Stability Under Different Stress Treatments of a Virus-like Particle Vaccine Based on a Recombinant Hepatitis E Vaccine

TL;DR

This study evaluates the stability of a hepatitis E virus-like particle vaccine under various stress conditions and identifies early signs of instability.

Contribution

The study introduces charge heterogeneity as an early and sensitive marker for detecting instability in virus-like particle vaccines.

Findings

Changes in charge heterogeneity were the earliest detectable sign of instability in the vaccine under thermal stress.

VLPs remained stable at 25 °C for 28 days but aggregated under high-intensity shaking after 3–6 days.

Freeze–thaw cycles altered charge variants but did not compromise structural integrity or function.

Abstract

Background/Objectives: Virus-like particles (VLPs) are effective vaccine platforms but are susceptible to degradation, which compromises stability and immunogenicity. A key challenge is the lack of sensitive early indicators of instability. This study aimed to systematically evaluate the stability of an aluminum-free recombinant hepatitis E virus VLP vaccine under various stresses and identify predictive markers of instability. Methods: The VLP vaccine was subjected to thermal stress (4 °C, 25 °C, 37 °C, 56 °C for up to 28 d), repeated freeze–thaw cycles (up to 30 cycles), and mechanical agitation (orbital shaking at 100 and 300 rpm for up to 12 d). Stability was assessed using a multi-parameter panel monitoring critical quality attributes: conformational and colloidal stability, formation of high-molecular-weight species, mean particle size, polydispersity index, charge heterogeneity, and antigen content. Results: Changes in charge heterogeneity were the earliest indicator of instability, detectable within 3 days at 25 °C, 8 h at 37 °C, and 4 h at 56 °C, preceding losses in structural integrity or antigen-binding function. The VLPs remained stable at 25 °C for 28 d. Freeze–thaw cycles induced a basic shift in charge variants without compromising structure or function, while high-intensity shaking (300 rpm) caused aggregation after 3–6 d. The effects of common excipients were also characterized. Conclusions: Charge-variant analysis serves as a sensitive and predictive marker for VLP vaccine instability. The study delineates the distinct impacts of different stress factors and provides critical data for optimizing formulation design and storage strategies to enhance VLP vaccine stability.

Figures

Full text

1. Introduction

Hepatitis E virus (HEV) infection is a significant global public health issue [1], particularly in countries with limited public health resources [2,3]. Vaccination is one of the most effective strategies for preventing HEV-associated morbidity and mortality [1,4]. Virus-like particle (VLP) vaccines, which mimic the structure of the native virus, exhibit excellent immunogenicity. These vaccines have therefore become a major focus in the development of novel immunization strategies. However, the complex multi-level structure of VLPs (such as polyhedral assembly and surface epitope conformational dependence) presents significant challenges for predicting their physicochemical stability [5].

The stability of vaccine formulations under different environmental and processing conditions is critical for ensuring their efficacy during storage, transportation, and use [6], particularly for temperature-sensitive VLP-based HEV vaccines, which are susceptible to inactivation due to temperature fluctuations or improper handling during storage and transit, leading to diminished immunogenicity or even a complete loss of efficacy [7,8]. Additionally, issues related to physical stability (such as aggregation and precipitation) and chemical stability (including protein degradation and oxidation) directly impact vaccine safety and potency [9]. Current strategies primarily rely on cold-chain systems to maintain vaccine activity, but these approaches are costly and difficult to implement globally, especially in resource-limited areas [6,10]. Therefore, conducting comprehensive stability studies and developing robust formulations for VLP vaccines are of paramount importance.

Given the broad population coverage of vaccination, any stability issues can be amplified in large-scale immunization programs. It is essential to systematically evaluate antigen stability early in the development process [11]. Stability studies are conducted throughout the entire lifecycle of biological products, requiring comprehensive evaluation of key factors affecting product quality and efficacy at different stages of the lifecycle [12]. However, most existing stability studies focus on late-stage changes such as aggregation or loss of potency, which fail to provide early warning. For structurally complex VLP vaccines, sensitive and forward-looking CQAs capable of predicting stability changes are still lacking [13]. Based on this, this study proposes a core hypothesis: changes in charge heterogeneity precede detectable structural and functional degradation. Guided by this hypothesis, we can systematically evaluate vaccine stability by monitoring multiple critical quality parameters, aiming to identify sensitive early indicators.

This study focuses on the HEV p239 VLP, a recombinant antigen derived from the pORF2 capsid protein (amino acids 368–606) that can self-assemble into particles approximately 20–30 nm in diameter [14] and serves as the core component of the commercially available vaccine Hecolin^®^ [15]. These VLPs lack regular icosahedral symmetry and exhibit significant structural heterogeneity. The resulting VLPs are porous spherical particles formed through non-covalent interactions among approximately 80–100 monomers, with an apparent molecular weight of around 2600 kDa [15,16].

In addition to identifying early instability indicators, optimizing formulation excipients represents another critical pathway to enhancing vaccine stability. Excipients in the formulation, such as sugars, PS 80, and human serum albumin (HSA), may also significantly affect the stability of the vaccine. In vaccine design, adjuvants such as aluminum salts are commonly used to enhance immune responses, but they may induce local or systemic toxic reactions [17,18]. Therefore, when developing aluminum-free or low-aluminum formulations, it is particularly important to optimize the excipient system to maintain or even improve antigen stability and immunogenicity. These excipients can not only directly maintain the physicochemical stability of the antigen but may also indirectly improve the safety profile of the vaccine by reducing aggregate formation and mitigating adjuvant-related inflammatory responses [19,20]. This study aimed to systematically evaluate the effects of various stabilizers at different concentrations on the conformational integrity, particle dispersion, and thermal stability of HEV p239 VLPs.

This study used the core antigen p239 of the hepatitis E vaccine as a model and employed a recombinant HEV vaccine concentrate (3.6 mg/mL, aluminum-free, phosphate-buffered system) to systematically investigate the effects of various stress conditions on HEV vaccine concentrate stability. The vaccines were subjected to a range of stress conditions, including thermal stress (4 °C, 25 °C, 37 °C, and 56 °C) for simulating long-term storage and accelerated degradation, freeze–thaw cycles to mimic transportation challenges, mechanical stress (orbital shaking at 100 and 300 rpm) for simulating handling disturbances, and formulation optimization with various stabilizers, such as HSA, trehalose, sucrose, and PS 80. These tests were conducted with the aim of comprehensively evaluating and enhancing vaccine stability. The stability of the VLP vaccine before and after accelerated stress was assessed using a suite of analytical techniques: differential scanning fluorimetry (DSF), static light scattering (SLS), dynamic light scattering (DLS), size-exclusion chromatography coupled with multi-angle light scattering (SEC-MALS), isoelectric focusing electrophoresis (icIEF), and enzyme-linked immunosorbent assay (ELISA). This comprehensive analysis evaluated changes in conformational stability, colloidal stability, particle size distribution, formation of HMWS, charge isoforms, and antigen content. These findings offer data support for cold chain transportation and storage conditions and provide guidance for optimizing formulation strategies to enhance vaccine shelf life and performance.

2. Results

2.1. The Effect of Thermal Stress on Antigen Stability

To systematically evaluate the impact of temperature, accelerated stability studies were conducted at 4 °C, 25 °C, 37 °C, and 56 °C for up to 28 days. A repeated measures two-way ANOVA was employed to systematically assess the effects of storage temperature and time. The statistical analysis confirmed that both temperature and time exerted highly significant main effects on the CQAs of the VLP vaccine (p < 0.0001), and a significant interaction was identified between the two factors (p < 0.0001). This indicates that the trend of change in the CQAs is strongly dependent on the storage temperature, revealing a pattern where alterations in charge heterogeneity preceded structural and functional degradation (Figure 1).

At 4 °C, the isoelectric point and relative content of charge variants in the sample changed by no more than 1.5% over 28 days, indicating no significant alteration of the sample at this temperature (Figure 1A). At 25 °C, over 28 days, the relative content of basic variants showed a decreasing trend (from 12.07% to 4.22%), and a shift exceeding 1.5% in the relative content of charge variants was observed by day 3, with the isoelectric point trending toward the acidic direction (Figure 1B). Under 37 °C conditions, the relative content of basic variants decreased rapidly within 7 days (from 11.59% to 4.15%), and the basic variants disappeared after 14 days, while new acidic variants emerged. A shift exceeding 1.5% in the relative content of charge variants was observed at 8 h (Figure 1C). At 56 °C, a significant acidic shift in the isoelectric point (pI below 5.2) was already evident after just 4 h of storage (Figure 1D). The acidic shift emerged earliest at 56 °C (4 h), followed by 37 °C (8 h), and 25 °C (3 d). Correspondingly, mass spectrometry (Figure S1) suggested the occurrence of deamidation.

Colloidal and conformational stability, as indicated by T_m_ and T_agg_, remained relatively stable at 4 °C and 25 °C. Specifically, T_m_ showed only a slight decrease (less than 1 °C) after 7 days and beyond, from 72.2 °C to 71.9 °C at 4 °C and from 72.2 °C to 71.2 °C at 25 °C. Meanwhile, Tagg exhibited no significant changes at 4 °C (from 66.6 °C to 66.0 °C) and 25 °C (from 66.6 °C to 65.10 °C). At 37 °C, T_m_ significantly decreased from 72.2 °C to 70.5 °C after 7 days and beyond. At 56 °C, both T_m_ and T_agg_ significantly decreased to 67 °C at 1 day. Notably, falling beyond the detection limit within 7 days at 56 °C (Figure 2A). DLS showed time-dependent increases in mean particle size and PDI at 25 °C and 37 °C, indicating aggregation; at 56 °C, aggregation was rapid, with the particle size exceeding the DLS detection range after 7 days (Figure 2D–G). SEC-MALS data further confirmed the progression of aggregation—the approximately 95-mer state was maintained at 4 °C for 28 days. It then gradually increased to approximately 152-mers at 25 °C and then doubled rapidly to approximately 190-mers within 3 days at 37 °C, reaching approximately 342-mers after 28 days and exceeded 380-mers after just 1 day at 56 °C (Figure 2B,H–K).

Finally, antigen-binding function (measured by ELISA) was completely lost after day 3 at 56 °C but remained stable at 4 °C, 25 °C, and 37 °C over 28 days (Figure 2C). Pearson correlation analysis further revealed a significant negative correlation between HMWS and antigen-binding activity (p = 0.02), supporting HMWS increase as a predictive indicator for functional decline. In contrast, no significant correlation was observed between charge variants and antigen-binding function (p > 0.05). These results indicate that the timepoint of functional loss was notably later than the significant changes in charge variants and the evident increase in HMWS. This establishes icIEF changes as an early warning signal for subsequent aggregation and functional loss. Specifically, at 37 °C, although shifts in charge variants and an increase in HMWS were observable, antigen-binding activity remained stable throughout the 28-day period. This indicates that at this temperature, the chemical modifications and physical aggregation had not yet progressed to an extent sufficient to disrupt the structure of critical epitopes. Together, these findings support that changes in charge variants or HMWS are necessary but not sufficient conditions for functional loss, their ultimate impact depends on both the specific sites affected and the extent of modification or aggregation.

2.2. Prediction of Long-Term Stability

Based on the growth kinetics of HMWS under thermal stress, an Arrhenius model for the aggregation process of the HEV VLP vaccine was successfully established to predict its long-term stability at 4 °C.

The HMWS growth data at 25 °C, 37 °C, and 56 °C were fitted to a first-order kinetic model, yielding rate constants (k) (Figure 3A). Arrhenius analysis (Figure 3B) revealed a strong linear relationship between ln(k) and 1/T, described by the regression equation:

$$[eqn]$$

The apparent activation energy ( $[eqn]$ ) calculated from this equation was 49.6 kJ/mol ( $[eqn]$ ).

Using the Arrhenius relationship established above, the degradation rate constant at 4 °C was extrapolated as 2.74 × 10^−6^ h^−1^. Based on the stability study, the determination of the endpoint was made in accordance with the analysis of significant changes in the molecular weight of high-molecular-weight substances (HMWS). Under different temperature conditions, the time points at which the molecular weight first showed statistically significant changes were 28 days at 25 °C, 3 days at 37 °C, and 8 h at 56 °C. Accordingly, the molecular weight data from the preceding time points before these significant changes, namely, 21 days at 25 °C, 1 day at 37 °C, and 4 h at 56 °C were selected, and their average value was calculated to be 3356.4 kDa, while the initial molecular weight of the sample was 2513.5 kDa. With a defined failure criterion, where the HMWS content must not exceed 133% [21,22] of the initial value, corresponding to a molecular weight of 3356 kDa or less. The predicted theoretical shelf-life would exceed two years (approximately 10 years).

This study successfully applied the Arrhenius model to predict the long-term stability of the HEV VLP vaccine under recommended storage conditions (2–8 °C), thereby providing a scientific foundation for proposing a preliminary expiration period. The high goodness-of-fit of the kinetic model, reflected by a coefficient of determination R^2^ of 0.9997, supports the reliability of this prediction.

2.3. Effect of Freeze–Thaw Cycles on Antigen Stability

The change in ΔT_m_/T_agg_ remained within approximately 1.5 °C throughout, with Tm ranging from 71.3 °C to 70.0 °C and Tagg ranging from 64.3 °C to 63.2 °C. Statistically significant changes were only detected at specific cycle points: for colloidal stability at cycles 15, 18, and 27; for conformational stability at cycles 6, 9, and 12. The mean particle size showed significant changes at certain cycles (12, 15, 27) but did not exceed twice the initial value. (Figure 4A,B,D). SEC-MALS analysis confirmed that the molecular weight of the main peak remained stable throughout the process. No significant decrease in antigen-binding activity was observed up to the 30th cycle. (Figure 4C). However, freeze–thaw cycles induced definitive chemical modification. The icIEF profiles showed a shift in charge variants towards the basic direction (increased pI), characterized by a decrease in the main peak proportion, an increase in basic variants, and the emergence of new basic peaks (Figure 4E).

This is likely due to oxidation of methionine or cysteine residues induced by increased local oxygen concentration during freeze–thaw via the “freeze-concentration effect [23,24]. A key conclusion is that while freeze–thaw cycles under the tested conditions did not lead to immediate loss of function or structural integrity, they induced a detectable chemical modification (basic shift), which may impact long-term vaccine stability or immunogenicity and should be considered in formulation development and storage strategies.

2.4. Effect of Mechanical Stress on Antigen Stability

Under orbital shaking at 100 and 300 rpm for up to 12 days, no significant changes were observed in colloidal and conformational stability (ΔT_m_ and T_agg_ less than 1.5 °C, with T_m_ varying from 71.4 °C to 70.1 °C and T_agg_ varying from 66.5 °C to 64.3 °C), charge heterogeneity, or antigen-binding activity (Figure 5A,D–F). However, under high-intensity stress (300 rpm), DLS detected increases in mean particle size and PDI between days 3 and 6 (Figure 5B). SEC-MALS further confirmed protein aggregation. After 3 days of shaking at 300 rpm, the main peak molecular weight increased from approximately 1850 kDa (corresponding to the expected oligomer) to approximately 4600 kDa, while it remained constant at 100 rpm (Figure 5C).

This indicates that for this type of mechanical stress, SEC-MALS was more sensitive than DLS in detecting early-stage aggregation. The 300 rpm condition used here represents a high-intensity, accelerated boundary test designed to assess the intrinsic shear tolerance of the VLPs, exceeding the severity of typical transport or handling scenarios.

2.5. Effect of Excipients on Antigen Stability

The T_m_ and T_agg_ values of samples containing different types and ratios of vaccine stabilizers were determined using DSF and SLS (Figure 6A–D). Upon the addition of HSA to the formulation, the overall T_m_ and T_agg_ increased significantly compared to the HSA-free group. However, since HSA itself contributes to both fluorescence and scattering signals, the measured signals reflect the characteristics of the overall system, making it difficult to attribute conformational changes specifically to the target protein. For formulations containing HSA, interpretation relies more definitively on orthogonal methods such as SEC-MALS, as the DSF and DLS signals represent a composite from both HSA and the target VLPs, making it difficult to attribute conformational changes specifically to the VLPs. In the groups supplemented with trehalose and sucrose, both T_m_ and T_agg_ showed an increasing trend of 2–4 °C, with trehalose increasing T_m_ from 70.6 °C to 72.9 °C and T_agg_ from 64.0 °C to 66.5 °C, sucrose increasing T_m_ from 70.6 °C to 74.1 °C and T_agg_ from 64.0 °C to 66.8 °C. Both sugars exhibited similar effects on the formulation. In contrast, supplementation with PS 80 not only failed to improve the conformational and colloidal stability of the recombinant protein vaccine but also disrupted its original stability. As shown in Figure 6D, as the proportion of PS increased, the T_m_ decreased by 5–10 °C (from 72.3 °C to 63.7 °C) and T_agg_ decreased by 2–3 °C (from 66.8 °C to 64.0 °C) compared to the PS-free group. For the excipient screening experiment, a two-way ANOVA revealed that both excipient type (p < 0.001) and treatment condition (p < 0.001) had significant main effects on conformational stability. Moreover, the interaction between excipient and treatment condition was also statistically significant (p < 0.001), indicating that the response of conformational stability to thermal stress differed significantly depending on the excipient present.

The mean particle size and PDI of samples containing different stabilizers were analyzed by DLS. In the HSA-supplemented group, with an increasing HSA ratio, the homogeneity of the particle size distribution decreased and the average particle size diminished, likely due to interference from HSA signals (Figure 7A). In the trehalose and sucrose groups, the average mean particle size increased with higher stabilizer ratios (Figure 7B,C). This is likely due to the formation of a thicker hydration or solvation layer around the VLPs, thereby increasing their apparent hydrodynamic radius. In the PS 80 group, the particle size initially increased and then decreased with increasing surfactant concentration: when the PS 80 concentration reached 5%, the particle size dropped sharply, suggesting that high surfactant concentrations may lead to the disassociation of self-assembled particles (Figure 7D).

The antigen-binding activity of the accelerated samples was characterized by ELISA (Figure 8A–D). Under different buffer conditions, the HEV antigen-binding activity was significantly enhanced, even exceeding that of the control group stored at 4 °C. Since the ELISA specifically detects HEV monomers, this increase in binding activity suggests that the self-assembled structures may have dissociated into a greater number of monomers.

2.6. Summary of Key Findings

This study employed a multi-parameter analytical framework to systematically evaluate the stability changes in the HEV p239 VLP vaccine under various stress conditions. The key findings are summarized as follows: charge heterogeneity was the earliest indicator of change under all stress conditions; thermal stress led to an increase in acidic charge variants, a rise in molecular weight, and eventual loss of antigen-binding activity; freeze–thaw cycles primarily induced an increase in basic charge variants but did not significantly affect structural or functional integrity; high-intensity mechanical stress could increase molecular weight without altering charge profiles or antigen activity; cavitation was destructively disruptive to VLP structure; and the choice of formulation excipients had significant and complex effects on stability.

3. Discussion

The stability of VLPs is a critical attribute determining their efficacy as vaccines or delivery vehicles, thus directly impacting drug safety and efficacy [6,10]. In this study, we systematically evaluated the stability of an aluminum-free HEV p239 VLP vaccine under various stress conditions using a multi-parameter analytical framework. Our key finding is that changes in charge heterogeneity, as detected by icIEF, serve as the earliest and most sensitive indicator of VLP instability, preceding measurable losses in conformational integrity, particle size growth, or antigen-binding function.

The T_m_ of antigen domains is generally lower than that of other biopharmaceutics such as monoclonal antibodies (mAbs). This difference stems from their distinct biological functions. Antigens require a degree of conformational flexibility to perform their biological roles (e.g., receptor binding, membrane fusion) [25]. This flexibility often comes at the cost of reduced structural rigidity (i.e., thermal stability). In contrast, the primary function of mAbs (especially the Fab region) is to bind a specific epitope with high affinity and specificity. This requires their antigen-binding regions (complementarity-determining regions, CDR loops) to maintain a relatively stable conformation before and after binding to achieve precise epitope engagement. The overall structure of a mAb, particularly the Fc region, is typically stabilized by numerous disulfide bonds and tightly packed constant domains, conferring higher thermal stability [26,27]. Although VLP assembly can enhance stability to some extent, this stability highly relies heavily on non-covalent interactions between subunits rather than chemical bond linkages [28]. As products of immune system selection and engineering optimization, mAbs possess inherently high stability and developability, and exist as monomers or simple complexes, thus exhibiting greater structural homogeneity [26]. In summary, in-depth investigation into vaccine stabilization mechanisms and the development of novel thermostable formulations are of significant importance for enhancing the stability and applicability of vaccine products.

Our investigations revealed that changes in charge heterogeneity occurred earlier than functional and structural changes. Under thermal, freeze–thaw, and agitation stresses, the content and distribution of charge variants exhibited acidic shifts, basic shifts, and no shift, respectively. Changes in icIEF appeared first, followed by changes in HMWS and mean particle size, collectively contributing to the decrease in antigen-binding activity. Under thermal stress, an acidic shift in charge variants was observed early. For instance, at the extreme temperature of 56 °C, this shift was detectable within just 4 h, preceding any measurable changes in conformational stability, particle size, or antigen-binding activity. This early alteration likely reflects rapid chemical modifications such as deamidation of surface-exposed asparagine residues, which are particularly susceptible to high-temperature degradation [29]. Although this initial acidic shift did not immediately impair antigen-binding function, which was lost only after 3 days at 56 °C, it serves as a critical early warning signal for subsequent structural and functional deterioration. The importance of this early charge shift lies in its predictive value for VLP stability [30,31]. In vaccine development and quality control, the ability to detect instability before functional loss is crucial [32]. The correlation between charge variant shifts and functional loss is mechanistically sequential but not instantaneous. The acidic shift observed under thermal stress correlates with subsequent deamidation (as suggested by mass spectrometry). This is consistent with the known susceptibility of residues like Asn573 in p239 to deamidation at elevated temperatures, a modification that can ultimately disrupt critical epitope-antibody interactions [33,34]. Surface-exposed deamidation sites might become neo-epitopes, potentially enhancing immunogenicity [35]. As such, the spatial location and charge characteristics of these residues are crucial for maintaining epitope integrity. While the initial charge variant shift is an early indicator, it is the ensuing structural perturbation that leads to reduced binding activity. This understanding underscores the utility of monitoring charge variants via icIEF as a sensitive and rapid means of assessing thermal stress impact in accelerated studies. Furthermore, identifying the specific chemical modifications underlying these shifts provides a scientific basis for designing formulations that protect sensitive amino acid residues, thereby effectively enhancing thermal stability. While 56 °C represents an extreme stress condition, the early detection of charge variant shifts underscores the necessity of considering site-specific chemical degradation pathways when developing robust VLP formulations, particularly for vaccines intended for use in regions with variable cold-chain compliance [10].

Under both freeze–thaw and agitation stress, no regular changes were observed in critical quality attributes (CQAs) such as T_m_, T_agg_, binding activity, HMWS, and mean particle size, only a basic shift in icIEF occurred after freeze–thaw, without affecting binding activity. This indicated that neither stress disrupted the overall VLP structure or antigen epitope integrity. Therefore, changes in charge heterogeneity do not necessarily lead to functional loss and require analysis based on the specific site. Although icIEF changes preceded functional decline, the basic shift induced by freeze–thaw was not accompanied by loss of binding activity, implying the charge change might occur in regions non-critical for the epitope. Studies suggest that during protein freezing, the “freeze concentration effect” occurs, leading to an increase in localized oxygen and protein concentrations [23,24], thereby making active thiol groups are susceptible to radical attack, leading to oxidation forming disulfide bonds or sulfoxides, thereby reducing total thiol content and increasing basic variants [16,36,37,38].The p239 sequence (aa 368-606) contains multiple Cys (e.g., Cys472, Cys552) and Met (e.g., Met515) residues [37], Methionine can be oxidized to methionine sulfoxide, leading to charge variants (basic shift) or conformational changes [38], However, neither cysteine nor methionine residues are part of the critical 8C11 epitope [37], explaining why the charge variant (basic shift) induced by freeze–thaw stress did not reduce binding activity.

Mechanical stress (agitation) did not induce significant charge variant shifts or functional loss within the tested range, though higher shear (300 rpm) led to detectable aggregation via SEC-MALS. This highlights the utility of orthogonal methods: while DLS and conformational stability assays (T_m_/T_agg_) were robust for thermal stress, SEC-MALS proved more sensitive for detecting early aggregation under mechanical stress.

Changes in charge heterogeneity represent the most sensitive indicator in this study, appearing earliest and exhibiting stress-specific responses. Structural and functional changes, in contrast, occur later and only under extreme conditions. Specifically, the formation of high-molecular-weight species shows a direct temporal correlation with the loss of antigen-binding activity, whereas early shifts in charge variants (e.g., the acidic shift observed at 4 h under 56 °C) are temporally dissociated from functional loss (e.g., after 3 days at 56 °C). This distinction reveals their distinct roles in the degradation cascade: shifts in charge variants reflect early-stage chemical modifications (e.g., deamidation), which alter the net charge of the protein but may not immediately disrupt its three-dimensional conformation or the integrity of critical epitopes [36,39]. In contrast, the accumulation of HMWS represents a later-stage physical aggregation process, which ultimately leads to functional loss by directly masking or distorting conformational epitopes [30,40]. Thus, charge variants serve as upstream early-warning indicators, providing signals at the initiation of degradation pathways and creating a critical time window for formulation optimization and quality control intervention. HMWS, on the other hand, act as downstream function-related indicators, directly linked to the final product potency. Monitoring charge variants is of significant importance for predicting long-term stability risks and screening stabilization strategies during formulation development [31,41].

Beyond identifying early indicators, the ability to quantitatively predict long-term stability is crucial for vaccine development and quality assurance. In this study, the growth of HMWS under elevated temperatures was found to follow apparent first-order kinetics, enabling the construction of an Arrhenius model [22,42,43]. The activation energy for HMWS formation was calculated to be 49.6 kJ/mol. This value is lower than the activation energies commonly observed for random, bulk hydrolysis or oxidation reactions in small-molecule drugs (typically >80 kJ/mol) [44]. The lower activation energy suggests that these chemical modifications are likely occurring at structurally sensitive, specific sites, where reactivity is enhanced by the higher-order structure of the protein, thus requiring a lower energy barrier. Therefore, for complex biologics such as VLPs, the primary risk often lies in low-energy, site-specific chemical modifications that trigger physical changes, rather than in high-energy bond cleavage. Consequently, monitoring early indicators of chemical change, like charge heterogeneity, is more predictive than solely tracking the final physical aggregates. It should be emphasized, however, that the prediction relies on extrapolation from accelerated data obtained at elevated temperatures and assumes an invariant degradation mechanism across the temperature range studied. Consequently, these results must be regarded as a preliminary stability assessment, and their accuracy will require final confirmation through ongoing real-time, long-term stability studies conducted at 2–8 °C.

Based on the above findings, the stability of VLP vaccines is influenced by multiple factors, with temperature deviation being a core risk. Consequently, we optimized the formulation of the VLP liquid vaccine by adding various stabilizers (e.g., sugars, surfactants, proteins) at different ratios and assessed their impact under thermal stress. In liquid formulations, HSA, a traditional protein stabilizer, can co-stabilize protein drugs or vaccines through synergistic effects. Quantitative analysis showed that the addition of HSA (e.g., 2% w/v) significantly increased the T_m_ and T_agg_ of the system by approximately 4–6 °C. However, introducing protein excipients significantly increased the formulation complexity, particularly for analytical methods. Since both HSA and VLPs are proteins, they overlap in terms of physicochemical properties such as fluorescence signals, light scattering signals, and size range. This makes it impossible for DSF and SLS techniques to establish whether signal changes originate from HSA or VLP conformational alterations. In DLS, HSA acts as an additional protein component, introducing particles of different sizes than VLPs into the solution, leading directly to heterogeneous particle distributions. A observed decrease in average size may result from HSA signal interference rather than representing a true change in VLP size [45]. Sugar excipients such as sucrose and trehalose stabilize proteins through preferential hydration. At higher ratios (e.g., mass ratio 100:1), both sugars increased the T_m_ and T_agg_ of VLPs by 2 to 4 °C. These highly polar excipients tend to interact strongly with water molecules, being excluded from the protein’s hydration layer, thus helping to maintain its integrity and stability. This helps prevent VLPs’ conformational unfolding or irreversible aggregation due to hydration layer disruption. Simultaneously, in the liquid state, high sugar concentrations can restrict VLPs’ molecular motion and reduce conformational fluctuations, further consolidating conformational and colloidal stability [46,47].

PS 80 stabilizes proteins by preferentially occupying the air–liquid interface, preventing protein adsorption and subsequent denaturation [48,49], while weakly associating with hydrophobic patches on the protein surface, inhibiting intermolecular hydrophobic interactions and thus reducing aggregation [49,50]. However, the stabilizing effect of surfactants must be strictly controlled at an optimal concentration (e.g., near the critical micelle concentration, CMC) [49,51,52]. Quantitative studies have shown that at lower concentrations (e.g., 0.01–0.1%), PS 80 provided a certain degree of stabilization. Excessively high PS 80 concentrations (e.g., 0.5%) might induce protein instability due to their own degradation (e.g., oxidation generating peroxides) or excessive micelle formation [53,54]. High concentrations might not act via “interface competition” or “hydrophobic patch binding” but rather destabilize VLPs’ structural integrity by altering solution phase behavior [50,53,54]. This result indicates that the subsequent application of surfactants in VLP vaccine formulations requires extensive pre-experimentation to optimize the concentration, ensuring stabilization while avoiding risks associated with excessive levels.

This study provides direct experimental evidence and risk warnings for the process of developing an HEV p239 VLP vaccine. Specifically, trehalose or sucrose at a mass ratio of approximately 100:1 (w/w, sugar: antigen) can significantly enhance the conformational stability (T_m_) of the VLP by 2–4 °C, thereby identifying effective stabilizer types and key concentration ranges for the development of aluminum-free liquid formulations. In contrast, the data clearly demonstrate that PS 80 at a concentration of 0.5% leads to severe conformational destabilization (T_m_ decreased by 5–10 °C) and poses a risk of particle dissociation. This highlights that its use in vaccine formulations must be based on rigorous concentration optimization and risk assessment, rather than routine addition. Therefore, in subsequent vaccine development, it is recommended to adopt sugars as core stabilizers and to establish clear safety limits for surfactant concentrations. These conclusions, grounded in quantitative CQAs, can directly guide the design and screening of robust formulations, thereby mitigating stability risks during development.

The rapid and sensitive nature of DSF and icIEF make them suitable for high-throughput screening of formulation candidates. We propose a practical two-step workflow: initial screening with DSF/icIEF to identify stabilizers that maintain conformational and charge homogeneity, followed by long-term stability validation using SEC-MALS, DLS, and ELISA. While this study focused on HEV p239 VLPs, the principles identified, particularly the role of charge heterogeneity as an early stability indicator [39], are likely applicable to other conformation-dependent VLP platforms, such as HPV [55] or norovirus VLPs [56]. Comparative studies with these VLP vaccines would further validate the generalizability of our findings.

This study systematically characterized the stability of HEV p239 VLP through a multi-parametric approach, yet some limitations indicate future research opportunities. Conclusions on aggregation and structural changes rely on solution-state analyses, and direct morphological evidence from techniques such as TEM is lacking. The specific chemical modifications underlying charge shifts (e.g., deamidation/oxidation) were not mapped at the molecular level, and the link between the observed physicochemical changes and in vivo immunogenicity remains to be established. Future studies integrating direct imaging (e.g., cryo-EM), advanced mass spectrometry, and functional immunogenicity assays will help translate these stability insights into robust vaccine development strategies.

4. Materials and Methods

4.1. Materials

The Recombinant Hepatitis E Vaccine drug substance (3.6 mg/mL, Lot AP-20220102-04) was provided by Beijing Wantai Biological Pharmacy Enterprise Co., Ltd. (Beijing, China). The national standard for antigen content (300051-202101) was obtained from the National Institutes for Food and Drug Control (Beijing, China). Key excipients and reagents included HSA, D-trehalose dihydrate, sucrose, PS 80, and Pharmalyte™ carriers for icIEF. Detailed information regarding the experimental materials is provided in the Supplementary Materials Table S1.

4.2. Rationale for Stress Conditions and Analytical Endpoints

To investigate the stability of the VLP formulation, we selected various types and intensities of conditions based on the scientific principles of forced degradation studies as outlined in ICH Q1A(R2). We employed a high-temperature forced degradation model based on Arrhenius kinetic extrapolation, aiming to simulate and predict long-term stability risks through short-term (28-day) high-temperature exposure, thereby rapidly revealing the primary chemical and physical degradation pathways of the formulation. Thirty freeze–thaw cycles constitute a severe stress case to evaluate robustness against phase changes and cryoconcentration beyond that encountered in typical transport scenarios.

The analytical endpoints were chosen to monitor CQAs pertinent to vaccine quality. Melting temperature (T_m_), aggregation temperature (T_agg_), mean particle size, PDI, HMWS, charge heterogeneity, and antigen content collectively assess attributes critical for potency, safety, and stability, consistent with the stability evaluation framework for biologics [ICH Q5C].

4.3. Investigating the Effect of Temperature on Antigen Stability

Aliquots of 500 µL of HEV vaccine drug substance (protein concentration 3.6 mg/mL) were dispensed into 1.5 mL sterile polypropylene centrifuge tubes (3810X, Eppendorf, Hamburg, Germany). The aliquoted samples were placed in an incubator (3907, Thermo Scientific, Waltham, MA, USA) set at temperatures of 4 ± 2 °C, 25 ± 2 °C, 37 ± 2 °C, and 56 ± 2 °C for a 4-week accelerated stability study. Samples were prepared at 8 time points: 4 h, 8 h, 1 d, 3 d, 7 d, 14 d, 21 d, and 28 d. The experiment was conducted using three independently aliquoted samples.

4.4. Long-Term Stability Prediction

To quantitatively assess the long-term stability of the HEV VLP vaccine, the Arrhenius model was applied to perform kinetic analysis of the accelerated stability data [57]. The content of high-molecular-weight species (HMWS) was selected as the key predictive indicator, as its change under accelerated conditions (25 °C, 37 °C, 56 °C) follows first-order or pseudo-first-order reaction kinetics. Using GraphPad Prism version 8.0.2 software, the data of HMWS content (Mw) over time (t) at each temperature were fitted via nonlinear regression to the first-order kinetic equation:

$$[eqn]$$

where $[eqn]$ is the initial molecular weight, $[eqn]$ is the fitted maximum molecular weight at the plateau, and $[eqn]$ is the apparent first-order degradation/aggregation rate constant (h^−1^). The goodness of fit was evaluated by the coefficient of determination R^2^.

The obtained $[eqn]$ values from the three accelerated temperatures and their natural logarithms, $[eqn]$ , were subjected to weighted linear regression against the corresponding reciprocal of the absolute temperature, $[eqn]$ , to establish the Arrhenius equation:

$$[eqn]$$

The slope of the regression line was used to calculate the apparent activation energy $[eqn]$ for the aggregation process. Using this equation, the rate constant at the target storage temperature of 4 °C (T = 277.15 K), denoted as $[eqn]$ , was extrapolated.

4.5. Investigating the Effect of Freeze–Thaw Cycles on Antigen Stability

Aliquots of 500 µL of HEV vaccine drug substance (protein concentration 3.6 mg/mL) were dispensed into 1.5 mL sterile polypropylene centrifuge tubes. The aliquoted samples were placed in an ultra-low temperature freezer (Forma 88600V, Thermo, Waltham, MA, USA) at −80 °C for approximately 1 h until the liquid was completely frozen and then thawed at room temperature (25 °C) for approximately 1 h. This freeze–thaw cycle was repeated 3, 6, 9, 12, 15, 18, 21, 24, 27, and 30 times. The experiment was conducted using three independently aliquoted samples.

4.6. Investigating the Effect of Mechanical Stress on Antigen Stability

Aliquots of 500 µL of HEV vaccine drug substance (protein concentration 3.6 mg/mL) were dispensed into 1.5 mL sterile polypropylene centrifuge tubes. The aliquoted samples were placed on a rotator (RH-18, Miulab, Hangzhou, China) at room temperature (25 °C) and subjected to end-to-end rotation at 100 rpm and 300 rpm for a 12-day mechanical stress stability study. Samples were collected at 4 time points (1 d, 3 d, 6 d, and 12 d) for further analysis. The experiment was conducted using three independent samples.

4.7. Investigation of Excipients on Formulation Stability

HSA, d-trehalose, sucrose, and PS 80 were added to the HEV vaccine drug substance at different ratios, as detailed in Table 1. The samples were then placed in an incubator (ICTHI-250T, Stik, Shanghai, China) at 40 °C and 75% RH for 48 h. The conformational stability and colloidal stability of the samples were assessed. The experiment was conducted using three independent samples.

4.8. Assessment of Protein Conformational Changes

T_m_ and T_agg_ of the HEV vaccine drug substance before and after stress was measured using the protein stability analyzer (Uncle, Unchained Labs, Pleasanton, CA, USA). Samples were centrifuged (5415R, Eppendorf, Hamburg, Germany) at 13,000× g for 5 min at 4 °C. A total of 9 µL of the supernatant was loaded into the analyzer. The experiment employed continuous heating from 25 °C to 95 °C at a rate of 0.5 °C/min. During this process. T_m_ was determined using DSF. Measurements were taken across excitation and emission wavelengths ranging from 250 nm to 720 nm. T_agg_ was determined using SLS. The sample was excited at a wavelength of 266 nm, and the scattered light intensity was detected at a 90° angle. Following data acquisition, Analysis V6.01 software was used for data analysis. The corresponding T_m_ values were calculated using the centroid wavelength method. The corresponding T_agg_ values were determined by identifying the temperature at which the scattering intensity curve reached 10% of the maximum peak height in its first derivative curve.

4.9. Measurement of Mean Particle Size and PDI

The mean particle size and PDI of the HEV vaccine drug substance before and after stress was measured using the protein stability analyzer (Unchained Labs, Pleasanton, CA, USA). Samples were centrifuged (5415R, Eppendorf, Hamburg, Germany) at 13,000× g for 5 min at 4 °C. A total of 9 µL of the supernatant was loaded into the analyzer. Measurements were performed using DLS at 25 °C with a wavelength of 660 nm, using a default refractive index of 1.332 and a viscosity of 0.89 cp to determine the particle size and PDI. Following data acquisition, Analysis V6.01 software was used for data analysis to obtain the Z-average and PDI values.

4.10. Quantification of HMWS by SEC-MALS

The HMWS in samples (protein concentration 3.6 mg/mL) before and after stress were determined by SEC-MALS. The experiment utilized an Agilent 1260 HPLC system equipped with a TSKgel G3000Wx, size-exclusion chromatography column (7.8 × 300 mm, 5 µm, Tosoh Bioscience, Tokyo, Japan), a DAWN multi-angle light scattering detector (Wyatt Technology, Santa Barbara, CA, USA), and an optilab differential refractive index detector. The MALS instrument employed a 658 nm laser, with light scattering measurements taken at 90°. The mobile phase was 1× PBS, filtered through a 0.22 µm filter before use. The flow rate was 0.5 mL/min, detection was performed using a UV detector at 280 nm, the run time was 30 min, and the injection volume was 25 µL. Data were collected and processed using ASTRA^®^ software version 7.2 (Wyatt Technology, Santa Barbara, CA, USA) to analyze the MW of the samples.

4.11. Charge Variant Detection

Charge heterogeneity of samples before and after stress was analyzed using a fully automated protein analysis system (Maurice S, ProteinSimple, San Jose, CA, USA) with an icIEF Cartridge. First, a master mix containing markers was prepared. The composition of the master mix included 0.35% methyl cellulose, 2.0% Pharmalyte™ pH 3–10, 2.0% Pharmalyte™ pH 5–8, 0.5% pI marker 4.05, 0.5% pI marker 8.4, 10 mM iminodiacetic acid (IDA), and 5 M urea. This master mix was combined with the sample to achieve a final sample concentration of 0.06 mg/mL. The mixture was gently inverted to mix and centrifuged at 10,000× g for 5 min. At least 150 µL of supernatant was used for charge variant analysis. The focusing parameters were: prefocusing time 1 min at 1500 V, focusing time 6 min at 3000 V.

4.12. Evaluation of Antigen-Binding Activity by ELISA

The recombinant Hepatitis E vaccine national standard (5.2 × 10^5^ U/mL) was first diluted to 520 U/mL using 1× PBS; this was the starting concentration for an 8-point serial dilution (see example below). The dilution series in a 96-well plate started at 520 U/mL and ended at 0.16 U/mL, resulting in 8 dilutions. Samples (3.6 mg/mL) before and after stress were diluted with 1× PBS to a final concentration of 11.25 ng/mL. A volume of 100 µL of the diluted sample was added to wells of a microplate pre-coated with HEV polyclonal antibody (Hepatitis E virus Antigen Detection Kit, Beijing Wantai Biological Pharmacy Enterprise Co., Ltd., Beijing, China). The plate was sealed and incubated at 37 °C for 45 min. After incubation, the plate was washed three times with 1× PBST (0.05% Tween 20, pH 7.4, CST, Danvers, MA, USA). Then, 100 µL of enzyme conjugate reagent was added to each sample well. The plate was sealed and incubated at 37 °C for 60 min. After incubation, the plate was washed three times with 1× PBST. Finally, 50 µL each of Chromogen Solution A and B were added to each well. The plate was sealed and incubated at 37 °C for 15 min. The reaction was stopped by adding 50 µL of Stop Solution per well. The absorbance was measured within 10 min using a SpectraMax M5 microplate reader (Molecular Devices, San Jose, CA, USA) at dual wavelengths of 450 nm and 630 nm. Each sample was tested in triplicate, and the average result was analyzed using SoftMax Pro 7.0 GxP software (Molecular Devices, San Jose, CA, USA).

4.13. Data Processing and Analysis

Data analyses and graphical representations were performed with GraphPad Prism version 8.0.2 (GraphPad Software, Boston, MA, USA). Data are presented as mean ± SD (n = 3). A two-way ANOVA was conducted to examine the effects of temperature (4, 25, 37, 56 °C) and time (0–672 h) on the measured parameters, followed by Tukey’s post hoc test for multiple comparisons between groups when significant effects were detected (p < 0.05).

5. Conclusions

This study established an efficient strategy for evaluating and predicting the stability of VLP vaccines through multi-stress and multi-parameter systematic analysis. The study confirmed that changes in charge heterogeneity detected by imaging capillary isoelectric focusing serve as the earliest and most sensitive early-warning signal of instability, identifiable prior to detectable structural or functional degradation. This indicator becomes detectable within 3 days at 25 °C, 8 h at 37 °C, and as early as 4 h at 56 °C, providing a critical tool for real-time quality monitoring during vaccine development. The research further clarified the distinct degradation pathways induced by different stress conditions: thermal stress triggered an acidic charge shift associated with deamidation and subsequent aggregation; freeze–thaw cycles resulted in a basic charge shift due to oxidation without functional impairment; and high-intensity mechanical agitation directly promoted aggregation without preceding changes in charge variants. Additionally, the effects of key excipients were quantitatively assessed: sugars (trehalose/sucrose) at a 100:1 mass ratio enhanced conformational stability, increasing T_m_ by 2–4 °C, whereas PS 80 at 0.5% severely destabilized conformation, lowering Tm by 5–10 °C and posing a risk of particle dissociation. Based on the mechanistic understanding of stabilizer effects, a two-step formulation screening workflow was proposed: initial rapid screening of conformation-stabilizing candidates using sensitive DSF followed by comprehensive long-term stability verification of optimized formulations via techniques such as SEC-MALS and ELISA, significantly enhancing development efficiency.

The evaluation framework and stability principles derived here possess broader applicability. For other self-assembled, conformation-dependent VLP platforms (e.g., HPV or norovirus VLPs), charge heterogeneity may likewise serve as a universal early stability indicator. This study provides direct data and scientific rationale for formulation development, Quality by Design (QbD) practices, and the design of robust strategies that extend beyond conventional cold-chain requirements. Ultimately, this work contributes to the acceleration of the development of more stable and accessible next-generation VLP vaccines.