Carlo
Pifferi
ab,
Ane
Ruiz-de-Angulo
b,
David
Goyard
a,
Claire
Tiertant
a,
Nagore
Sacristán
b,
Diego
Barriales
c,
Nathalie
Berthet
a,
Juan
Anguita
*cd,
Olivier
Renaudet
*a and
Alberto
Fernández-Tejada
*bd
aDépartement de Chimie Moléculaire, Université Grenoble Alpes, UMR 5250, CNRS, 38000 Grenoble, France. E-mail: olivier.renaudet@ujf-grenoble.fr
bChemical Immunology Lab, CIC bioGUNE, Biscay Science and Technology Park, Building 801A, 48160 Derio, Spain. E-mail: afernandeztejada@cicbiogune.es
cInflammation and Macrophage Plasticity Lab, CIC bioGUNE, Biscay Science and Technology Park, Building 801A, 48160 Derio, Spain. E-mail: janguita@cicbiogune.es
dIkerbasque, Basque Foundation for Science, Maria Diaz de Haro 13, 48009 Bilbao, Spain
First published on 8th April 2020
Tumor associated carbohydrate antigens (TACAs), such as the Tn antigen, have emerged as key targets for the development of synthetic anticancer vaccines. However, the induction of potent and functional immune responses has been challenging and, in most cases, unsuccessful. Herein, we report the design, synthesis and immunological evaluation in mice of Tn-based vaccine candidates with multivalent presentation of the Tn antigen (up to 16 copies), both in its native serine-linked display (Tn-Ser) and as an oxime-linked Tn analogue (Tn-oxime). The high valent vaccine prototypes were synthesized through a late-stage convergent assembly (Tn-Ser construct) and a versatile divergent strategy (Tn-oxime analogue), using chemoselective click-type chemistry. The hexadecavalent Tn-oxime construct induced robust, Tn-specific humoral and CD4+/CD8+ cellular responses, with antibodies able to bind the Tn antigen on the MCF7 cancer cell surface. The superior synthetic accessibility and immunological properties of this fully-synthetic vaccine prototype makes it a compelling candidate for further advancement towards safe and effective synthetic anticancer vaccines.
Aberrantly glycosylated versions of the protein mucin-1 (MUC1) are overexpressed in most human epithelial cancers,20,21 making this glycoprotein a target of high interest for both diagnostic and immunotherapeutic applications.22 Cancer-related MUC1 tandem repeats display truncated carbohydrate moieties at O-glycosylation sites, exposing glycan epitopes such as the Tn antigen that are normally hidden in mucins expressed in non-transformed cells. The Tn antigen consists of an N-acetyl-D-galactosamine (GalNAc) unit α-O-linked to the serine (Ser) or threonine (Thr) of a peptide backbone,23,24 and is overexpressed in approximately 90% of breast carcinomas,25 and in 70–90% of colon, bladder, cervix, ovary, stomach and prostate cancers.26–28 As such, it represents an excellent target for vaccine development, and we and other research groups have focused efforts on the design of anticancer vaccine candidates based on the Tn antigen.29–38 The inherent low immunogenicity of carbohydrate antigens is even more critical in TACAs, as they are self-derived antigens and therefore not prone to being recognized as foreign molecules by the immune system.39,40 Thus, a fundamental challenge for TACA-based vaccines involves the ability to produce functional, isotype-switched, IgG antibodies that are able to recognize native antigens expressed on cancer cells and selectively promote their clearance. Advances in the chemical-immunology field have provided evidence that through an accurate structural design, it is possible to develop effective TACA-based anticancer vaccines with promising preclinical outcomes.10,41–48 To overcome important limitations associated with protein–hapten conjugates,49–51 fully synthetic vaccine approaches are being developed that enable the assembly of structurally defined and easily characterizable constructs via modular chemical strategies.15,52–55 In this context, we report herein the design, synthesis and immunological evaluation of unprecedented fully synthetic vaccines with high Tn-antigen valency that are able to elicit robust and functional immune responses in mice.
Thus, multivalency-driven recognition of carbohydrate antigens by BCRs followed by receptor-mediated internalization of the vaccine construct incorporating a CD4+ epitope would allow antigen-presenting B cells to load the TH epitope onto MHC-II molecules on the cell surface. Direct B-T cell contact via T-cell receptor (TCR) on CD4+ T lymphocytes would provide bidirectional signaling, ultimately leading to B-cell proliferation and differentiation (Fig. 1b).69–71
Initially, we generated a small series of multivalent glycodendrimers as B-cell epitope carriers (Scheme 1, compounds 2, 5–8) to investigate whether a higher order multivalency could be beneficial to antibody binding in an anti-Tn mAb interaction assay (Fig. 2).74 We first synthesized Tn-based glycodendrimers 2 and 5, in which GalNAc units are α-O-linked to the Ser hydroxyl groups (referred to as “Tn-Ser” along the text), as in the native Tn antigen. Starting from intermediate 1 (see the ESI, Scheme S1†), bearing four protected Tn-Ser residues, global O-acetyl and Fmoc removal gave tetravalent Tn-Ser glycodendrimer 2 (Scheme 1a). Meanwhile, functionalization of the free amino group of the lower lysine side chain in 1 with Boc-aminooxyacetic acid N-hydroxysuccinimide ester (Boc-Aoa-NHS),75 followed by base-mediated deprotection of the four GalNAc-Ser residues afforded intermediate 3.
Scheme 1 (a) Synthesis of tetravalent and hexadecavalent Tn-Ser glycodendrimers 2 and 5. Reagents and conditions: [a] 5% piperidine in CH3CN, r.t., 15 min; [b] NaOMe/MeOH (pH ≈ 10), r.t., 20 min; [c] Boc-Aoa-NHS (1.2 eq.), DIPEA (2.0 eq.), CH3CN/DMF (1:1), 20 min; [d] TFA/CH2Cl2 (1:1), r.t., 30 min; [e] 4 (ref. 72) (0.17 eq.), 0.1% TFA in H2O, 37 °C, 45 min. (b) Tn-oxime glycodendrimers 6 (ref. 61) and 8 (ref. 73), and tetravalent GlcNAc-oxime control 7 (ref. 73). |
Fig. 2 Interaction curves of the anti-Tn mAb 9A7 (ref. 74) with glycodendrimers 5 (□), 8 (■), 6 (●), 2 (○), and 7 (▲). ELISA plates were coated with decreasing concentrations of glycodendrimers, starting from 100 μM. Results expressed as in absorbance values at 490 nm. |
Deprotection of the Boc-aminooxy group on 3 and subsequent oxime ligation with core scaffold 4,72 which displays four α-oxo-aldehyde groups,76 provided hexadecavalent Tn-Ser glycodendrimer 5 in four steps from intermediate 1 following a convergent strategy involving unprotected moieties (see the ESI, Scheme S4†). Conversely, the Tn-analogue glycodendrimer 8, in which the GalNAc units are attached via oxime linkages (referred to as “Tn-oxime” along the text), was synthesized in a divergent fashion by grafting the aminooxy-GalNAc moiety onto a hexadecavalent, α-oxo-aldehyde functionalized scaffold as the last step in the route.73 With our set of compounds in hand featuring hexadecavalent Tn-Ser/oxime structures (Scheme 1, compounds 5 and 8, respectively) and their tetravalent analogues (Scheme 1, compounds 2 and 6 (ref. 61), respectively), we then performed direct interaction assays to evaluate their ability to be recognized by anti-Tn antibodies (anti-Tn mAb clone 9A7).74 Tetravalent construct bearing GlcNAc-oxime residues (Scheme 1, compound 7 (ref. 73)) was used as a negative control (Fig. 2).
While hexadecavalent Tn-Ser compound 5 exhibited the strongest binding, the interaction curve of the Tn-oxime glycodendrimer 8 indicated that this multivalent system can serve as an effective analogue of the native Tn antigen. In contrast, tetravalent Tn-containing compounds 2 and 6 showed absorbance values close to the baseline, comparable to those of negative control 7. These initial results prompted us to pursuit the synthesis of the complete vaccine structures based on selected glycodendrimers 5 and 8 for immunogenicity studies in mice. In addition to the hexadecavalent vaccine prototypes, the corresponding tetravalent vaccine constructs derived from 2 and 6 were also synthesized to evaluate the impact of antigen valency in the elicited immune response.
To complete the design of our synthetic vaccine prototypes, we incorporated T helper CD4+ and CD8+ epitopes from ovalbumin77–80 (OVA) into the previous glycodendrimers to generate potent immune responses with strong and long-lasting production of IgG antibodies against the T-cell independent Tn carbohydrate antigen. OVA323–339 CD4+ T helper and OVA257–264 CD8+ T cell epitopes were synthesized “in-line” incorporating a cysteine residue at the C-terminus for further chemoselective conjugation to the core cyclopeptide scaffold (Scheme 2). The synthesis started with protected peptide sequence 9, which was obtained using standard Fmoc-based automated solid phase peptide synthesis (SPPS) on a Rink amide resin. Treatment with a TFA/TIS/H2O (96:2:2) cocktail resulted in the removal of all acid-labile side-chain protecting groups, with concomitant cleavage from the resin, affording peptide 10 in a 41% overall yield.
Scheme 2 Synthesis of immunostimulant peptide 10 containing in-line CD4+ and CD8+ epitopes from OVA. Reagents and conditions: [a] TFA/TIS/H2O (96:2:2), r.t., 3 h (×3 cycles), 41% overall yield. |
We first focused our efforts on the synthesis of the “native-like” Tn-Ser vaccine prototypes based on glycodendrimers 2 and 5. Starting from building block 1 (see Scheme 1), which displays protected Tn-Ser peripheral residues and a free amino group of the lysine of the scaffold, coupling with the NHS ester of Boc-[S-(3-nitro-2-pyridinesulfenyl)]-cysteine (Boc-Cys(NPys)-NHS) in DMF provided intermediate 11 (Scheme 3a).81,82 Taking advantage of the thiol activating nature of the NPys group, Boc removal (1:1 TFA/CH2Cl2) from the cysteine residue, followed by disulfide bridge formation with cysteine-containing peptide 10 gave OVA-functionalized intermediate 12. Finally, global deprotection under Zemplén conditions (NaOMe/MeOH, pH ≈ 10) with concomitant removal of the O-acetyl and Fmoc protecting groups provided the tetravalent vaccine candidate 13 (Scheme 3a). The same modular strategy was then applied to the synthesis of the hexadecavalent Tn-Ser vaccine construct based on glycodendrimer 5 (Scheme 1). Hexadecavalent intermediate 14, obtained by convergent attachment of four copies of protected Tn-Ser peripheral glycodendrimer S5 to the α-oxo-aldehyde-bearing central scaffold 4 through oxime linkages (see the ESI, Scheme S6†), was functionalized at its lower domain with OVA peptide 10via disulfide bridge formation by using the three-step procedure described above (Scheme 3b). The resulting Tn-Ser hexadecavalent construct 15, however, was found to be unstable to the various deprotection conditions used to remove the O-acetyl and Fmoc moieties (see the ESI, Scheme S8†), which presumably affected the internal oxime linkages, failing to afford the fully-deprotected vaccine candidate 16.
Scheme 3 (a) Synthetic strategy for tetravalent Tn-Ser vaccine candidate 13. (b) Synthetic strategy towards hexadecavalent Tn-Ser vaccine candidate 16 (see the ESI, Scheme S8†). Reagents and conditions: [a] Boc-Cys(NPys)NHS (1.5–1.6 eq.), DIPEA (1.5 eq.), DMF, r.t., 30 min; [b] TFA/CH2Cl2 (1:1), r.t., 30 min; [c] 10 (1.1–1.2 eq.), DMF/NaOAc buffer (2:1, pH 4.5, 40 mM), r.t., 30 min. |
To address this problem, we designed an alternative strategy towards a modified vaccine construct (19) in which the internal oxime linkages were replaced with 1,4-triazoles (Scheme 4). Since this modification involves the “internal” part of the final glycosylated structure and not the external B-cell epitope display, the binding ability of the construct should not be affected. In contrast to the synthetic route towards 16, the new strategy allows for a more convergent assembly via late-stage CuAAC and also enables O-acetyl and Fmoc removal from the Tn-Ser moiety at an earlier stage in the synthesis. In the event, CuAAC reaction between a slight excess of fully-deprotected peripheral glycodendrimer 17 (see the ESI, Scheme S9†) equipped with an alkyne handle on the lower domain and OVA-functionalized, azide-bearing core scaffold 18 (see the ESI, Scheme S10†) was carried out in a degassed DMF/PBS mixture using our previously reported protocol in the presence of copper(II) sulfate, THPTA and sodium ascorbate.83 Hexadecavalent Tn-Ser vaccine construct 19 was thus obtained in 75% yield after RP-HPLC purification.
With Tn-Ser vaccines 13 and 19 in hand, we next directed our efforts towards the synthesis of the Tn-oxime vaccine candidates. Unlike for the Tn-Ser constructs, the presence of a single free amino acid (Lys on the lower domain of the scaffold) in glycodendrimers 6 and 8 enabled the use of a versatile divergent strategy involving late-stage chemoselective functionalization at this position followed by installation of the T cell OVA epitopes in the last step of the route. Thus, starting from glycodendrimer 6 (see Scheme 1b), vaccine construct 20 was synthesized in 53% yield over three steps via Boc-Cys(Npys) installation, and disulfide bridge formation with OVA peptide 10 (Scheme 5). Following this three-step sequence, vaccine candidate 21 was obtained analogously from hexadecavalent glycodendrimer 8, although its poor solubility in DMF required the use of a DMF/PBS mixture (1:1, pH 7.4) to achieve the coupling of the Boc-Cys(NPys) residue with a satisfying 41% yield after RP-HPLC purification (Scheme 5).
Three weeks after the last immunization, blood was collected for serological analysis and the mice were sacrificed to assess cellular immunity. Notably, no toxic side effects (e.g. local inflammation, systemic reactions, mouse weight loss or death) were observed over the course of the immunizations (data not shown), indicating the non-toxicity of the synthetic vaccine constructs. First, we evaluated the ability of the constructs to generate antibody responses and of the antisera to bind the Tn-antigen in different presentation modes [native Tn-Ser residues (5, 2) and unnatural Tn-oxime analogues (8, 6), as well as in higher (5, 8) and lower (2, 6) valency, see Scheme 1]. Microtiter plates were coated with the corresponding Tn glycodendrimers lacking the OVA peptide and the total anti-Tn IgG levels in blood sera were detected by ELISA. Group A mice, immunized with hexadecavalent oxime-linked Tn construct 21 in combination with QS-21, exhibited the highest IgG levels against its glycodendrimer counterpart 8 (Fig. 4a). Moreover, this group was the only one in which all five mice were able to generate humoral responses. In contrast, Group B (hexadecavalent Tn-Ser 19 + QS-21) showed variable but lower IgG levels, while Groups C (20 + QS-21) and D (13 + QS-21) in which mice were immunized with the corresponding tetravalent constructs, exhibited IgG levels similar to the no adjuvant control (Group E). Interestingly, IgG antibodies produced by Group A mice (hexadecavalent Tn-oxime 21 + QS-21) were also able to recognize hexadecavalent Tn-Ser glycodendrimer 5 (Fig. 4b), showing therefore no clear preference for the native or unnatural Tn presentation on the scaffold. However, antisera from these mice (Group A) were less efficient in binding the Tn antigen presented through low-valency glycodendrimers Tn-oxime 6 and Tn-Ser 2 (Fig. 4c and d). Therefore, increased Tn antigen valency was found to be crucial for high-affinity binding of the IgG antibodies to the Tn antigen construct. On the other hand, while Group B mice showed antisera (IgG antibodies) able to bind both high-valency glycodendrimers 8 and 5 (Fig. 4a and b), their levels were considerably lower than those exhibited by Group A, and were not able to bind tetravalent glycodendrimers 6 and 2 (Fig. 4c and d). In these assays, tetravalent compounds 20 and 13 (Groups C and D, respectively) elicited IgG antibody levels similar to the no adjuvant Group E, highlighting the importance of the increased multivalency of vaccine candidates 21 and 19 to generate potent humoral responses. In addition to the presence of the OVA CD4+ T helper epitope, co-administration of QS-21 as an adjuvant was found to be essential for antibody class-switching and elicitation of IgG antibodies, with IgM antibody levels being negligible at the last time point (see the ESI, Fig. S45†).85 Antibody titration using oxime-linked glycodendrimer 8 for coating was carried out for the two groups showing the highest OD values, i.e. Group A (mice immunized with hexadecavalent Tn-oxime 21 + QS-21) and B (mice vaccinated with hexadecavalent Tn-Ser 19 + QS-21) (Fig. 5).
Fig. 4 Analysis of total IgG antibodies elicited against the synthetic glycodendrimers. Microtiter plates were coated with the four different glycosylated scaffolds lacking the OVA epitopes: (a) 8, (b) 5, (c) 6, and (d) 2 (see Scheme 1); and the antisera from mice A to E were probed. Statistical significance compared to no-adjuvant control group (Group E) was assessed using two-tailed unpaired Student's t-test: ***p < 0.001. |
Antibody subtyping of the anti-8 IgG isotypes revealed that Group A mice showed not only high total IgG titers but also elevated levels of the IgG1, IgG2b and IgG2c antibody subtypes. Interestingly, the IgG2c antibody titers in this group were as high as those observed for IgG1 antibodies, suggesting that 21 in combination with QS-21 elicits a balanced Th1/Th2 immune response (Fig. 5e). The IgG2c subtype is of particular interest since it is associated with potent antitumor effect such as complement- and antibody-dependent cell toxicity in mice.86 In contrast, IgG antibody titers from Group B mice were not significantly higher than those of the no adjuvant control Group E, and subtyping of these antibodies revealed a bias towards the IgG2b subclass. These results indicate that hexadecavalent vaccine construct 21 bearing oxime-linked Tn antigen analogue is more efficient than its native Tn-Ser counterpart 19 in generating potent humoral responses, and was therefore selected for further immunological studies.
We then evaluated the ability of the antisera to recognize the Tn antigen in a native context and analyzed the binding of the vaccine-induced serum antibodies to a human cancer cell line (MCF7) expressing the natural Tn antigen by using fluorescence microscopy. Notably, IgG antibodies elicited by immunization with compound 21 plus QS-21 (Group A mice) were able to specifically bind Tn-expressing MCF7 cells. Moreover, indirect immunofluorescence images showed high signal coming from sera from Group A mice, whereas antisera from mice immunized with 21 without QS-21 (Group E) showed no Tn-specific IgG binding (Fig. 6a). Antibody binding of sera from mice Group A showed broad membrane surface localization (Fig. 6b). In contrast, although compound 21 alone (Group E) was able to elicit IgG antibodies to a small extent (Fig. 5e), these antibodies were not able to specifically recognize natural Tn antigen expressed in this tumor cell. These results highlight that antibodies elicited by the Tn-oxime analogue vaccine construct 21 coadministered with QS-21 are functional and able to recognize the native Tn antigen expressed on the cancer cell surface, confirming the tumor specificity of the generated antibody responses.
Next, we also investigated the cellular immune responses elicited by Tn-oxime construct 21 by testing its ability to specifically activate T cells. At the time of sacrifice (three weeks after the last immunization, day 49), whole splenocytes were harvested and assayed for T cell restimulation in the presence of full-length OVA protein. After 48 h stimulation, splenocytes were analyzed for activation markers using flow cytometry by assessing the percentage of CD4+CD44high (ref. 87 and 88) and CD8+CD107a+ (ref. 89) in the restimulated splenocyte pools. Notably, vaccine construct 21 in combination with QS-21 activated CD4+ as well as CD8+ T cells after stimulation with the specific antigen, confirming the immunogenicity of compound 21 for both B- and T-cells (Fig. 7). Therefore, construct 21 stands out as a potent and safe synthetic Tn analogue vaccine candidate that, coadministered with QS-21, is able to induce robust humoral and cellular immune responses without toxic side effects.
Footnote |
† Electronic supplementary information (ESI) available. See DOI: 10.1039/d0sc00544d |
This journal is © The Royal Society of Chemistry 2020 |