Functionalized Solid-Sphere PEG-b-PCL Nanoparticles to Target Brain Capillary Endothelial Cells In Vitro

Nanoparticles are increasingly used to implement drug targeting strategies. In the present study, solid-sphere nanoparticles (SNPs) made of poly(ethylene glycol)-b-poly(ε-caprolactone) (PEG-b-PCL) were covalently linked to a monoclonal antibody (8314mAb) targeted against the human insulin receptor that is highly expressed on human brain microvascular endothelial cells. Resulting targeted SNPs were characterized using transmission electron microscopy (TEM), cryo-TEM, dynamic light scattering, and fluorescence correlation spectroscopy. The critical aggregation concentration was determined using a fluorescence approach. Interaction with a well-characterized human in vitro model of the blood-brain barrier (hCMEC/D3) was analysed using an array of methods (flow cytometry, confocal laser scanning microscopy, and TEM). The toxicity on hCMEC/D3 cells and in addition on the human liver cell line HepG2 was assessed using the MTT assay. SNPs with a diameter of 80 nm and a homogeneous size distribution were obtained. Successful conjugation of 83-14mAb using a heterobifunctional linker resulted in 5-6 molecules of fluorescently labeled 83-14mAb per SNP. Functionalized SNPs were taken up by hCMEC/D3 cells efficiently without showing a significant toxic effect on cells of the blood-brain barrier and HepG2 cells. These results indicate that functionalized PEG-b-PCL SNPs are a promising candidate to deliver drugs to the CNS.


Introduction
Despite the presence of potent active pharmaceutical ingredients (API), efficient treatment of central nervous system-(CNS-) related diseases remains a major challenge in pharmacotherapy.The blood-brain barrier (BBB) that maintains the homeostasis of the brain excludes many small and large molecule drugs from the CNS [1,2].To reach therapeutic concentrations of systemic administered drugs in the CNS, high doses are often required.This is linked with an increased risk for toxic side effects.Therefore, different strategies for effective drug transfer across the BBB are currently under investigation [3].Besides chemical optimization of the API and invasive delivery techniques, targeting of transporters and receptors expressed on BBB endothelial cells is a promising way to specifically deliver drugs to the CNS [3].Endogenous macromolecules can cross the BBB by receptormediated transport and thus reach the brain despite their size as shown for insulin [4,5] and transferrin [6].Antibodies can mimic these endogenous ligands and transport bound drugs, macromolecules, or nanoparticles across the BBB in vivo.For example, the enzyme N-sulfoglucosamine sulfohydrolase (SGSH) was conjugated to a monoclonal antibody (mAb) directed against the human insulin receptor (HIR).Recombinant SGSH was not able to cross the BBB.However, it was possible to reach therapeutic concentrations in the brain after injection of the mAb fusion protein in Rhesus monkeys [7].While antibody drug conjugates (ADCs) provide specific targeting to a distinctive tissue or cell type, their drug carrying capacity is very limited.Nanocarriers can exceed the drug payload of ADCs by three to four orders of magnitude [8].Within these nanocarriers, self-assembled micelles consisting of biodegradable diblock copolymers are extensively studied as drug delivery systems (DDS) [9].Their hydrophobic core serves as a reservoir for poorly soluble drugs, sterically stabilized by a hydrophilic corona [10].Polymeric micelles can prevent premature drug degradation, reduce off-target effects, and increase bioavailability.In addition, these systems can be directed towards specific organs, tissues, or cells by coupling targeting moieties onto their surface [11].In recent years, various approaches were investigated to combine NPs with BBB targeting mAbs.The neuropeptide NK-1900, for example, improved scopolamine-induced learning impairments and memory deficits in rats by i.v.administration of NK-1900-loaded polymersomes conjugated to an antitransferrin receptor mAb (OX-26) [12].Previous studies by our group showed that polymersomes consisting of poly(dimethylsiloxane)-b-poly(2-methyloxazoline) (PDMSb-PMOXA) modified with 83-14 mAb were taken up specifically by human brain capillary endothelial cells in vitro [13].With   of 0.45 ± 0.10 nM, 83-14 mAb has a strong affinity to the -subunit of the insulin receptor expressed on human brain capillary endothelial cells (hBMEC).Its binding to the HIR was shown to trigger receptor-mediated endocytosis in vivo and 83-14 mAb was shown to cross the primate BBB 10x more efficiently as compared to an antitransferrin receptor mAb [14,15].
With respect to polymer-based NPs, however, the question arises if these particles can be safely degraded after delivering their drug payload, since accumulation and persistence within the target tissue are of potential safety concern.For example, PDMS-b-PMOXA vesicles used in our previous studies are characterized by a high chemical and mechanical stability [16].Polyacrylamide-based nanocarriers used previously to implement drug targeting strategies could give rise to potential toxicity due to degradation products [17].Polylactic acid-based NPs have been shown to be readily degradable but lactic acid, a major degradation product, can lead to acidification of the metabolizing cell and organ [18][19][20].
To address these potential issues, this study aimed to design and use PEG-b-PCL SNPs ([PP-SNP]) as a safe alternative to existing technologies.This polymer is already approved by regulatory authorities such as the US Food and Drug Administration (FDA) for tissue engineering and drug delivery.PEG-b-PCL is known to be biocompatible and biodegradable and was therefore used to implement an active targeting strategy to the brain [21].83-14 mAb was covalently conjugated to the surface of [PP-SNP] by a PEG-spaced heterobifunctional linker.Resulting [PP-SNP]-[mAb] was then analysed using an array of analytical methods (i.e., TEM, cryo-TEM, dynamic light scattering [DLS], pyrene encapsulation, and fluorescence correlation spectroscopy [FCS]).
Interactions with eukaryotic cells and potential toxic effects were studied in a well-characterized in vitro BBB model (hCMEC/D3) and the human hepatocellular carcinoma cell line HepG2 [13,22].

Preparation of [PP-SNP].
[PP-SNP] composed of a mixture of amphiphilic diblock copolymers (95% mol/mol PEGb-PCL and 5% mol/mol NH 2 -PEG-b-PCL) was prepared using the cosolvent method, also called the solvent displacement method [23].In brief, the block copolymer (5.3 mg) was dissolved in THF (50 L) and stirred at 700 rpm.Phosphatebuffered saline (PBS; 157 mM Na, 140 mM Cl) EDTA 5 mM, pH 7.4 (1 mL) was added dropwise under constant stirring.The mixture was stirred at room temperature (RT) overnight.THF was removed by two consecutive steps: THF was allowed to evaporate during overnight stirring [24].Buffer was subsequently exchanged by gel filtration chromatography using Superose 6 Prep column eluting with PBS, pH 7.4 [25].DiI-loaded [PP-SNP] ([PP-SNP-DiI]) was produced by adding 1 g DiI per mg of PEG-b-PCL to THF and particles were produced using the same protocol as described above.

Conjugation of 83-14 mAb to
Prep, PBS pH 7.4).For all experiments, modified SNPs were used immediately after preparation.

Gold-Nanohybrid SNP Formation. To visualize [PP-SNP]
in TEM, gold-nanohybrids [PP-SNP-Au] were prepared as described recently [26].In brief, 5.3 mg of PEG-b-PCL was dissolved in THF (50 L) and stirred at 700 rpm.The AuRsolution (1 mM HAuCl 4 , 4.1 mM citrate) was added dropwise under constant stirring.The mixture was heated to 70 ∘ C under constant agitation at 300 rpm on the thermomixer (Thermomixer Comfort; Eppendorf, Hamburg, Germany).Free gold NPs were separated by gel filtration chromatography (Superose 6 Prep) using PBS pH 7.4 as an eluent.Free gold particles did not elute during gel filtration chromatography.They had a high affinity for the stationary phase and were retained in the chromatography column [26].[PP-SNP-Au] and [PP-SNP-Au]-[mAb] were collected after 8-12 min using a column volume of 30 mL and a flow rate of 1 mL/min.

Characterization of [PP-SNP].
Morphology and size distribution of [PP-SNP] were analysed using TEM, cryo-TEM, and DLS.For TEM analysis, 5 L of nonconjugated and conjugated SNPs was mounted on a carbon-coated copper grid, negatively stained with 2% uranyl acetate solution, and dried overnight.Samples were visualized using a Philips Morgagni 268D transmission microscope.Mean SNP size was determined based on  = 50 particles.For cryo-TEM analysis, 4 L of the polymer suspension was adsorbed onto glow-discharged holey carbon-coated grid (Quantifoil, Großlöbichau, Germany), blotted with Whatman filter paper, and vitrified into liquid ethane at −178 ∘ C using a vitrobot (FEI Company, Eindhoven, Netherlands).Frozen grids were transferred onto a Philips CM200-FEG electron microscope using a Gatan 626 cryoholder.Electron micrographs were recorded at an accelerating voltage of 200 kV and a nominal magnification of 50 000x, using a low-dose system (10 e − / Å2 ) and keeping the sample at −175 ∘ C. Defocus values were −4 m.Micrographs were recorded at 4 K × 4 K CMOS camera (TVIPS, Gauting, Germany).DLS measurements were performed using a Delsa Nano C Particle Sizer (Beckman Coulter, Brea, CA, USA) operated in back scattering mode (165 ∘ angle).

Critical Aggregation Concentration (CAC).
To assess the stability of [PP-SNP] upon dilution, the CAC of PEG-b-PCL was determined.Pyrene was used as a fluorescent probe as described previously [27,28].Pyrene in acetone (0.5 mL, 2.4 × 10 −6 M) was mixed with 2 mL of Milli-Q water and 0.05, 0.5, 5, 50, or 500 g/mL of PEG-b-PCL.The mixtures were heated to 60 ∘ C for 1 h and allowed to cool down to RT.After degasing with N 2 , the fluorescence of pyrene was measured (Spectramax M2 plate reader, Molecular Devices, Sunnyvale, CA, USA; excitation:  ex 332 nm; emission:  em1 373 nm,  em2 384 nm).The ratio of  em2 / em1 was plotted against the logarithm of the PEG-b-PCL concentration to determine the CAC.

Density Measurements and Specific Volume Calculations.
Density of particle suspension was measured at 20 ∘ C using an Anton Paar Density Meter DMA 4500 M (Anton Paar GmbH, Graz, Austria).The measured density (g/cm 3 ) of a suspension containing nanoparticles was used to calculate the apparent specific volume (cm 3 /g) of the nanoparticles (V PP-SNP ) according to [29] where  suspension is the measured density of the suspension,  blank is the measured density of the buffer, and  polymer is the weight fraction of the polymer.The apparent volume (cm 3 ) of one polymer molecule in the assembled state was calculated according to where   is the molecular weight of the polymer and   is the Avogadro constant.The aggregation number ( agg ) was then calculated according to where V polymer is the apparent volume of one polymer molecule and  PP-SNP is the volume of one solid-sphere nanoparticle ([PP-SNP]) and  is the nanoparticle diameter as determined by DLS or TEM measurements.
2.9.Gold Labeling of 83-14 mAb.To visualize conjugated 83-14 mAb, antibodies were stained using an anti-mouse IgGgold antibody.In brief, 0.25 mg of [PP-SNP]-[mAb] was incubated with 50 L of the anti-mouse IgG-gold antibody for 1 h at RT. Free anti-mouse IgG-gold antibodies were separated by gel filtration chromatography (Superose 6 prep) eluting with PBS pH 7.4.Samples were negatively stained as described above and were analysed by TEM.To exclude unspecific binding of the anti-mouse IgG-gold antibody, [PP-SNP] was stained using the same protocol.2.12.MTT Assay.Cell viability was determined using the MTT assay as described previously [22].Toxicity of [PP-SNP] and [PP-SNP]-[mAb] was tested on the target cell line (hCMEC/D3).In addition, the well-characterized and frequently used human hepatocellular carcinoma cell line HepG2 was used to confirm cell viability results in another cell line and to rule out cell line specific effects.In brief, cells were seeded at a density of 2 × 10    [ were confirmed by DLS.A mean hydrodynamic diameter of 78.2 ± 1.7 nm and a narrow size distribution (0.069 ± 0.021) were determined (Table 1).The zeta potential was −4.13±0.24(Table 1).In buffer (PBS, pH 7.4, room temperature), [PP-SNP] was stable for at least 2 months as shown by a constant particle size and a PDI below 0.2.To characterize the core structure of the NPs, cryo-TEM was performed.In cryo-TEM, a well-defined membrane can be observed for polymersomes, while no such structure is visible in micelles and solidsphere NPs [30].No membrane was observed in micrographs of [PP-SNP] as shown in Figure 1 classified as solid-sphere nanoparticles (SNPs).The CAC was determined using a hydrophobic fluorescent probe [27,28].By this technique, changes in fluorescence characteristics of fluorescent dyes, induced by their environment, are analysed.Thus, lipophilic dyes will be redistributed to the hydrophobic PCL-core of [PP-SNP] at concentrations above the CAC.The CAC of PEG-b-PCL determined by pyrene encapsulation during this study was 5 g/mL (2.8 × 10 −7 M).In this study, in vitro cell uptake experiments were always performed at 0.1 mg/mL, a concentration that is about 20 times higher than the determined CAC.The specific volume V PP-SNP of the [PP-SNP] suspension was determined to be 0.87 ± 0.01 cm 3 /g.3.3.Toxicity.SNPs composed of the FDA-approved, biodegradable PEG-b-PCL show high biocompatibility [21,31].This is particularly important with respect to brain delivery of SNPs since accumulation of polymeric material could lead to toxic side effects within the CNS.To confirm these findings, hCMEC/D3 cells and a human hepatocellular carcinoma cell line (HepG2) were incubated with [PP-SNP] and [PP-SNP]-[mAb] (0.01-1 mg/mL) to analyse cell viability using the MTT assay.Even high polymer concentrations of up to 1 mg/mL caused only a slight decrease in cell viability.Uptake experiments were performed with 0.1 mg/mL PEG-b-PCL.This concentration does not show any significant cytotoxic effects on HepG2 and hCMEC/D3 cells.

Discussion
In this study, PEG-b-PCL diblock copolymer was used to produce a nanosized drug delivery system.SNPs were prepared using the cosolvent method at a final concentration of 5.3 mg/mL polymer.Particles assembled immediately after stirring at 700 rpm and adding PBS dropwise.This method is easy to handle, fast, and highly reproducible and uses no expensive or complex equipment.A fast preparation protocol is of huge advantage for different applications, for example, for scale-up or when radiolabeled NPs are prepared for pharmacokinetics studies.To target the NPs towards a specific tissue or cell type, ligands have to be attached to the NP surface.Different strategies can be used for such modifications.Since different classes of targeting moieties (e.g., mAbs, peptides, or small molecules) may require different conjugation strategies, a universal coupling tool is advantageous.Therefore, 5% mol/mol of amine-terminated PEG-b-PCL (NH 2 -PEG-b-PCL) was used to covalently couple targeting moieties to [PP-SNP] by using the heterobifunctional linker SM(PEG)  .SM(PEG)  contains an amine-reactive NHS ester functionality and maleimide that reacts to a metabolically stable thioether in the presence of sulfhydryl groups.A PEG-spacer of variable length links the two reactive groups.While a very low degree of functionalization was observed using SM(PEG) 2 with a spacer length of 17.6 Å, conjugation with 24 PEG-subunit long SM(PEG) 24 (spacer length: 95.2 Å) resulted in an increased mAb/NP ratio per [PP-SNP]-[mAb] using the same reaction conditions.Furthermore, PEG-spacers also can prevent the covalently attached mAb from being covered by particle-adsorbed biomolecules [32].FCS was used to determine the mAb/NP ratio.On average, 5 molecules of fluorescently labeled 83-14 mAb were conjugated to the surface of one [PP-SNP]-[mAb-DL] (Figure 2(c)).FCS was used similarly in previous studies to determine the amount of fluorescently labeled molecules on SNPs [13,33].Radiolabeling, as typically used for the quantification of antibodies per NP, can be avoided by using FCS [34].The number of targeting moieties on the NP surface influences the interaction with the target cell and can promote cellular uptake but may lead to increased clearance from the blood circulation [35].In addition, a defined and narrow size distribution and the surface charge are important determinants for the fate of nanoparticles upon systemic exposure.NPs with a size below 6 nm can easily be excreted by renal filtration, while NPs with >300 nm in diameter or with a high surface charge will be recognized and removed by the reticuloendothelial system [36,37].Noncharged, pegylated NPs with a diameter below 200 nm show reduced protein opsonisation and remain in the systemic circulation for hours [38].
In cryo-TEM micrographs, no membrane was observed for [PP-SNP] indicating a solid-sphere structure of [PP-SNP] (Figure 1(c)).In comparison to low-molecular weight surfactants, SNPs show a low CAC, which is important for stability and drug retention.A high CAC can lead to a micelle collapse and subsequent burst release of the incorporated drug upon dilution in the blood compartment [10,39].Whereas typical low-molecular weight surfactants have CAC values in the range of 10 −3 -10 −4 M, CACs of diblock polymers can typically be found in range 10 −6 -10 −7 M. Additionally, the kinetic stability of the core structure of SNPs is high and thus the disassembly of SNPs after dilution below the CAC occurs slowly allowing an efficient retention of their drug payload [40].The CAC of [PP-SNP] was determined using a fluorescent probe-based approach as described previously [28].The CAC of PEG-b-PCL is 5 g/mL (2.8 × 10 −7 M), comparable to previously published data characterizing similar polymers [27].
The apparent specific volume V PP-SNP of the [PP-SNP] suspension was calculated from the measured density as described by Sommer et al. [29].V PP-SNP was 0.87±0.01cm 3 /g, which is close to the specific volume determined for micelles formed by the PEG-derived nonionic surfactant "Brij" [29].Using this result, one can calculate the apparent volume of a single PEG-b-PCL polymer molecule (V polymer ) according to (2).In our experiment, V polymer was 51 nm 3 .The aggregation number ( agg ) of [PP-SNP] can then be estimated according to (3).A similar approach using apparent volumes to calculate aggregation numbers was proposed previously to determine  agg of NPs consisting of a cell-penetrating peptide [41].Since this calculation depends on the diameter of the particle,  agg covers a range of ∼2200 to ∼10 000 for [PP-SNP] with diameters of 60 to 100 nm, respectively.These numbers are higher than results obtained by, for example, static light scattering (SLS) or fluorescent techniques where typical  agg for diblock copolymer micelles are lower than 1000 [42,43].
Among the broad range of polymer-based DDS, NPs composed of the FDA-approved, biodegradable PEG-b-PCL show high biocompatibility [21,31,44].This is an important advantage regarding organ delivery of NPs since accumulation of polymeric material might lead to toxic side effects, for example, in the CNS.Because toxic effects of polymeric NPs on cells are driven not only by the polymeric material but also by many other characteristics including particle size, charge, or impurities, each NP formulation needs to be tested [45].Toxicity of [PP-SNP] and [PP-SNP]-[mAb] on hCMEC/D3 cells was tested using the MTT assay.[PP-SNP] induced only a slight decrease in cell viability on hCMEC/D3 cells if incubated in concentrations up to 0.2 mg/mL (Figure 3).Cells incubated with [PP-SNP]-[mAb] showed as well a high cell viability.Even at high concentrations of 1 mg/mL, more than 60% of the incubated cells were viable.For cellular uptake experiments performed in this work, a concentration of 0.1 mg/mL was used.At this concentration the decrease in cell viability was negligible.The MTT assay was repeated with a second cell line to confirm the biocompatibility of the nanoparticles and to rule out cell line specific effects.The human hepatocellular carcinoma derived cell line HepG2 was used.Again, only a slight decrease in cell viability was observed (Figure 3).We conclude from these experiments that PEG-b-PCL SNPs are well tolerated within a typical dose range used in nanomedicine [22].
[PP-SNP-DiI]-[mAb] was taken up by hCMEC/D3 cells.These cells express similar HIR protein levels as freshly isolated hBMECs (1 fmol HIR/g protein) and were therefore successfully used as an in vitro targeting model in previous studies [13,[46][47][48].A time dependent increase in the fluorescence signal was observed (Figure 4).Conjugation with 83-14 mAb leads to an increase in MFI.Furthermore, uptake was specific and mediated by the 83-14 mAb since competition for the target receptor (HIR) with free mAb leads to a decrease of intracellular fluorescence (Figures 4    To this end, a protocol to prepare gold-nanohybrids ([PP-SNP-Au]) was developed by our group [26].This opens attractive possibilities to monitor intracellular trafficking of SNPs.hCMEC/D3 cells were grown on transwell inserts for 8 days as it was reported that hCMEC/D3 cells grown under these conditions show BBB characteristics and develop cell polarity [49].This was furthermore demonstrated by their use for quantitative drug transport studies and qualitative analysis of albumin-gold-conjugates by TEM [50,51].[PP-SNP-Au]-[mAb] targeting the HIR was found in intracellular vesicles after 15 min of incubation at 37 ∘ C (Figure 6(c)).Accumulation in multivesicular bodies (MVBs) was detected after 60 min of incubation (Figure 7).Cellular uptake of gold-labeled insulin was studied previously using retinal vascular endothelial cells.It was shown that insulin was taken up fast and efficiently and accumulated in MVBs after 30 min of incubation [52].However, cellular uptake of insulin seems to be faster and more efficient as compared to [PP-SNP-Au]-[mAb].Similarly, previous studies using OX-26immunoliposomes targeting the transferrin receptor showed that size is a crucial factor with respect to cellular uptake kinetics and efficiency.The permeability surface area product, a factor describing the BBB permeability, of the free OX-26 mAb was significantly higher as compared to the immunoliposomes in vivo [53].In the present study, single colloidal gold particles were detected at all time-points, indicating that gold particles were taken up while incorporated into [PP-SNP-Au]-[mAb].Free colloidal gold particles prepared without PEG-b-PCL agglomerated very fast and were found as clusters of particles when analysed by TEM [26].To the best of our knowledge, this is the first study using gold-loaded targeted polymeric SNPs to visualize their uptake in a cellbased model.Further experiments such as quantitative in vitro transport studies and colocalization studies by fluorescent imaging, as performed previously, could help to identify the uptake mechanism of 83-14 modified SNPs [13,[54][55][56].

Conclusion
SNPs of the biodegradable diblock copolymer PEG-b-PCL were prepared and characterized in vitro.Our SNPs are characterized by a hydrodynamic diameter of 79.6 nm, a homogeneous size distribution, a slightly negative zeta potential, and a low CAC in the submicromolar range.In second step, [PP-SNP] was conjugated to anti-HIR mAb to enhance the cellular uptake by hCMEC/D3 cells.This resulted in targeted [PP-SNP]-[mAb] decorated with 5 molecules of mAb per SNP.
[PP-SNP]-[mAb] was taken up by human BBB endothelial cells in vitro as shown by confocal microscopy and flow cytometry.Furthermore, the intracellular localization of gold-nanohybrids [PP-SNP-Au]-[mAb] in hCMEC/D3 cells grown on transwell inserts was analysed by TEM.Accumulation in multivesicular bodies of target cells was observed after 60 min of incubation.A similar behaviour was described for gold-labeled insulin [52].Our in vitro data suggest that SNPs consisting of PEG-b-PCL are well tolerated.They are promising candidates for the implementation of a targeted drug delivery strategy.Further experiments will be needed to explore their potential in experimental animals.
-SNP] showed spherical particles with a homogenous size distribution (Figure 1(a)).Mean diameter of  = 50 particles determined by TEM was 67 ± 15 nm.Size and size distribution of [PP-SNP]