Tuning micelle size was achieved by tailoring the molecular weight of their constituent polymers. To this end, we synthesized mPEG-b-p(HPMAm-Bz) block copolymers with a fixed molecular weight of the hydrophilic PEG block (5 kDa) and a varying molecular weight of the hydrophobic HPMAm-Bz block. The hydrophobic blocks of the resulting polymers had degrees of polymerization of 42, 71, and 102 (here referred to as small, medium, and large, respectively) based on NMR (Figs. 1A, C and S1), which corresponded with an overall number average molecular weight of 15.4, 22.5, and 30.2 kDa. The three polymers demonstrated similar dispersity based on GPC analysis (Đ = 1.5–1.6, Fig. 1B and C). The small, medium, and large polymers were separately used to prepare non-drug loaded micelles of different sizes, resulting in self-assemblies of 47, 60, and 130 nm size, respectively (Fig. 1D). All the micelles had a rather narrow size distribution with PDI around or below 0.2. Our DLS results corroborated a unimodal size distribution for all of them (Fig. S2). TEM images confirmed that the micelles, regardless of the size, had similarly spherical morphology and a narrow size distribution (Fig. 1F).

Preparation of polymeric micelles of different sizes. A Chemical structure of mPEG‐b‐p(HPMAm-Bz) block copolymers. B GPC chromatograms of small, medium, and large polymers. C Characterization of small, medium, and large polymers by 1H NMR and GPC, where degree of polymerization refers to the number of HPMAm-Bz units in the hydrophobic block. D, E Size (D) and polydispersity index (PDI) (E) of micelles prepared from small, medium, and large mPEG‐b‐p(HPMAm-Bz) copolymers. F TEM images of micelles prepared from small, medium, and large mPEG‐b‐p(HPMAm-Bz) copolymers. G Critical micelle concentration (CMC) of the three polymers. To prepare 1 mL of micellar dispersion, 30 mg of polymer were used. Data are presented as mean ± SD (N = 3)

Colloidal stability of the micelles was assessed by determining the CMC of their polymers. Low CMC values are crucial to ensure preservation of the self-assembled structure and retention of drugs upon dilution in the bloodstream after formulation administration [32]. Micelles prepared from the three different polymers demonstrated high stability with CMC values between 1 and 2 µg/mL (Fig. 1G), which were similar to previous observations [26, 33].

Subsequently, we prepared PTX and DEX single- and co-loaded micelles of different sizes (small, medium, and large) and evaluated the effect of co-encapsulation on key pharmaceutical properties such as size, size distribution, and drug encapsulation. To do so, we comparatively assessed the impact of increasing drug feed amounts and drug type in both single- and co-loaded setups, initially using medium-sized micelles (Fig. 2A).

Preparation and characterization of paclitaxel and dexamethasone single- and co-loaded medium-sized micelles. A, B Size, polydispersity index (PDI), encapsulation efficiency (EE), and loading capacity (LC) of paclitaxel (PTX) single-loaded (A) and dexamethasone (DEX) single-loaded (B) micelles at different drug feed amounts. C Size, polydispersity index (PDI), drug encapsulation efficiency (EE), and loading capacity (LC) of PTX and DEX co-loaded micelles with different feed amounts. D TEM images of PTX and DEX co-loaded micelles at different drug feed amounts. To prepare 1 mL of micellar dispersion, 30 mg of polymers were used in all cases. Data are presented as mean ± SD (N = 3)

For single-loaded PTX micelles, a wide range of 1 to 20 mg of PTX feed amount was examined. The micelles efficiently solubilized PTX with encapsulation efficiencies (EE) of about 80% for up to 15 mg of PTX and a maximum loading capacity (LC) of 30%. For higher PTX feed amounts (20 mg), the EE dropped to about 65%. Increasing PTX amounts was also accompanied by a gradual increase in micelle size from 60 to 80 nm, while the narrow size distribution of the micelles was maintained with PDI values < 0.1. For single loading DEX into the medium-sized micelles (Fig. 2B), a feed amount range of 0.3 to 10 mg was used, as the clinically used doses for DEX are remarkably lower for DEX compared to PTX [34,35,36]. Contrary to the single-loaded PTX micelles, increasing DEX feed amount had no effect on the size and size distribution of the formulations (~ 60 nm and PDI < 0.1). While high EE values of 80–90% were achieved at low DEX feed amounts (0.3–2.5 mg), the EE values drastically decreased to about 20% when feed amounts above 5 mg were used, due to significant DEX precipitation. The maximum LC of DEX-loaded micelles was approximately 10%. Single-loaded micelles with different drug feed amounts were also prepared from small and large polymers (Figs. S3 and S4). The different loaded micelles showed similar sizes and PDI as their empty counterparts. Except for 15 mg PTX feed in small polymers, all the formulations had EE values of 80% or higher.

For the PTX and DEX co-loaded micelles using medium-sized polymers, we employed a fixed PTX-to-DEX feed amount ratio of around 7 to 1 (w/w). This ratio was selected considering that the commercially available PTX formulations have drug concentrations between 5 and 6 mg/mL and that the clinical dose of DEX is lower than that of PTX [34,35,36,37]. We used three different PTX/DEX feed amounts of 2.5/0.3, 7.5/1, and 15/2 (mg/mg) (Fig. 2C, D). The size of the micelles slightly increased from 65 to 75 nm with increasing total drug feed amount, while PDI remained below 0.1 in all the cases (Fig. 2C). The increase in size is likely caused by the higher feed amount of PTX. Furthermore, the zeta potential of the PTX-DEX co-loaded micelles (with 7.5/1 (mg/mg) PTX/DEX feed amounts) was measured and compared to that of the empty formulation. In both cases, the micelles were found to be slightly negatively charged, with zeta potential values in the range of − 1 to − 2 mV (Fig. S5). Nanoparticles with neutral or slightly negative surface charge have been reported to circulate longer in the bloodstream and show improved tumor accumulation as compared to highly (positively) charged particles [38].

Regarding drug encapsulation, DEX was efficiently loaded (EE of 80%) in all three co-loaded formulations, whereas EE values for PTX only slightly decreased at high feed amounts (15 mg) with respect to the corresponding single-loaded micelles (from about 80 to 70%). LC for PTX and DEX in the assessed range had maximum values of around 25% and 5%, respectively. TEM images of all three co-loaded formulations confirmed homogeneous spherical morphology (Fig. 2D). Overall, PTX-DEX co-loaded micelles demonstrated analogous physiochemical properties to the PTX single-loaded ones, probably due to the significantly higher feed amount of PTX as compared to DEX. Furthermore, EE values for both drugs in the co-loaded formulations were comparable to the single-loaded counterparts.

To assess the influence of the polymer molecular weight on the pharmaceutical properties of the corresponding co-loaded formulations, micelles of different sizes with a PTX/DEX feed amount of 7.5/1 (mg/mg) were prepared. As observed for empty and single-loaded micelles, the size of the co-loaded formulations was mostly driven by the molecular weight of their constituting block copolymers, with values of about 50, 70, and 150 nm for small, medium, and large polymers, respectively (Fig. 3A). Co-loaded micelle sizes were similar to PTX single-loaded formulations and slightly larger than the empty ones, likely due to the solubilization of the drugs in the hydrophobic core of the micelles. PDI values followed a similar trend in both non-loaded and single-loaded micelles, indicating that all the formulations had narrow size distributions (Fig. 3B). Both PTX and DEX were efficiently loaded in the three different micelles, with EE of about 80% for PTX and 90% for DEX (Fig. 3C).

Comparison of paclitaxel and dexamethasone co-loaded micelles of different sizes. A, B Size (A) and polydispersity (PDI) (B) of paclitaxel (PTX) and dexamethasone (DEX) co-loaded small-, medium-, and large-sized micelles. C PTX and DEX encapsulation efficiency (EE). D–F PTX and DEX release profile of micelles prepared from small (D), medium (E), and large (F) polymers in simulated physiological conditions (sink conditions, PBS pH 7.4 containing 45 mg/mL BSA). For all the formulations, 30 mg of polymers and PTX/DEX feed amount of 7.5/1 (mg/mg) were used to prepare 1 mL of micellar dispersion. Data are presented as mean ± SD (N = 3)

Drug release in physiological conditions was studied by placing different formulations in a dialysis setup under sink conditions and using 45 mg/mL BSA in PBS (pH 7.4) solution as medium over the course of 72 h (Fig. 3D–F). A similar experimental setup was recently reported to be highly representative of in vivo drug retention in mice [39]. Drug retention capabilities of the micelles increased as the molecular weight of hydrophobic block became larger (i.e., large > medium > small polymers). After 24 h, about 55, 45, and 30% of PTX was released from small, medium, and large micelles, respectively, and a similar trend was observed for DEX (from 95% for small, to 65% for large micelles), which is in line with previous observations [31, 33]. Interestingly, DEX was released about two times faster than PTX regardless of the micelle size, which can be caused by a weaker interaction of DEX with the core of the micelles. This kinetically controlled release pattern can contribute to achieving sequential pharmacological effects, which may be favorable in multiple therapeutic setups. The released DEX could prime the TME to promote deeper micelle penetration, thereby enabling more effective PTX delivery to and action at the pathological site. Such sequential effects, achieved through different release rates for two drugs co-loaded in a single carrier, have been previously shown to be beneficial for improving antitumor response and therapeutic index [40].

To understand whether co-loading two drugs influences the release profile of the individual compounds, PTX and DEX single-loaded micelles of the three different sizes were also assessed for drug retention in PBS medium with BSA (Figs. S6–S8). Both PTX and DEX in all single-loaded formulations demonstrated similar release behavior in comparison to their co-loaded counterparts. Furthermore, the feed amount of the encapsulating drug did not play a remarkable role in the drug release profile. All in all, a modular nanoplatform for kinetically controlled co-delivery of PTX and DEX was successfully developed. Additionally, the data obtained for the co-loaded versus single-loaded micelles suggest that co-encapsulation of both drugs does not influence the retention of the individual PTX and DEX molecules in the hydrophobic core of the micelles and thereby does not impact the pharmaceutical properties as compared to the single-loaded formulations.

As the next step, we aimed to assess the versatility of the co-loaded platform beyond PTX and DEX, to two other clinically approved taxanes (DTX and CTX) and corticosteroids (PRD and CIC). To this end, we chose the least and the most hydrophobic taxane-corticosteroid combinations based on log P (Table 1), i.e., DTX-PRD and CTX-CIC, respectively. Medium-sized polymers with a feed of 7.5 mg for taxanes and 1 mg for corticosteroids were used to prepare micelles. The resulting micelles showed analogous size (around 70 nm) and PDI (below 0.1), as well as similarly homogenous spherical morphology among both drug combinations (Fig. 4A–C), as well as to the PTX-DEX micelles. The zeta potentials of DTX-PRD and CTX-CIC co-loaded micelles were also measured. Nanoparticles from both formulations showed slightly negative surface charges, with values in the range of − 1 to − 2 mV, similar to empty and PTX-DEX co-loaded micelles (Fig. S5). The loading of different drugs in the micelles, thus, did not influence the surface charge of the formed nanoparticles. Moreover, both DTX-PRD and CTX-CIC were also efficiently co-loaded in the micelles, with EE higher than 80% in all the cases (Fig. 4D). Drug release profiles of the micelles were evaluated under sink conditions in a medium containing BSA (Fig. 4E). As for PTX-DEX co-loaded micelles, the release rate for PRD (corticosteroid) was substantially faster than that of DTX (taxane), endowing the formulation with a pharmacologically favorable release feature that can result in advantageous sequential therapeutic effects. Conversely, CTX-CIC co-loaded micelles displayed a different release pattern between the two drug classes as compared to the previous co-encapsulated formulations; while CTX showed similar release kinetics to the two other taxanes, CIC release from the micelles was remarkably slower than the two other corticosteroids (DEX and PRD) and similar to CTX. This behavior may be ascribed to the high hydrophobicity of CIC (log P value of 4).

Preparation and characterization of two other taxane-corticosteroid co-loaded micelles. Docetaxel with prednisolone (DTX-PRD) and cabazitaxel with ciclesonide (CTX-CIC) co-loaded micelles were prepared using mPEG-b-p(HPMAm-Bz) of medium molecular weight. A, B Size (A) and polydispersity index (PDI) (B) of DTX-PRD and CTX-CIC co-loaded micelles. C TEM images of DTX-PRD and CTX-CIC co-loaded micellar dispersions. D Drug encapsulation efficiency (EE) for DTX-PRD and CTX-CIC micelles. E Drug release profile of DTX-PRD and CTX-CIC micelles. A total of 30 mg of polymers and a taxane/corticosteroid feed amount of 7.5/1 (mg/mg) were used to prepare 1 mL of micellar dispersions. Data are presented as mean ± SD (N = 3)

To explore whether different biologically relevant pH conditions influence polymeric micelle stability and drug retention, the three co-loaded formulations (prepared using medium polymers with 7.5 mg of taxane and 1 mg of corticosteroid as feed amounts) were incubated in PBS-containing media at pH 7.4 (representing bloodstream) and 6 (representing the TME and endosomes) for 7 days (Fig. S9). Our results showed that all co-loaded micelles were stable in terms of size and PDI over time in both pH conditions, without any signs of aggregation. Drug retention over time followed a similar trend to that observed previously in the sink condition release study, without a clear effect of the pH on the release behavior. For both pH 7.4 and 6, taxanes were better retained in micelles than corticosteroids, except for the CIC-CTX pair.

Taken together, we show that the mPEG-b-p(HPMAm-Bz)-based polymeric micelles comprise a versatile and tunable nanoparticle platform for corticosteroids and taxanes co-delivery. Therefore, understanding which drug properties affect the retention in the delivery system can assist in optimizing nanomedicine design. A recent study by Varela-Moreira et al. [39] aimed to shed light on this matter by evaluating the micellar retention of 4 different compounds of various drug categories. Their results suggest that while the log P is important for drug encapsulation into and retention in the micelles, the number of aromatic rings of the drug is more strongly associated with its retention. Given the high retention observed for CIC in the present work, we aimed to deepen into this question using the findings from the three taxanes and three corticosteroids.

Considering that neither the feed amount of the loaded drugs nor their state of being single- or co-loaded into the micelles influenced their retention, we analyzed the obtained drug release data in order to evaluate the association between different structural and physicochemical properties of the drug and its retention in the micelles. The properties of the used taxanes and corticosteroids are provided in Table 1. HPLC chromatograms of all the 6 compounds using a non-polar C18 column (Fig. S10) confirmed the highly hydrophobic nature of CIC compared to the other compounds. To understand which drug property contributes best to its micellar retention, we plotted the percentage of drug retained in the micelles after 24 h as a function of different drug characteristics (log P, water solubility, molecular weight, number of aromatic rings, and number of π electrons) using single linear regression (Fig. 5A–E). Among the properties evaluated, log P and MW showed the strongest associations with drug retention in the micelles (R2 > 0.7), followed by water solubility (R2 > 0.6). Interestingly, our findings point to a weaker effect of the number of aromatic rings on the drug retention (R2 = 0.42) for taxanes and corticosteroids compared to previous observations for another group of drugs [39]. In our case, this result is greatly influenced by CIC, which does not contain aromatic rings but is efficiently retained in the micelles. Further analysis using drug retention values at a shorter time (after 6 h) also confirmed similar trends, with drug retention being best associated with log P values (R2 > 0.8, Fig. S11).

Associations between taxane and corticosteroid properties and their retention in micelles. A–E Drug retention in mPEG-b-p(HPMAm-Bz) polymeric micelles after 24 h as a function of log P (A), water solubility (B), molecular weight (MW) (C), number of aromatic rings (D), and number of π electrons (E). F Drug retention after 24 h as a function of both log P and MW (colored bar indicates drug retention)

Based on these data, we subsequently aimed to assess the combined effect of multiple drug properties on its retention in the micelles using multiple linear regression. The resulting analysis showed high multi-collinearity among the different properties assessed, which made the eventual modeling unreliable. To avert the problem, we only focused on log P and molecular weight, mainly as they are non-related properties and they both individually showed the best association with drug retention at 24 h. The effect of the two properties together on drug retention was analyzed using multiple linear regression (Fig. 5F), and the results showed a very high coefficient of determination (R2 > 0.9) and a low variance inflation factor (VIF) value (1.49). While having a high number of aromatic rings contributes to retention of the drugs in the mPEG-b-p(HMPAm-Bz) micelles, our findings demonstrate that it is certainly not a prerequisite. In addition, while it is important to take into account that statistically robust and predictive models cannot be obtained using our relatively small sample size (with only 6 drugs), our data do point towards a major contribution of molecular weight and particularly hydrophobicity (mainly log P, but also water solubility) to better drug retention; more prominently than previously assumed [39]. Collectively, it seems evident that drug retention in the micelles is not determined by only one individual feature. A combination of different structural and physiochemical properties modulates the interaction of the drug with the polymeric micelles, and the extent of each factor’s individual contribution can differ between various drug classes.