F(3,24)=4.5; P=0.012). These findings suggest that CD26 is involved in mediating NK cytotoxicity against syngeneic tumor targets.

Immunohistological characterization: antibody specificity effector-to-target colocalization, and CD26-like immunoreactivity in NK cells and MADB106 cells

Representative parallel stainings for MADB106 tumor targets (CFSE/DE1 mAb), NK effector cells (CD161/3.2.3. mAb) and CD26 protein (CD26/OX61 or 6A3 mAbs) in lung tissue derived from F344HAN, F344USA and mutant F344JAP substrains are shown in Fig. 2 (first, second and third columns, respectively).

As shown in photomicrographs of lung tissue, only the functional enzymatically active CD26 was detected by the mAb OX61 (Fig. 2A, B; blue precipitate), whereas lungs of F344JAP rats lacked OX61-like immunoreactivity (Fig. 2C). Anti-CFSE and anti-CD161 double staining for MADB106 tumor target cells and NK effector cells are shown in the photomicrographs of Fig. 2D–F. As previously reported [22, 28], the CFSE staining of vital dye labeled tumor cells appeared as a cytoplasmic and cell membrane-associated highly sensitive and specific blue precipitate with virtually no background, while NK cells were characterized by red membrane-associated staining (Fig. 2D–F). Colocalization of effector with target cells was frequently observed (Fig. 2D–F; insets).

In accordance with CD26 staining using the mAb OX61, binding of the CD26-specific mAb 6A3 (Fig. 2G–I) was markedly reduced in tissue from F344JAP rats [7] (Fig. 2I). Furthermore, double staining for CD26 and NK cells in Fig. 2G–I demonstrated that NK cells (red precipitate) also express CD26 protein at levels detectable by immunohistology.

In order to finally address the question as to whether soluble CD26 clusters at tumor cells during in vivo passage, double staining for intracellular CFSE-positive MADB106 tumor cells (red precipitate) and CD26 (blue precipitate) using the mAb 6A3 (Fig. 2J–L) was also performed. As described above, CD26-like immunoreactivity was found in capillaries of lungs (Fig. 2J–L). In vitro-cultured MADB106 tumor cells did not show CD26-like enzymatic activity (data not shown) nor demonstrate CD26-like immunoreactivity (see Fig. 4F). However, when injected in vivo, these cells appeared to be clearly positive for CD26 in tissue from the F344HAN and F344USA substrains (Fig. 2J, K, see also Fig. 4G). Even in the otherwise CD26-negative lungs of the CD26-low expressing F344JAP rats, MADB106 tumor cells were found to be positive for CD26 (Fig. 2L, see also Fig. 4H).

Soluble CD26: lower expression of mutant CD26 in F344JAP rat serum

To clarify the role of soluble CD26 protein in F344 rat substrains, Western blotting analysis using serum from wild-type-like F344USA rats and mutant CD26-negative F344JAP substrains was performed (Fig. 5). Serum from the wild-type F344USA substrain expressed a 110 kDa band. In contrast, serum from the F344JAP substrain showed a weak band near 110 kDa. Protein (Coomassie R250) staining of a corresponding SDS-PAGE gel demonstrated almost equivalent protein loading in each lane (data not shown).

Aggregated pFN and soluble CD26 on tumor cells incubated with F344JAP rat serum

Photomicrographs of cytospin-centrifuged MADB106 tumor cells and PBMC stained for FN and CD26 (mAb 6A3) are shown in Fig. 4. As shown in Fig. 4A, MADB106 tumor cells originally expressed cell surface associated FN. When these tumor cells were incubated with rat serum, pFN adhered to the cell surface resulting in the formation of large FN globules by FN self-association as previously reported [2, 8, 11, 21] (Fig. 4B, C, arrows). No difference was observed in the appearance of FN in the incubated serum of the different F344 substrains (Fig. 4B, C). In contrast to tumor cells, cell surface-associated FN was not observed on PBMC (Fig. 4D, E). Furthermore, CD26 immunoreactivity was not detected on MADB106 tumor cells derived from cell culture (Fig. 4F), but after incubation with rat serum, CD26-like immunoreactivity was found on these tumor cells (Fig. 4G, H, arrows). Notably, MADB106 cells incubated with F344JAP rat serum also showed clear immunoreactivity in comparison with the other substrain (Fig. 4G, H). However, CD26-like immunoreactivity was found to be more intense on PBMC from the F344USA substrain (Fig. 4I) than on PBMC from the CD26 mutant F344JAP substrain (Fig. 4J). Interestingly, PBMC of the F344JAP substrain showed intense intracellular staining (Fig. 4J, arrow), suggesting the expression and intracellular retention of mutant CD26-like protein [26].