Novel sialoglycan linkage for constructing adjuvant-protein conjugate as potent vaccine for COVID-19

Abstract

Self-adjuvanting protein vaccines have been proved to be highly immunogenic with efficient codelivery of adjuvant and antigen. Current protein vaccines with built-in adjuvants are all modified at the peptide backbone of antigen protein, which could not achieve minor epitope interference and adjuvant multivalency at the same time. Herein, we developed a new conjugate strategy to construct effective adjuvant-protein vaccine with adjuvant cluster effect and minimal epitope interference. The toll-like receptor 7 agonist (TLR7a) is covalently conjugated on the terminal sialoglycans of SARS-CoV-2-S1 protein, leading to intracellular release of the small-molecule stimulators with greatly reduced risks of systemic toxicity. The resulting TLR7a-S1 conjugate elicited strong activation of immune cells in vitro, and potent antibody and cellular responses with a significantly enhanced Th1-bias in vivo. TLR7a-S1-induced antibody also effectively cross-neutralized all variants of concern. This sialoglycoconjugation approach to construct protein conjugate vaccines will have more applications to combat SARS-CoV-2 and other diseases.

Graphical abstract

A new conjugating strategy was developed to construct effective self-adjuvanting protein vaccines through sialoglycoconjugation. The resulting adjuvant-glycoprotein conjugate (TLR7a-S1) induced effective anti-virus humoral and cellular immunity.

1. Introduction

Vaccination is the most effective strategy for the prevention of many infectious diseases. Developing various vaccine platforms with high potency and low toxicities, is still a pressing need to combat emerging threats such as severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2). Subunit vaccines based on homogeneous components offer more precise targeting and enhanced safety profile than traditional whole-pathogen vaccines [1,2]. And compared with nucleic-acid-based vaccines (DNA and mRNA), protein subunits could offer a promising alternative with relatively low-cost manufacture and few side effects as the next-generation vaccine platform for prevention and treatment of viral infections [[1], [2], [3]].

Protein-based vaccines are often combined with powerful adjuvants to improve the poor immunogenicity of antigen proteins, and adjuvants functioned as ‘danger signals’ are usually physically mixed with antigens as external immune stimulators in conventional protein-based vaccines [[4], [5], [6], [7]]. However, free small-molecule adjuvants are prone to diffuse away from the injection site that causing systemic toxicity [8,9]. In addition, this simple formulation could not ensure the codelivery of antigen and adjuvant to the same antigen-presenting cell (APC) for optimal immune response [[10], [11], [12]]. Therefore, to attach adjuvants on protein antigens is an attractive and novel strategy of vaccine design, and self-adjuvanting conjugate vaccines have been proved to be highly immunogenic in our previous studies [[13], [14], [15], [16], [17]]. For the use of small-molecule stimulators, conjugation provides efficient codelivery of adjuvants and antigens and greatly reduces potential toxicities by impeding adjuvant diffusion to the whole body. To date, most of the conjugate vaccines have used chemical reactions to randomly conjugate adjuvants on the residues of protein, such as lysine side chains [[15], [16], [17], [18], [19], [20]], but the immunogenic epitopes of antigen protein might be interfered with by residue modification. Besides, our group has previously developed an adjuvant-protein conjugate vaccine that site-specifically attached the adjuvant to the N-terminus of protein antigen [[21], [22], [23]], but this method limits the number of conjugated adjuvants thus could not equip small-molecule adjuvants with a cluster effect. Therefore, it would still be necessary to develop new conjugating approaches with multiple linkage sites and minor epitope interference for constructing effective conjugate vaccines (Fig. 1 ).

Fig. 1.

Different constructs of conjugate vaccines. A) Vaccine constructs of previous protein conjugate vaccines. B) Vaccine construct of protein conjugate vaccine in this work.

As an immunodominant domain of SARS-CoV-2 spike (S) protein that elicits potent neutralizing antibody responses, S1 is an ideal protein antigen for developing subunit vaccines against SARS-CoV-2 [24,25]. Like other viral surface proteins, S protein and its S1 subunit are heavily glycosylated for the evasion of antibody neutralization by shielding immunogenic epitopes (Fig. 2A,B) [26,27]. Studies have reported that different glycosylation of the receptor-binding domain (RBD) of S1, the key part for binding to the angiotensin-converting enzyme 2 (ACE2) receptor [28], did not impact ACE2-RBD binding [29]; and vaccination with S protein lacking glycan shields elicited enhanced protective responses with broad neutralizing activity in animal models [30,31]. These results suggest that there might be minor impact of glycan alterationon the immunogenicity of S protein antigens. As the terminal sialic acids of glycan chains of antigen glycoproteins are typically remote from the peptide backbone, we hypothesized that the sialic acids of glycans could be employed as linkage sites for conjugating adjuvants with minimal interference on the peptide epitopes (Fig. 1B).

Fig. 2.

Design of the TLR7a-S1 conjugate vaccine platform. A) Surface representation of the glycosylated SARS-CoV-2 S glycoprotein trimer and S1 glycoprotein with N-linked glycans shown in red. B) Schematic representation of the S glycoprotein. The positions of N-linked glycosylation sequons are shown as branches. Protein domains are illustrated: N-terminal domain (NTD), receptor binding domain (RBD), fusion peptide (FP), heptad repeat 1 (HR1), heptad repeat 2 (HR2), S1 domain of S protein (S1); S2 domain of S protein (S2). C) Covalent binding scheme of TLR7a-S1. D) Schematic illustration of processing and presentation of TLR7a-S1 components. Adjuvant and antigen in TLR7a-S1 conjugate are codelivered to the same dendritic cell (DC), where they are internalized by DC. The antigen S1 is broken down into peptide fragments inside endosomes then presented on the cell surface by MHC class II proteins, or transferred into the cytosol and degraded and presented via MHC I cross-presentation pathway. During antigen processing in the endosome compartments, the released TLR7a stimulates endosomal TLR7 and activate myeloid differentiation primary response protein 88 (MyD88), leading to the nuclear factor-κB (NF-κB) translocation to the nucleus and production of inflammatory cytokines. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Toll-like receptors (TLRs) are a category of pattern recognition receptors (PRRs) critical for recognizing invading pathogens [32]. Among the TLR family, TLR7 is a promising target for the development of small-molecule adjuvants, including some purine-like small molecules that have been reported as potent TLR7 agonists [9,33,34]. Endosomal TLR7 signals in a myeloid differentiation primary response protein (MyD88) pathway, leading to the activation of transcriptional factors such as nuclear factor-κB (NF-κB) [9,32,33]. Stimulation of TLR7 results in rapid activation of innate immunity, thereafter promotes effective adaptive immunity [[33], [34], [35]]. Unfortunately, as systemic distribution of these small-molecule stimulators results in severe toxicity, the applicability of TLR7a compounds is largely limited in immunotherapies [36,37]. To overcome these problems, one promising strategy is to conjugate the TLR7a onto peptides or proteins for direct co-localization of adjuvant and antigens, which has been reported with enhanced immune responses compared with simple co-administration [38,39].

Herein, to explore a new approach for constructing conjugate vaccines, we developed a conjugate vaccine platform that specifically attaches adjuvants on the sialoglycans of glycoprotein. Using mild periodate oxidation, an aldehyde was introduced on sialic acids of glycans in the S1 glycoprotein, followed by oxime ligation with alkoxyamine modified TLR7 agonist (TLR7a, compound 1) (Fig. 2C). In contrast to the unconjugated TLR7a with rapid diffusion, the conjugated TLR7a could be efficiently recognized and taken up by APCs followed by intracellular release during antigen processing (Fig. 2D). To the best of our knowledge, this is the first study that covalently attached adjuvants on protein antigen through sialoglycoconjugation: (1) the built-in adjuvant reduces the risk of systemic toxicity; (2) multiple conjugating sites provide a cluster effect of adjuvants; (3) the sialic acid site-specific conjugation has minimal interference on immunogenic peptide epitopes. In this study, the resulting conjugate vaccine TLR7a-S1 was evaluated for its ability to activate the immune cells in vitro, and to elicit antibody and cellular responses in vivo. The antibody neutralizing activity against SARS-CoV-2 and variants of concern (VOCs) was also assessed for its potential as an effective vaccine candidate for COVID-19.

2. Materials and methods

2.1. Materials

AffiniPure goat anti-mouse kappa antibody IgM, IgG were purchased from Jackson ImmunoResearch, and peroxidase-conjugated AffiniPure goat anti-mouse kappa, IgG1, IgG2a, IgG2b, and IgG3 antibodies were purchased from Southern Biotechnology. Reagents used were RPMI-1640, DMEM and FBS (Fetal Bovine Serum) (Gibco). Alum adjuvant were purchased from Thermo Fisher Scientific. The SARS-CoV-2 Spike S1 protein (VISC2-S1) was provided by Jiangsu East-mab Biomedical Technology. All animal experiments were carried out at Laboratory Animal Centre of Huazhong Agricultural University (Wuhan, China). Animal experiments were conducted according to ethical guidelines and follow the recommendations concerning institutional experimental animal welfare.

2.2. Procedure for DMB derivatization of sialic acid and HPLC analysis

As previously reported [47], a Neu5Ac standard sample (40, 20, 10, 5 or 2.5 μg) was mixed with the DMB reaction solution (200 μL; 6.9 mM DMB⸱2HCl, 18 mM Na₂S₂O₄, 1.4 M acetic acid, 0.75 M β-mercaptoethanol in Milli-Q water). The mixture was stirred at 50 °C for 3 h in a thermomixer (300 rpm) in the dark. For the determination of sialic acid of the S1 glycoprotein, S1 protein (295 μg) was mixed with 4 M aqueous acetic acid solution, and the mixture was kept at 80 °C for 2.5 h, which was then derivatized with DMB in the same manner as described above. After the mixture was cooled and filtered off through a 0.22 μm film, the filtrate (50 μL) was analyzed by RP-HPLC (C18 column, 250 × 4.6 mm, 5 μm) at a flow rate of 1 mL/min using gradient eluents of solvents A (0.1% TFA in water) and B (0.1% TFA in acetonitrile). Elution started with 5% B, which was increased gradually to 25% B within 10 min, and 25% B was kept for 10 min. The elution was then increased gradually to 70% B in 15 min, which was kept under this condition to the end of an experiment (UV detection at 370 nm). HPLC peaks containing the anticipated DMB derivative of sialic acid (retention time: 8.2 min) were separately collected from each sample, and verified by ESI-MS [m/z: calculated, 425.14; observed, 426.47 (M + H)⁺].

2.3. Procedure for SGP oxidation and oxime ligation

Isolation and purification procedure of the sialylglycopeptide (SGP) was based on previous studies [44]. Pure SGP after preparative RP-HPLC column purification was treated with 1 mM NaIO₄ in dark at 4 °C for 30 min. Then the oxidation reaction was quenched by adding 20 mM Na₂S₂O₃ for 20 min at rt. Followed by RP-HPLC column purification, oxidized SGP 15 was treated with O-ethylhydroxylamine hydrochloride (30-fold molar excess of 15) in PBS (pH 5.5) and incubated at rt. for 48 h to give 16. HPLC peaks containing the anticipated compounds were collected from each sample, and verified by ESI-MS.

2.4. Procedure for conjugation of adjuvant and protein

Vaccine candidate TLR7a-S1 was prepared by the conjugation of TLR7a-linker (compound 1) and oxidized S1 glycoprotein following previously reported protocols [21]. The S1 protein solution was oxidized by adding NaIO₄ with the final concentration of 1 mM in dark at 4 °C for 0.5 h. Then the oxidation reaction was quenched by adding 20 mM Na₂S₂O₃ for 20 min at rt. Following the reaction, the unreacted small molecules (NaIO₄, Na₂S₂O₃ etc.) were removed by Amicon Ultrafiltration 0.5 mL units with a molecular weight cut-off of 30 kDa. The resulting concentrated reacted protein was incubated with compound 1 (30-fold molar excess of the protein) in 1 mL PBS/DMF (4:1, v/v, pH 5.5) at rt. for 48 h. The reaction was stopped and the excess small molecules were removed by using Amicon Ultrafiltration 0.5 mL units again. The resulting TLR7a-S1 conjugate was freeze-dried and confirmed through MALDI-TOF mass spectrometry.

3. Results and discussion

3.1. Preparation of TLR7a-spacer-ONH₂

To synthesize an alkoxyamine-modified linker for TLR7a and S1 conjugation, compound 9 was prepared by reacting bis[2-(2-chloroethoxy)ethyl] ether with sodium azide. Subsequent reduction of the azide groups of 9 with Pd(OH)₂/C and H₂ gave intermediate 10, following reaction with NHS-activated N-Boc-aminooxyacetic acid 12 to obtain compound 13 (Scheme 1A). The small molecule TLR7a was synthesized based on previously reported approaches, as outlined in Scheme 1B [15]. TLR7a 6 was then reacted with compound 13 to give TLR7a-spacer-ONH₂ 1, with high purity of 95% by HPLC (Fig. S6).

Scheme 1.

Synthesis of TLR7a-spacer-ONH₂. Reagents and conditions are as follows: A: (a) NaN₃ (3.0 equiv), DMF, 60 °C, 5 h, 91%; (b) Pd(OH)₂/C, H₂, CH₂Cl₂/MeOH, rt., 3 h; (c) NHS (3.0 equiv), DCC (1.2 equiv), CH₂Cl₂, rt., 2 h, 95%; (d) 10 (2.0 equiv), DIPEA (2.0 equiv), rt., 2 h, 43%; B: (a) 2-methoxyethanol (1.2 equiv), Na (7.2 equiv), 140 °C, 17 h, 80%; (b) methyl 4-(bromomethyl) benzoate (2.0 equiv), K₂CO₃ (7.0 equiv), DMF, 60 °C, 5 h, 68%; (c) Br₂ (2.0 equiv), CHCl₃, rt., 8 h, 78%; (d) NaOH, MeOH, 100 °C, 4 h, 76%; (e) 13 (1.5 equiv), DCC (2.0 equiv), CH₂Cl₂/DMF, rt., 3 h, 87%; (f) CH₂Cl₂, TFA, rt., 1 h, quant.

3.2. Conjugation of TLR7a and S1 glycoprotein

To introduce aldehydes into S1 glycoprotein for the oxime ligation with TLR7a-ONH₂ 1, mild periodate oxidation was used to generate an aldehyde on sialic acid-occupied glycans (Fig. 2C), which has been widely known to specifically oxidize the dihydroxy of polyhydroxy side chain of sialic acids. This reaction has been applied to label sialylated glycoproteins on living cells without altering the cellular proteins [[40], [41], [42]], and to introduce aldehyde groups on antibodies for developing antibody-drug conjugates (ADC) without affecting antibody integrity, functionality, and aggregation [43,44]. Therefore, to assess potential linkage sites for TLR7a conjugation, the amount of sialic acids present on the S1 glycoprotein was quantified by HPLC analysis. The most common member of sialic acid family, N-acetylneuraminic acid (Neu5Ac), was derivatized to be photometrically detectable by 1,2-diamino-4,5-methylenedioxybenzene (DMB) [45,46]. Different concentrations of standard Neu5Ac samples were treated with DMB, and the UV absorbance at 370 nm was measured by HPLC to establish proper calibration curve (Fig. S1) [47]. The HPLC peak areas were in good linear relationship (R ² value = 0.9956) with the Neu5Ac concentrations. The determined standard curve [HPLC peak area = 9917 [Neu5Ac] – 12,883] was used to calculate Neu5Ac concentration in the sample derived from S1 glycoprotein (Fig. S1B). Sialic acids of the S1 glycoprotein were first detached from the glycoprotein under mild acidic conditions, followed by DMB derivatization and HPLC-UV analysis. The HPLC results showed the DMB derivative of Neu5Ac from S1 glycoprotein had the same acquisition time (8.21 min) as that of the standard samples, and the peak was identified by ESI-MS [m/z: calculated, 425.14; observed, 426.47 (M + H)⁺]. The quantification result indicated that there were about 5–6 sialic acids on the S1 glycoprotein, which were subjected to mild periodate oxidation and conjugation with TLR7a-ONH₂. HPLC and MALDI-TOF mass spectrometry analysis of the final TLR7a-S1 conjugate indicated that about 6 TLR7a molecules covalently linked with the S1 protein, suggesting the TLR7a-ONH₂ conjugation on the oxidized sialic acids was highly efficient (Fig. S8).

To further evaluate the applicability and yield of the sialoglycan-mediated conjugation reaction, we employed the egg-yolk sialylglycopeptide (SGP) [44,48,49] as a model glycopeptide to monitor the oxidation and oxime ligation reactions by RP-HPLC analysis. The purified SGP was treated with 1 mM NaIO₄ at 4 °C and then conjugated with O-ethylhydroxylamine hydrochloride (Fig. S2). The HPLC and ESI-MS analysis demonstrated >95% yields of the periodate oxidation and conjugation reaction respectively (Fig. S2). Therefore, this periodate-mediated oxime ligation is a simple means to chemically conjugate adjuvant molecules onto glycoproteins with sialic acids, which is achieved by the stringent specificity of the mild periodate oxidation of sialic acids and an efficient oxime ligation that allows reaction at mild conditions with inexpensive and commercially available reagents.

3.3. Vaccination

The conjugate vaccine TLR7a-S1 was immunologically evaluated in 6- to 8-week-old female BALB/c mice (n = 5), and control groups of mice were injected with S1 alone (S1), S1 plus Alum adjuvant (Al/S1), and S1 mixed with TLR7a (TLR7a/S1), with the same dose of S1 (10 μg) and optimized doses of adjuvants (100 μL Alum, 3.6 μg TLR7a). All groups were injected subcutaneously (S.C.) on days 1, 15, and 29. Mouse sera were collected on days 14, 28, and 42 after vaccinations, and splenocytes were isolated from PBS treated and immunized mice on day 42 (Fig. 4A).

Fig. 4.

Mouse vaccination schedule and antibody responses induced by candidate vaccines. A) Mouse vaccination schedule. B) Anti-S1 IgG antibody titers induced by different vaccines on day 42. C) Ratio of IgG2a/IgG1 antibody titers on day 42. D) Anti-S1 IgG1 antibody titers induced by different vaccines on day 42. E) Anti-S1 IgG2a antibody titers induced by different vaccines on day 42. F) Anti-S1 IgG2b antibody titers induced by different vaccines on day 42. G) Anti-S1 IgG3 antibody titers induced by different vaccines on day 42. H) Neutralization titers (pVNT₅₀) against wild-type (WT) pseudovirus after the last immunization. Data are shown as the mean ± SEM of 5 mice per group and are representative of three separate experiments. Statistical significance was determined using one-way ANOVA with Dunn's multiple comparison test. I) Neutralization titers (pVNT₅₀) against variant pseudoviruses on day 42. Statistical significance was determined using unpaired two-tailed t-test. No significant difference: ns, P < 0.05: *, P < 0.01: **, P < 0.001: ***, P < 0.0001: ****.

3.4. Evaluation of immune cell activation in vitro

To evaluate the ability of TLR7a-S1 to initiate innate immunity, bone marrow-derived DCs (BMDCs) and spleen-derived lymphocytes were incubated with S1, TLR7a/S1 and TLR7a-S1, and the expression of secreted cytokines and costimulatory surface proteins were measured by flow cytometry and ELISA (Fig. 3A). The results indicated that TLR7a-S1 augmented the expression of costimulatory molecules CD80 (Fig. 3B) and CD86 (Fig. 3C) on BMDCs compared with TLR7a/S1, S1 and untreated controls. Besides, the IL-6 and IFN-γ secretion remained at low levels when BMDCs and splenocytes were incubated with S1 alone, but increased significantly with the treatment of TLR7a/S1 and TLR7a-S1 (Fig. 3D,E). Moreover, the levels of IL-6 and IFN-γ in the TLR7a-S1-treated group were 4.3- and 3.2-fold higher than that in the TLR7a/S1-treated group respectively, suggesting the enhanced efficacy of TLR7a to activate the immune cells. Therefore, conjugation of TLR7a onto S1 antigen boosts the BMDC-activating prowess of TLR7a in vitro, which might result from the improved recognition and uptake of clustered TLR7a by immune cells.

Fig. 3.

Evaluation of the immune cell activation in vitro. A) Schematic illustration of the immune cell activation experiment. B) Mean fluorescence intensity (MFI) and histogram profiles for the CD80 expression on CD11c+ DCs. C) MFI and histogram profiles for the CD86 expression on CD11c+ DCs. D) Secretion of IL-6 by the BMDCs after different treatments (n = 3). E) Secretion of IFN-γ by the splenocytes after different treatments (n = 3). BMDCs or splenocytes were stimulated for 18 h with TLR7a-S1 at a final concentration of 1 μM. S1 or TLR7a/S1 was added at the same molar quantity. Data are shown as the mean ± SEM of 3 mice per group. Statistical significance was determined using one-way ANOVA with Dunn's multiple comparison test. P < 0.05: *, P < 0.01: **, P < 0.001: ***, P < 0.0001: ****.

3.5. Evaluation of specific antibody response

To evaluate the humoral immunity induced by the conjugate vaccine, S1-specific antibody titers after the last immunization were measured by ELISA. Evaluation of anti-S1 antibody responses showed that the TLR7a-S1 group elicited the highest level of IgG antibody, with about 2- to 3-fold higher than that of S1-, Al/S1-, and TLR7a/S1 controls on day 42 (Fig. 4B). The traditional Alum adjuvant and physically mixed TLR7a did not significantly improve the immunogenicity of S1 protein, whereas the conjugated TLR7a effectively promoted antibody responses after the third vaccination. The TLR7a-S1 also showed higher efficacy in promoting antibody class switching from IgM to IgG, with a relatively low level of IgM elicited after the last vaccination (Fig. S3). In addition, a candidate vaccine should also favor the production of Th1 over Th2 responses, as the Th1-biased immune response has been shown to be related to enhanced protection against viral infection [50,51]. The assessment of IgG subtype distribution showed that the IgG subtype elicited by S1, Al/S1, and TLR7a/S1 groups was primarily IgG1 (Fig. 4C,D), with significantly lower titers of IgG2a and IgG2b compared to that of TLR7a-S1 group (Fig. 4E,F), and all the groups induced comparable amount of IgG3 subtype (Fig. 4G). As shown in Fig. 4C, the ratios of IgG2a/IgG1 of S1, Al/S1, and TLR7a/S1 groups were < 0.01 that indicated typical Th2-biased immune responses. In contrast, the TLR7a-S1 achieved a significantly higher ratio of IgG2a/IgG1 (>1) than other control groups, which was indicative of a Th1-biased antibody response. These results demonstrated that the conjugation of TLR7a and S1 not only enhanced the efficacy of antibody responses, but also promoted the humoral immunity to a Th1 bias. As a broad IgG subclass distribution is crucial for the protective immunity, TLR7a-S1 conjugate has great potential to be developed as an effective vaccine candidate against viral infections.

3.6. Pseudovirus neutralizing activity and cross-neutralization of variants

Strong viral neutralizing activity of antibodies is an important indicator of the immunological protection conferred by a vaccine candidate. The neutralization capacity of the vaccine-elicited antisera was evaluated against wild-type (WT) pseudotyped SARS-CoV-2 after the last immunization. As indicated in Fig. 4H, TLR7a-S1 vaccination were able to potently neutralize pseudovirus with the mean 50% pseudovirus neutralization titer (pVNT₅₀) of 6293 after the last boost, significantly higher than that of the TLR7a/S1 (mean pVNT₅₀ = 2437), Al/S1 (mean pVNT₅₀ = 3748), and S1 alone (mean pVNT₅₀ = 1717) groups. Compared to the S1 protein administered with Alum adjuvant, the TLR7a-S1 induced approximately 2-fold higher neutralizing antibody titer against the WT pseudovirus.

Furthermore, in the context of the emergence of new SARS-CoV-2 variants worldwide and widespread circulation of Omicron variant, it is necessary to evaluate the neutralization breadth of the antibodies elicited by the candidate vaccines [52]. Cross-neutralization assay against all VOCs (B.1.1.7/Α, B.1.351/Beta, P.1/Gamma, B.1.617.2/Delta and B.1.1.529/Omicron) was performed for the TLR7a-S1 and Al/S1 groups using different pseudoviruses. As shown in Fig. 4I, the neutralization induced by TLR7a-S1 and Al/S1 groups against the B.1.1.7 variant were comparable to that against the WT, respectively. But the neutralizing titers of TLR7a-S1 group against B.1.351, P.1, B.1.617.2, and B.1.1.529 were 2.6-, 1.6-, 2.3-, and 15.8-fold lower compared to the WT, respectively. The mean pVNT₅₀ of Al/S1 group against variants were lower than that of TLR7a-S1 group, with a similar neutralization pattern of the five VOCs. These results are in accordance with previous reported results of mRNA, adenovirus vector-based or protein-based vaccines [[53], [54], [55]]. Although neutralization against B.1.351/Beta, P.1/Gamma, B.1.617.2/Delta and B.1.1.529/Omicron suffered large reductions, especially the Omicron variant which has been reported with the greatest magnitude of evasion from antibody neutralization [55,56], the TLR7a-S1 conjugate vaccine still elicited effective neutralization against these variants, with more potent neutralizing activity than that of Al/S1 control. Therefore, the TLR7a-S1 conjugate vaccine might has the potential for protecting against SARS-CoV-2 and muted viruses as a vaccine candidate.

3.7. T-cell responses induced by vaccines using different adjuvants

To investigate the S1-specific cellular immunity, splenocytes were collected from immunized mice two weeks after the final vaccination and stimulated with overlapping peptide pool (spanning SARS-CoV-2-S1) for 18 h. The S1-specific cytokine-secreting cells were assessed by IFN-γ ELISPOT and intracellular cytokine staining (ICS) assays. As indicated in Fig. 5A, control groups from mice immunized with PBS or S1 did not show any IFN-γ spots. In contrast, the TLR7a-S1 vaccination elicited the highest level of IFN-γ spots, with ∼7- and ∼3-fold higher compared to that after injection with Al/S1 and TLR7a/S1, respectively. In the ICS assay, cytokine-producing CD4+ and CD8+ T cells of splenocytes were measured to further explore the S1-specific T-cell immunity. As shown in Fig. 5B, CD4+ T cells derived from TLR7a-S1-injected mice elicited higher frequency (1.75%) of IFN-γ and TNF-α cytokine-producing cells, compared with control groups of <1% frequencies. Similarly, the TLR7a-S1 vaccination induced the highest frequency (3.22%) of IFN-γ and TNF-α double positive CD8+ T cells, significantly higher than that of S1 (1.20%), Al/S1 (0.87%), and TLR7a/S1 (1.66%) controls (Fig. 5C). These results suggests that TLR7a-S1 provoked not only potent humoral specific antibodies, but also effective T-cell immune responses as a promising COVID-19 vaccine candidate.

Fig. 5.

Specific cytokine-producing T cell immune responses indeuced by different vaccines. Spleen cells were collected from vaccinated mice after the last immunization and stimulated with overlapping peptide pool (spanning SARS-CoV-2-S-S1) for 18 h. A) IFN-γ ELISpot assay. Images are representative ELISpot wells. B) Flow cytometry assay of IFN-γ+ TNF-α+ cells in CD4+ T cells. C) Flow cytometry assay of IFN-γ+ TNF-α+ cells in CD8+ T cells. D) Example flow cytometry plots for CD3+ CD4+ and CD3+ CD8+ T cells expressing both IFN-γ and TNF-α. Data are shown as the mean ± SEM of 5 mice per group, each sample being performed in triplicate. Statistical significance was determined using one-way ANOVA with Dunn's multiple comparison test. P < 0.05: *, P < 0.01: **, P < 0.001: ***, P < 0.0001: ****.

3.8. Safety evaluation of vaccines

To further address the biosafety of the vaccine candidates, the potential toxicity of the vaccines was evaluated in mice (n = 5). As shown in Fig. 6 , no adverse events were observed including ruffling of fur, behavioral changes, decreased appetite, etc. Histopathological assessment showed that no pathologic changes were observed in hearts, livers, spleens, lungs, kidneys, and brains of injected mice. These results showed that all vaccine candidates did not cause any obvious toxicity.

Fig. 6.

Safety evaluation of vaccines. Histological sections (H&E staining) of the major organs from immunized mice after the last vaccination, including heart, liver, spleen, lung, kidney and brain, with the PBS group serving as control. Scale bar = 100 μm.

4. Conclusions

In summary, we developed a new conjugating strategy to construct a biosafe adjuvant-protein conjugate vaccine, in which the sialylated S1 glycoprotein was selectively periodate-oxidized to specifically introduce the TLR7a at the sialic acid of S1. The TLR7a-S1 conjugate significantly enhanced the immune efficacy and could release the small-molecule adjuvant intracellularly for limited systematic toxicity. It is the first time to covalently attach adjuvant on protein antigen through sialoglycoconjugation. Compared with the unconjugated TLR7a/S1 mixture and S1 plus Alum adjuvant, the resulting TLR7a-S1 conjugate vaccine provoked significantly stronger activation of immune cells in vitro, and more potent humoral and cellular responses in vivo. Meanwhile, the conjugate vaccine induced Th1-biased immune responses desirable for anti-virus vaccines. Moreover, the antisera from TLR7a-S1-immunized mice induced potent neutralizing responses and effective cross-neutralization of different SARS-CoV-2 VOCs. These results suggest that the sialoglycoconjugation strategy provides a useful method to develop effective self-adjuvanting protein vaccines, and could be used to attach adjuvants on any glycoprotein antigen that contains terminal sialic acids. We expect that the strategy of adjuvant-glycoprotein conjugates will have more applications in prevention and therapeutic vaccines against viral infections and other diseases as well.

CRediT authorship contribution statement

Yu Wen: Investigation, Formal analysis, Writing – original draft. Ru-Yan Zhang: Investigation, Formal analysis. Jian Wang: Investigation, Formal analysis. Shi-Hao Zhou: Investigation. Xiao-Qian Peng: Investigation. Dong Ding: Investigation. Zhi-Ming Zhang: Investigation. Hua-Wei Wei: Resources. Jun Guo: Conceptualization, Writing – review & editing, Supervision, Funding acquisition.

Declaration of Competing Interest

The authors declare no competing financial interest.

Appendix A. Supplementary data

Supplementary material 1

Supplementary material 2

Data availability

Data will be made available on request.