3.4.4. Common Subsets of Hard Corona Proteins in Plasma and Vitreous Samples

There were 22 common species among all the enriched proteins found in the plasma HCs of more than one liposome (Supplementary File 6). Among the seven proteins enriched in the HCs of all formulations, in addition to FGA and FGB, C3, and immunoglobulin constant fragments, were serotransferrin (TF) and alpha-2-macroglobulin (A2M), an iron transfer protein and protease inhibitor that are both highly abundant in plasma. In vitreous humor, only three proteins were common to the four formulations, including two structural crystallins and lactotransferrin (LTF), which is involved in iron transfer and immune functions (Supplementary File 6). Overall, the vitreous common subset included 14 proteins, such as glyceraldehyde-3-phosphate dehydrogenase (GADPH), clusterin (CLU), albumin (ALB), and alpha-enolase (ENO1), the latter being associated with glucose metabolism and a number of functions including coagulation and immune stimulation.

Figure 8 displays the commonality between plasma and vitreous for the common subsets for the four liposomes tested in both biological matrices. It demonstrates that the liposomes interact with 30–34 common proteins across the matrices, 74% of which are common to all formulations tested in both plasma and vitreous samples. These 28 proteins are listed in Figure 8B along with the corresponding top enriched biological functions in Figure 8C. Plasma HC common subset proteins carry a negative charge at physiological pH and are dominantly hydrophilic, similar to the top 20 most abundant proteins (Supplementary File 6). The same holds for the vitreous common subset, which was composed of smaller-molecular-weight proteins with fewer hydrophobic residues on average. Among both common subsets, the retinol-binding protein 3 (RBP3) in the vitreous subset was the only one with a positive hydropathicity value. It is a visual cycle glycoprotein with a significantly higher molecular weight at 135 kDa compared to the 49 kDa mean. Although it also had the lowest pI, it was highly enriched on the 50 nm anionic liposome HCs, despite being mostly present in the source and SCs of the other formulations (Figure *Pharmaceutics* **2020** 6). , *12*, x FOR PEER REVIEW 16 of 25

**Figure 8.** (**A**) Venn diagram of the commonalities between all proteins identified in the plasma and vitreous HCs of liposome formulations F2, F4, F6, and F7 without accounting for relative enrichment. (**B**) Proteins of the common subset. (**C**) Enriched biological functions in the common subset. AC: protein accession code; FDR: false discovery rate. **Figure 8.** (**A**) Venn diagram of the commonalities between all proteins identified in the plasma and vitreous HCs of liposome formulations F2, F4, F6, and F7 without accounting for relative enrichment. (**B**) Proteins of the common subset. (**C**) Enriched biological functions in the common subset. AC: protein accession code; FDR: false discovery rate.

(Figure 9). Interestingly, an analysis of the *outlier* proteins that were specific to HA-coated liposomes with ICG (F4) in plasma showed the enrichment of collagen binding and extracellular matrix organization, with 67% of the predicted network explained by physical interactions (Figure S8; Table S3, Supplementary Materials). The small anionic liposomes were enriched with glycerolipid metabolism and actin filament organization functions, whereas the top enriched functions on neutral liposomes in plasma were related to glucose metabolism and immune activation. Overall, the top 20 HC enrichment patterns in plasma are mainly associated with blood microparticle and plasma lipoprotein particle-related functions, suggesting that they are secreted proteins that interact with lipid membranes (Figure S9; Table S4, Supplementary Materials). While complement activation did not rank highly, many of the proteins present in the common subset are associated with blood microparticle or plasma lipoprotein particles, or both, in addition to their immune system role (e.g., clusterin, CLU). Many of these proteins are also physical interaction partners. Glycolysis, visual perception, and blood microparticle were among the highest ranked functions in vitreous humor

(Tables S5 and S10, Supplementary Materials).

All of the common subset proteins were also among the top 20 most abundant proteins in plasma

All of the common subset proteins were also among the top 20 most abundant proteins in plasma and vitreous samples (Figure 9). A list of the proteins and their properties in plasma and vitreous is provided in Supplementary Files 3 and 4. There is a high overlap between HC and SC identifications (Figure 9). Interestingly, an analysis of the *outlier* proteins that were specific to HA-coated liposomes with ICG (F4) in plasma showed the enrichment of collagen binding and extracellular matrix organization, with 67% of the predicted network explained by physical interactions (Figure S8; Table S3, Supplementary Materials). The small anionic liposomes were enriched with glycerolipid metabolism and actin filament organization functions, whereas the top enriched functions on neutral liposomes in plasma were related to glucose metabolism and immune activation. Overall, the top 20 HC enrichment patterns in plasma are mainly associated with blood microparticle and plasma lipoprotein particle-related functions, suggesting that they are secreted proteins that interact with lipid membranes (Figure S9; Table S4, Supplementary Materials). While complement activation did not rank highly, many of the proteins present in the common subset are associated with blood microparticle or plasma lipoprotein particles, or both, in addition to their immune system role (e.g., clusterin, CLU). Many of these proteins are also physical interaction partners. Glycolysis, visual perception, and blood microparticle were among the highest ranked functions in vitreous humor (Tables S5 and S10, Supplementary Materials). *Pharmaceutics* **2020**, *12*, x FOR PEER REVIEW 17 of 25

**Figure 9.** Venn diagrams of the commonalities between the neutral 100 nm (F1–F3 or F2), anionic 100 nm HA-coated (F4–F5 or F4), and anionic 50 nm (F6–F7) groups. Plasma: the top 20 most abundant proteins in the plasma HC (Top 20 HC) and SC (Top 20 SC), and their comparison with the common subsets (Top 20 vs common subsets), and for all HC proteins in found in plasma samples (HC intersection). Vitreous: the top 20 most abundant proteins in the vitreous HC (Top 20 HC) and SC (Top 20 SC), and their comparison with the common subsets (Top 20 vs common subsets), and for all HC proteins in found in vitreous samples (HC union). **Figure 9.** Venn diagrams of the commonalities between the neutral 100 nm (F1–F3 or F2), anionic 100 nm HA-coated (F4–F5 or F4), and anionic 50 nm (F6–F7) groups. Plasma: the top 20 most abundant proteins in the plasma HC (Top 20 HC) and SC (Top 20 SC), and their comparison with the common subsets (Top 20 vs common subsets), and for all HC proteins in found in plasma samples (HC intersection). Vitreous: the top 20 most abundant proteins in the vitreous HC (Top 20 HC) and SC (Top 20 SC), and their comparison with the common subsets (Top 20 vs common subsets), and for all HC proteins in found in vitreous samples (HC union).

#### *3.5. Vitreal Mobility 3.5. Vitreal Mobility*

The vitreal mobility of HA-coated liposomes was compared against their PEGylated counterparts (Table 2) [29]. While there were no significant differences in the mean track speeds between the PEG- and HA-coated liposomes, both formulations had higher overall speeds than the uncoated liposome (*p* < 0.05; Figure 10B). The vitreal diffusion coefficient (Dv), determined based on the mean square displacement by single-particle tracking analysis, showed that the HA-coated liposome diffused 1.5 times less than the PEG-coated liposomes despite their negative surface charge (Figure 10A). The diffusivity of the HA-coated liposomes was reduced 20-fold compared to their theoretical diffusion in water (Dw), indicating that they encounter more resistance in the vitreous humor compared to the PEGylated and uncoated liposomes (Dw/D<sup>v</sup> = 13 and 16, respectively). The vitreal mobility of HA-coated liposomes was compared against their PEGylated counterparts (Table 2) [29]. While there were no significant differences in the mean track speeds between the PEGand HA-coated liposomes, both formulations had higher overall speeds than the uncoated liposome (*p* < 0.05; Figure 10B). The vitreal diffusion coefficient (Dv), determined based on the mean square displacement by single-particle tracking analysis, showed that the HA-coated liposome diffused 1.5 times less than the PEG-coated liposomes despite their negative surface charge (Figure 10A). The diffusivity of the HA-coated liposomes was reduced 20-fold compared to their theoretical diffusion in water (Dw), indicating that they encounter more resistance in the vitreous humor compared to the

formulations. D<sup>v</sup> was obtained experimentally by single-particle tracking for 1 s, and the D<sup>w</sup> was

HA 116.1 0.008 0.28 ± 0.14 5.68 20.2 PEG\* 107.6 0.086 0.47 ± 0.25 6.12 13.0 Uncoated\* 125.8 0.035 0.33 ± 0.17 5.27 16.0

**/s) D<sup>w</sup> (μm<sup>2</sup>**

**/s) Dw/D<sup>v</sup>**

calculated based on the Stoke–Einstein equation (*N* = 3). \* Data from Reference [29].

**Formulation Mean Diameter (nm) PdI D<sup>v</sup> (μm<sup>2</sup>**

PEGylated and uncoated liposomes (Dw/D<sup>v</sup> = 13 and 16, respectively). Nonetheless, the mobility hindrance observed with the HA-coated liposome was minimal, as evident by the representative Brownian trajectories (Figure 10C).

**Table 2.** Comparison of the vitreal diffusion of HA-coated, PEG-coated, and uncoated liposome formulations. D<sup>v</sup> was obtained experimentally by single-particle tracking for 1 s, and the D<sup>w</sup> was calculated based on the Stoke–Einstein equation (*N* = 3). \* Data from Reference [29].


18 of 25

**Figure 10.** Comparison between (**A**) vitreal diffusion of HA- and PEG-coated light-triggered liposomes, analyzed by unpaired *t*-test (\* *p* < 0.05); (**B**) ensemble-average speed of HA-coated, PEGylated, and uncoated light-activated liposomes (τ = 1 s). Means with significant differences are shown as analyzed by ordinary one-way ANOVA with Tukey's multiple comparison test (\*\* *p* < 0.01). Error bars indicate SD. (**C**) Representative trajectories of the HA-coated liposome in the intact porcine vitreous. Inset shows the zoomed-in view of tracks obtained using the single-particle tracking **Figure 10.** Comparison between (**A**) vitreal diffusion of HA- and PEG-coated light-triggered liposomes, analyzed by unpaired *t*-test (\* *p* < 0.05); (**B**) ensemble-average speed of HA-coated, PEGylated, and uncoated light-activated liposomes (τ = 1 s). Means with significant differences are shown as analyzed by ordinary one-way ANOVA with Tukey's multiple comparison test (\*\* *p* < 0.01). Error bars indicate SD. (**C**) Representative trajectories of the HA-coated liposome in the intact porcine vitreous. Inset shows the zoomed-in view of tracks obtained using the single-particle tracking technique.

#### technique. **4. Discussion**

PEGylated counterparts [13,39].

**4. Discussion** DSPE was conjugated with HA in-house to prepare hyaluronated liposomes, since there was no pre-existing conjugate available commercially. The successful reductive amination method was based on the work of Yao et al. [32]. The HA-coating on the liposomes did not significantly change the size, ICG encapsulation efficiency, or phase transition temperature of the liposomes (Table 1; Figure 1A). In the presence of hydrophilic coatings such as PEG, the ICG is stabilized by polymer clusters on the external side of the lipid bilayer [21]. Doubling the amount of both HA and ICG doubled the encapsulated amount of ICG without affecting the bilayer stability, while doubling ICG alone destabilized the membrane, causing passive leakage at 37 °C. This is in accordance with our previous results with PEG-coated ICG liposomes and suggests that HA is at least as efficient as PEG in stabilizing ICG [20]. The change in the surface coating of the 100 nm liposomes did not affect the temperature-induced release of calcein, although the small liposomes (~70 nm) had slightly higher release temperatures compared to the larger HA-coated liposomes (Figure 1B). A similar observation was previously made with PEG-coated ICG liposomes, which may be caused by curvaturemodulated phase separation in the lipid bilayer [18]. This observation is supported by the light-DSPE was conjugated with HA in-house to prepare hyaluronated liposomes, since there was no pre-existing conjugate available commercially. The successful reductive amination method was based on the work of Yao et al. [32]. The HA-coating on the liposomes did not significantly change the size, ICG encapsulation efficiency, or phase transition temperature of the liposomes (Table 1; Figure 1A). In the presence of hydrophilic coatings such as PEG, the ICG is stabilized by polymer clusters on the external side of the lipid bilayer [21]. Doubling the amount of both HA and ICG doubled the encapsulated amount of ICG without affecting the bilayer stability, while doubling ICG alone destabilized the membrane, causing passive leakage at 37 ◦C. This is in accordance with our previous results with PEG-coated ICG liposomes and suggests that HA is at least as efficient as PEG in stabilizing ICG [20]. The change in the surface coating of the 100 nm liposomes did not affect the temperature-induced release of calcein, although the small liposomes (~70 nm) had slightly higher release temperatures compared to the larger HA-coated liposomes (Figure 1B). A similar observation was previously made with PEG-coated ICG liposomes, which may be caused by curvature-modulated phase separation in the lipid bilayer [18]. This observation is supported by the light-activated release study, where small (~70 nm) HA-coated liposomes released less of their contents than the larger ones in response to a 5 s light exposure (Figure 1C).

activated release study, where small (~70 nm) HA-coated liposomes released less of their contents than the larger ones in response to a 5 s light exposure (Figure 1C). Notably, while light-activated drug release from 100 nm HA-coated liposomes was comparable to the PEGylated liposomes in buffer, plasma, and vitreous samples, the HA coating improved liposomal stability in plasma at 37 °C (Figure 2C–D). Vitreal stability was comparable for both Notably, while light-activated drug release from 100 nm HA-coated liposomes was comparable to the PEGylated liposomes in buffer, plasma, and vitreous samples, the HA coating improved liposomal stability in plasma at 37 ◦C (Figure 2C,D). Vitreal stability was comparable for both formulations over the period of one week. Following a 3-h incubation period in vitreous and plasma samples, the 15 s light

bilayer-destabilizing effect of the biocorona on thermosensitive liposomes [37]. Therefore, the better plasma stability of the HA-coating especially within the first 12 h permits longer delays between intravenous (i.v.) administration and light activation, which is a beneficial property from a therapeutic perspective. The HA coating also made the liposomes significantly anionic (−11 mV), caused by the deprotonation of disaccharide carboxyl groups in an aqueous environment [38]. The analysis of the top 20 most abundant corona protein properties showed that the HA-coated liposomes were predominantly enriched with proteins that carry a negative charge in both plasma and vitreous pH, suggesting that the liposome anionic charge is also retained after coronation (Figure 7). This may partly explain the improvement in the stealth properties of HA-coated liposomes compared to their exposure increased release by as much as 44% and 69% compared to passive leakage. The light-activated release was more pronounced in plasma, possibly due to the lipid bilayer-destabilizing effect of the biocorona on thermosensitive liposomes [37]. Therefore, the better plasma stability of the HA-coating especially within the first 12 h permits longer delays between intravenous (i.v.) administration and light activation, which is a beneficial property from a therapeutic perspective. The HA coating also made the liposomes significantly anionic (−11 mV), caused by the deprotonation of disaccharide carboxyl groups in an aqueous environment [38]. The analysis of the top 20 most abundant corona protein properties showed that the HA-coated liposomes were predominantly enriched with proteins that carry a negative charge in both plasma and vitreous pH, suggesting that the liposome anionic charge is also retained after coronation (Figure 7). This may partly explain the improvement in the stealth properties of HA-coated liposomes compared to their PEGylated counterparts [13,39].

In human plasma, ~5 nm HCs and ~10 nm SCs formed on the HA and PEG-coated liposomes, which correspond to sparse monolayers [28,40]. The formulation-dependent compositions also clustered based on liposomal surface charge (Figure 5). The influence of nanoparticle surface charge on the HC composition and the enrichment of fibrinogens under in vitro conditions were reported [41,42]. ICG also influences corona composition, as evidenced by the higher relative enrichment of complement and immunoglobulin proteins on the 8–15 kDa HA-coated liposome with ICG (F4) than without ICG (F5). This is likely due to trace endotoxins, since <400-kDa HA is not immunogenic [43]. The enrichment of immunoglobulin variable regions may be due to the particular epitope repertoire of the plasma donors [44].

In undiluted vitreous, the thinner ~2 nm HCs and ~3 nm SCs had formulation-independent compositions (Figure 4C,D). The slightly higher SC thicknesses but significantly different composition compared to the HCs suggest that it consists of only a few loosely interacting molecules. Interestingly, the 100 nm HA-coated formulation bound more proteins than the PEG-coated liposome in vitreous humor, while the reverse was seen in plasma. Interactions with the structural HA–collagen meshwork of the eye involve alpha and beta-crystallins, both abundant in vitreal coronas, as well as fibronectin and the inter-alpha trypsin inhibitors found in both plasma and vitreous HCs (Figure 8B). In plasma, the proteins that were only found on the HA-coated liposomes interact with fibrillar collagen type I and cell-adhesion protein collagen type XIV (Figure S8; Table S3, Supplementary Materials). These functions were not enriched in vitreous humor, possibly due to the low number of identified proteins on the HA-coated liposomes (Figure S10, Table S5, Supplementary Materials). Therefore, the "biological identity" ensuing from the HA coating may result in physical interactions with the structural collagen of the eye [45].

Consequently, these interactions may explain the slightly slower vitreal mobility of the HA-coated liposomes compared to PEG-coated liposomes. Since the increase in liposome diameter as a result of corona formation did not influence the diffusion of 50 nm anionic pegylated liposomes [29], it is inconsequential for neutral 100 nm liposomes at similar corona thicknesses. Although anionic nanoparticles diffuse freely in the vitreous humor [46], PEG-coated nanoparticles diffuse faster, possibly as a result of a "shielding" effect against charge-mediated interactions with the meshwork [47]. The HAand PEG-coated liposomes diffused at comparable speeds, but significantly faster than the uncoated formulation (Figure 10). The random walk diffusion patterns indicate that particles with higher speeds do not necessarily relocate longer distances (Figure 10C). This is evident based on the comparable diffusion coefficients of the HA-coated and uncoated liposomes, despite the faster mean track speed of HA-coated liposomes. The more heterogeneous track speed distribution of the HA-coated liposomes may be explained by vitreous meshwork interactions.

The HA-coated liposomes had a low number of identified proteins with a narrow range of fold-changes in vitreous humor (Supplementary File 2). However, as a result of median normalization, proteins with high abundances in individual replicates could be identified. The consistency (36–38% of the highest number of identifications) suggests that this is a consequence of physical interactions, such as protein complex binding to the liposomes. Stable physical interactions between proteins are

needed for cellular processes, and evolutionarily conserved soluble complexes are abundant in the human proteome [48]. The vitreous humor is abundant in energy metabolism-related proteins, such as GAPDH, which interact with other proteins to exert their functions [49]. Earlier, we found that 20–22% of the highly enriched vitreal corona proteins on 50 nm anionic liposomes were physical interaction partners of GAPDH [29]. These showed a preference for the PEGylated formulation, as can also be seen in Figure 6A (F7). It is, therefore, possible that some of the less enriched proteins interact with GAPDH or other HC proteins instead of the nanoparticle surface, as demonstrated with complement protein C3 [50]. Protein complexes should be considered a factor that influences corona composition and structure.

It was proposed that protein complexity of the HC increases with nanoparticle size, and small nanoparticles bind proteins with more aromatic residues [51]. In our study, the 50 nm liposomes did not bind proteins above 200 kDa, which may reflect the effect of surface curvature, but this also decreased the number of aromatic residues. In vitreous humor, the uncoated anionic 50 nm liposome (F6) had the most distinct HC (50% unique protein identifications), and the small anionic liposome group had slightly more outliers than the other groups in plasma and vitreous samples (Figure 9). Therefore, our findings suggest an inverse relationship with HC complexity and liposome size. In addition, the properties of the top 20 most abundant proteins in HC or SC did not correlate with liposome size or charge (Figure 7 and Figure S7, Supplementary Materials). As expected based on earlier studies [29,30], there was also no correlation with liposome properties and the corona composition in vitreous, although these liposomes had formulation-dependent coronas in plasma. However, our analysis confirmed that the HC and SC proteins are predominantly hydrophilic and carry an anionic charge in both plasma and vitreous samples. Similar observations were made in diluted plasma [42], as well as in diluted and undiluted vitreous humor with more limited sample sets [29,30]. Our results indicate that this trend is more universal than previously known.

Biocorona formation is influenced by nanoparticle-related and experimental factors [27,44] together with the biological environment, such as the nature and amount of its constituents and ionic strength [52]. However, the volume available to proteins influences all aspects of their behavior, including adsorption processes [53]. As a result, incubation in diluted and undiluted solutions results in different corona compositions and thicknesses [54]. Except for Digiacomo et al. [54], molecular crowding received little attention in the context of biocorona formation. Since there is a significant difference in plasma (~76 mg/mL) and vitreous (~1.5 mg/mL) protein concentrations, crowding agents such as structural HA are important modulators of ocular protein interactions [52,53,55]. Molecular crowding directly influences thermodynamics and diffusion, the drivers of protein adsorption and corona formation [56,57]. It is, therefore, likely that the differences in corona formation between plasma and vitreous samples, at physiological concentrations due to the use of undiluted fluids, are explained partly by structural HA. This would also explain the need for longer LC separation gradients and the lower number of protein identifications on HA-coated liposomes in vitreous humor. It should be noted that porcine vitreous humor is the best substitute to human vitreous humor in terms of its protein and HA concentrations [29] and viscoelastic properties [58].

The most abundant corona proteins most likely determine the ensuing biological responses, due to the higher probability of encountering cellular receptors in the correct orientation [26,59]. It is noteworthy that there is a common subset of 30–34 proteins in both plasma and vitreous samples, 74% of which were found in all of the HCs, including on 100 nm PEG-coated and HA-coated liposomes and on 50 nm uncoated and PEG-coated liposomes (Figure 8). Functional analysis showed an association with blood microparticle and immune activation functions (Figure 8C), which are also among the enriched functions for the top 20 HC proteins in plasma (Figure S9; Table S4, Supplementary Materials). Considering that nanoparticle aggregation and protein adsorption are driven by surface free energy, it is interesting to note the "double role" of lung surfactant proteins, where they participate in immune surveillance in addition to reducing alveolar surface tension [52]. The common subset, therefore, labels the liposomes "non-self" lipid particles regardless of the biological environment (plasma or

vitreous), a possible consequence of combined surfactant and immune activity in response to the "abnormally" high surface energy of cell-mimetic lipid bilayers. This may also explain why the lipid corona is insensitive to nanoparticle properties [60,61].

The two-fold increases in hydrodynamic diameters measured with LALS results indicate aggregation (Figure 4B) [57]. Although the HA-coated liposomes showed comparable or improved stability in vitreous and plasma samples, respectively (Figure 2), they may be more susceptible to clustering and aggregation than the PEG-coated liposomes. This may also be influenced by the higher abundance of immunoglobulins [52,62]. However, protein adsorption did not influence the liposome stability or mobility in vitreous humor, regardless of the formulation. Future studies should explore the mechanism of biocorona formation in the ocular environment together with the ensuing biological responses, including elimination by ocular phagocytes and retinal cell uptake [63].
