Results and discussion

Formation and characterization of emulsions and antibubbles

In this study, the term “parent double emulsion” refers to the W1/O/W2 emulsion before freeze-drying, whereas “antibubbles” refers to the W1/A/W2 structures obtained after freeze-drying and rehydration. The microscopic image of the parent double emulsion (Fig. 2a) of formulation V3 (Table 1) depicts most of the droplets as perfectly spherical, having bright, orange-colored inner cores conferring the presence of daunorubicin. After freeze-drying and rehydration, the corresponding antibubbles retained a multicore structure, indicating the presence of daunorubicin-containing aqueous cores within the antibubbles (Fig. 2b). The mean diameter of the final antibubbles after reconstitution was 25.7 μm (Fig. 2c). The entrapment efficiency (Fig. 2d) of all antibubble variants (V1, V2, V3) shows an increasing trend when the concentration of AEROSIL® R972 particles increases in the oil phase during double emulsion formation. V1 and V3 showed a statistically significant difference in entrapment efficiencies, i.e. 86% and 94%, respectively. A higher particle concentration at the oil-water interface results in a denser interfacial layer, which effectively inhibits droplet aggregation and coalescence, thereby maintaining smaller droplet sizes and enhancing the drug-encapsulating ability of the Pickering double emulsion (Frelichowska et al., 2009; Ribeiro et al., 2023; Yin et al., 2025). A high particle concentration could also be linked with forming an aerogel network inside the antibubbles, keeping the inner droplets of the antibubbles confined in space, as seen in particle-stabilized emulsions (Poortinga and Van Nostrum, 2025).

Fig. 2: Daunorubicin-loaded formulations: (a) parent double emulsion (V3, Table 1), (b) rehydrated antibubbles at neutral pH (V3), (c) size distribution of antibubbles (V3), and (d) entrapment efficiency of different antibubble variants (V1, V2, V3, and V4, Table 1), where the error bars represent standard deviation (n = 3).

A notable decrease in entrapment efficiency was observed for V4 when relatively hydrophilic particles, AEROSIL® R816, were used alone at the outer interface, as depicted in Fig. 2d. Tukey's test revealed a significant difference in the entrapment efficiencies of V4 compared to V2 (91% vs. 73%), despite having the same particle concentration at the inner interface. This indicates that, during double emulsion formation, the interfacial adsorption of particles in V4 formulation is relatively slow, allowing escape of some inner droplets, likely due to the lower affinity of relatively hydrophilic particles (R816) for the oil-water interface than the outer water phase. In contrast, V2, with a combination of hydrophobic (R972) and relatively hydrophilic (R816) particles, ensures rapid interface establishment during the second emulsification step.

Factors responsible for drug release from daunorubicin-loaded antibubbles

Effect of pH and bile extract

The results presented in Fig. 3a demonstrate that the initial daunorubicin detected after rehydration of antibubbles remained low at all tested pH values (2, 5, and 7). This initial fraction corresponds to the non-encapsulated drug fraction indicated by the entrapment efficiency results in Section 3.1. The subsequent time-dependent release remained limited, confirming the stability of the silica nanoparticle-stabilized antibubbles under varying pH conditions. The slightly higher release at pH 2 is more likely due to increased daunorubicin solubility under acidic conditions, as has been reported for doxorubicin systems (Jafarzadeh-Holagh et al., 2018; Sultan et al., 2023), rather than pH-induced antibubble destabilization. These findings indicate that pH was not an effective trigger for drug release from the silica-stabilized antibubbles, whereas acidic pH can destabilize CaCO3-based antibubbles and trigger drug release (Zia et al., 2023).

Fig. 3: Daunorubicin release from antibubbles of type V3 (Table 1) under varying rehydration conditions: (a) Effect of pH (2, 5, and 7) on drug release, and (b) Effect of different concentrations of sodium taurocholate (NaTc) and bile extract on drug release. After two hours, Tween 20 (at 1% of the mixture) was added to break antibubbles for maximum drug release. The error bars represent standard deviation (n = 3).

Fig. 3a also shows a complete drug release upon the addition of Tween 20. The addition of Tween 20 displaces the silica particles at the interface, causing the collapse of the antibubble structure and subsequent release of its content. This forced release further supports that most of the drug was retained inside a protected antibubble structure and became available only after disruption of the particle-stabilized interfaces. For accurate quantification, the residual daunorubicin binding to silica after Tween 20 treatment was also considered. Although daunorubicin binding to silica was ∼35% before treating the mixture with Tween 20, this decreased to ∼5% after Tween 20 addition (as detailed in Supplementary Information, Section S.1). Since the total releasable drug was determined after Tween 20 treatment, only this residual ∼5% bound fraction was relevant for correction of the release percentages.

Fig. 3b demonstrates the critical role of bile salts (bile extract or NaTC) in destabilizing antibubbles and releasing daunorubicin. The drug release was minimal (less than 5%) when antibubbles were rehydrated in pH 7 alone. However, as bile salts were introduced, drug release increased significantly. NaTC impacted the release in a concentration-dependent manner, with 20 mM NaTC inducing substantially greater release than 5 mM. As mentioned in Section 2.4.2, these concentrations were selected to reflect physiologically relevant intestinal bile salt levels (Santos et al., 2025). Notably, bile extract (0.5%) produced the highest initial burst release, whereas 20 mM NaTC resulted in the highest overall drug release, followed by 0.5% bile extract; in contrast, 5 mM NaTC gave an appreciably lower release.

This destabilizing effect of bile extract was further confirmed by additional microscopy experiments performed to identify the minimum effective concentration required for antibubble disruption over a longer incubation time (4 h). Fig. S1 (Section S.2 of supplementary information) shows microscopic images of antibubbles exposed to increasing concentrations of bile extract. A gradual morphological transition from well-defined spherical structures to irregular, elongated particles was observed, with complete collapse occurring at around 1.5 mM after 4 h incubation at 37 °C. This concentration corresponds to approximately 0.1% (w/v) bovine bile and indicates that the complete bile extract is a more potent destabilizing agent than NaTC alone. A similar effect was also examined for other individual bile salts, and sodium deoxycholate, sodium glycocholate, and sodium cholate also required higher concentrations of around 20 mM or more to induce similar effects (data not shown).

These observations are consistent with the known role of bile salts, which are naturally present in the human intestine but absent in the stomach, and play a critical role as biological emulsifiers in the digestion of dietary fats (Shulpekova et al., 2022). They reduce interfacial tension and promote fat disruption, essential for efficient lipid absorption. In the case of silica-stabilized antibubbles, bile salts destabilize the particle-stabilized silica interfaces, thereby triggering the release of encapsulated drugs. Our findings highlight the physiological relevance of bile salts as a natural trigger for intestinal drug release while confirming that pH changes are not responsible for destabilizing the silica-stabilized antibubble.

Mechanism of drug release

To further visualize how bile salts affect antibubble structure, a higher NaTC concentration (20 mM) was used to more clearly observe the detailed structural changes associated with antibubble destabilization. Fig. 4b & c show the impact of NaTC on the structural integrity of antibubbles compared to when they were rehydrated in pH 2 (Fig. 4a) or pH 7 (Fig. 2b). The antibubbles remain intact upon rehydration at pH 2 and pH 7, exhibiting well-defined structures with inner cores. Sudden changes in antibubbles' appearance were observed upon exposure to 20 mM of NaTC. In Fig. 4b, the antibubbles exhibit fewer or no inner cores immediately after NaTC exposure, indicating rapid changes (disruption) of the internal structure. Fig. 4c specifically shows some residues and imprints of disappearing antibubbles. These microscopic observations were further validated through size measurements of antibubbles under varying pH conditions and in the presence of NaTC. Fig. 4d demonstrates that antibubbles in pH 2 buffer exhibit minimal changes in size over 3 h, with only a slight increase in mean size, showcasing the stability of silica-based antibubbles in acidic conditions. Similarly, antibubbles rehydrated in pH 7 buffer (Fig. 4e) showed a minimal increase in mean size and standard deviation over time. In contrast, Fig. 4f reveals that rehydration in a NaTC solution led to a marked and rapid decrease in antibubble size, reflecting the impact of NaTC on silica-stabilized interfaces and the resulting reduction of mean antibubble size.

Fig. 4: Microscopic images (top row) illustrating the structural changes in antibubbles (variant V3, Table 1): a) antibubbles after approximately 5 min of rehydration at pH 2, and b) & c) antibubbles upon sudden exposure of already rehydrated antibubbles to sodium taurocholate (20 mM) at pH 7, showing destabilization in antibubble structure. The scale bars represent 30 μm. The box plots (bottom row) show the size distribution from image analysis of antibubbles upon rehydration in different conditions: d) pH 2, e) pH 7, and f) after 10 min exposure to sodium taurocholate at pH 7.

These observations can be interpreted considering how surfactants interact with particle-stabilized interfaces. Surfactants like Tween 20 can disrupt particle-stabilized interfaces either by displacing the particles entirely to occupy the interface themselves or by adsorbing onto the particle surface, increasing their wettability and promoting their detachment from the interface (Vashisth et al., 2010). As NaTC is also a surface-active agent, it can similarly destabilize the antibubbles. Indeed, the structure collapse (drug release rate) depends on NaTC concentration, suggesting a rapid or partial (but gradual) destabilization of the antibubbles as a function of NaTC concentration. The antibubbles became relatively small and contained fewer inner cores immediately after treatment with NaTC. NaTC likely caused the initial detachment of some particles from the interface, resulting in localized defects (e.g., cracks) or a more porous interface. The phenomenon of crack formation in particle-stabilized interfaces was explained in detail by Vella et al. (2006). They explained that a local surface tension reduction occurs when a surfactant is introduced (with the help of a needle) on a densely packed monolayer of colloidal particles. That results in a tensile stress at the interface, creating a crack in the particulate layer. So, whatever the outcome of the interaction between antibubbles and NaTC (i.e., the formation of either cracks or disruption of particles from the interface), this would cause leakage of the entrapped air, leading to a reduction in antibubble size. Consequently, the external phase (containing NaTC) inrushes to the antibubble to replace the gap created by air leakage. Hence, the inner cores also begin to coalesce with the outer aqueous phase and subsequently disappear. Here, it is essential to note that the remains of the antibubbles are smooth and spherical in the presence of NaTC (Fig. 4b & c) as compared to when they were rehydrated in pH 2 or pH 7. In fact, the interface behaves initially like a solid due to a layer of adsorbed close-packed particles. After adding the bile salt, it behaves like a traditional air-liquid interface because the bile salt (NaTC) displaces (part of) the adsorbed particles. This could explain the increased sphericity of the antibubbles in the presence of NaTC.

Impact of inner interface particle concentration and outer interface particle type on the release of daunorubicin at pH 2

The objective was to develop antibubbles that provide maximum protection for daunorubicin in the stomach while ensuring efficient release in the intestine. To achieve this, the formulation was optimized by varying the particle concentration at the inner interface and adjusting the particle composition at the outer interface (Fig. 5a & b, respectively). The release behavior was determined in a pH 2 buffer, mimicking the acidic gastric pH. Fig. 5a illustrates the release of daunorubicin from antibubbles prepared from emulsions with increasing concentrations of silica R972 particles in the oil phase. V1 variant (with the lowest silica content) exhibited significantly higher drug release compared to V2 and V3 variants, where minimal release was observed. The highest stability was achieved with the highest silica content (V3), where drug release remained below 5% even after 3 h in the pH 2 buffer. These results indicate that increased particle concentration at the inner interface enhances core stability and thereby prevents drug release. A higher particle concentration is associated with the formation of thicker and more cohesive interfacial layers, which provides better mechanical resistance against environmental stress, effectively trapping the core droplets within the gas shell. This behavior is consistent with previous observations in particle-stabilized foams and emulsions (Kargar et al., 2012; Lam et al., 2014b; Wu and Ma, 2016). Conversely, insufficient interfacial coverage, as seen with variant V1, leads to loosely packed particles resulting in a weaker interface.

Fig. 5: Daunorubicin release from antibubbles in pH 2 buffer as influenced by the type and concentration of AEROSIL® functionalized silica nanoparticles used during the fabrication of antibubbles: (a) Effect of varying concentrations of R972 particles in the middle oil phase (cyclohexane) of parent double emulsion and (b) Effect of different combinations of R816 and R972 particles in the outer aqueous phase of the parent double emulsions. After three hours, Tween 20 (at 1% of the mixture) was added to destroy the antibubbles entirely. The error bars represent standard deviation (n = 3).

While the inner interface of the antibubbles was stabilized with R972 silica particles, this is not possible for the outer interface because this led to poorly dispersible antibubbles. Therefore, the more hydrophilic R816 type of silica was introduced. Fig. 5b demonstrates the importance of outer interface composition in controlling drug release. Antibubbles with two different particle compositions at the outer interface were compared, i.e., with R816 alone (V4) or with a 1:1 mixture of R816 and R972 (V2), at a fixed particle concentration (2.5% R972) at the inner interface. At 0 h, the release with V4 starts at 25%, significantly higher than the 6% observed with V2. This difference corresponds to different entrapment efficiencies in Fig. 1d (Section 3.1). Then, after 3 h, the release for V4 approached 46%, whereas V2's release was still 18%. The release rates, calculated as the slopes of the linear regression, reveal a significant difference in cumulative drug release per hour i.e., from 3.5% for V2 (R2 = 0.98) to 6.8% for V4 (R2 = 0.92). A difference in release over some time of 3 h demonstrates that the synergistic effect of combining hydrophobic and hydrophilic particles at the outer interface (mainly due to a stronger interface) results in more robust antibubbles.

Simulated gastrointestinal drug release from antibubbles

To better position the release behavior of the present system, the release profile of daunorubicin from antibubbles (V3 variant) was compared with selected oral doxorubicin delivery formulations reported in the literature (Fig. 6). Direct studies reporting daunorubicin release under simulated gastrointestinal conditions could not be found; therefore, doxorubicin formulations were selected as the closest anthracycline-based systems for comparison. The selected systems included free doxorubicin solution, liposomal doxorubicin, taurine-chitosan-coated liposomal doxorubicin, hyaluronic acid-based doxorubicin nanoparticles, and chitosan/sodium alginate hydrogel beads, as summarized in Table 2. Most of these systems were designed either to improve oral absorption or to delay drug release toward the colon.

Fig. 6: Comparison of drug release from daunorubicin-loaded antibubbles (black line) and selected oral doxorubicin formulations reported in the literature. (a) Release profiles up to 48 h, and (b) release profiles during the first 6 h, highlighting the gastric-to-intestinal transition. The error bars for the daunorubicin-loaded antibubbles represent standard deviation (n = 3), while literature data are shown as reported mean values. The details of the selected literature formulations are provided in Table 2.

Code used in Fig. 6 | Formulation name/type | Description/detail | Reference
AB-DON | Antibubbles | Silica nanoparticle-stabilized antibubbles loaded with daunorubicin | Present study
DOX | Free drug solution | Doxorubicin hydrochloride solution used as control | Qin et al. (2024)
LIP-DOX | Liposomes | Doxorubicin-loaded liposomes prepared from egg yolk lecithin and cholesterol | Qin et al. (2024)
CS-LIP-DOX | Coated liposomes | Doxorubicin-loaded liposomes coated with taurine-grafted chitosan | Qin et al. (2024)
HGS-DOX | Self-assembled nanoparticles | Doxorubicin-loaded hyaluronic acid-glycerol monostearate nanoparticles | Wu et al. (2024)
CS/SALG-DOX | Hydrogel beads | Doxorubicin-loaded double cross-linked chitosan/sodium alginate hydrogel beads; CS/SALG ratio 1:4 | Wu et al. (2020)

During the gastric phase (0−2h), the antibubbles demonstrated excellent stability, with limited daunorubicin release (<15%). This indicates that the particle-stabilized interface remained largely intact under acidic conditions, in accordance with the stability behavior discussed in Section 3.2.3. However, a sharp increase in drug release occurred upon transitioning to SIF (supplemented with bile extract), with approximately 60% release observed immediately after exposure to the bile salts in the intestinal fluid. Afterwards, the release continued gradually, reaching nearly 80% after 6 h.

As discussed in Section 3.2.1, this burst release is mainly linked with the presence of bile salts in the intestinal phase. Bile salts, being amphiphilic molecules, interact directly with the silica particles at the air-water interface of the antibubbles (Zhang et al., 2021). The hydrophobic tails of the bile salts likely adsorb onto the hydrophobic domains of the silica particles while their hydrophilic heads remain oriented toward the surrounding aqueous medium. This interaction alters the particle wettability, reducing their affinity with the interface versus outer water. As a result, the silica particles detach from the antibubble structure, destabilizing and subsequently rupturing the antibubbles, thereby releasing the encapsulated drug. The effect of bile salts on an air-water interface was described comprehensively in another study (Torcello-Gómez et al., 2012).

Compared with the reported doxorubicin systems, the present antibubbles showed a distinct release pattern. The reported formulations generally exhibited a more gradual release, consistent with diffusion-controlled, sustained, or colon-targeted delivery. In contrast, the antibubbles retained most of the drug during the gastric phase and then released a major fraction upon transition to bile-containing intestinal fluid. Therefore, the main advantage of the present system is not only protection during gastric exposure but also the ability to achieve a bile salt-triggered burst release in the intestinal environment. Here, it should be noted that the intended function of antibubbles is not their uptake by the intestinal epithelium, but protection of daunorubicin during gastric exposure followed by its release in the intestinal lumen. Therefore, any subsequent absorption would involve the released daunorubicin, which should be further studied.

Preservation of encapsulated daunorubicin by antibubbles

As discussed in Section 3.2.2, antibubbles effectively maintain structural integrity and prevent the premature release of daunorubicin under gastric pH conditions. Nevertheless, an optimal encapsulation system should also ensure protection of the encapsulated drug from chemical degradation. To demonstrate this, we rehydrated antibubbles at pH 9 or above, a condition under which daunorubicin degradation has already been documented (Maniez-Devos et al., 1986; Piekarski et al., 2014). Although this extreme pH does not mimic the intestinal pH, it was selected to accelerate pH-dependent changes in daunorubicin structure, serving as a proof of concept for the ability of antibubbles to protect an encapsulated active from such changes. The degradation of daunorubicin was assessed spectrophotometrically by monitoring the decrease in absorbance at 480 nm, as described in Section 2.3.1.

Fig. 7 depicts the degradation percentage of daunorubicin after 24 h under two different storage conditions: pH 9 at 4 °C in the dark and pH 9.5 at 23 °C at ambient light conditions, the latter representing a harsher storage condition. The comparison includes the degradation of pure (free) daunorubicin in aqueous solution and its encapsulated form within antibubbles. Under the more severe condition of pH 9.5 at 23 °C, the degradation of the free daunorubicin was around 90% in 24 h, corresponding to a pseudo-first order degradation rate constant (k) of 3.8 × 10−5 s−1. In contrast, the encapsulated drug showed significantly lower degradation, i.e., approximately 8% after 24 h. The latter value is close to the fraction of free drug present in the antibubble sample given the entrapment efficiency was around 94% (see Fig. 1d). A fair conclusion is then that the encapsulated fraction of daunorubicin is (almost) fully protected against degradation. At pH 9 at 4 °C, the degradation of the free daunorubicin reached around 12% (k = 1.7 × 10−6 s−1), while the encapsulated drug exhibited minimal degradation, approximately 1–2%, again most likely related to part of the free (unencapsulated) drug fraction in the antibubble sample.

Fig. 7: (a) Degradation (%) of unencapsulated (free drug in solution) and encapsulated daunorubicin (in antibubbles) at two storage conditions after 24 h. Data represent mean ± standard deviation (n = 3).

For comparison, Piekarski et al. (2014) reported degradation rate constants of 7 × 10−6 s−1 and 2 × 10−5 s−1 at 40 °C for pH 9 and 9.5, respectively. Applying the general rule that degradation rates approximately halve with every 10 °C decrease in temperature, we extrapolated their values to estimate rate constants of 5.8 × 10−7 s−1 at pH 9 and 4 °C, and 6.16 × 10−6 s−1 at pH 9.5 and 23 °C. The experimentally observed rate constants in our study (1.7 × 10−6 s−1 and 3.8 × 10−5 s−1, respectively) are higher than these estimates but remain less than an order of magnitude different. This indicates that the values are reasonably comparable, with the observed differences likely stemming from variations in experimental conditions, such as light exposure at pH 9.5 and 23 °C, which may accelerate degradation. Overall, the comparison supports the consistency of our experimental data with established degradation behavior.

These findings demonstrate that antibubbles effectively protected daunorubicin from hydroxyl ion-induced degradation under both conditions, almost all of the drug remaining intact within the antibubble structure even under the severe conditions, where most of the free drug had degraded in solution within 24 h.