Carbohydrate-based vaccines with a glycolipid adjuvant for breast cancer

Abstract

Globo H (GH) is a hexasaccharide specifically overexpressed on a variety of cancer cells and therefore, a good candidate for cancer vaccine development. To identify the optimal carrier and adjuvant combination, we chemically synthesized and linked GH to a carrier protein, including keyhole limpet hemocyanion, diphtheria toxoid cross-reactive material (CRM) 197 (DT), tetanus toxoid, and BSA, and combined with an adjuvant, and it was administered to mice for the study of immune response. Glycan microarray analysis of the antiserum obtained indicated that the combination of GH-DT adjuvanted with the α-galactosylceramide C34 has the highest enhancement of anti-GH IgG. Compared with the phase III clinical trial vaccine, GH–keyhole limpet hemocyanion/QS21, the GH-DT/C34 vaccine elicited more IgG antibodies, which are more selective for GH and the GH-related epitopes, stage-specific embryonic antigen 3 (SSEA3) and SSEA4, all of which were specifically overexpressed on breast cancer cells and breast cancer stem cells with SSEA4 at the highest level (>90%). We, therefore, further developed SSEA4-DT/C34 as a vaccine candidate, and after immunization, it was found that the elicited antibodies are also IgG-dominant and very specific for SSEA4.

The aberrant glycosylation associated with tumor progression was first described by Meezan et al. (1) in 1969 with the demonstration that many glycans on cancer cells differ from normal cells. The aberrations include loss or overexpression of certain glycan structures, persistence of truncated glycans, and emergence of novel glycans. Some of the aberrant glycans on malignant tissues were identified by immunohistochemical staining with lectins (2, 3) or mAb or by MS (4, 5). To date, numerous tumor-associated antigens expressed on cancer cells in the form of glycolipids or glycoproteins have been characterized and correlated to specific types of cancers (6, 7). However, the role of surface carbohydrates in malignant tumor cells remains obscure. In breast cancer and small cell lung carcinomas, many studies have suggested that expression of the carbohydrate CaMBr1 is associated with tumor aggressiveness and poor prognosis (8, 9). CaMBr1 was identical to GL6, which was isolated from the human teratocarcinoma cell line 2102Ep and originally identified as a globo-series glycosphinolipid with an H-like determinant at its terminus, hence subsequently designated as Globo H (GH; Fucα1→2Galβ1→3 GalNAcβ1→3Galα1→4Galβ1→4Glcβ1) (10). GH has been shown to be overexpressed on a variety of epithelial tumors such as colon, ovarian, gastric, pancreatic, endometrial, lung, prostate, and breast cancers by the staining with mAb MBr1 (11–13) and VK-9 (14). GH is also expressed weakly in normal tissue but only restricted to apical epithelial cells at lumen borders, a site that seems to be immune-inaccessible (9, 15–17). Moreover, higher levels of anti-GH Ab were detected in the plasma of breast cancer patients (18–20). Our previous study also revealed that the expression of some glycans, such as GH and Stage-specific embryonic antigen 3 (SSEA3) (also called Gb5), was observed on breast cancer cells and breast cancer stem cells (BCSCs) (21). All these findings support a rationale for development of carbohydrate-based vaccines based on these cancer-specific glycans.

However, because carbohydrates generally exhibit poor immunogenicity, some issues need to be resolved before practically using carbohydrate antigens for cancer immunotherapy. A multivalent glycan vaccine or carbohydrate conjugated with a carrier protein and administered with an immunological adjuvant has been adopted to increase the immunogenicity of carbohydrates. As an example, a GH vaccine with ∼700 GH units conjugated to one keyhole limpet hemocyanin (KLH) protein was prepared and administered together with a mixture of saponins called QS21 adjuvant (18, 22–25), and it had been shown that treatment of mice with GH-KLH/QS21 could induce antisera against GH-positive cell lines without autoimmune reactions (14, 23). Furthermore, the vaccine was fairly well-tolerated in a phase I clinical trial for metastatic breast cancer patients. This vaccine is currently in phase 3 clinical trials, and the GH epitope is manufactured by the programmable one-pot method (26). Although GH-specific IgM was generated in most patients, the level of IgG was low, and patients often reported inflammatory response at the injection site with occasional systemic complaints such as fever, arthralgias, and myalgias (18). All these undesirable symptoms are similar to the reported side effects of QS21 (27–29). Therefore, we decided to develop a GH vaccine with different carriers and adjuvants to improve the immunogenicity and safety profile.

The α-Galactosylceramide (α-GalCer or C1), isolated from marine sponges, and analogs exhibit antitumor activity (30–32) and also are considered as possible adjuvant candidates. CD1d on dendritic cells is the receptor of C1 (33). When C1 is bound to CD1d and presented to T-cell receptors on the invariant Natural Killer T (iNKT) cells (34), the iNKT cells are activated to produce cytokines such as IL-4 with immunosuppression activity and IFN-γ with adjuvant effect as well as cell proliferation and cytotoxicity capacity of iNKT cells (31, 35–37). It was found that high avidity of CD1d–glycolipid–T-cell receptor complex would selectively enhance the Th1 pathway and thus, the adjuvant activity (38). In this study, we evaluated the immunogenicity of different GH vaccines using various carrier proteins, such as KLH, diphtheria toxoid cross-reactive material (CRM) 197 (DT), tetanus toxoid (TT), and BSA, and coadministrated with various C1 analogs as adjuvants. Our results showed that GH conjugated to the DT protein and coadministrated with C34, an analog of C1, induced a robust IgG response against GH and GH-related structures specifically on breast cancer cells and the cancer stem cells.

Results and Discussion

Synthesis of GH Conjugated with Different Carrier Proteins.

GH A was synthesized by a new programmable one-pot strategy developed by our laboratory (19) and reacted with a p-nitrophenyl ester homobifunctional linker (39, 40) to yield the half ester B, which after purification, was incubated with a carrier protein as shown in Scheme 1.

Scheme 1.

Synthesis of GH half ester and glycoconjugates.

The molecular weights of GH–protein conjugates including C (GH-BSA), D (GH-DT), and E (GH-TT) were determined by MALDI-TOF to calculate the average number of GH epitope on each carrier protein. Glycoconjugate D (GH-DT) showed an average of two to four GH incorporated to the protein, whereas C (GH-BSA) and E (GH-TT) showed an average of eight and six GH epitopes, respectively, per protein molecule (SI Materials and Methods, Fig. S1, and Table S1). GH-KLH showed the greatest (∼700) GH incorporation among these GH–protein conjugates because of its multimeric structure and numerous lysine residues.

GH-DT and GH-KLH Induced Comparable Anti-GH Ab Response.

A previous study showed that the fully synthetic GH-KLH vaccine could induce Ab against human breast cancer cells, MCF7 (22). To understand if other carrier proteins could induce a better Ab response, mice were i.m. immunized with synthetic GH conjugated to different carriers (2 μg GH) in the presence or absence of the glycolipid adjuvant C1 (2 μg). Three vaccinations were given at 2-wk intervals; 10 d after the last injection, mice sera were collected and subsequently tested with a GH-coated glycan microarray (19, 42, 43) to evaluate the level and diversity of anti-GH related Ab. We found that GH-KLH and GH-DT could induce higher levels of anti-GH IgM than GH-TT and GH-BSA (Fig. 1A). The levels of anti-GH IgM were significantly enhanced when coadministered with C1. A similar trend was also observed in the production of anti-GH IgG (Fig. 1B). Therefore, we conclude that (i) despite lower carbohydrate density than GH-KLH, GH-DT exhibited comparable immunogenicity with GH-KLH and (ii) C1 significantly enhanced Ab response.

Fig. 1.

Anti-GH Ab elicited by different GH–carrier protein conjugates. Groups of three mice were immunized s.c. with 1 μg synthetic GH conjugated to various carrier proteins (KLH, BSA, TT, and DT) and mixed with or without 2 μg α-GalCer (C1) as the adjuvant. Blood samples were collected 10 d after the last immunization. Mouse sera were analyzed by glycan microarray: (A) 1:60 dilution for IgM and (B) 1:240 dilution for IgG. The levels of anti-GH Ab were detected by Cy3-conjugated anti-mouse IgG or IgG, and the fluorescence intensities were read at 532 nm. Each bar represents the average fluorescence intensity ± SEM.

Coadministration of GH-DT with Glycolipid C34 as a Potent Adjuvant.

Because C1 was shown to be an effective adjuvant for GH-KLH and GH-DT, three C1 analogs (C17, 7DW8-5, and C34) (Fig. 2A) (40), aluminum phosphate, and MF59 were tested for their adjuvant activities and compared with C1. Groups of BALB/c mice were immunized i.m. with GH-DT (2 μg GH) and GH-KLH (2 μg) with or without 2 μg glycolipid at 2-wk intervals for three injections. Sera were obtained 10 d after the third vaccination, and the elicited Abs were profiled by a glycan microarray created by GH, GH analogs, and GH fragments, including SSEA3 and SSEA4 (SI Materials and Methods and Fig. S2A). In general, mouse anti-GH IgG titers increased as immunization proceeded, but the IgM levels were almost independent of vaccination times. The results showed that, when immunized with GH-DT or GH-KLH alone without any adjuvants, mice generated only low titers of anti-GH IgG and IgM. When mice were coadministed with any of these C1 glycolipid analogs, they produced significant amounts of anti-GH IgG, and C34 gave the best adjuvant efficacy (Fig. 2 B and C).

Fig. 2.

The adjuvant efficacy of α-GalCer (C1) and its analogs for GH-KLH and GH-DT. (A) Structures of C1 and its analogs including 7DW8-5, C34, and C17. (B) Female BALB/c mice were immunized i.m. with GH-KLH (2 μg GH) either alone or with 2 μg C1, 7DW8-5, C34, C17, AlPO₄, or MF59. (C) Female BALB/c mice were immunized i.m. with GH-DT (2 μg GH) either alone or with 2 μg C1, 7DW8-5, C34, or C17. Mouse sera were collected 10 d after the third immunization, and the production of IgG and IgM against GH, SSEA3, and SSEA4 was analyzed by glycan microarray. Each bar represents the mean fluorescence intensity ± SEM.

Ab Response Elicited by GH-DT Adjuvanted with C34 or QS21.

Because GH-DT exhibited comparable immunogenicity with GH-KLH with C1 as adjuvant (Fig. 1) and the adjuvant effect of C34 is the best (Fig. 2C), we evaluated the vaccine efficacy of GH-DT adjuvanted with C34 or QS21. Mice were immunized with GH-DT/C34 or GH-DT/QS21, and the immune sera were tested for the binding intensity to GH, SSEA3, and SSEA4. The IgM from both GH-DT/C34- and GH-DT/QS21-immunized mice showed a strong binding to GH and a lesser extent to SSEA3 and SSEA4, with C34 more effective in inducing high IgM response (Fig. 3). For IgG response, a strong binding to GH and weaker binding to SSEA3 and SSEA4 were observed in both C34- and QS21-adjuvanted vaccines. Comparing with GH-DT/QS21, GH-DT/C34 induced a lot more anti-GH and anti-SSEA4 IgG.

Fig. 3.

The GH, SSEA3, and SSEA4 binding profiles of the serum Ab collected after immunization with GH-DT/C34 or GH-DT/QS21. Mice (n = 6) were immunized i.m. with GH-DT (2 μg GH) with or without 2 μg C34 or 2 μg QS21 as adjuvant. Mouse sera were collected 10 d after the third immunization, and the binding intensities of anti-GH, SSEA3, or SSEA4 IgG and IgM were analyzed by glycan microarray. Each bar represents the mean fluorescence intensity ± SEM.

Search for the Best Epitope Ratio of GH-DT/C34 Vaccine and Comparison of GH-DT/C34 with GH-KLH/QS21.

Because GH-DT/C34 showed better vaccination results, it was of our interest to find the best composition of vaccine by changing the amount of GH attached to each carrier protein DT at pH 7.6. When the amount of DT (1 mg) was fixed, adding 1 mg GH monoester to the carrier gave 1.6 molecules GH on 1 molecule DT. When the amount of GH monoester was increased to 5, 10, and 22 mg, the epitope ratio increased to 5.1, 9.8, and 15.6, respectively (detailed synthetic procedures are discussed in SI Materials and Methods, Scheme S2). The epitope ratio was determined by MALDI-MS (SI Materials and Methods, Fig. S3, and Table S2). With these various compositions in hand, groups of BALB/c mice were immunized i.m. with the same amount of protein of these four vaccine candidates with 2 μg C34 at 2-wk intervals for three injections. Sera were obtained 10 d after the third vaccination. The results showed that 5.1 GH conjugated to 1 DT induced the strongest antibodies titers against GH. When we set the signal to noise (S/N) ratio > 3, the induced GH binding IgG still showed a good signal after dilution to 25,000-fold (Fig. 4A); however, the GH binding IgM can only be diluted 60-fold (Fig. 4B) with significant signal, and the IgM signals totally reached to the background when the sera were diluted 1,000-fold (SI Materials and Methods and Figs. S4 and S5).

Fig. 4.

The induced GH-IgG and GH-IgM titers and glycan binding profile of IgG collected from different epitope ratios of GH-DT/C34-immunized mice. Mice (n = 6) were immunized i.m. with the same protein amount of different epitope ratios of GH-DT and 2 μg C34. Mouse sera were collected 10 d after the third immunization. The data showed the highest dilution fold when S/N > 3. The subscript figures represent the number of GH molecules conjugated to 1 DT molecule. (A) 25,000-fold dilution of GH binding IgG. (B) 60-fold dilution of GH binding IgM. The binding profiles of IgG induced by GH_5.1-DT/C34 were analyzed by the glycan microarray with 49 glycans.

As shown above, GH_5.1-DT/C34 exhibited the best immunogenicity for the production of anti-GH antibody (Fig. 4). It is important to ascertain that the antibodies generated will not react with normal cell surface glycans. Therefore, BALB/c mice were immunized with GH_5.1-DT/C34 as described above, and the immune sera (60-fold dilution) were tested on the glycan microarray containing 49 chemically synthesized glycans, including GH, GH analogs, GH fragments, sialosides, and other tumor-associated carbohydrate antigens (SI Materials and Methods and Fig. S2B), to examine the glycan binding profiles of the immune sera. For comparison, we further examined the glycan binding profiles of immune sera (60-fold dilution) from GH-KLH/QS21 vaccine, which is under phase 3 clinical trials now. The results showed that the GH-KLH/QS21 vaccine-induced antibodies recognized more glycans compared with the GH-DT/C34 group (Fig. 5). Furthermore, analyses of the induced IgG antibodies by GH_5.1-DT/C34 vaccine were largely specific to three glycans, including GH, SSEA3, and SSEA4 (Fig. 5C). These data suggest that the GH-DT/C34 is more effective than KLH in eliciting immune response against GH and generates fewer IgM reactivities to unrelated glycans than KLH.

Fig. 5.

The induced antibodies titers and glycan binding profile from GH-DT/C34 and GH-KLH/QS21 immunized mice. (A) GH-DT/C34-induced IgM antibodies titers and glycan binding profile. (B) GH-KLH/QS21-induced IgM antibodies titers and glycan binding profile. (C) GH-DT/C34-induced IgG antibodies titers and glycan binding profile. (D) GH-KLH/QS21-induced IgG antibodies titers and glycan binding profile.

It was reported that DT would induce antigen-specific T-cell proliferation and elevate production of IL-2, IFN-γ, and IL-6, suggesting its role in a Th1-driven pathway (41–44). Despite the fact that the cytokine profile was predominantly Th1, subtypes of anti-DT–associated antibodies were IgG1 with no detectable IgG2a. These results, together with the Th1-driven property of C34, prompted us to evaluate the IgG subtype profiles of our GH vaccines, and it was found that GH-DT/C34 induced mainly IgG1 antibody, a trace amount of IgG2b, and IgG2c antibodies (SI Materials and Methods and Figs. S6 and S7), indicating a selective Th1 response. DT has been widely used for human vaccination against diphtheria for decades. Most importantly, it has been approved by the US Food and Drug Administration (FDA) for various glycoconjugate vaccines. Moreover, the GH-DT/C34 vaccine has better defined molecular entity and thus, is easier to manufacture.

Expression Profiles of GH-Related Structures in Breast Cancer Cells and BCSCs.

Previously, we have shown the expression of GH and SSEA3 in breast cancer cells and the CD44⁺/CD24^low/− BCSC population from clinical specimens (21). However, this CD44/CD24 BCSC system is not applicable to all specimens. Recent studies showed that the breast cancer cells characterized with high expressions of epithelial specific antigen (ESA) and protein C receptor, endothelial (PROCR) are potential BCSCs (45). Therefore, we decided to investigate the expression of GH-related structures, including GH, SSEA3, and SSEA4, in either CD44⁺/CD24^low/− or ESA^hi/PROCR^hi BCSCs. Four breast cancer cell lines were stained with either the CD44/CD24 (for cancer cell lines MCF-7 and SK-BR3) or ESA/PROCR (for cancer cell lines MDA-MB-231 and MDA-MB-361) system to identify the BCSC population. In the ESA/PROCR system, the expressions of GH and its related structures were consistent with both MDA-MB-231 and MDA-MB-361 cell lines, and BCSCs showed higher percentages of GH- and SSEA3-positive populations compared with the entire population (Fig. 6, SI Materials and Methods, and Fig. S8). In the CD44/CD24 system, there were lower percentages of GH- and SSEA3-positive populations in BCSCs than in the total cell population in MCF-7 (Fig. 6B). However, the expressions of the structures in BCSCs subpopulation of SK-BR3 were comparable with the entire cell population. It is noted that almost all of the breast cancer cells and stems cells tested here were highly SSEA4-positive. Therefore, we can conclude that the GH vaccine, shown to elicit the Ab against GH, SSEA3, and SSEA4, is potentially effective against not only the general breast cancer cells but also BCSCs.

Fig. 6.

The expressions of GH, SSEA3, and SSEA4 in breast cancer and BCSCs. MDA-MB-231 and MCF-7 cells were triple-stained by BCSCs markers (A) ESA/PROCR or (B) CD44/CD24, respectively, along with SSEA3, SSEA4, or GH and subjected to flow cytometry for expression. The percentages of GH-, SSEA3-, and SSEA4-positive cells in the entire population vs. ESA^hi/PROCR^hi or CD44⁺/CD24^−/low BCSCs were analyzed. Cells in boxes are positive in GH, SSEA4, and SSEA3, respectively.

Synthesis of SSEA4-DT as a Vaccine Candidate.

The results of the expression profile studies identified SSEA4 as another important antibreast cancer target because of its high expression level not only on general breast cancer cells but also on BCSCs. Thus, we used the programmable one-pot strategy (46) to synthesize SSEA4 that bears the aminopentanyl spacer arm suitable for protein conjugation (Scheme 2). To conjugate carrier protein DT to SSEA4, our first attempt was to use the same homobifunctional linker strategy, but the yield was very low. To increase the conjugation efficiency, we modified the aminopentanyl group of SSEA4 to the thiopentanyl group and also modified the lysine NH₂ group of DT to the maleimine functional group. Followed by a Michael addition under different reaction conditions, we were able to conjugate, on average, 2.79, 4.31, 5.98, and 7.01 SSEA4 to 1 DT (Scheme 2 and Schemes S2, S3, and S4 show detailed synthetic procedures). Using the same vaccination protocol mentioned previously, sera were obtained 10 d after the third vaccination, and the glycan binding profiles of immune sera were examined by 49-glycans microarray. The results showed that, on average, 4.31 and 5.98 SSEA4 conjugated to 1 DT induced the strongest IgG antibodies titers against SSEA4 (SI Materials and Methods and Fig. S9A). To our surprise, specificity analysis of the induced IgG antibodies by SSEA4-DT/C34 vaccine showed that the induced antibodies specifically bound to SSEA4 and its tetrasaccharide epitope (Fig. 7). The induced antibody subtypes are similar to those from the GH-DT/C34 vaccine, mainly IgG1, IgG2b, and IgG2c antibodies (SI Materials and Methods and Fig. S9B).

Scheme 2.

Synthesis of SSEA4-DT with different epitope ratio.

Fig. 7.

The glycan binding profile of IgG collected from different epitope ratios of SSEA4-DT/C34-immunized mice. Mice (n = 6) were immunized i.m. with the same protein amount of different epitope ratios of SSEA4-DT and 2 μg C34. Mouse sera were collected 10 d after the third immunization. The binding profiles of IgG (800-fold dilution) induced by different epitope ratios of SSEA4-DT/C34 were analyzed by the glycan microarray with 49 glycans.

Conclusion

Glycan microarray offers a powerful platform for testing Ab specificity and is useful for monitoring patients’ immune responses after immunization with carbohydrate-based vaccines. In this study, we used glycan microarrays to show that GH-DT, could elicit a comparable Ab response in mice compared with GH-KLH. Moreover, we found that the chemically synthesized glycolipid C34, which stimulates the immune system to produce IgG instead of IgM, could serve as a promising adjuvant for the GH vaccine. This new GH-DT/C34 vaccine elicited specific Ab, especially IgG, against GH, SSEA3, and SSEA4, which were specifically observed in breast cancer cells and cancer stem cells, and the immune response profile is more desirable than that of GH-KLH/QS21 vaccine, which induces more IgM than IgG with less specificity. In addition, we have developed SSEA4-DT/C34 as a vaccine candidate, which can induce more specific antibodies against SSEA4, that is highly expressed on breast cancer and BCSCs.

Materials and Methods

Detailed methods are provided in SI Materials and Methods.

These describe detailed synthetic procedures of GH-DT and SSEA4-DT vaccines, MALDI-TOF MS analysis of glycoconjugate, mice immunization schedule, serologic assay with glycan array, cell culture, and flow cytometry. Additional research materials and assays including synthesis of different epitope ratios for GH-DT and SSEA4-DT vaccine, glycan microarray fabrication, and analysis of the induced antibody subtypes of GH-DT and SSEA4-DT vaccines.