All animals were bred and maintained as specific pathogen-free condition (SPF) at the Institute of Experimental Animal Sciences, Osaka University Medical School. All animal care and procedures described in the present study were approved by the Ethics Review Committee for Animal Experimentation of Osaka University (experimental number 20-055-0), and animal wellbeing was taken into consideration in the study design. All animal experiments were performed in accordance with the Guidelines for proper conduct for animal experiments from Scientific Council of Japan.

Mice used in the present study had disrupted α1,3-galactosyltransferase (α1,3GT) genes and are referred as α1,3GT knockout (KO) mice. The α1,3GT KO mice were generated on a C57BL/6×BALB/c genetic background (H-2bxd) (15,16). Prior to the experimental procedure, anti-Gal antibody (Ab) production was elicited in 6- to 8-week-old α1,3GT KO mice by four weekly intraperitoneal (i.p.) injections with 100 mg of pig kidney membrane homogenate (9). The amount (titer) of anti-Gal Ab was confirmed to be similar to that observed in humans (1:400–1:2,000, designated as high anti-Gal KO mice) by enzyme-linked immunosorbent assay (ELISA) with synthetic α-gal epitopes linked to bovine serum albumin (BSA) (Dextra Laboratories Ltd., Berkshire, UK) as the solid phase antigen (9,12,15).

The human pancreatic cancer cell line, PANC1 (ATCC, Manassas, VA, USA), which intrinsically expressed the Mucin1 (MUC1) molecule, was employed (12,17). We established stable PANC1-transfected cells, expressing α-gal epitopes, by mouse α1,3GT gene transfection (called α-gal PANC1) as previously described (12). To generate PANC1 tumors, 2×10⁶ live cells (either parental or α-gal PANC1) were injected subcutaneously into the back of non-obese diabetic severe combined immunodeficiency (NOD/SCID) mice (NOD. CB17-Prkdc^scid/J mice; Charles River, Tokyo Japan). The grown PANC1 tumors were enucleated and homogenized under sterile conditions, washed with 200 ml of PBS and centrifuged at 30,000 × g. The tumor membranes were resuspended at 100 mg/ml (weight/volume) in saline, and were subsequently irradiated with 50 Gy and frozen until needed (Fig. 1A).

The high anti-Gal KO mice were vaccinated by i.p. injection five times at 1-week intervals with 10 mg of 50-Gy-irradiated parental or α-gal PANC1 tumor lysates (abbreviated here as pt-lysate or α-gal-t-lysate, respectively). One week after the 5th vaccination, the mice were assessed for immune response induced by tumor lysate vaccination as described below (Fig. 1A). To compare the effectiveness of the α-gal PANC1 whole cell (abbreviated here as α-gal-whole-c) vaccine with that of α-gal-t-lysate vaccine, the mice received five i.p. injections of 1×10⁶ cells of 50 Gy-irradiated α-gal PANC1 whole cell vaccine in a manner similar to the tumor lysate vaccine (Fig. 1A) (12).

To determine whether the studied tumor lysates expressed α-gal epitopes, the tumor homogenates were assayed by ELISA using the monoclonal anti-Gal IgM Ab, M86, as previously described (17–19). The expression level of MUC1 in tumor lysates was assessed by ELISA using anti-MUC1 monoclonal antibody (mAb) (clone VU4H5; Santa Cruz Biotechnology, Santa Cruz, CA, USA; cat. no. sc-7313, lot no. B1611). Anti-MUC1 IgG production, was detected by ELISA using MUC1-BSA as the solid phase antigen, as previously described (12). Anti-PANC1 IgG production was detected by ELISA using dried-up PANC1 cells as the solid phase antigen, as previously described (20).

An enzyme-linked immunospot (ELISPOT) assay was used to identify the expansion of anti-MUC1 secreting B cells and MUC1-specific activated T cells (i.e., IFN-γ secreting T cells), using a previously described method (12).

Parental PANC1 and α-gal PANC1 tumor specimens, generated in NOD/SCID mice, were cut into small blocks, fixed in formalin and then embedded in paraffin. Tissue sections (4 μm thick) were incubated with either mouse anti-human MUC1 mAb (1:100; Santa Cruz Biotechnology; clone VU4H5, cat. no. sc-7313, lot. no. B1611) or M86 anti-Gal mAb (1:2) (21) in PBS Tween-20 (0.05% w/v) for 16 h at 4°C. The sections were then incubated with appropriate antibodies (for anti-MUC1 Ab, HRP-conjugated goat anti-mouse IgG, dilution 1:1,000; for M86 mAb, HRP-conjugated goat anti-mouse IgM, dilution 1:1,000). Immunostaining was visualized with 0.02% diaminobenzidine (DAB; Sigma-Aldrich) as the chromogen. The specificity of the primary Abs was verified using control sections prepared as described above but without the use of the primary Abs.

To evaluate the expression of CD44 and CD24, which are CSC markers of pancreatic cancer, on parental and α-gal PANC1 tumors, tissue sections were incubated with either rabbit anti-human CD44 mAb (dilution 1:100; Abcam, Cambridge, MA, USA; cat. no. ab97478) or rabbit anti-human CD24 mAb (dilution 1:100, Santa Cruz Biotechnology; cat. no. FL-80, sc-11406), respectively, followed by incubation with Alexa Fluor 555 goat anti-rabbit IgG Ab (A21429, dilution 1:1,000; Invitrogen). Fluorescence signals were observed with a Biozero fluorescence microscope (Keyence Corporation of America, Elmwood Park, NJ, USA). The α-gal epitopes in PANC1 tumors were detected by incubating the sections with M86 anti-Gal mAb (1:2 dilution) (21) in PBS Tween-20 for 16 h at 4°C, followed by incubation with Alexa Fluor 488 goat anti-mouse IgM Ab (A21042; dilution 1:1,000; Invitrogen). Fluorescence signals were assessed by fluorescence microscopy.

To investigate Ab production against differentiated pancreatic cancer cells (isolated differentiated cancer cells from PANC1 cells; i.e., CD44⁻CD24⁻ PANC1 cells) and pancreatic CSCs (isolated cancer stem cells from PANC1 cells; i.e., CD44⁺CD24⁺ PANC1 cells), cells were stained with sera from KO mice vaccinated with pt-lysate, α-gal-t-lysate, or α-gal-whole-c, as previously described (12). To determine whether or not splenocytes from the vaccinated α1,3GT KO mice can be specifically stimulated by MUC1 peptide, PANC1 cells or PANC1 tumor lysate, a carboxyfluorescein diacetate succinimidyl ester (CFSE; Invitrogen, Carlsbad, CA, USA; CellTrace™ CFSE Cell Proliferation kit, cat. no. C34554) assay was performed according to the manufacturer’s recommended protocol. Human embryonic kidney HEK293 cells were also employed as stimulatory cells. The CFSE labeled mouse splenocytes were cultured in 96-well round bottom plates (cat. no 3870-096; Iwaki, Japan) at 2×10⁵ cells/well with 1×10⁴ stimulatory cells (irradiated PANC1 or HEK293 cells), and 10 μg/well of MUC1 peptide, 10 mg/well of PANC1 tumor lysate or 3 μg/ml of ConA. The stimulated cells were cultured for 72 h. Proliferation of either CD4⁺ or CD8⁺ responder T-cells was measured with a FACSCalibur and analyzed with CellQuest software (BD Biosciences).

As shown in Fig. 1A, high anti-Gal KO mice (n=90) were generated by immunization with pig kidney fragments, then vaccinated with pt-lysate (n=30), α-gal-t-lysate (n=30) or α-gal-whole-c (n=30). One week after the last vaccination, splenocytes were prepared from successfully vaccinated donor KO mice and then suspended in warm (37°C), sterile RPMI complete medium containing 50 μM of 2-mercaptoethanol. For adoptive transfer, these isolated splenocytes were transferred by i.p. injection into NOD/SCID mice three times at 3-day intervals (75-150×10⁶ cells/vaccinated KO mouse). Splenocytes obtained from pt-lysate-, α-gal-t-lysate- or α-gal-whole-c-vaccinated KO mice were injected in equal amounts into NOD/SCID recipient mice (in total, 90×10⁶ splenocytes were transferred; each group, n=10 transferred NOD/SCID mice). One day after adoptive transfer, all NOD/SCID mice were challenged with subcutaneous injection of either 10×10⁶ live PANC1 cells or 5×10⁶ CD44⁺CD24⁺ PANC1 cells (i.e., the pancreatic CSC fraction of PANC1 cells) (14). Subsequently, these mice were examined for both tumor growth and survival. All mice were monitored every day after injection to detect the changes of general signs. Mice were sacrificed at the humane endpoints defined as following changes: i) physical appearance (self-injury, soiling of hair with urine of faces, bleeding, severe body weight loss defined by >20% loss in maximal body weight and loss of appetite); ii) clinical physiology (tachypnea and low body temperature). When remarkable increase of tumor size, defined by >10% increase in body weight was observed, mice were humanely sacrificed. The mice were induced deep anesthesia by isoflurane and subsequently sacrificed by cervical dislocation.

Data were collected from at least five independent experiments. Quantitative data were expressed as the mean ± SD. Statistical analysis was performed using the Student’s t-test. Kaplan-Meier curves of estimated survival were generated, and comparisons between parental PANC1 tumor lysate, α-gal PANC1 tumor lysate, and α-gal PANC1 whole cell vaccine groups were performed using a two-sided log rank test. A P-value <0.05 was considered significant.

The histological findings of the PANC1 tumors originating from parental and α-gal PANC1 cells were compatible with those of human pancreatic cancer (Fig. 2A). Low expression levels of α-gal epitopes were observed in the parental PANC1 tumor, whereas high expression levels were detected on the cell surface of α-gal PANC1 tumors (Fig. 2A). The low expression of α-gal epitopes in parental PANC1 tumor was likely dependent on the migration of stromal tissues, including vascular and fibrous cells that originated from recipient NOD/SCID mice.

We previously reported the expression of 5×10¹³ α-gal epitopes/mg-lysate in pig kidney fragments (9,19). ELISA determined that approximately 2×10¹⁴ α-gal epitopes/mg-lysate were expressed in α-gal-t-lysate (Fig. 2B). For α-gal-whole-c, similar levels of α-gal epitope expression were detected (~2×10⁹ α-gal epitopes/cell). A BCA protein assay was performed to assess the accurate protein concentration of tumor lysates or α-gal-whole-c. The protein concentration was approximately 1 mg/ml for both tumor lysates and α-gal-whole-c. Therefore, 100 mg of glycoprotein/i.p. injection, expressing 2×10¹⁵ α-gal epitopes, was contained in either the 10 mg α-gal-t-lysate or 1×10⁶ α-gal-whole-c vaccination given to high anti-Gal KO mice. Similar levels of MUC1 expression were observed in parental and α-gal PANC1 tumors by both immunohistochemical staining and ELISA (Fig. 2C and D). Similar levels of MUC1 expression were also observed in α-gal-whole-c (data not shown).

As shown in Fig. 3 (A and B; anti-PANC1 IgG response, C and D; anti-MUC1 IgG response), repeated vaccinations (five times) with 10 mg/i.p. injection of α-gal-t-lysate elicited strong responses of anti-PANC1 IgG and anti-MUC1 IgG. Vaccinations with pt-lysate did not induce these Ab responses. Vaccination with the α-gal-t-lysate elicited an ~16-fold increase in both anti-PANC1 IgG and anti-MUC1 IgG production, compared with the pt-lysate vaccination. There was ~2–4-fold higher production of anti-PANC1 IgG observed in sera from α-gal-t-lysate vaccinated KO mice than detected after vaccination with α-gal-whole-c; however, there were no differences in anti-MUC1 IgG production (data not shown).

To further investigate the subclass of immunoglobulin reactivity of either anti-PANC1 IgG or anti-MUC1 IgG, we performed an ELISA using HRP-conjugated goat anti-mouse IgG1, IgG2a, IgG2b and IgG3 mAbs as secondary antibodies. All secondary antibodies were purchased from Bethyl Laboratories Inc. Sera from α-gal-t-lysate-vaccinated KO mice showed large amounts of all IgG subclasses, including IgG1, IgG2b, IgG2a and IgG3. The IgG1 subclass of both anti-PANC1 and anti-MUC1 IgG was especially expressed, and induces strong antitumoral cytolysis through antibody-dependent cell-mediated cytotoxicity and complement-dependent cytotoxicity (Fig. 4A and C). On the other hand, sera from pt-lysate-vaccinated KO mice produced only a small amount of IgG1 and did not produce the IgG2a, IgG2b and IgG3 subclasses (Fig. 4B and D). With the α-gal-whole-c vaccination, there was no production of the IgG2a subclass of either anti-PANC1 or anti-MUC1 IgG in sera despite production of large amounts of the other IgG subclasses (data not shown) (12).

As shown in Fig. 5A, splenocytes isolated from pt-lysate-vaccinated KO mice displayed 136.7±13.2 spots/1×10⁶ splenocytes of anti-MUC1-secreting B cells. In contrast, α-gal-t-lysate-vaccinated KO mice had 305.3±44.0 spots/1×10⁶ splenocytes (P=0.0071). In pt-lysate-vaccinated KO mice, we detected 181.7±27.5 and 44.3±6.5 spots of IFN-γ secreting T cells in the presence and absence of the MUC1 stimulator peptide, respectively; thus, a significant increase in the number of spots was observed with MUC1 peptide stimulation (P=0.0011; Fig. 5B). In α-gal-t-lysate-vaccinated α1,3GT KO mice, 626.7±118.6 and 76.3±12.9 spots were detected with or without MUC1 peptide stimulation, respectively, and the difference in the number of spots was also significant (P=0.0013; Fig. 5B). Furthermore, the number of spots in the presence of the MUC1 peptide was significantly higher in the α-gal-t-lysate-vaccinated group than in the pt-lysate group (P=0.0032; Fig. 5B).

As shown in Fig. 6, the negative control showed no significant difference in the percentage of proliferated T cells, which appeared as CFSE-low responder T cells (i.e., the CFSE intensity was <400), between α-gal-t- and pt-lysate-vaccination. The positive control (Concanavalin A stimulation) also showed no significant differences in proliferated T cells. Proliferation of T cells was significantly induced in the presence of PANC1 whole cells, PANC1 tumor lysate and MUC1 peptide; whereas, no proliferation was elicited by HEK293 whole cell stimulation. Lymphocytes were also stimulated with other kinds of irradiated cells, including monkey COS7 cells and mice fibroblast NIH3T3 cells. However, these types of stimulatory cells failed to induce significant proliferation (data not shown). Moreover, the proliferation rate of T cells in α-gal-t-lysate-vaccinated α1,3GT KO mice was significantly higher than in pt-lysate-vaccinated α1,3GT KO mice (Fig. 6).

The experimental design of in vivo studies is shown in Fig. 1A. To confirm the production of anti-PANC1 and anti-MUC1 IgG Abs in adoptively transferred NOD/SCID mice, we performed an ELISA prior to tumor challenge (Fig. 1B and C). Sera from control NOD/SCID mice (without adoptive transfer of splenocytes) showed no anti-PANC1 and anti-MUC1 IgG Ab production; while, NOD/SCID mice who received pt-lysate-vaccinated splenocytes showed small amounts of anti-PANC1 IgG Ab (Fig. 1B and C). In contrast, extremely large amounts of both anti-PANC1 and anti-MUC1 IgG Abs were noted in NOD/SCID mice who received α-gal-t-lysate-vaccinated-splenocytes (Fig. 1B and C). Representative pictures of mice treated with α-gal-whole-c, α-gal-t-lysate or pt-lysate are shown in Fig. 7A. Compared with untreated control mice (data not shown), pt-lysate- and α-gal-whole-c-vaccinated mice developed large tumors; while, no tumors were noted in the α-gal-t-lysate-vaccinated mice (Fig. 7A and B). The in vivo results, including survival time, are summarized in Table I. As shown in Fig. 7B, we monitored tumor growth in splenocyte-transferred mice. No significant differences in the time to appearance of palpable tumor after tumor challenge were observed in either the untreated control or pt-lysate group (untreated, 10.6±2.5 days; pt-lysate, 11.9±2.1 days). In contrast, the development of tumors in the α-gal-whole-c-vaccination group was significantly delayed compared with the untreated and pt-lysate groups (α-gal-whole-c: 16.0±2.8 days, P=0.018 vs. control; P=0.004 vs. pt-lysate). In the untreated control group, the maximum tumor size was 100 mm² within 29 to 34 days (mean, 31.4±2.1 days). In comparison, tumor growth to a similar size was markedly delayed in both the pt-lysate group (40.3±6.9 days, P=0.007 vs. control) and α-gal-whole-c group (45.6±8.3 days, P=0.0013 vs. control). The beneficial effects of vaccination with pt-lysate, α-gal-t-lysate, or α-gal-whole-c were also noted in the prolongation of survival after tumor challenge (Fig. 7C). As shown in Fig. 7C and Table I, the mean survival time of KO mice vaccinated with α-gal-t-lysate was markedly prolonged (82.5±21.9 days) compared with non-vaccinated (41.0±5.7 days, P<0.001), pt-lysate-vaccinated (48.0±6.7 days, P<0.001), and α-gal-whole-c-vaccinated KO mice (57.0±12.6 days, P=0.01). The final cause of death for adoptively transferred NOD/SCID mice from non-vaccinated, pt-lysate-vaccinated and α-gal-whole-c-vaccinated KO mice were indicated as cancer death, whereas mice transferred from α-gal-t-lysate-vaccinated KO mice died a natural death without the appearance of cancer. Notably, the mean survival time was significantly improved in the pt-lysate-vaccinated group compared with the non-vaccinated group (P=0.02), despite the lack of synthesis of α-gal epitopes in the tumor lysate vaccine.

Compared with untreated control mice (data not shown), pt-lysate- or α-gal-whole-c-vaccinated mice developed large tumors, but the tumorigenesis of pancreatic CSCs was completely prevented in all α-gal-t-lysate-vaccinated mice (Fig. 8A and B). With the exception of the α-gal-t-lysate group, there were no significant differences in the time to appearance of palpable tumors after tumor challenge among the groups (untreated, 13.1±3.3 days; pt-lysate, 14.4±3.4 days; α-gal-whole-c, 17.0±3.8 days) (Table II). The tumor size reached 100 mm² in 40.6±1.8 and 48.0±4.4 days in the untreated and pt-lysate groups, respectively; while, tumor growth to a similar size was significantly delayed in the α-gal-whole-c group, (60.5±7.9 days; P<0.001, vs. control; P=0.033, vs. pt-lysate) (Fig. 8B, Table II). However, vaccination with pt-lysate and α-gal-whole-c did not prolong the survival time after tumor challenge (49.3±14.3 and 60.0±16.8 days, respectively), compared with the non-vaccinated control mice (46.5±11.8 days) (Fig. 8C, Table II). The final causes of death for these mice were indicated as cancer death. In contrast, vaccination using α-gal-t-lysate significantly improved survival after tumor challenge and these treated mice died a natural death without the appearance of cancer (85.0±20.8 days; P<0.001 vs. control; P=0.002 vs. pt-lysate; P=0.018 vs. α-gal-whole-c) (Fig. 8C, Table II). The mean survival time was not significantly different between mice vaccinated with pt-lysate and the non-vaccinated control group (Fig. 8C, Table II), despite the beneficial effects seen with live parental PANC1 cells.

As shown in Fig. 9A, sera from both the α-gal-whole-c and α-gal-t-lysate groups more strongly bound to CD44⁻CD24⁻ PANC1 cells than those from the pt-lysate group, as judged by the mean fluorescence intensity. There was strong Ab production against pancreatic CSCs (i.e., CD44⁺ CD24⁺ isolated PANC1 cells) elicited by vaccination with α-gal-whole-c and α-gal-t-lysate (Fig. 9B). Importantly, vaccination with α-gal-t-lysate induced better Ab production against both CD44⁻CD24⁻ PANC1 cells and pancreatic CSCs than with α-gal-whole-c, as judged by the mean fluorescence intensity (Fig. 9A and B).

Evaluation of α-gal epitope expression (M86 staining) revealed there was abundant expression of α-gal epitopes in α-gal PANC1 tumors; whereas, expression of α-gal epitopes was scarcely observed in parental PANC1 tumors (Fig. 9C and D). Both CD44 and CD24 molecules were expressed in >90% of cancer cells in PANC1 tumors on NOD/SCID mice. The expression levels of these CSC markers in PANC1 tumor cells were markedly upregulated in comparison with the levels in either α-gal or parental PANC1 cells (12,13). Thus, the CSC components in PANC1 cells were enriched upon tumor formation in NOD/SCID mice. In α-gal PANC1 tumor tissue, merged microphotographs showed tissues stained positive for both M86 and CD44 as well as for both M86 and CD24; thus, the tissues simultaneously expressed the α-gal epitopes and CD44 or CD24 on the cell surface (Fig. 9C, yellow regions). However, no yellow regions were observed in parental PANC1 tumor tissue (Fig. 9D). These results suggest that the buildup of α-gal epitopes on the carbohydrates of CSC-related molecules allows the internalization and antigen-presentation of these molecules by APC. Furthermore, the concentration of CSC-related molecules in the tumor lysate vaccine seems greater than in the whole cell vaccine (i.e., α-gal PANC1 whole cell vaccine).