Disulfide-Mediated Bioconjugation - ACS Publications - American

Jul 8, 2019 - tone) bearing or not a 2-pyridyl disulfide (PDS) terminal group, which quantitatively ...... are consistent with soft spheres for the 30...
0 downloads 0 Views 4MB Size
Research Article www.acsami.org

Cite This: ACS Appl. Mater. Interfaces 2019, 11, 26607−26618

Disulfide-Mediated Bioconjugation: Disulfide Formation and Restructuring on the Surface of Nanomanufactured (Microfluidics) Nanoparticles Mike Geven,† Hanying Luo,‡ Donghun Koo,‡ Gangadhar Panambur,‡ Roberto Donno,† Arianna Gennari,† Roberto Marotta,§ Benedetto Grimaldi,∥ and Nicola Tirelli*,†,⊥

Downloaded via BUFFALO STATE on August 1, 2019 at 10:02:10 (UTC). See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.



Laboratory of Polymers and Biomaterials, §Electron Microscopy Facility, and ∥Laboratory of Molecular Medicine, Fondazione Istituto Italiano di Tecnologia, 16163 Genova, Italy ‡ MilliporeSigma Materials Science, 6000 N Teutonia Avenue, Milwaukee, Wisconsin 53209, United States ⊥ Division of Pharmacy and Optometry, School of Health Sciences, Faculty of Biology, Medicine and Health, The University of Manchester, M13 9PT Manchester, U.K. S Supporting Information *

ABSTRACT: This study is about (1) nanomanufacturing (focusing on microfluidic-assisted nanoprecipitation), (2) advanced colloid characterization (focusing on field flow fractionation), and (3) the possible restructuring of surface disulfides. Disulfides are dynamic and exchangeable groups, and here we specifically focus, first, on their use to introduce biofunctional groups and, second, on their re-organization, which may lead to variable surface chemistries and uncontrolled cell interactions. The particles were obtained via microfluidic-assisted (flowfocused) nanoprecipitation of poly(ethylene glycol)-b-poly(ε-caprolactone) bearing or not a 2-pyridyl disulfide (PDS) terminal group, which quantitatively exchanges with thiols in solution. In this study, we have paid specific attention to size characterization, thereby also demonstrating the limitations of dynamic light scattering (DLS) as a stand-alone technique. By using asymmetric flow field flow fractionation coupled with DLS, static light scattering (SLS), and refractive index detectors, we show that relatively small amounts of >100 nm aggregates (cryogenic transmission electron microscopy and SLS/DLS comparison suggesting them to be wormlike micelles) dominated the stand-alone DLS results, whereas the “real” size distributions picked 99% purity) were obtained as N-terminal acetamides in a hydrochloride form from Biomatik (Cambridge, Canada). The assays used in cell culture, bicinchoninic acid (BCA, Pierce) and MTS (CellTiter 96 Aqueous One Solution Proliferation Assay), were obtained from Thermo Fisher Scientific and Promega, respectively. All other chemicals and solvents were obtained from Sigma-Aldrich. Ultrapure water was prepared using a Merck Millipore Advantage A10 system. 2.2. Physicochemical Characterization. 2.2.1. Nuclear Magnetic Resonance. 1H NMR experiments were performed on a Bruker AVANCE III 400 MHz spectrometer equipped with a Broad Band Inverse probe. The polymers were dissolved at a concentration of 10 mg/mL in deuterated chloroform and measurements were performed at 27 °C. 2.2.1.1. PEG-PCL. δ: 4.06 (2H, PCL, CH2O), 3.64 (4H, PEG, CH2CH2O), 3.37 (3H, PEG end-group, OCH3), 2.30 (2H, PCL, (OC)CH2), 1.65 (4H, PCL, CH2CH2CH2CH2CH2), 1.38 ppm (2H, PCL, CH2CH2CH2CH2CH2). 26609

DOI: 10.1021/acsami.9b07972 ACS Appl. Mater. Interfaces 2019, 11, 26607−26618

Research Article

ACS Applied Materials & Interfaces

Figure 1. 1H NMR spectra in CDCl3 (A) and GPC chromatograms in THF (B, RI signal) for PEG-PCL and PDS-PEG-PCL. Please note that, behind the resonance of CHCl3 at 7.24 ppm, these polymers typically present also that of (probably PEG-bound) water at 1.8−1.9 ppm, which can only be eliminated through azeotropic drying with toluene. The number average molecular weight of each blocks was calculated as the ratio between the resonance of the functional end group [PDS (average of k to n peaks) or methoxy (f′ peak)] and that of the PEG (e peak) or of the PCL (average of a to d peaks) chain. a Poroshell 120 EC-C18 column (4.6 × 150 mm, 4 μm packing) and an Agilent Technologies 1260 II Infinity UV detector was used for HPLC analysis. Water and acetonitrile, both containing 0.1% (v/v) of trifluoroacetic acid, were used as eluents. Samples were eluted with a variable acetonitrile/water gradient that was increased from 0 to 37.5% (v/v) over 10 min, maintaining the flow rate at 1 mL/min and estimating the peptide concentration through its absorbance at 215 nm. The peptide concentrations were calculated using appropriate calibration curves (Figure 7) of each peptide in 5× concentrated PBS (0.05 M phosphate) containing 75 mM TCEP, with a concentration range between 10 and 156 μM. 2.6. Nanoparticle Toxicity and Uptake. 2.6.1. Nanoparticle Labeling. In order to quantify their uptake in cells, nanoparticles were loaded with pyrene (see Supporting Information, Figure S5); the very low water solubility of this fluorophore minimizes its leaching from nanoparticles, and therefore, its fluorescence permits the quantification of the nanoparticles in cell lysates or supernatants. Pyrene-loaded PEGPCL and PDS-PEG-PCL nanoparticles were prepared using 0.25% polymer solutions in acetone (polymer/pyrene mass ratio of 360:1) and, respectively, a TFR of 125 and 2000 μL/min (FRR = 0.5). Pyreneloaded PEG-PCL-GCDDG and -GCRRG particles were prepared by reaction of pyrene-loaded PDS-PEG-PCL particles with a 3-fold excess of GCDDG or GCRRG (3:1 thiol/PDS) in PBS at 37 °C. These particles were subsequently purified as described in Section 2.5. All particles were finally mixed with 10× PBS in order to obtain suspensions in PBS (1×). These suspensions were sterile-filtered through a 0.45 μm PES syringe filter before application in cell culture experiments. As shown in Supporting Information, Figure S6, the fluorescence of functional (peptide-bearing) nanoparticles is lower than that of the original particles due to the additional preparative/ purification steps, which are associated to significant pyrene leakages. However, this does not affect the biological readings because all particles are calibrated independently. 2.6.2. General Cell Culture. A human colon carcinoma cell line (HCT116 ATTC CCL-247) was used to assess both cytotoxicity and nanoparticle uptake. HCT116 were cultured at 37 °C under 5% CO2 in McCoy’s medium supplemented with 1% glutamine, 10% fetal bovine serum (FBS), and 1% penicillin/streptomycin. In all experiments, cells were cultured in medium for 1 day to allow the cells to adhere to the cell culture substrates (polystyrene tissue culture plastic, BD Falcon). Thereafter, medium was exchanged with fresh medium only or fresh medium containing 12 mM GSH. Pyrene-loaded nanoparticle suspensions in PBS, pyrene PBS solutions, or just PBS as controls were subsequently added to the medium in an amount corresponding to 1/6 (17%) of the final volume. 2.6.3. Cytotoxicity Experiments. Suspensions of pyrene-loaded nanoparticles (and pyrene solutions, see Supporting Information,

2.4. Nanoparticle Functionalization. Concentrated phosphatebuffered saline (PBS) (10×, 0.1 M phosphate) was mixed with PDSPEG-PCL nanoparticle suspensions at a particle concentration estimated to 1.25 mg/mL (directly after acetone removal), using a suspension/PBS volume ratio of 9:1. The resulting nanoparticle concentration in PBS was 1.13 mg/mL (0.11 mM PDS). Solutions (0.11 mM) of the various peptides in PBS were added in volumes appropriate to obtain 3:1, 1:1, and 1:3 thiol/PDS molar ratios, while keeping the cumulative concentration of the two reaction partners constant ([thiol] + [PDS] = 0.11 mM). The reactions were performed in a 96 well plate (200 μL per well) at 37 °C while shaking at 560 rpm in a BioTek Synergy H1 plate reader. The progress of the reaction was followed by recording the absorbance of 2-pyridinethione (2PT) at 340 nm. The extent of 2PT formation was calculated from a calibration curve of 2PT in PBS at a concentration range of 15.6−250 μM (linear with R2 = 0.999; see Supporting Information, Figure S3) and used to calculate the conversion of PDS groups over time (see Supporting Information, Figure S4). All data are presented as averages ± standard deviation (n = 6) of the conversion. The maximum (asymptotic) yield and a rate constant for each reaction was determined by fitting these data with a y = a(1 − e−bx) one-phase exponential model (Origin 2018 software; OriginLab Corp., Northhampton, U.S.A.), where a is the asymptotic conversion and b is employed as a rate parameter expressed in M−1 s−1. 2.5. Peptide Release. For peptide release studies, nanoparticle suspensions of PDS-PEG-PCL were prepared at a TFR of 500 μL/min and an FRR of 0.5 and were subsequently dried and mixed with PBS (10×) as described (v.s.). The resulting nanoparticle suspensions in PBS were reacted with each peptide at a thiol/PDS ratio of 3:1 for 2 h at 37 °C in an orbital shaker operating at 500 rpm. Completion of the reaction was thereafter confirmed by absorbance measurements of 2PT at 340 nm and 37 °C using a BioTek Synergy H1 plate reader. The functionalized nanoparticles were purified from any unreacted peptide and from 2PT via dialysis against ultrapure water using regenerated cellulose tubing (8−13 kDa MWCO) for 24 h, exchanging the dialysate after 2, 4, and 20 h. The purified particle suspensions were diluted 20× in PBS or PBS supplemented with 10 mM GSH, both containing 0.02% sodium azide (particle concentration of 56 μg/mL, theoretically corresponding to 5.5−6 μM disulfides). The diluted suspensions were thereafter incubated at 37 °C while shaking at 300 rpm. At predefined time points (2, 8, 16, 24, 72, and 168 h), samples (n = 3) were removed and centrifuged at 10 000g for 15 min at 20 °C. Of the supernatant, 1 mL was lyophilized and redispersed in 200 μL of a 75 mM tris(2carboxyethyl)phosphine (TCEP) solution in ultrapure water. To reduce any disulfides by TCEP, samples were left for 30 min prior to submitting the samples to high-performance liquid chromatography (HPLC) analysis. An Agilent Technologies 1260 Infinity II system with 26610

DOI: 10.1021/acsami.9b07972 ACS Appl. Mater. Interfaces 2019, 11, 26607−26618

Research Article

ACS Applied Materials & Interfaces

Figure 2. (A) Z-average hydrodynamic size (left axis) and PDI (right axis) of PEG-PCL nanoparticles as a function of the TFR and FRR. (B) As in (A) for nanoparticles prepared at variable TFR (FRR = 0.5) and with different PDS-PEG-PCL/PEG-PCL weight ratios. In both graphs, bars refer to Zaverage hydrodynamic size (average diameter), symbols to PDI. All data (n = 3) are obtained employing DLS as a stand-alone instrument. Figure S7) were added to cells at five different concentrations (ranging from 12 to 190 μg/mL for nanoparticles, corresponding to 66−530 ng/ mL for pyrene either alone or in the nanoparticles). After 4, 24, and 48 h of culture, cells were washed with PBS and an MTS assay was performed according to the supplier’s protocol. Cells were thereafter washed with PBS three times and lysed using a radioimmunoprecipitation assay (RIPA) buffer containing 0.4% protease and phosphatase inhibitors (Sigma-Aldrich, Protease Inhibitor Cocktail and Phosphatase Inhibitor Cocktail 2 and 3). Following lysis, a BCA assay was performed according to supplier’s protocol. Average viability of cells (n = 3) was expressed as their metabolic activity (MTS) normalized to protein content (BCA), relative to controls. 2.6.4. Uptake Experiments. Nanoparticle suspensions and pyrene solution (as control) were added to cells at a final concentration of 190 μg/mL and 530 ng/mL, respectively. After 30 min, 1 h, and 2 h, cells were washed with PBS three times and thereafter soluble lysates were prepared in RIPA buffer. Fluorescence (350 nm excitation, 390 nm emission) of the lysates was read in black wells-plates with clear bottom (Polystyrene, Falcon). Average nanoparticle and pyrene contents (n = 3 for 30 min and 1 h time points, n = 6 for 2 h time point) of these lysates were calculated using fluorescence standard curves prepared in lysates of cells cultured in medium supplemented with 17 vol % PBS. Uptake was expressed as the calculated mass of nanoparticles or pyrene in cell lysates relative to the initially added amounts. Significance was assessed using a two-tailed Welch’s t-test, setting the minimum significance level to 0.05.

The polymers were dissolved in acetone and produced in the form of nanoparticles through nanoprecipitation in water within a Micromixer Chip. Different from previous studies of our group,1 the amphiphilic nature of the polymers allows the production of nanomaterials also in the absence of a surfactant.2 Using PEG-PCL, we have investigated the effect of FRR (i.e., the acetone/water ratio in the final dispersion) over particle size and PDI. FRR = 0.5, that is, the highest acetone content, corresponded to the lowest PDI at any value of the TFR tested (Figure 2A) and was, therefore, chosen for all further experiments; this higher homogeneity may be due to a slower growth of the nanoprecipitates in the presence of large amounts of the organic solvent: a sort of “equilibrated” process possibly based on aggregation of smaller particles through collisions. We have then recorded the effect of the TFR, showing that the nanoparticle size decreased with increasing flow rate for PEGPCL, for PDS-PEG-PCL and for a 1:1 mixture of the two (Figure 2B). This effect may be due to faster laminar mixing at higher flow rates, leading to accelerated nucleation and thus to a larger number of smaller particles. Hydrophobic PDS groups increased the particle size: their presence can allow PEG chains to bridge between hydrophobic domains, thereby aggregating particles during/immediately after nucleation. It is noteworthy that the identity of the polyester block plays a major role in nanoprecipitation experiments. For example, PEGPLGA with the same end-functionality (methoxy and PDS) and a comparable, albeit larger, size ( M n = 13.3 kDa for the PLGA and 5.1 kDa for the PEG block from NMR experiments; see Supporting Information, Section S1) behaved very differently: when PDS-terminated, macroscopic aggregation would occur in the chip, while the methoxy-terminated polymer would still provide slightly larger particles (always >80 nm; see Supporting Information, Figure S1) than PEG-PCL. As shown in Figure 2, it would appear that nanoparticles with different and controlled average size can be obtained solely on the basis of the TFR, with a general higher TFRsmaller particles relation. The only apparent exception is seen at FFR = 0.1 and TFR 2000 μL/min; this is, however, the most dispersed sample and the one where the process is possibly most out of control. It is noticeable that the particles retained their dimensions upon freeze drying and storage (for min. 1 week in the presence of sucrose as the cryoprotectant, see Supporting Information, Figure S2), suggesting the possibility of them to be still useable and sterile-filterable upon dry storage. However, these size values may be largely affected by the well-known high

3. RESULTS AND DISCUSSION 3.1. Nanoparticle Production and Characterization. In this study, we have produced biofunctional particles using PEGPCL as a macromolecular substrate that displayed PDS groups at the PEG end; non-functional (methoxy-terminated) PEGPCL was employed as a control in order to highlight any effect of the terminal group on the nanoparticle characteristics. First, we have confirmed that the presence of the PDS group was indeed the only significant difference between the two polymers. 1H NMR spectroscopy (Figure 1A) allowed the identification of all end groups and of the average composition of the polymers, which corresponded to M n values of 5.8 (PDSterminated) and 5.4 kDa (methoxy-terminated) for the PEG block and of 5.8 (PDS) and 6.4 kDa (methoxy) for the PCL block. Additionally, the two polymers showed very similar traces in triple detection GPC (Figure 1B), only differing for a high molecular weight small shoulder presented by PDS-PEG-PCL; the M n values of the block copolymers obtained via GPC [10.0 (PDS) or 11.2 kDa (methoxy)] were broadly in agreement with those calculated via NMR [11.7 kDa (PDS) or 11.8 kDa (methoxy)]. 26611

DOI: 10.1021/acsami.9b07972 ACS Appl. Mater. Interfaces 2019, 11, 26607−26618

Research Article

ACS Applied Materials & Interfaces

Figure 3. Size distributions (center) as determined by stand-alone DLS (dashed black line), DLS (AF4-DLS, red line), and RI (AF4-RI, solid black line) used as in-line detectors after AF4 separation, and cryo-TEM pictures (left and right) for “large” (TFR = 125 μL/min) PDS-PEG-PCL (A) and PEG-PCL (B) nanoparticles.

Figure 4. (A) Comparison of size distribution of “small” (TFR = 2000 μL/min), “medium” (TFR = 500 μL/min), and “large” (TFR = 125 μL/min) PEG-PCL and PDS-PEG-PCL nanoparticles as determined by stand-alone DLS (top graphs), AF4-DLS (middle graphs), and AF4-RI (bottom graphs). Circles (empty = “small”, black = “medium”, red = “large”) mark the peaks of each distribution, with corresponding numerical values reported in a tabular form (top middle, below graph legend; peaks in the 30−40 nm region reported in red). (B) Rg/RH values (Rg from MALS, RH from DLS detection; both after AF4 separation) recorded separately for the objects with RH < 100 nm and >100 nm. (C) Cryo-TEM pictures for “small” (TFR = 2000 μL/min) and “large” (TFR = 125 μL/min) PDS-PEG-PCL nanoparticles highlight the presence of elongated structures, which are present in more significant amounts in the latter particles. In each panel, arrows point to the features (size distribution, aspect ratio, and images) of large objects.

sensitivity of DLS to larger particles because of their higher scattering intensity compared to smaller particles. In order to confirm the control that the microfluidic-assisted preparation seems to exert on nanoparticle dimensions, we have conducted a more detailed analysis, by comparing DLS as a stand-alone instrument on unfractionated samples, with DLS

and RI employed as in-line detectors, respectively, for the determination of hydrodynamic size and concentration of sizefractionated samples (AF4-DLS and AF4-RI), further validating the results via cryogenic transmission electron microscopy (cryo-TEM, Figure 3). 26612

DOI: 10.1021/acsami.9b07972 ACS Appl. Mater. Interfaces 2019, 11, 26607−26618

Research Article

ACS Applied Materials & Interfaces

Figure 5. (A) Disulfide exchange reaction between terminal PDS groups and cysteine residues in peptides can be monitored by recording the absorbance of 2PT, which is obtained through rapid tautomerization of the 2-pyridine thiol and is the only species with a significant absorption in the 300−400 nm spectral region. On the right, the reaction between “large” PDS-PEG-PCL nanoparticles and GCRRG with a 3:1 Pept-SH/PDS stoichiometry (all curves reported in Supporting Information, Figure S4). (B) Kinetic parameters for PDS-PEG-PCL particles reacting with four peptides characterized by thiols with different pKa (values in the list from ref 27; they correspond to around 20, 10, 5, and 3% thiol deprotonation at pH = 7.4). The gray dashed line for 1:3 Pept-SH/PDS indicates the maximum achievable conversion (33.3 mol % PDS) (n = 6).

With both polymers and at any TFR, stand-alone DLS always showed monomodal size distributions, whereas AF4-DLS revealed bimodal populations (compare dotted black lines with solid red lines Figure 3, center; compare also the different graphs in Figure 4A); in the latter analysis, 30−50 nm objects were accompanied by larger materials, up to 200 nm in size, however, which could not be detected in AF4-RI. The AF4-RI signal, although less sensitive, is broadly proportional to the mass of the material, which would indicate the 30−50 nm nanoparticles as the major components of the suspensions. In terms of the identity of these colloids, the two important considerations are as follows:

is of relatively little importance because a 30 nm nanoparticle and a 15 nm micelle would be hardly distinguishable from most points of view. Rather, it is worth noting that this nanomanufacturing process designed to yield a kinetically controlled product (→ nanoparticles, depending on mixing conditions) probably still features minor amounts of a thermodynamically controlled one (→micelles; determined by molecular features). (b) The larger and highly scattering colloids appear to always be a minor component in any particle suspensions; we interpret them as wormlike micelles, or similar highaspect ratio aggregates: (1) elongated structures can be clearly seen in cryo-TEM (in higher amounts for “large” particles; see arrows in Figure 4C); (2) the Rg/RH values obtained from using both static and DLS as AF4 detectors are consistent with soft spheres for the 30−50 nm objects, but indicate a high aspect ratio shape for those sized >100 nm (Figure 4B).

(a) Under equilibrium conditions, these PEG-PCL macromolecules are likely to form (spherical) micelles in water, probably sized around 10−15 nm due to the low molecular weight of these polymers. Indeed, cryo-TEM appears to show objects sized considerably below 30 nm and even down to 10−15 nm. Their apparent absence in most AF4 traces may be ascribed to a combination of low concentration and low scattering/RI difference from the from the medium, although we cannot exclude that they may also cross the AF4 semipermeable membrane. However, the precise lower limit in size of these colloids

In conclusion, materials were prepared predominantly in the form of 30−50 nm nanoparticles, with variable but small amounts of wormlike micelles. At this point, we would like to emphasize the striking difference between stand-alone DLS and 26613

DOI: 10.1021/acsami.9b07972 ACS Appl. Mater. Interfaces 2019, 11, 26607−26618

Research Article

ACS Applied Materials & Interfaces

Figure 6. (A) Size distributions of peptide-functionalized “medium” particles (TFR = 500 μL/min) as obtained through stand-alone DLS, AF4-DLS, and AF4-RI. For better visibility, the four AF4-DLS curves are not overlapped; please note that, although all show a rather noisy >100 nm part due also to inherent fluctuation of the DLS-based in-line detection, they coherently provide a bimodal picture; there, the comparison with SLS allows us to identify smaller objects as rather spherical, the larger ones as elongated. (B) ζ-potential (n = 3) determined by electrophoretic mobility experiments of the peptide-functionalized particles in comparison to PEG-PCL and PDS-PEG-PCL nanoparticles. The non-functional nanoparticles (OMeterminated) and the PDS-terminated ones have mildly negative ζ-potential values, which are maintained after the conjugation with the peptides containing DR or DG sequences. On the contrary, the presence of double charged peptides bearing RR and DD sequences clearly shifted the nanoparticle ζ-potential respectively to slightly positively or considerably more negative values (bars outlined in red), thereby confirming the presence of these peptides at the nanoparticle surface.

Figure 7. (A) Scheme of the possible exchange between nanoparticle surface disulfides and soluble thiols (GSH), which would lead to the decoration of particles with GSH and the release of soluble peptides. (B) HPLC traces of the four different peptides as a function of their concentration, and an example (right) of HPLC quantitation of GCDDG release from Pept-PEG-PCL nanoparticles (prepared from “medium” PDS-PEG-PCL nanoparticles) after 7 days with (red curve) or w/o (black curve) GSH. (C) Release of the peptides over time from Pept-PEG-PCL “medium” nanoparticles (TFR = 500 μL/min and FRR = 0.5; functionalization performed at 3:1 Pept-SH/PDS stoichiometry) in PBS. (D) As in (C), but using 10 mM GSH in PBS (n = 3).

obtain kinetically “easier” structures (roughly spherical aggregates) at the expense of likely more thermodynamically stable ones. 3.2. Nanoparticle Functional Behavior: Differential Possibility of Disulfide Restructuring. We have shown that “large,” “medium,” and “small” particles have a largely similar size distribution but appear to differ in the amount of larger aggregates. Here, we have compared their behavior, in order to

AF4-DLS: the two techniques use the same kind of signal (scattered intensity), but the excessive sensitivity of the former to the presence of larger objects makes DLS substantially unreliable for the analysis of complex colloidal suspensions. Lastly, the apparent higher TFRsmaller particles seen before should be revisited as higher TFRfewer wormlike micelles; although hypothetical, the explanation is that the higher the TFR, the more rapid the mixing, and the easier to 26614

DOI: 10.1021/acsami.9b07972 ACS Appl. Mater. Interfaces 2019, 11, 26607−26618

Research Article

ACS Applied Materials & Interfaces

Figure 8. HCT116 viability (mitochondrial reductase activity per mg of cell protein in lysates, relative to control samples of HCT116 cultured in medium containing 17 vol % PBS) after exposure to nanoparticles of PEG-PCL terminated with OMe groups (A), with PDS groups (B), and with GCDDG (C) and GCRRC (D) peptides. It is worth mentioning that all particles were loaded with pyrene, which in itself at these concentrations had also negligible effects on cell viability, see Supporting Information, Figure S7 (n = 3).

highlight any possible effect of these aggregates during the surface functionalization. The disulfide exchange reaction was monitored using 2PT absorbance as a reporter, by fitting it as a function of time with an exponential model (y = a(1 − e−bx)), where a is a maximum conversion, therefore related to the PDS availability, and b expresses the reactivity of the partners (Figure 5A). Because thiolates are the active species in disulfide exchange,28 as much as in other thiol-based reactions such as Michael-type addition,27 it is expected that the kinetics can be controlled through the peptide primary structure. Therefore, we have employed four cysteine-containing peptides as thiols, where neighboring charges (arginine, R, and aspartic acid, D) determine the thiol pKa and thus the thiolate concentration at neutrality. Indeed, our results show that, while the peptide identity did not significantly affect the overall amount of exchanged disulfide (essentially quantitative in all cases), it significantly affected the kinetics (Figure 5B). In particular, the following two points are clear:

disulfide exchange can be considered valid, although probably modulated by local factors; for example, this was shown in 2011 by the group of Gauthier and Leroux,29 who demonstrated that the local thiolate concentration can increase or decrease depending on electrostatic attraction/repulsion with vicinal groups. Finally, DLS/AF4 experiments showed that the disulfide exchange did not significantly affect the size distribution of the nanoparticles (Figure 6A); on the contrary, the surface charge strongly depended on the charge of the peptide (Figure 6B), which indicates the mild character of the biofunctionalization reaction and at the same time confirms its effectiveness. Therefore, having ascertained that disulfides can be readily formed from PDS groups, with a kinetics controlled by the features of the thiols, we have proceeded to evaluate whether these newly formed disulfides may further exchange in the presence of other thiols, specifically using GSH (Figure 7A). Understanding the likelihood of this exchange is most important to evaluate the potential of disulfides other than PDS to restructure in a biologically relevant and thiol-rich environment. It is noteworthy that this exchange is different from protein glutathionylation;30 the latter process, which is thought to regulate the activity of several enzymes, is typically based on the exchange between peptidic (protein) thiols and GSH disulfide (GS-SG). Here, on the contrary, the reaction occurs between GSH in its free thiol form (GSH, pKa = 9.1 in water) and a mixed disulfide (on the particles). In this equilibrium, the less-acidic thiol would be typically released, although the large excess of GSH (10 mM vs 5−6 μM for particle-bound disulfides) may also at least partially overcome the difference in acidity. We have followed the peptide release in solution, monitoring their concentration via HPLC (Figure 7B): in PBS, DR-, DG-, and DD-containing peptides were released in amounts ranging from 10 to 18% (calculated on the basis of the quantative conversion of PDS groups into disulfide-bound peptides as seen in Figure 5B) reaching a plateau at about 24 h (Figure 7C); the amount of released peptide inversely scaled with the thiol

(a) The full availability of PDS groups. Employing three different disulfide−thiol stoichiometric ratios, the parameter a was always close to its theoretical value (100% with thiol in excess or 1:1, 33.3% with disulfide in excess). Even if the nanoparticle size is always much larger than the dimension of the polymer coils, and therefore a significant number of the PEG termini must necessarily be located within the particles, their reactive groups are still fully accessible to the peptidic thiols. (b) The quicker reaction of more acidic thiols. The only peptide with a net positive charge next to the thiol (GCRRG, two arginine residues) reacted considerably faster than the other three. The rate parameter of these noncationic peptides scaled with the thiol pKa at a 1:1 thiol−disulfide stoichiometric ratio, whereas with ratios of 1:3 and 3:1 no clear dependency could be seen; nevertheless, the general trend thiol acidity/reactivity in 26615

DOI: 10.1021/acsami.9b07972 ACS Appl. Mater. Interfaces 2019, 11, 26607−26618

Research Article

ACS Applied Materials & Interfaces

Figure 9. (A) Nanoparticle uptake (evaluated using pyrene fluorescence in cell lysates; see Supporting Information, Figures S5 and S6) in HCT116 in 10% FBS. The uptake is expressed as the relative % of mass in lysates. The value of pyrene alone (provided to cells in PBS, at the same dose as when is loaded in PEG-PCL nanoparticles) is presented at 2 h as a reference (horizontal red line). (B) As in (A), but with 10 mM GSH in the cell culture medium. (C) Comparison of cell uptake at 2 h in the absence (left) and in the presence (right) of 10 mM GSH (n = 3 for the kinetics, n = 6 for the 2 h time points). P-values determined by a Welch’s two-tailed t-test: *≤0.05, **≤0.01, ***≤0.001, absence of symbols indicates an absence of statistically significant differences (P > 0.05).

Incubation with nanoparticles at concentrations up to 190 μg/mL and for times up to 48 h, with or without GSH (Figure 8), showed negligible effects on cell viability, independent of the polymer terminal group (methoxy for PEG-PCL, PDS, GCRRG, and GCDDG; the latter were chosen not only because of their stronger effect on the nanoparticle charge, see Figure 6B, but also because of their different disulfide exchange kinetics). Considering the absence of toxicity from these particles, we have used their uptake in HCT116 as a cellular test to confirm whether the disulfide-rich surfaces would undergo any significant restructuring in the presence of GSH (negligible peptidic disulfide restructuring in the first few hours of exposure, as seen in Figure 7). We have evaluated nanoparticle uptake by assessing the fluorescence of cell lysates,31 which take into account both surface-bound and internalized particles; within the duration of the experiments (2 h), the former are more likely than the latter, but it is noteworthy that in no case saturation was reached, independent of the presence of GSH (Figure 9A,B). Two points are apparent, which are as follows: (A) PDS-bearing nanoparticles were taken up quite rapidly in the absence of GSH, likely due to the direct reaction with thiols on the cell surfaces; indeed, it has been demonstrated that GSH-thiols exported from the cytosol

acidity, and the most acidic RR peptide was not released at all. This would suggest that the release to be controlled by disulfide (kinetic) stability, which is highest with the RR peptide, although we cannot completely exclude a fraction of not covalently bound (adsorbed) peptides. In the presence of GSH (Figure 7D), the peptide disulfides were more significantly exchanged, although with a similar kinetics (plateau in 24 h) and again more easily for the less acidic peptides, whereas the GCRRG was substantially not released. Therefore, we can conclude that a higher thiol acidity produces (a) a more rapid reactivity with PDS and (b) a less restructuring disulfide. It is also noteworthy that also for restructuring peptides, their exchange kinetics was relatively slow; this means that whatever be the peptide present and with a large GSH excess, the nanoparticle surfaces do not appreciably restructure for at least a few hours. 3.3. Nanoparticle Toxicity and Uptake in HCT116. We have employed the colorectal tumor HCT116 cell line as an in vitro model in order to assess first the potential toxicity of the nanoparticles after functionalization and then the influence of the surface chemistry (the nature of the disulfide-bonded group) on the kinetics of nanoparticle uptake. 26616

DOI: 10.1021/acsami.9b07972 ACS Appl. Mater. Interfaces 2019, 11, 26607−26618

Research Article

ACS Applied Materials & Interfaces to the cell membrane32 or membrane-associated thiolbearing proteins33 can attack proximal (pericellular) disulfides in order to bind them to the cell surface. When these cell-surface thiols are converted to thioethers or pre-oxidized to disulfides, the uptake of disulfidemodified fluorophores is reduced.34 It has also been shown that small disulfide-bearing molecules bind to cellsurface thiols to be dependent on disulfide stability (enhanced binding at lower stability) and is followed by uptake which is not necessarily dependent on endosomal/ lysosomal pathways.32,34 Here, the nanoparticle uptake dropped dramatically in the presence of GSH and was statistically indistinguishable from that of the nonfunctional, OMe-terminated control (compare Figure 9A,B, and see the statistical analysis of the 2 h data in Figure 9C). Because of the reactivity of PDS, the particles in the latter experiments are assumed to be completely covered by GSH-containing rather stable disulfides, thereby reducing the reactivity of the nanoparticle surface with cell thiols. This surface functionality can therefore be seen as a substantial passivation. (B) The uptake of both peptide-bearing particles (about 10% of the dose after 2 h) was larger than both the control and the GSH-bearing particles (about 5% in the same time period), independent of the presence of GSH (compared to the PEG-PCL control) and at any time point.

extracellular space, while still allowing (slow) thiol release under intracellular reducing conditions.



ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acsami.9b07972. Nanoparticles produced from PEG-PLGA, nanoparticle stability in the dry form, additional experimental data for nanoparticle functionalization, and additional data for cell culture (toxicity and analysis of cell lysates) (PDF)



AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. ORCID

Nicola Tirelli: 0000-0002-4879-3949 Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS The authors would like to thank Sigma-Aldrich for funding this project and for supplying the polymers. The authors are also much indebted to Dr. Chiara Parodi for her help in cell culture experiments and to Dr. Nora Francini for the GPC measurements.



These data, therefore, align to what is seen at a molecular level: the uptake in the first 2 h is affected by GSH only when nanoparticles display the rapidly restructuring PDS disulfide, but not the slowly restructuring GCDDG or the non-restructuring GCRRG.

REFERENCES

(1) Donno, R.; Gennari, A.; Lallana, E.; De La Rosa, J. M. R.; d’Arcy, R.; Treacher, K.; Hill, K.; Ashford, M.; Tirelli, N. Nanomanufacturing through microfluidic-assisted nanoprecipitation: Advanced analytics and structure-activity relationships. Int. J. Pharm. 2017, 534, 97−107. (2) Lallana, E.; Donno, R.; Magrì, D.; Barker, K.; Nazir, Z.; Treacher, K.; Lawrence, M. J.; Ashford, M.; Tirelli, N. Microfluidic-Assisted Nanoprecipitation of (PEGylated) Poly(D,L-Lactic Acid-co-Caprolactone): Effect of Macromolecular and Microfluidic Parameters on Particle Size and Paclitaxel Encapsulation. Int. J. Pharm. 2018, 548, 530−539. (3) Baby, T.; Liu, Y.; Middelberg, A. P. J.; Zhao, C.-X. Fundamental studies on throughput capacities of hydrodynamic flow-focusing microfluidics for producing monodisperse polymer nanoparticles. Chem. Eng. Sci. 2017, 169, 128−139. (4) Kamaly, N.; Fredman, G.; Fojas, J. J. R.; Subramanian, M.; Choi, W. I.; Zepeda, K.; Vilos, C.; Yu, M.; Gadde, S.; Wu, J.; Milton, J.; Carvalho Leitao, R.; Rosa Fernandes, L.; Hasan, M.; Gao, H.; Nguyen, V.; Harris, J.; Tabas, I.; Farokhzad, O. C. Targeted Interleukin-10 Nanotherapeutics Developed with a Microfluidic Chip Enhance Resolution of Inflammation in Advanced Atherosclerosis. Acs Nano 2016, 10, 5280−5292. (5) Chen, D.; Love, K. T.; Chen, Y.; Eltoukhy, A. A.; Kastrup, C.; Sahay, G.; Jeon, A.; Dong, Y.; Whitehead, K. A.; Anderson, D. G. Rapid Discovery of Potent siRNA-Containing Lipid Nanoparticles Enabled by Controlled Microfluidic Formulation. J. Am. Chem. Soc. 2012, 134, 6948−6951. (6) Lim, J.-M.; Cai, T.; Mandaric, S.; Chopra, S.; Han, H.; Jang, S.; Il Choi, W.; Langer, R.; Farokhzad, O. C.; Karnik, R. Drug loading augmentation in polymeric nanoparticles using a coaxial turbulent jet mixer: Yong investigator perspective. J. Colloid Interface Sci. 2019, 538, 45−50. (7) Surdo, S.; Geven, M. A.; Donno, R.; Diaspro, A.; Tirelli, N.; Duocastella, M. Cavitation-assisted micromixing for polymeric nanoparticle generation. Proceedings 2018, 2, 942. (8) Stetefeld, J.; McKenna, S. A.; Patel, T. R. Dynamic light scattering: a practical guide and applications in biomedical sciences. Biophys. Rev. 2016, 8, 409−427.

4. CONCLUSIONS First, here we have proven the feasibility of the microfluidicassisted preparation of thiol-reactive polymer nanoparticles and the good control over particle size (predominantly