Biophysical evaluation

The S glycoprotein or its domains are resolved by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) to infer the correct size, purity, and stability of the antigenic determinants. Further, Western blot using S-specific antibodies isolated from convalescent samples and size-exclusion chromatography (SEC), or size-exclusion chromatography multi-angle light scattering (SEC-MALS) for absolute molar mass and size determination are employed to remove candidates with the propensity to aggregate. The samples for such analyses are subunit vaccines by themselves and/or antigen extracted/secreted from model cell lines like HEK293 transfected/transduced with the viral strain (for an inactivated vaccine), vector-based or nucleic acid vaccine. SEC in combination with gel-shift assays is also used to confirm the multimeric display of antigens (like S trimer) and self-assembly of the nanoparticle vaccines [23, 25–32, 35, 38, 39, 42, 45]. The monodispersity of the subunit/nanoparticle and/or nucleic acid vaccines correlates with vaccine stability as determined by dynamic light scattering (DLS) and SEC-MALS to differentiate between a monomer, dimer, and oligomer [26, 29, 30, 41, 45]. In addition, nucleic acid vaccines are checked for their length, purity, and integrity of DNA/mRNA fragments before lipid encapsulation by microfluidic capillary or agarose gel electrophoresis, thus ensuring a contaminant-free (including protein, genomic DNA, nucleases, and enzymatic inhibitors) nucleic acid sample [39, 43–45]. Besides, the efficiency of encapsulating mRNA in lipid nanoparticles is a measure of mRNA vaccine stability usually interpreted in terms of particle size using DLS or ribogreen-based fluorescence assay [43–45]. Further, zeta potential measurements of mRNA vaccines give insight into the effect of varying pH on the vaccine stability or/and endosomal release of mRNA [45] and confocal imaging to study vaccine localization in endoplasmic reticulum/Golgi [39].

Functional evaluation

For a stable and effective vaccine, the antigenic determinants must have a physiologically relevant configuration to drive an immune response. The conformation of S glycoprotein expressed/displayed by vaccine candidates is studied structurally to confirm the prefusion state of S glycoprotein using electron microscopy (EM) and/or negative staining cryo-EM [23, 25, 26, 29–32, 39, 45]. The conformational flexibility (like up/down state of RBD) and multimeric display (like S trimer) of S glycoprotein mimics that of the native virus. Further, the correctly folded S glycoprotein on/expressed by vaccine preparations should bind hACE2 with high-affinity like SARS-CoV-2, which is studied using surface plasmon resonance (SPR) or biolayer interferometry (BLI) [26, 27, 29, 30, 32, 35, 39]. Further, the vaccine expressed S glycoprotein or its domains with correct folding and conformation alike native virus, should bind to the SARS-CoV-2 nAbs. SARS-CoV-2 nAbs from convalescent serum or a panel of recombinant nAbs help to evaluate functionally active S glycoprotein expressed/displayed by vaccine candidates [23, 29–31, 39, 41]. Further, N-linked glycans on S glycoprotein have an important role proper folding of trimers and modulating interaction with host proteases and nAbs [4, 67–69]. Analysis of site-specific N- and O-linked glycosylation on expressed S glycoprotein and glycoform distribution for subunit vaccines is analysed using glycoproteomics approach by recording mass spectra [27, 28, 30]. Another important aspect is the COVID-19 vaccine biodistribution upon administration which is a quantification of S glycoprotein in various organs and tissues of vaccinated animals [45]. Further, immune response against vector/nanoparticle can be detrimental to the vaccine stability upon administration. It must be characterised by test immunisations in a small group of animals [37].

Evaluation of inactivated vaccines

Central to an effective and broadly neutralising immune response from an inactivated whole virus vaccine is the shortlisting of most prevalent SARS-CoV-2 viral strain(s) in a phylogeny-based analysis using all available circulating genomes in Global Initiative on Sharing All Influenza Data (GISAID, https://www.gisaid.org/) [23–25]. The shortlisted viral strains are amplified in Vero cells for 7–10 passages to monitor the proliferation efficiency, high rate of infection, and genetic stability of the progeny for several generations by high throughput next-generation sequencing (NGS) methods [23–25]. This process ensures that no amino acid change should occur in the candidate strain compared to the parent and seed strain. Suitable candidate strain is assigned a clade using GISAID classification. The beta-propiolactone-mediated viral inactivation process is standardised. Complete elimination of infectivity and repeatability of the inactivation process for candidate strains is desirable [23–25]. Further, contaminating viruses or mycoplasma in the vaccine preparations are tested by NGS and/or infecting cells to isolate/characterise contaminating microorganisms [23–25].

Animal models for vaccine efficacy study

The WHO Research & Development Blueprint Team established a multi-location research consortium for COVID-19 modeling in animals, the initiative known as World Health Organization-COVID-19 Disease Modelling (WHO-COM) [70]. An overview of COVID-19 animal models established under WHO-COM initiative is available elsewhere [55]. Here, we only focus on mouse and NHP models useful for vaccine evaluation. Mice are the most sought-after animal models, which can be housed in a smaller space to make a relatively larger experimental group for meaningful statistical analysis of therapeutic efficacy. Although mice do not get efficiently infected by SARS-CoV2 due to a low-affinity interaction between S glycoprotein and mouse ACE2 (mACE2), they are the primary choice for COVID-19 vaccine evaluation [23, 25–32, 34, 35, 37, 39–41, 44, 45]. Mice susceptible to SARS-CoV-2 infection and mimicking severe COVID-19 could be more beneficial in such evaluation studies. Several approaches can make mice more susceptible to SARS-CoV-2 infection like (1) mutating SARS-CoV-2 RBD for enhanced binding to mACE2 [71]; (2) forced passaging of SARS-COV-2 in mouse lung tissue to allow enrichment of mutant virus capable of causing severe COVID-19 [72]; (3) infecting mouse with engineered adenovirus (Ad5-hACE2) capable of transducing hACE2 gene to sensitise for SARS-CoV-2 infection [54, 73]; (4) genetic modification of mouse to express hACE2 using tissue-specific promoter (K18-hACE2), universal promoter (CMV enhancer and chicken β-actin promoter), and endogenous mouse ACE2 promoter [73–75]; and (5) developing humanised mouse with hACE2 gene supporting SARS-CoV-2 replication in respiratory and brain tissue [76]. Although these strategies allow enhanced viral replication in mouse tissue, COVID-19 symptoms are still mild to moderate, similar to the wild-type animals.

Before ascertaining vaccine efficacy in humans, protective ability and adverse effects are evaluated in NHPs. The majority of COVID-19 vaccine candidates reviewed here have been evaluated in rhesus macaques (Macaca mulatta), cynomolgus macaques (Macaca fascicularis) and African green monkeys (Chlorocibus aethiops) [23–28, 31–33, 35–39, 41–43, 53, 56, 62, 64, 66]. During COVID-19 modeling studies, NHPs exposed to SARS-CoV-2 through either mucosal, intrabronchial, or aerosol exposure showed early radiographic abnormalities in lungs, high level of viral replication in both upper and lower respiratory tract for up to 2 weeks, virus shedding in respiratory and gastrointestinal tracts, and pathological features associated with viral pneumonia [53, 55, 56, 62, 64]. However, clinical disease in these models was still mild, and infection usually self-resolved within 11–14 days. Importantly, older macaques typically presents high viral load in the lungs, enhanced viral shedding, and more prominent radiological and histopathological changes [62]. Further, the SARS-CoV-2 challenge elicited natural protective immune responses, which offered robust protection against the virus rechallenge [53]. Alike mouse models, NHPs also do not recapitulate COVID-19 disease as in humans, although pathological manifestations have still allowed vaccine evaluation with the emergence of meaningful data [23, 24, 26–28, 31–33, 35–39, 42, 43, 66].

COVID-19 is usually a mild to moderate disease [77, 78]. However, lymphopenia and higher levels of the pro-inflammatory cytokine among older patients can cause severe disease. These conditions may lead to uncontrolled viral infection, severe pneumonia, acute respiratory distress syndrome (ARDS), septic shock, and multiple organ failure [77–79]. Further, clinical data show a high fatality rate among males (7.3%) compared to females (4.4%) due to COVID-19 [80, 81]. Pre-existing comorbidities like hypertension (16%), cardiovascular disease (12.11%), and diabetes (7.87%) among hospitalized COVID-19 patients can further predispose individuals to a greater risk of death from COVID-19 [82, 83]. Therefore, vaccine candidates must be evaluated in animal models closely mimicking the physiology linked to severe COVID-19 related fatality. Most vaccine candidates referred here were evaluated in normal mouse or NHP models, which may not be very useful in knowing vaccine effectiveness in the individuals with age-related dysfunctions or comorbidities.

COVID-19 vaccine-induced S1/RBD binding serum antibodies

An effective COVID-19 vaccine must trigger a robust serum antibody response against RBD and a complete seroconversion [23–45, 54, 66]. Therefore, all COVID-19 vaccine strategies strive to elicit high titers of RBD binding serum antibodies. The first evaluation strategy determines the immunogenic strength of COVID-19 vaccine candidates based on two prime indicators: (1) serum immunoglobulin G (IgG)/IgM/IgA titers (represented as arithmetic and/or geometric mean titers (GMTs) which are an estimate of S1/RBD-specific antibody class/IgG subclass/isotype in serially diluted serum samples [23–45, 54, 66] and (2) seroconversion rate implicating the percentage of vaccinated animals, which S1/RBD-specific serum antibodies [23–45]. The most prevalent method employed to determine antibody titers in serum samples is an enzyme-linked immunosorbent assay (ELISA) using IgG/IgM/IgA standards; serum antibody titers can also be determined by flow cytometry using anti-IgG/IgM/IgA beads. Among referred COVID-19 vaccine candidates, serum IgG titers were estimated generally every week after prime and boost vaccinations, and at an interval of two days post-virus infection. Both in mice (BALB/c, C57BL/6J, B6C3F1/J strains) and monkeys, serum IgG titers against S1/RBD and seroconversion rate depend on the vaccine dose and immunisation program (“prime only” and “booster”) without exception [23–45]. A booster dose in most cases further expands the serum IgG titers against S1/RBD and is responsible for complete seroconversion. During SARS-CoV-2 infection, the immune system also generates IgM response which peaks within three weeks of infection and protects against the virus in nasal tissue [84–87]. COVID-19 vaccines (RBD subunit, S-Fer nanoparticle, S-I53-50 NP, ChAdOx1nCoV19, and Ad26-S.PP) show elicitation of IgM antibodies recognising RBD in serum of BALB/c and CD1 mice and NHPs in second or third week after prime immunisation [27, 29, 31, 37, 38]. Besides, RBD-specific secretory IgA response generated by S-I53-50 NP, ChAd-SARS-CoV-2-S, and MVA/S vaccinations may be responsible for protecting the mucosal epithelia in respiratory, intestinal, and urogenital tracts against SARS-CoV-2 invasion [31, 33–35]. SARS-CoV-2 specific IgA remains neutralising in the saliva of infected subjects [88].

Most COVID-19 vaccine candidates referred to in this review were administered through the intramuscular (IM) route, which predominantly induces humoral immunity (IgG-directed) and limited mucosal immunity (IgA-directed) against the antigenic composition [89]. It can compromise the recognition/inhibition of SARS-CoV-2 in the nasopharyngeal tract, which is the initial site of virus contact. Therefore, IM administration might not provide sterilizing immunity against SARS-CoV-2 [33, 34, 89–91]. The potential of few vaccine candidates in generating mucosal immunity via intranasal (IN) administration in preclinical studies is known. Vaccine candidates like ChAdOx1 nCoV-19/AZD1222, ChAd-SARS-CoV-2-S, and lentiviral vector (LV) encoding full length, membrane anchored S [LV: :S_FL] via IN administration enhances protection against SARS-CoV-2 in the upper respiratory tract [33, 34, 90, 91]. Besides eliciting systemic immunity, mucosal IgA response from IN-administered vaccines (ChAdOx1 nCoV-19/AZD1222 and ChAd-SARS-CoV-2-S) reduces virus shedding in the nasopharynx and can halt further spread [33, 34, 91]. Intranasal administration of booster dose can mediate a robust expansion of systemic and mucosal immunity in mice primed via the intraperitoneal (IP) route [90]. It may imply that additional IN administration in IM/IP vaccinated individuals can expand protective immunity or even provide sterilizing immunity. These studies were performed in hACE2 transduced mice, golden hamsters, and rhesus macaques [33, 34, 90, 91].

Previously, severe symptoms of vaccine-associated enhanced respiratory disease (VAERD) were present in measles and respiratory syncytial virus infected subjects vaccinated with a non-protective formulation [92]. Of some concern in the current scenario is that SARS-CoV vaccines like vaccinia virus-based Lc16mOrVV-N either alone or in combination with Lc16mOrVV-S, modified Vaccinia virus-based ADS-MVA, and double-inactivated vaccine (DIV) induce VAERD-like pathology (T_h2 dominated inflammatory infiltrates and increased numbers of eosinophils) in animal models [93–95]. Therefore, the second evaluation strategy focuses on the physiological nature of the COVID-19 vaccine-induced serum IgG response concerning possible VAERD-like pathology. These involve analysing the (1) serum IgG isotypes; (2) Th cell subset; (3) eosinophil counts; (4) proinflammatory cytokines; or (5) anti-inflammatory cytokines after vaccination [92]. Non-neutralizing fraction of serum IgG can lead to ADE [92]. Serum IgG isotyping in COVID-19 vaccinated animals suggests a VAERD free immune response characterized by a balanced T_h1/T_h2 or slightly T_h1 biased responses even with higher dose as inferred by a higher ratio of RBD binding serum IgG2a to IgG1 [41, 43, 96].

COVID-19 vaccine-induced nAbs in serum

Elicitation of nAbs upon SARS-CoV-2 exposure can protect from reinfection. SARS-CoV-2 nAbs preferentially recognise the epitopes within S1/RBD, [4, 9, 11, 23–35, 37–46, 48, 54, 66] while antibodies binding to the S2 domain and/or NC are non-protective [25, 27]. The nAb response against SARS-CoV-2 is an outcome of multiple clinical/immunological components which must be carefully evaluated. These include (1) optimisation of the vaccine dose since nAb titers are known to increase with a higher vaccine dose [23, 25, 27, 28, 39, 41, 43–45]. However, higher dose of BBV152 resulted in weaker neutralising response which was capable of resolving the lung pathology completely at the experimental endpoint [24]. It is difficult to interpret outcome as both low and high doses were formulated using the same adjuvant. Lower dose induced stronger IL-8 response which seems to be responsible for immune cell recruitment and vaccine efficacy [24]. However no such dose effect on neutralising antibody titers was evident in BBIBP-CorV inactivated vaccine [23]; (2) administration of booster dose as it has been widely proven that booster dose increases nAb titers and render a clear therapeutic benefit with complete seroconversion [23–35, 37–45]. In a particular case, a third vaccine dose resulted in maximal humoral response in animals [23]; (3) dosing interval as increasing the gap between a primary and booster dose can also enhance nAb titers in animals [25]; (4) presence of RBD binding antibodies since a strong correlation exists between the titers of anti-RBD binders and serum neutralising activity in comparison to S2 or N-specific antibodies, affirming the role of RBD in humoral immune response [23, 25]; (5) a higher ratio of “neutralising” to “binding only” RBD serum antibodies which lowers the risk of ADE upon vaccination, which has been previously reported in related infections caused by other members of the Coronaviridae family [13, 45, 97, 98]; and (6) immune effector responses which is the ability of nAbs to activate effector responses via FcγR2A-3, FcγR2A-1, IgM, and complement engagement to further enhance anti-SARS-CoV-2 immunity [31, 35, 38, 42]. The neutralising activity increases via activation of several effectors functions including antibody-dependent complement deposition (ADCD), antibody-dependent cellular phagocytosis (ADCP), antibody-dependent neutrophil phagocytosis (ADNP), and antibody-dependent natural killer cell activation (ADNKA) [31, 35, 38, 42, 99]. Therefore, all these parameters are crucial in finding the most effective vaccines.

The protective efficacy of nAbs is validated on (1) SARS-CoV-2 pseudo-typed virus (PSV) [27–32, 35, 38, 40, 42, 44, 45] or/and (2) live wild type (WT)/engineered SARS-CoV-2 isolates [23–27, 30–35, 37–39, 41–45, 54] produced/amplified in HEK293T and Vero-E6/CCL-81 cells respectively.

COVID-19 vaccine-induced cellular response

Cellular immune components resulting from COVID-19 vaccination can provide long-term immunity against SARS-CoV-2 infection. Upon viral exposure, lung epithelial cells, alveolar macrophages, and neutrophils activate innate immunity, which in turn triggers adaptive immunity against the virus via activation of B lymphocytes (B-cell), helper T-cells (CD4⁺), and cytotoxic T-cells (CD8⁺) [18–22, 25, 30, 32, 34, 38, 39]. COVID-19 vaccines are shown to generate an immune response which is T_h1 biased with neutralising serum IgGs [18–22, 26, 35, 40, 41, 44], enhanced drainage of B-cells into the lymphoid compartments [39], activation of T follicular helper (T_fh) cells, long-lived plasma cells (LLPCs), memory B-cells (MBPs), germinal center (GC) maturation in secondary lymphoid tissues [26, 32, 34, 35, 39, 43, 66], and S-specific INF-γ⁺/IL-2⁺/TNF-α⁺ CD4⁺/CD8⁺ T-cells [23, 24, 26–28, 31–35, 37–45].

Anamnestic immune response after SARS-CoV-2 challenge

The anamnestic response is a natural reaction by the host against infecting pathogen (SARS-CoV-2 in this case) involving the memory compartments such as circulating memory B/T-cells capable of protecting from infection and/or disease [53, 54, 100, 102]. Such anamnestic responses can lead to low titers of binding/nAbs and CD4⁺/CD8⁺ T cells specific to S1, S2, NC, and non-structural (NS6, NS7a, and NS8) proteins of SARS-CoV2 within a short period of virus challenge [35, 38, 43]. Anamnestic responses can occur in sham control/unvaccinated and a minority of vaccinated animals after the live SARS-CoV-2 challenge [35, 38, 43]. Vaccinated animals (e.g., with S-I53-50NP, MVA/S, Ad26-S.PP, and mRNA1273) show no anamnestic expansion of nAbs and only low titers of S2 and NC specific CD4⁺/CD8⁺ T-cells for a shorter duration post-challenge compared to control animals [35, 38, 43]. Lack of NC-specific CD8⁺ T-cells expansion and nAbs in vaccinated animals suggests the nonavailability of NC protein produced by actively replicating virus [35]. A much lower viral replication in vaccinated animals after SARS-CoV-2 challenge points towards a protective vaccine-induced immune response. A low level of anamnestic response in vaccinated animals after the SARS-CoV-2 challenge indicates reduced viral replication [31, 35, 38, 43]. Weak anamnestic response after AD26-S.PP vaccination correlated with the immunophenotype of BAL cells indicating vaccine effectiveness [38]. An inefficient or even non sterilising vaccine candidate can also give rise to anamnestic response, and vaccine non-responders may also mount such a response [35, 38, 42, 43].

Therefore, anamnestic responses must be differentiated from a vaccine-induced immune response by accurately profiling the precise components of the SARS-CoV-2 immune response. Such analysis incorporates the determination of neutralising antibody titers against S1, S2, S, and NC protein in the serum from control and vaccinated animals before and after the SARS-CoV-2 challenge. Further, the neutralising antibody titers must be evaluated at frequent intervals after the virus challenge to understand the kinetic of anamnestic antibody response [31, 43]. For example, antibody levels in mRNA1273 vaccinated animals after the SARS-CoV-2 challenge remained stable while control animals showed increase within two weeks after the challenge [43]. Similarly, in S-I53-50NP vaccinated animals, nAbs continued to wane following SARS-CoV-2 challenge suggesting vaccine-induced immunity compared to controls [31]. Anamnestic expansion of CD8⁺/CD4⁺ T-cells recognising S1, S2, S, and NC in the spleen, lymph nodes, and BAL fluid of vaccinated and control animals after SARS-CoV-2 challenge must be accurately determined [31, 35, 38, 43]. As mentioned, anamnestic antibody response can correlate with the cellular subsets in BAL fluid; such correlations may help in unravelling the immune protective mechanism of vaccine candidates [38]. Further, a correlation between anamnestic CD8⁺/CD4⁺ T cells response and viral subgenomic RNA (sgRNA) may suggest effective vaccine candidates [35].

COVID-19 vaccine-induced protection/pulmonary pathology

A less severe lung pathology is an accepted indicator of vaccine efficacy. Most COVID-19 vaccines result in lower pulmonary pathological score in mice and NHP models of SARS-CoV-2 challenge. Pulmonary pathological score is the cumulative outcome of (1) thoracic/chest radiograph with evidence of pneumonia; (2) the number of upper, middle, lower lobes of lung affected; (3) extent of tissue lesions by necroscopy; (4) extent of bronchointerstitial pneumonia; (5) type 2 pneumocyte hyperplasia; (6) peribronchiolar hyperplasia; (7) alveolar septal thickening; (8) alveolar macrophage infiltration; (9) haemorrhages; (10) pulmonary edema (increased lung: body weight ratio) and fibrin deposits; and (11) formation of perivascular cuffs due to infiltration of lymphocytes [23–28, 31–35, 37, 39, 43–45]. A lower lung pathology score correlates with disease protection [23–28, 31–35, 37, 42–45]. The protective outcome of COVID-19 vaccine candidates correlates with significantly lower viral load in lung tissue and/or BAL fluid as measured by quantitative reverse transcription polymerase chain reaction (qRT-PCR) using primers specific for SARS-CoV-2 genomic RNA (gRNA; indicative of inoculated virus) and sgRNA; indicative of replicating virus) transcripts [24, 27, 28, 31–35, 38, 39, 42, 43, 45]. Lower sgRNA transcripts indicate incapability/reduced capability of SARS-CoV-2 replication and protection from COVID-19. Protective immunity can also be assessed by quantifying sgRNA in nasal turbinates/oropharyngeal swabs [35–37, 39, 43, 44]. Further, lung histopathology score and pulmonary viral load both correlate with vaccine-induced protection from COVID-19 [24–28, 31–35, 42–45]. However, discrepancies also exist, for example, in a study involving Macaca mullata, vaccination completely protected animals from lower lung infection with no detectable SARS-CoV-2 transcripts in BAL fluid but similar lung histopathology as in control animals [39]. Besides, in a few studies, extrapulmonary tissues from the heart, spleen, brain, lymphoid organs, GI tract, testes, etc., are also used for histopathological analysis and viral load quantification in vaccinated and control animals [24, 32, 34, 35, 37].

COVID-19 vaccine-induced protection/pulmonary immunopathology

SARS-CoV-2 infection alters the balance of innate and adaptive immunity in pulmonary tissues leading to a compromised immune status in controls compared to vaccinated animals [35, 43]. The balance between the kinetics of innate and adaptive immune responses in pulmonary tissues of vaccinated animals correlate with the protection from COVID-19 as inferred by lower viral load [35, 43]. The immunopathology status of the pulmonary tissue is analysed by monitoring various parameters such as (1) increased frequency of immune cells in BAL fluid and in the GC of draining lymph nodes of vaccinated animals suggests unstable kinetics in control animals [35]; (2) increased cytokine/chemokine levels (transcriptomics analysis by singe-cell RNA seq of IFN-stimulated genes (ISGs) in lung macrophages of BAL fluid in controls compared to vaccinated animals indicates lesser infection [35, 107]; (3) immune cell infiltration markers, for example, analysis of population shift in macrophage receptor with collagenous structure (MARCO⁺) resident alveolar macrophages compared to MACRO⁻ interstitial macrophages/infiltratory monocytes in vaccinated animals suggest reduced recruitment of immune cells in pulmonary tissues and correlates with protection/reduced viral load [35, 107]; (4) generation of local lymphoid tissues as vaccines (for example, MVA/S) can also induce the formation of local lymphoid tissue like inducible bronchus-associated lymphoid tissue (iBALT) [35]. Bronchus-associated lymphoid tissues (BALTs) like areas induced by natural infection or vaccination are in direct contact with mucosal epithelium and involved in lymphocyte priming and B/T-cell proliferation to respond to infection in lung and lower respiratory tract [98, 108]. These can help in the expansion of lung immunity following virus infection [35, 97, 98, 107, 109]. Extent of BALT induction may depend on antigenic composition of the vaccine (for example, full length S in MVA/S was more effective in BALT induction compared to S1-subunit in MVA/S1) [35]. Such iBALT formation is evident upon infection with respiratory viruses in the absence of peripheral lymphoid organs and protein cage nanoparticles instillation in lungs [98, 107, 109].