Surface Topography of Polyethylene Glycol Shell Nanoparticles Formed from Bottlebrush Block Copolymers Controls Interactions with Proteins and Cells

Abstract

Although poly(ethylene glycol) (PEG) is commonly used in nanoparticle design, the impact of surface topography on nanoparticle performance in biomedical applications has received little attention, despite showing significant promise in the study of inorganic nanoparticles. Control of the surface topography of polymeric nanoparticles is a formidable challenge due to the limited conformational control of linear polymers that form the nanoparticle surface. In this work, we establish a straightforward method to precisely tailor the surface topography of PEGylated polymeric nanoparticles based on tuning the architecture of shape-persistent amphiphilic bottlebrush block copolymer (BBCP) building blocks. We demonstrate that nanoparticle formation and surface topography can be controlled by systematically changing structural parameters of BBCP architecture. Furthermore, we reveal that the surface topography of PEGylated nanoparticles significantly affects their performance. In particular, the adsorption of a model protein and the uptake into HeLa cells were closely correlated to surface roughness and BBCP terminal PEG block brush width. Overall, our work elucidates the importance of surface topography in nanoparticle research as well as provides an approach to improve the performance of PEGylated nanoparticles.

Graphical Abstract

Nanoparticle-based therapeutics and diagnostics offer immense potential in a wide range of biomedical applications. For example, drug-loaded nanoparticles have garnered great interest in drug delivery as a strategy to improve the pharmacokinetics of drugs.¹ However, the impact of nanoparticle-based drug delivery in clinical medicine has been limited with only a small number of FDA-approved formulations to date.^2,3 Most formulations tested in clinical trials ultimately fail to show beneficial effects due to their rapid blood clearance and low drug delivery efficiency.^4,5 A widely used method to enhance the blood circulation time of nanoparticles is based on coating the surface with a dense layer of poly(ethylene glycol) (PEG), which reduces opsonization and subsequent clearance by the mononuclear phagocytosis system.⁶ While this stealthy PEG-coating effectively inhibits the adsorption of serum proteins on the nanoparticle surface via steric repulsion, the same characteristic inherently reduces their uptake into cells.⁷ Several physicochemical parameters of PEG-coated nanoparticles such as size,⁸ shape,⁹ rigidity,^10,11 and surface functionality^12,13 have been investigated as methods to modulate cell uptake and protein adsorption. However, only a small number of studies have focused on surface topography as a strategy to optimize nanoparticle design.

Tuning the surface topography of inorganic nanoparticles has resulted in improvements in nanoparticle pharmacokinetics, including cell uptake.^14–19 For example, Wang et al. observed greatly enhanced circulation half-lives and cellular internalization of mesoporous silica nanoparticles with virus-mimicking spiky surface topography compared to a smooth-surfaced control.²⁰ But, despite their frequent use in biomedical applications, the effect of surface topography is poorly understood in the context of PEGylated polymeric nanoparticles. This may be in part due to the difficulty of controlling the surface topography and morphology of polymeric nanoparticles. Linear polymer building blocks readily undergo conformational changes if exposed on the nanoparticle surface due to their flexible nature. Thus, it is challenging to modulate the surface of nanoparticles without affecting the polymer conformation. In particular, PEG chains transition from mushroom to brush conformation as their surface density increases, leading to drastically reduced serum protein adsorption and macrophage uptake.⁶ The outermost shell surface PEG density has been found to correlate to the biological fate of nanoparticles highlighting the importance of controlling the surface properties of PEG-shell nanoparticles.^21,22 Zhou et al. demonstrated that a second PEG layer can be conjugated onto PEGylated nanoparticles to create a dynamic topographical structure that prolongs circulation times in vivo.²³ However, their results were attributed to the combined effect of conformational fluctuations of the low-density outer PEG layer and the steric repulsion of the dense inner PEG layer rather than surface topography. To independently evaluate the effect of surface topography on the nanoparticle’s biological fate, it is critical to alter surface topography without changing the density and conformation of PEG chains on the nanoparticle surface.

Here, we propose a method to control the surface topography of PEGylated polymeric nanoparticles based on amphiphilic bottlebrush block copolymers (BBCPs) that are composed of linear poly(lactic acid) (PLA) and PEG side chains, respectively, attached to the two blocks of a polymeric backbone. Bottlebrush polymers exhibit a shape-persistent worm-like morphology due to their semi-rigid backbone, with side chains in an extended brush-like conformation given their high grafting density.²⁴ In an aqueous environment, amphiphilic BBCPs readily self-assemble into micellar nanoparticles in a well-controlled manner that can be predicted from a cone-filling perspective based on the length of the hydrophobic block bottlebrush backbone and cross-sectional area of the hydrophilic-hydrophobic interface.^24–26 The average number of BBCPs forming a nanoparticle (aggregation number) has been shown to increase with decreasing hypothetical cone angle caused by either a longer hydrophobic block backbone or a smaller interfacial cross-sectional area. Additionally, recent studies demonstrated the efficient encapsulation of various molecules inside the core of nanoparticles assembled from BBCPs.^25,27,28 We hypothesized that by manipulation of the BBCP architecture we would be able to tailor the nanoparticle surface topography without changing the PEG conformation thanks to the shape-persistent nature of the BBCP backbone and the fixed PEG side chains density. In this study, we identified BBCP terminal PEG block brush width, hydrophilic/hydrophobic block backbone length, and interfacial asymmetry as the main structural parameters of BBCP architecture to control surface topography (Figure 1). We used this architectural framework in a systematic study of diblock BBCPs with 1) similar PEG/PLA side chain length (symmetric), 2) dissimilar PEG/PLA side chain length (asymmetric), and 3) triblock BBCPs with additional interfacial PEG side chain of different length (triblock).

Figure 1.

Effect of BBCP architecture on nanoparticle surface topography. Structural parameters of BBCP architecture and types of BBCPs investigated in this study. Blue and green segments represent PEG and PLA side chains, respectively, attached to a polynorbornene backbone (red spheres).

To underscore the biological relevance of these structures, we observed that nanoparticle cell uptake and the adsorption of a model protein is closely correlated to surface topography. Furthermore, our findings provide a facile strategy to improve the performance of PLA-PEG nanoparticles in drug delivery as well as shed light on the importance of surface topography in the PEGylation of nanomaterials.

RESULTS AND DISCUSSION

Polymer Design.

BBCP synthesis was based on ring-opening metathesis polymerization (ROMP) due to its excellent control of BBCP composition, nearly quantitative monomer conversion, and narrow molecular weight distribution.^28,29 PEG macromonomers (PEG-MM) and PLA macromonomer (PLA-MM) containing polymerizable norbornene anchor groups were polymerized using sequential graft-through ROMP (Figure 2a). Given the controlled process of ROMP, the BBCP length and the backbone degree of polymerization (DP) of each block was precisely tuned by adjusting the molar ratio of macromonomer to initiator, i.e., the third-generation bispyridyl Grubbs catalyst (G3). PLA-MM with a number-average molecular weight (M_n) of 2.6 kDa was synthesized via 1,8-diazabicyclo[5.4.0]-undec-7-ene (DBU)-catalyzed ring-opening polymerization of D,L-lactide initiated from a hydroxy-norbornene derivative. Three different lengths of PEG-MM (S = 1.1 kDa, M = 2.3 kDa, L = 3.5 kDa) were prepared by attaching methoxy-PEG-amine to a norbornene-N-hydroxysuccinimidyl-ester following a previously reported method.³⁰ All BBCPs were designed with a target backbone DP of 80 to obtain nanoparticles of approximately similar size. Samples were named according to the BBCP type (Symmetric, Asymmetric, Triblock), their PEG side chain length (Short, Medium, Long) followed by the targeted PEG block backbone DP. In particular, we focused on four symmetric BBCPs (Sy-M20, Sy-M40, Sy-M60, Sy-M70), two asymmetric (Asy-S60, Asy-L60) as well as three triblock BBCPs with terminal and interfacial PEG block side chain of different length (Tri-S40-M20, Tri-S40-L20, Tri-L40-M20). To evaluate the effect of BBCP architecture on nanoparticle surface topography, we further categorized the BBCPs based on four structural parameters: 1) ratio of backbone DP of the PEG and PLA blocks (backbone ratio), 2) side chain asymmetry of diblock BBCPs (asymmetry), 3) interfacial side chain asymmetry of triblock BBCPs with similar terminal block side chain (interfacial asymmetry), and 4) brush width of the terminal PEG block (brush width).

Figure 2.

Synthesis of PLA-PEG BBCPs. (a) Schematic representation of BBCP synthesis via ROMP of PEG-MM and PLA-MM followed by BBCP self-assembly in water as well as BBCP architectures investigated (bottom left). Samples are categorized based on the structural parameter that is varied. The approximated simplified structure of each BBCP is depicted next to the sample name where the blue part represents the PEG segments and the green part the PLA segment. (b) SEC chromatograms of Tri-S40-L20 after formation of first block (ROMP I), second block (ROMP II), and final BBCP (ROMP III).

Polymer Characterization.

BBCPs were characterized via size exclusion chromatography (SEC) and proton nuclear magnetic resonance (¹H NMR). SEC analysis revealed successful chain extension with monomodal molecular weight distribution and a negligible amount of unreacted intermediate bottlebrush polymer after the addition of the second and third block, respectively (Figure 2b). Near-complete consumption of macromonomer after quenching the polymerization with ethyl vinyl ether was confirmed by the absence of the norbornene vinyl peak at 6.3 ppm in ¹H NMR (Figure S1). The BBCP molecular weight and backbone DP was determined via a combination of SEC equipped with a multi-angle light scattering detector (SEC-MALS) in DMF (1 g/L LiBr) of the PEG block (dn/dc = 0.055) before the addition of PLA-MM and ¹H NMR of the final BBCP (Table 1). All BBCPs were obtained at close to the targeted DP for the individual block backbone length, as well as the total backbone length, with low dispersity.

Table 1.

Characterization of PLA-PEG BBCPs

^aSubscript denotes side chain number-average molecular weight (M_n) followed by the targeted backbone DP for each block.

^bM_n of BBCP obtained via combination of SEC-MALS and ¹H NMR.

^cNumber of side chains in BBCP backbone for each block.

^dInterfacial asymmetry based on M_n of PEG/PLA side chain at hydrophilic-hydrophobic block interface.

Nanoparticle Formation.

BBCP self-assembly was first attempted via a direct dissolution method proposed by Fenyves et al. which involves the hydration of a polymer film above the glass transition temperature (T_g) of PLA (48 °C) to facilitate micelle formation.²⁴ However, BBCPs with large PLA content did not form nanoparticles as made evident by dynamic light scattering (DLS) and visible aggregates. We hypothesized that BBCPs with unfavorable symmetry do not readily self-assemble into spherical particles due to the chain stretching required for the hydrophobic block to adopt a cone-shaped morphology.³¹ Thus, an ultrasonication step was added to kinetically drive the self-assembly of BBCPs into their thermodynamically stable morphology through acoustic cavitation-induced reorganization of BBCPs in a metastable state.^32,33 In particular, short sonication (15 min) resulted in the lowest polydispersity (PdI) and size without affecting the BBCP whereas signs of polymer degradation were observed in SEC and NMR at longer sonication times (Figure S2). Using this modified method, highly monodisperse nanoparticles around 100 nm in size according to DLS (PdI < 0.2) were obtained except for Sy-M20 (Table 2). Despite extensive sonication of Sy-M20, large aggregates remained in solution and no nanoparticle formation was observed in DLS. We suspect that the small amount of PEG in combination with the long backbone of the PLA block could not accommodate a spherical morphology and alternative structures of larger size formed.³⁴ Notably, a diameter of 100 nm corresponded to approximately twice the length of a BBCP with 80 macromonomers polymerized (~0.6 nm per repeat unit of the backbone), implying that nanoparticles with uniform micellular structure were formed and that their size can be well controlled by the BBCP backbone DP.³⁵ After nanoparticle formation and DLS analysis, the samples were concentrated and washed with water three times using a centrifuge filter which effectively removed any trace amount of macromonomer and intermediate PEG bottlebrush polymer that did not get incorporated into the nanoparticle as indicated by the absence of low molecular weight peaks in the SEC traces of purified nanoparticles (Figure S3).

Table 2.

Characterization of nanoparticles assembled from PLA-PEG BBCPs in water

^aZ-average DLS hydrodynamic diameter.

^bDetermined from cryo-TEM imaging (mean ± SD).

^cObtained from static light scattering and Zimm plot analysis.

^dDetermined by ¹H NMR (PEG proton peak intensity of nanoparticles in D₂O before and after addition of acetone-d₆ (mean ± SD of n = 3 independent replicates).

Nanoparticle Characterization.

A slightly negative zeta potential was observed for all nanoparticles with increasingly more negative values for nanoparticles with lower PEG content (Table 2) in agreement with previous reports.³⁶ Static light scattering was used to examine the effect of BBCP architecture on nanoparticle molar mass and aggregation number via Zimm plot analysis. Furthermore, the aggregation number was determined from the ratio of the weight-averaged nanoparticle molar mass and the number-averaged molecular weight of the BBCP from SEC and NMR analysis. Despite only small variations in side chain length and block backbone DP ratio, a wide range of aggregation numbers was observed (Table 2). The aggregation number for the series of symmetric BBCPs with varying backbone ratio (Sy-M40, Sy-M60, Sy-M70) increased proportionally as predicted from the increase in PLA backbone DP according to the cone-filling model. For diblock BBCPs with constant PLA backbone length (Asy-S60, Sy-M60, Asy-L60), increasing the length of the interfacial block PEG side chain led to a drastic decrease in aggregation number from 1958 to 59 following the increase in side chain asymmetry. A similar trend was observed for the series of BBCPs with short 1.1 kDa PEG as the terminal block side chain (Asy-S60, Tri-S40-M20, Tri-S40-L20) where the aggregation number decreased as a longer interfacial block side chain was used. It is noteworthy that triblock BBCPs had larger aggregation numbers than corresponding diblock BBCPs with a side chain length similar to the triblock interfacial block side chain, implying that the hydrophilic terminal block side chain on the nanoparticle surface has an effect on self-assembly not considered in the cone-filling model but predicted from dissipative particle dynamics simulations.^34,37,38 Our results also suggest that controlled interfacial side chain asymmetry can be successfully applied to increase the aggregation number for nanoparticles made from BBCPs of large terminal PEG block brush width by incorporating a shorter PEG side chain block at the PLA interface, as in Asy-L60 and Tri-L40-M20, respectively.

Cryogenic transmission electron microscopy (cryo-TEM) was used to image nanoparticles in their native hydrated state showing highly monodisperse nanoparticles with individual PEG bottlebrush polymers protruding from the solid PLA core (Figure 3). Furthermore, differences in brush width between 1.1 kDa, 2.3 kDa, 3.5 kDa PEG side chains as well as nanoparticle surface topography were visible. The cryo-TEM images closely match DLS and SLS analysis in terms of nanoparticle size and aggregation number with the exception of Asy-S60. A mixture of spherical and elongated nanoparticles was observed, whereas a unimodal peak in DLS suggests a monodisperse size distribution with a hydrodynamic diameter close to that of the spherical particles. We suspect that Asy-S60 exhibits uniform spherical morphology after self-assembly but coalesces into elongated nanoparticles during cryo-TEM sample preparation. In fact, the high sample concentration required for cryo-TEM, along with the shear stress caused by the blotting process, have been observed to induce morphology transitions.^39–41

Figure 3.

Representative cryo-TEM images of nanoparticles assembled from PLA-PEG BBCPs. The scale bars represent 100 nm.

Nanoparticle core diameters obtained via cryo-TEM ranged between 15 and 50 nm. For symmetric BBCPs Sy-M40, Sy-M60, and Sy-M70, the core size is directly proportional to the PLA block backbone DP and approximately doubles as the PLA block backbone length is doubled. However, the other nanoparticles with comparable PLA block backbone DP of approximately 20 varied in their core diameter beyond what can be expected solely from the differences in backbone DP according to SEC analysis. Core size and hydrodynamic diameter seem to be correlated to the aggregation number where a larger aggregation number leads to an increase in size presumably due to the increased stretching of the bottlebrush backbone and PLA tail.

We further confirmed the micellar core-shell structure of the nanoparticles observed in cryo-TEM by determining the fraction of PEG that is located in the solvated shell as compared to PEG that is trapped inside the solid core via ¹H NMR spectroscopy. ¹H NMR spectra of nanoparticles in D₂O containing trimethylsilylpropanoic acid as an internal standard before and after the addition of four parts acetone-d₆ were obtained and the solubilized PEG signals were compared to quantify the relative PEG exposure. As shown in Table 2, no considerable differences among nanoparticles were observed with nearly all PEG located in the nanoparticles shell indicating a highly homogeneous PLA core and PEG shell.

NMR Relaxometry.

To learn more about the nature of the nanoparticle shell, we turned our attention to NMR relaxometry as a powerful and readily available method that provides information about polymer chain flexibility and local chain dynamics in polymer aggregates.⁴² De Graaf et al. observed that linear PLA-PEG micelles in D₂O exhibit a biphasic T₂ relaxation for PEG protons which was attributed to rigid, fast-relaxing PEG segments close to the hydrophobic core and flexible, slow-relaxing segments in the outer shell.⁴³ We hypothesized that similar T₂ relaxation measurements could be used to distinguish between the densely crowded inner shell and the more flexible outer shell of nanoparticles assembled from BBCPs based on their different relaxation behaviors. A recent small-angle neutron scattering (SANS) study of amphiphilic BBCP micelles concluded that the micellular structure most closely resembles a core-shell-shell model consisting of a dense less hydrated inner shell and a highly hydrated outer shell.⁴⁴

First, nanoparticles were washed with D₂O several times to completely remove non-deuterated H₂O and NMR T₂ relaxation measurements were obtained. As shown in Figure 4a, only protons in the solvated shell were visible, corresponding to PEG (3.6 ppm) and PEG methoxy end group (3.3 ppm) protons. Fitting the relaxation data to a standard monoexponential fit did not yield acceptable R² values. However, biexponential fitting assuming a fast relaxing (T_2fast) and a slow-relaxing (T_2slow) component resulted in much better fitting statistics based on the previously published Eq 1, where I represents the peak intensity, τ the relaxation delay, and f_fast and f_slow the mole fraction of fast and slow-relaxing protons, respectively.⁴³

$$I_{\tau }=\left(f_{\text{fast}}{e}^{-\frac{\tau }{T_{2\text{fast}}}}+f_{\text{slow}}{e}^{-\frac{\tau }{T_{2\text{slow}}}}\right)I_0$$ (1)

Figure 4.

NMR T₂ relaxation of biphasic nanoparticle shell. (a) Schematic illustration of NMR T₂ relaxation measurement procedure for biexponential fitting of methoxy (3.3 ppm) and PEG (3.6 ppm) proton signals. (b) Schematic representation of nanoparticle assembled from PLA-PEG BBCPs consisting of fast-relaxing inner shell (red) and slow-relaxing outer shell (blue). (c) Summary of fast/slow-relaxing PEG-methoxy end groups obtained from biexponential fitting. Data are presented as mean ± SD (ANOVA with post-hoc Tukey’s multiple comparison test, n = 3 independent replicates). *P < 0.05, **P < 0.01, ****P < 0.0001.

We attributed the fast-relaxing rigid component to the dense inner shell with substantial side chain overlap between adjacent bottlebrush polymers and the slow-relaxing component to the flexible highly hydrated outer shell (Figure 4b). Notably, a PEG bottlebrush homopolymer (PEG_2.3k60) that was also investigated exhibited a monoexponential relaxation behavior for both PEG and methoxy protons (Figure S4) strongly supporting the idea that the biphasic relaxation is caused by the restricted flexibility of PEG side chains in the nanoparticle shell rather than the structure of the bottlebrush polymer.

While fitting parameters for PEG protons and methoxy protons showed similar trends (Table S1), we focused on the relaxation of methoxy end group protons to obtain information about the extent and thickness of the inner and outer shell (Figure 4c). In particular, the number of end group protons is directly proportional to the bottlebrush backbone DP of the hydrophilic block whereas the number of PEG protons varies with side chain length and overemphasizes the contribution of longer side chains in triblock BBCPs as well as their location within the nanoparticle shell.

The fraction of fast-relaxing end group protons decreased for symmetric BBCPs (Sy-M40, Sy-M60, Sy-M70) as the ratio of PEG to PLA backbone DP increases following the trend in aggregation number. Asy-S60, the nanoparticle with the largest aggregation number, exhibited the largest fraction of fast-relaxing end groups with 51%, suggesting a highly crowded shell with considerable side chain overlap. However, increasing the side chain asymmetry of diblock BBCPs (Sy-M60 and Asy-L60) significantly reduced the fraction of fast-relaxing protons. A similar effect was observed for triblock BBCPs with larger interfacial asymmetry (Tri-S40-M20 and Tri-S40-L20), indicating that side chain asymmetry can be used to effectively control the crowding and density of the nanoparticle shell. Consequently, BBCPs with different terminal block brush widths but similar interfacial asymmetry (Tri-S40-M20, Sy-M60, Tri-L40-M20) only experienced small differences in their fraction of fast-relaxing end group protons.

The trends observed in NMR relaxation are in agreement with the SANS study of amphiphilic BBCP micelles by Matson and coworkers.⁴⁴ Specifically, the ratio of inner to outer shell thickness decreased with increasing ratio of hydrophilic to hydrophobic backbone length consistent with the results obtained for the series of symmetric BBCPs in this study. Furthermore, the relative extent of inner and outer shell for BBCPs with comparable architecture is within the range of values for the fraction of fast/slow-relaxing end group protons reported herein. Our findings suggest that NMR relaxometry can provide a readily accessible and straightforward alternative to SANS in the study of nanoparticles formed from BBCPs.

Surface Topography.

Quantitative characterization of nanoparticle surface topography (i.e., surface roughness and surface area) is a challenge due to the small characteristic length scale of the surface. Standard surface characterization methods such as atomic force microscopy and nitrogen adsorption isotherms that have been used for nanoparticles in the past are inadequate to analyze soft polymeric nanoparticles in their native hydrated state as their shell is prone to deformation and collapse upon dehydration.^45,46 Here, we use a simple model to approximate the surface topography by depicting the nanoparticle shell as a number of individual cylinders with an overlapping portion close to the solid core (Figure 5). The radius of cylinders representing the outer segment of the PEG bottlebrush polymer (R_brush) was determined from cryo-TEM images and follows the trend in terminal block PEG side chain degree of polymerization (N) (Table 3). In particular, the bottlebrush radius scales with approximately N^0.7 in agreement with the predicted scaling of N^3/4 for bottlebrush polymers in good solvent.²⁸ The radius of the inner shell sphere (R_ISS) containing the overlapping cylinder portion was estimated based on the fraction of fast-relaxing PEG end group protons from NMR T₂ relaxation (Table S2). The mean line sphere radius (R_MLS) representing a sphere of equal volume was calculated as shown in Figure S5. The arithmetic mean roughness was determined based on the sum of height deviations from the mean line sphere (MLS) divided by the MLS surface area (A) and the MLS radius (r) to account for the nanoparticle curvature according to previously reported Eq 2, where |Z(x, y)| represents the integral of the cylinder volume above the MLS and valley volume below.⁴⁷

$$roughness=\frac{1}{Ar}{\iint }_0^A|Z(x,y)|dxdy$$ (2)

Figure 5.

Schematic illustration of nanoparticle with hydrophobic core (green) and hydrophilic shell (blue) approximated as overlapping cylinders. The mean-line surface radius (R_MLS) represents the radius where the brush volume above equals the valley volume below the MLS. The inner shell sphere radius (R_ISS) corresponds to the radius of the overlapping portion of the bottlebrush cylinder shell.

Table 3.

Structural parameters of nanoparticle surface topography

^aDetermined from cryo-TEM image analysis presented as mean ± SD.

Roughness calculations show that Sy-M40 is the nanoparticle with the smoothest surface and the mean roughness substantially increases as the PEG/PLA block backbone ratio becomes larger for Sy-M60 and Sy-M70 (Table 3). Furthermore, larger interfacial PEG/PLA side chain asymmetry results in increased roughness for diblock BBCPs as well as triblock BBCPs with similar terminal block PEG side chain. Specifically, the roughness of nanoparticles based on BBCPs with short 1.1 kDa PEG side chain as the terminal block increases from 0.40 for Asy-S60 to 0.67 for Tri-S40-M20 and 0.89 for Tri-S40-L20 being the nanoparticle with the roughest surface. A larger terminal block brush width in Tri-L40-M20 and Sy-M60 resulted in a smoother surface compared to Tri-S40-M20. Notably, the surface roughness of nanoparticles studied herein is approximately one order of magnitude higher than values obtained via a similar method for rough silver-coated silica nanoparticles of equivalent size.⁴⁷

The nanoparticle surface area was calculated from the surface area of the non-overlapping portion of bottlebrush cylinders as follows:

$$\mathit{surface\; area\; per\; nanoparticle}=N_{\text{agg}}\left(2\pi R_{\text{brush}}\left(R_{\text{h}}-R_{\text{ISS}}\right)+\pi R_{\text{brush}}^2\right)$$ (3)

The surface area varied considerably between nanoparticles despite only small differences in diameter. The aggregation number strongly affected the calculated surface area, as expected from the increasing number of individual bottlebrush polymers exposed on the nanoparticle surface. For example, Asy-S60 has a ~14 times larger surface area than Asy-L60. Additionally, when we compared the nanoparticle surface area to that of an equivalent smooth nanoparticle with homogeneous shell density based on the MLS radius, we observed an increase in surface area by a factor of 4.1 to 11.3 depending on the BBCP architecture (Table S2). We do not expect bottlebrush polymers in the nanoparticle shell to behave as impenetrable hard cylindrical surfaces, but these results demonstrate the ability of our approach to generate nanoparticles with controlled surface topography as judged by their increase in surface area and roughness.

Protein Adsorption.

An important predictor of nanoparticle in vivo performance and pharmacokinetics is the adsorption of serum proteins and formation of a protein corona on the nanoparticle surface.^7,48 In particular, the total amount of protein adsorption has been found to directly correlate to blood circulation times with reduced adsorption resulting in prolonged circulation.⁶ We developed a simple protein adsorption assay, based on FITC-labeled bovine serum albumin (BSA), which is similar to the most abundant human serum protein HSA. Isothermal titration calorimetry (ITC) studies of Sy-M70 titrated with BSA-FITC did not show any significant exothermic or endothermic signal (Figure S6) comparable to literature reports of nanoparticles with high density PEG shell titrated with unlabeled BSA suggesting that the FITC label does not lead to increased interaction with the nanoparticle surface.⁴⁹ Nanoparticles in PBS were incubated with BSA-FITC at 37 °C and washed with fresh PBS four times to remove any protein not adsorbed onto the nanoparticle surface (Figure S9). The fluorescence signal of protein remaining in the pellet was then measured to calculate the amount of BSA adsorbed. It is noteworthy that DLS analysis and cryo-TEM imaging after incubation of Sy-M70 and Tri-S40-M20, respectively, indicate that the nanoparticles maintain their structural integrity and size in the presence of BSA (Figure S7b and Figure S8). We observed considerable differences in BSA adsorption between nanoparticles with the best-performing nanoparticles exhibiting virtually no adsorption, similar to a ‘stealthy’ smooth nanoparticle based on linear PLA-PEG with high PEG surface density (0.38 PEG/nm², D_h = 87 nm) (Figure 6a). The lowest adsorption was obtained for nanoparticles with long 3.5 kDa PEG terminal block side chain (Asy-L60 and Tri-L40-M20) whereas Asy-S60 adsorbed by far the most protein per particle followed by Tri-S40-M20. Nanoparticles based on symmetric BBCPs exhibited slightly reduced BSA adsorption with larger PEG/PLA block backbone ratio. However, increasing the side chain asymmetry for diblock BBCPs as well as the interfacial asymmetry for triblock BBCPs resulted in substantial lower protein adsorption with highly significant differences between samples. Furthermore, the terminal block brush width played a significant role with the narrow Tri-S40-M20 BBCP showing increased adsorption compared to Sy-M60 and Tri-L40-M20.

Figure 6.

Effect of nanoparticle surface topography on BSA adsorption and cell uptake. (a) BSA-FITC adsorption of nanoparticles assembled from PLA-PEG BBCPs and a smooth nanoparticle control based on linear PLA-PEG. (b) Impact of nanoparticle roughness on the number of BSA-FITC adsorbed per nanoparticle. Trendlines are based on the molecular weight of the terminal block PEG side chain used. (c) Number of BSA-FITC per nanoparticle compared to the number of accessible slow-relaxing groups obtained from NMR relaxation. (d) Mean fluorescence intensity (MFI) of HeLa cells incubated with DiO-loaded nanoparticles. (e) Comparison of nanoparticle uptake into HeLa to the number of BSA adsorbed for nanoparticles with 1.1 kDa PEG side chain as the terminal block (red) and 2.3 kDa PEG (blue). Data are presented as mean ± SD (ANOVA with post-hoc Tukey’s multiple comparison test, n = 4 independent replicates in (a) and n = 3 in (d)). **P < 0.01, ***P < 0.001, ****P < 0.0001.

The importance of terminal block brush width was also evident when we compared BSA adsorption to nanoparticle surface roughness. The number of BSA proteins adsorbed per nanoparticle linearly decreased with surface roughness with the extent dependent on the terminal block side chain length (Figure 6b). Specifically, a shorter side chain increases the contribution of surface roughness in reducing protein adsorption. While it has been previously reported that the local curvature of rough patchy nanoparticles inhibits protein adsorption, we speculate that the short terminal block side chain increases the flexibility of the bottlebrush backbone and may allow proteins to penetrate the nanoparticle shell more easily.⁵⁰ However, this effect gradually diminishes as a longer terminal block side chain is incorporated and the bottlebrush backbone becomes more rigid.

In general, it is well established that protein adsorption of PEGylated nanoparticles decreases with PEG surface density due to increased steric repulsion, as PEG chains transition from mushroom to brush conformation.⁷ However, as the shell of nanoparticles composed of BBCPs does not consist of a homogenous layer of flexible PEG that readily undergoes conformational changes, the protein adsorption cannot be accurately represented by simple calculation of PEG surface density. The extraordinary high density of PEG in the nanoparticle shell dictated by the bottlebrush backbone (i.e., one PEG side chain per five backbone carbon atoms) is expected to provide sufficient shielding to prevent any substantial hydrophobic interactions between proteins and the PLA core. Thus, we hypothesized that protein adsorption is primarily driven by direct interaction between PEG and proteins. Han et al. reported that increasing the density of PEG grafted onto a hydrophobic substrate beyond a critical grafting density results in increased protein adsorption via intermolecular protein-PEG interactions such as hydrogen-bonding.⁵¹ In addition, these interactions primarily occur between proteins and hydrophilic PEG methoxy end groups due to the directional dependency of hydrogen bonding.⁵² As a result, protein adsorption should be directly correlated to the number of PEG end groups per nanoparticle that are accessible for proteins to interact with rather than the actual surface area. We estimated the number of accessible PEG end groups in the outer nanoparticle shell based on the fraction of slow relaxing PEG end group protons (f_slow), the number of BBCPs per nanoparticle (N_agg), and number of PEG side chain per BBCP (n_PEG,backbone) (Eq 3 and Table S2).

$$\mathit{accessible\; end\; groups\; per\; nanoparticle}=\left(N_{\mathrm{agg}}\right)\left(n_{\mathrm{PEG},\mathrm{backbone}}\right)\left(f_{\text{slow}}\right)$$ (4)

Comparison of BSA adsorption to the number of accessible PEG end groups resulted in a positive linear correlation (Figure 6c) supporting the hypothesis that BSA directly interacts with PEG end groups whereas nanoparticle surface area failed to give a conclusive trend. To our surprise, this relationship suggests that it is desirable to reduce the amount of PEG chains per nanoparticle to minimize protein adsorption for nanoparticle assembled from BBCPs. This behavior is drastically different from other PEGylated nanoparticles where protein adsorption is typically inhibited as the number of PEG chains increases.⁷

Cell Uptake.

The effect of nanoparticle surface topography on cell uptake was investigated using a human cervical cancer HeLa cell line (Figure 6d) and a murine macrophage RAW264.7 cell line (Figure S6). Nanoparticles were loaded with fluorescent DiO dye to measure their uptake into cells via flow cytometry. Cells were incubated with a nanoparticle concentration normalized based on their fluorescence signal. Similar trends in nanoparticle uptake were observed for both cell lines, with all tested nanoparticles experiencing considerably increased uptake compared to the linear PLA-PEG nanoparticle. Nanoparticles with narrow terminal block brush width (Asy-S60, Tri-S40-M20, Tri-S40-L20) showed significantly enhanced uptake relative to nanoparticles with wider terminal block (Sy-M40, Sy-M60, Sy-M70). Notably, the uptake of Tri-S40-M20 and Tri-S40-L20 into HeLa cells approximately tripled compared to Sy-M40 and increased by a factor of ~7 relative to the smooth PLA-PEG nanoparticle control. Within the series of nanoparticles with similar terminal block an increase in surface roughness resulted in enhanced cell uptake (Figure S7). This observation is in agreement with previous literature reports that nanoparticle surface asperities greatly lower repulsive interactions between negatively-charged cell membranes and hydrophilic polymers such as PEG resulting in enhanced adhesion to cells.^53,54 Furthermore, the effect was more pronounced with smaller asperity size explaining the higher uptake of nanoparticles with narrow brush width.

Next, we compared the uptake into HeLa cells to the amount of BSA adsorbed. Typically, low protein adsorption of PEGylated nanoparticles is correlated with reduced cell uptake as steric effects and hydrophilicity of a high-density PEG shell also inhibits interactions with cell membranes.⁶ However, for nanoparticles tested in this study, a high cell uptake was observed even for nanoparticles that experienced low BSA adsorption (Figure 6e). Specifically, Tri-S40-L20 showed a comparable high level of HeLa cell uptake to that of Tri-S40-M20 while being among the lowest BSA adsorbing nanoparticles. In addition, Tri-S40-L20 exhibited substantially enhanced uptake compared to the conventional smooth PLA-PEG nanoparticle but similarly low BSA adsorption. These results imply that reducing protein adsorption and improving cell uptake do not necessarily present a trade-off as commonly observed for PEGylated nanoparticles. In fact, we were able to independently control protein adsorption and cell uptake through surface roughness and terminal block brush width, demonstrating the importance of surface topography in nanoparticle research. While in vitro results of nanoparticles investigated in this study suggest that this approach can potentially be used to generate long-circulating nanoparticles with high drug delivery efficiency, we note that in vivo protein corona formation and cell uptake are much more complex. Additional studies of the effect of surface topography on protein corona composition and protein binding affinity will be necessary to explain potential differences in their in vivo fate.

CONCLUSION

We report a method to control the surface topography of PEGylated nanoparticles based on PLA-PEG BBCP building blocks. Tuning of BBCP architecture produced monodisperse nanoparticles with highly predictable surface topography controlled by a number of structural parameters. In vitro experiments demonstrate that nanoparticle surface topography represents an important factor in controlling protein adsorption and cell uptake that has previously not been adequately considered in the study of polymeric nanoparticles. Nanoparticles with rough surface and narrow terminal PEG block brush width based on triblock BBCPs exhibited low protein adsorption while still maintaining high cell uptake compared to conventional smooth nanoparticles assembled from linear PLA-PEG block copolymers. As long-circulating PEGylated nanoparticles with high-density PEG shell typically suffer from low cell uptake and limited drug delivery efficiency, optimization of nanoparticle surface topography may provide a strategy to improve the performance of PEGylated nanoparticles in drug delivery. Furthermore, our approach enables the facile generation of nanoparticles with hierarchically functionalized surfaces that can be potentially utilized for other biomedical applications.

METHODS

Materials.

Methoxy-PEG-amine (750 Da and 2 kDa), methoxy PEG (5 kDa), albumin–fluorescein isothiocyanate (BSA-FITC), and Grubbs second-generation catalyst were purchased from Sigma-Aldrich while methoxy-PEG-amine, HCl salt (3 kDa) was obtained from JenKem Technology USA. DiO dye (3,3′-dioctadecyloxacarbocyanine perchlorate) was purchased from Biotium and poly(lactic acid) (6.7 kDa and 14.3 kDa) from LACTEL. PEG-MM³⁰ and N-(hydroxyethyl)-cis-5-norbornene-exo-2,3-dicarboximide⁵⁵ were prepared according to previously reported procedures. All other solvents and reagents were purchased from commercial suppliers and used without additional purification unless otherwise noted.

Instrumentation.

¹H NMR spectra were recorded on either a 400 MHz or 500 MHz Agilent DD2 NMR spectrometer. SEC was performed on a ThermoFisher Ultimate 3000 UHPLC equipped with two Agilent InfinityLab PolyPore columns (7.5 × 300 mm) connected in series at 60 °C. Dimethyl formamide (DMF) with 1 g/L LiBr added was used as the eluent at a flow rate of 1 mL/min. The polymer molecular weight was obtained using a T-rEX refractive index detector (Wyatt Technology) and a Dawn Heleos II (Wyatt Technology) eight angle light scattering detector. DLS including zeta potential measurements were performed using a Zetasizer Pro (Malvern Panalytical).

PLA-MM Synthesis.

Inside a nitrogen-filled glovebox, a round-bottom flask was charged with N-(hydroxyethyl)-cis-5-norbornene-exo-2,3-dicarboximide (0.106 g, 0.51 mmol) and D,L-lactide (1.010 g, 7.00 mmol). After complete dissolution of reagents in dry DCM (10 mL), DBU catalyst (5.9 µL, 0.07 mmol) was added and the mixture stirred for 1 hour. The reaction was quenched by adding benzoic acid (54.9 mg, 0.45 mmol). The crude product was purified via precipitation in hexanes (2×) and 30:70 water:methanol (2×) to obtain 829 mg white powdery polymer (74 % yield).

BBCP Synthesis.

G3 was prepared from second-generation Grubbs catalyst by stirring in toluene following published literature.⁵⁶ PEG-MM (1.1 kDa, 2.3 kDa, and 3.5 kDa) and PLA-MM stock solutions (0.05 M) were prepared by addition of dry DCM inside a nitrogen-filled glovebox. BBCPs were synthesized via ROMP based on the following representative procedure for Tri-S40-L20: PEG-MM stock solution (1.1 kDa, 364 µL, 18.2 µmol, 20 equiv.) was transferred to a 1.4 mL vial charged with a stir bar and freshly prepared dark-green G3 (0.33 mg, 0.91 µmol, 1 equiv., 0.01 g/mL) in DCM added under constant stirring. After 45 min, a small aliquot (40 µL) was removed for SEC and NMR analysis followed by the dropwise addition of the second PEG-MM (3.5 kDa, 164 µL, 8.2 µmol, 10 equiv.) and additional stirring for 1 h. Another aliquot (30 µL) was taken before the dropwise addition of PLA-MM (153 µL, 7.6 µmol, 20 equiv.). The reaction mixture was allowed to stir for 90 min and then quenched with 100 µL ethyl vinyl ether.

BBCP Self-Assembly.

BBCPs were dissolved in acetone and transferred to a round-bottom flask. For dye-loaded samples, DiO dye (0.2 % w/w) in acetone was added to the BBCP solution. The solvent was slowly removed via rotary evaporation to form a polymer film at the bottom of the flask. The polymer film was dried in a vacuum oven before ultrapure water was added to obtain a polymer concentration of 1 mg/mL. The samples were stirred overnight at 65 °C followed by 15 min sonication in a sonication bath (Branson 2510) heated to 65 °C. Resulting nanoparticles were filtered through a 0.45 µm polyethersulfone syringe filter and washed with ultrapure water three times through a Amicon centrifuge filter tube (100 kDa MWCO).

Linear PLA-PEG Assembly.

PLA-PEG block copolymers were synthesized via ring-opening polymerization using methoxy-PEG (5 kDa) as macroinitiator and DBU catalyst. Synthesized PLA_8k-PEG_5k, PLA_16k-PEG_5k, and purchased PLA_14.3k, PLA_6.7k were dissolved in acetonitrile with a weight ratio of 40, 15, 25 and 20, respectively, to result in a total polymer concentration of 50 mg/ml. The polymer solution (1 mL) was added dropwise to a vial containing 7 ml of distilled water under vigorous stirring. After nanoparticle formation, the solution was dialyzed against water to remove organic solvents.

Static Light Scattering.

Nanoparticle molar mass and aggregation number were measured via static light scattering using a Dawn Heleos II (Wyatt Technology) with detector voltages normalized with dextran sulfate (15 – 20 kDa). Eight sample concentrations (0.08, 0.07, 0.06, 0.05, 0.04, 0.03, 0.02, 0.01 mg/mL) were injected via a syringe pump (KD Scientific, Gemini 88) at 0.4 mL/min. The weight-averaged molar mass was obtained via Zimm plot analysis in Astra with dn/dc values estimated based on the PEG mass fraction according to published literature.⁵⁷ Aggregation numbers were calculated by dividing the nanoparticle molar mass by the BBCP molecular weight.

¹H NMR T₂ Relaxation.

Nanoparticles (4 mg) were washed with D₂O three times through a Amicon filter tube (100 kDa MCWO), reimmersed in 700 µL D₂O, and transferred to a 5 mm NMR tube. NMR T₂ relaxation time measurements were performed on a 500 MHz Agilent DD2 NMR spectrometer using the Carr-Purcell-Meiboom-Gill pulse sequence with a recycle delay of at least five times T₁ at 25 °C. The peak intensity at 3.6 ppm (PEG backbone protons) and 3.3 ppm (methoxy end group protons) were fitted via nonlinear least-square fitting to a biexponential function to obtain relaxation times and fraction of fast and slow relaxing protons in Matlab (Mathworks).

PEG Exposure Measurement.

A ¹H NMR spectrum with 10 s relaxation delay of purified nanoparticles in D₂O containing 0.1 % w/w trimethylsilylpropanoic acid (TSP) was obtained. The samples were transferred to a separate vial and 300 µL nanoparticle solution was mixed with 1.2 mL acetone-d₆ followed by sonication for 1 min. The resulting mixture was transferred back to a 5 mm NMR tube and the ¹H NMR spectrum was recorded. The amount of PEG exposure was determined by calculating the ratio of the PEG peak integral at 3.6 ppm normalized to the TSP peak at 0 ppm before and after addition of acetone-d₆.

Cryo-TEM.

Nanoparticles were concentrated to approximately 20 mg/mL using Amicon filter tubes (100 kDa MWCO). After glow-discharging Quantifoil holey carbon 300 mesh copper grids (Electron Microscopy Sciences) using a PELCO easiGlow, a FEI Vitrobot cryo plunger (ThermoFisher Scientific) was used to vitrify nanoparticles in liquid ethane. Sample grids were kept in liquid nitrogen at all times and mounted onto a cryogenic holder (Gatan 626) before imaging. Cryo-TEM images were obtained on a FEI Talos L120C (ThermoFisher Scientific) operated at an acceleration voltage of 120 kV and under low dose conditions at −4 µm defocus. Nanoparticle core diameters were determined via the ParticleSizer ImageJ plugin. The outer brush radius was obtained by measuring the width of at least 50 individual bottlebrush polymers per sample in ImageJ.

BSA Adsorption.

Concentrated nanoparticles in water were diluted with 10× PBS to a final concentration of 2 mg/mL and 1× PBS. BSA-FITC was dissolved in PBS (2 mg/mL) and 200 µL mixed with 1 mL of nanoparticle solution in an Eppendorf tube and incubated at 37 °C for 1 h. The samples were washed with PBS four times by centrifuging at 21,130 g at 4 °C for 1 h in a microcentrifuge and carefully removing the supernatant before the pellet was redispersed in 1 mL cold PBS. After the last washing step, 1.5 mL DMF was added and pellets redispersed by 5 min sonication in a sonication bath. The fluorescence emission intensity was measured at 530 nm and an excitation wavelength of 480 nm using a fluorimeter (Perkin Elmer, LS55). Calibration standards were prepared by serial dilution of BSA-FITC in DMF. The experiment was performed in quadruplicates.

Cell uptake.

RAW 264.7 and HeLa were cultured in T75 cell culture flasks with high-glucose Dulbecco’s modified Eagle’s medium (DMEM) containing 10% fetal bovine serum (FBS), and 1% penicillin-streptomycin (10,000 U/ml of penicillin and 10,000 g/ml of streptomycin, Gibco-BRL) at 37 °C in 5% CO₂ humidified atmosphere for 24 h. The cells were detached from the culture flask using a scrapper for RAW 264.7 or trypsin-EDTA solution for HeLa cells during their exponential growth phase. Cell suspensions (0.5 mL per well) at a concentration of 10⁵ cells/ml in growth medium were transferred to 24-well plate and incubated overnight. The cellular medium was removed and 0.45 ml of fresh DMEM was added before nanoparticle treatment. Nanoparticle solutions were prepared in PBS with their concentration normalized based on their fluorescence intensity relative to Sy-M60 (200 µg/ml in growth medium) and 50 µl sample was added to each well. The cells were further incubated at 37 °C for 24 h and rinsed with PBS to remove residual nanoparticles. Following treatment with trypsin-EDTA, cells were collected for flow cytometry analysis using a flow cytometer (Attune NxT, Invitrogen).