Intracellular Drug Release from Curcumin-Loaded PLGANanoparticles Induces G2/M Block in Breast Cancer CellsPaolo Verderio,†,‡ Paolo Bonetti,§,‡ Miriam Colombo,†,§ Laura Pandolfi,†,§ and Davide Prosperi*,†,§

†Dipartimento di Biotecnologie e Bioscienze, Universita di Milano-Bicocca, piazza della Scienza 2, 20126 Milano, Italy§Dipartimento di Scienze Biomediche e Cliniche “Luigi Sacco”, Universita di Milano, Ospedale L. Sacco, via G.B. Grassi 74, 20157Milano, Italy

*S Supporting Information

ABSTRACT: PLGA nanoparticles are among the most studiedpolymer nanoformulations for several drugs against different kindsof malignant diseases, thanks to their in vivo stability and tumorlocalization exploiting the well-documented “enhanced permeationand retention” (EPR) effect. In this paper, we have developeduniform curcumin-bearing PLGA nanoparticles by a single-emulsion process, which exhibited a curcumin release following aFickian-law diffusion over 10 days in vitro. PLGA nanoparticleswere about 120 nm in size, as determined by dynamic lightscattering, with a surface negative charge of −30 mV. The loadingratio of encapsulated drug in our PLGA nanoformulation was 8 wt%. PLGA encapsulation provided efficient protection of curcuminfrom environment, as determined by fluorescence emissionexperiments. Next, we have investigated the possibility to studythe intracellular degradation of nanoparticles associated with a specific G2/M blocking effect on MCF7 breast cancer cells causedby curcumin release in the cytoplasm, which provided direct evidence on the mechanism of action of our nanocomplex. Thisstudy was carried out using Annexin V-based cell death analysis, MTT assessment of proliferation, flow cytometry, and confocallaser scanning microscopy. PLGA nanoparticles proved to be completely safe, suggesting a potential utilization of thisnanocomplex to improve the intrinsically poor bioavailability of curcumin for the treatment of severe malignant breast cancer.

■ INTRODUCTION

Turmeric Curcuma longa L. rhizomes have been used for a longtime in Indian medicine for the treatment of a large variety ofdiseases, including prostate and breast cancer and othercommon malignancies in humans.1,2 The medicinal propertiesof this plant have been attributed to the main componentpresent in the rhizome, chemically termed diferuloylmethane or1,7-bis (4-hydroxy-3-methoxyphenyl)-1,6-hepadiene-3,5-dione,which is commonly referred to as curcumin. This compoundhas been shown to possess a wide range of pharmacologicalactivities, including anti-inflammatory,3,4 antioxidant,5,6 andanticancer effects.7−11 In the present work, we have focused onantiproliferative effects of curcumin in a breast cancer model(MCF7 cell line).In recent years, in vivo and in vitro studies have

demonstrated curcumin ability to inhibit carcinogenesis atthree stages: tumor promotion, angiogenesis, and tumorgrowth. While this difference among specific cell types is notclear, this phytochemical has been shown to selectively killtumor cells even in the presence of normal cells. Curcuminactivity, indeed, has been correlated to the downregulation ofvarious transcription factors, such as nuclear factor NFkB.12

Mackenzie et al. demonstrated that curcumin is taken up by H-RS cells and inhibits the NFkB-DNA interaction as well as the

expression of NFkB target genes.13 In addition, curcumininteracts with TrxR, which is overexpressed in tumor cells,resulting in its conversion mediated by NADPH oxidase, whichleads to an increased production of H2O2 in tumor cells causingcancer cell death.14 This compound is also active towardepidermal growth factor receptor (EGFR), in particular, thehuman epidermal growth factor receptor 2 (HER2).15 For thisreason, curcumin is capable of inhibiting the proliferation andsurvival of HER2-positive breast cancer cells, includingMCF7,16−18 by triggering the downregulation of AP-1, cyclinD1, cyclin E, and, in the same context, the upregulation of p21,p27, and p53, resulting in cell proliferation repression, cell cyclearrest, apoptosis, and inhibition of cell migration.16,19−21

Finally, glutathione levels in tumor cells are generally lowerthan in normal cells, thus enhancing the sensitivity of tumorcells to curcumin.22 However, despite the great potential of thisnutraceutic agent for cancer treatment, studies over the pastyears related to absorption, distribution, metabolism, andexcretion (ADME) have revealed extremely rapid metabolismand poor absorption of this molecule (low serum levels, limited

Received: November 7, 2012Revised: January 6, 2013Published: January 27, 2013

Article

pubs.acs.org/Biomac

© 2013 American Chemical Society 672 dx.doi.org/10.1021/bm3017324 | Biomacromolecules 2013, 14, 672−682

tissue distribution, apparent rapid metabolism, and short half-life) that severely reduces its bioavailability.23,24 To overcomethis matter, modern medicine requires new drug deliverysystems (DDS) to provide longer circulation times, increasedpermeability, and resistance toward metabolic presystemicdegradation. Two possible approaches to improve thebioavailability of curcumin are (1) altering its structural featuresresulting in more soluble analogues or (2) creating promisingnovel nanoparticle formulations, including liposomes, micelles,and phospholipid complexes.25,26 Besides these examples,encapsulation of curcumin in poly(lactic-co-glycolic acid)(PLGA) nanoparticles could also increase bioavailability.Shaikh et al. demonstrated that curcumin entrapped in PLGAnanoparticles increased drug oral bioavailability at least 9-foldin rats when compared with curcumin administered withpiperine as absorption enhancer.27 In another work, Xie et al.found that 200 nm PLGA nanomicelles decreased drug uptakeby liver and spleen and enhanced drug diffusion in lung andbrain in mice.28

Based on the promising data in in vivo experiments fromprevious reports, in the present work we have developedcurcumin-loaded PLGA nanovectors to investigate the growthinhibition of breast cancer cells (MCF7), aiming to disclose therelevant factors related to the pharmacological activity ofcurcumin-loaded nanocomplexes. In particular, we haveevaluated the in vitro loading capabilities and release kineticsof our PLGA nanoformulation and drug local release inside thecytoplasm of MCF7 breast cancer cells leading to a controlledinhibition of cell proliferation induced by curcumin-triggeredcycle arrest.

■ EXPERIMENTAL SECTIONMaterials and Methods. PLGA [poly(lactic-co-glycolic acid), Mw

= 45−70 kDa with an inherent viscosity of 0.59 dL g−1]/polyvinylalcohol (PVA, Mw = 9−10 kDa, 80% hydrolyzed), 50:50, andcurcumin were purchased from Sigma-Aldrich (St. Louis, MO) andused as received without further purification. Other reagents andsolvents were supplied by Sigma-Aldrich (St. Louis, MO), Fluka (St.Gallen, Switzerland), and Riedel-de Haen (Seelze, Germany) and usedas received. Water was deionized and ultrafiltered by a Milli-Qapparatus (Millipore Corporation, Billerica, MA).Drug-Loaded PLGA Nanoparticle Preparation. Polymeric-

curcumin loaded nanoparticles (PCNPs) were prepared using asingle-emulsion process,29,30 which is also referred to as the “solvent-evaporation” technique, an ideal method for the encapsulation ofhydrophobic drugs. The optimized formulation is briefly reported. Adrug/polymer mixture was prepared dissolving PLGA (50 mg) andcurcumin (10 wt % polymer) in ethyl acetate (500 μL) in a glass vial,and the mixture allowed to dissolve for 45 min, with intermittentstirring. Into another glass tube was prepared a 5 wt % PVA aqueoussolution (2 mL). The drug/polymer solution was added dropwise tothe surfactant solution under vortexing and then it was left stirring at ahigh setting for additional 30 s. To create an oil-in-water emulsion, theformulation was treated with 3 cycles (30 s) of sonication andvortexing. Ultrasounds were generated by S15H Elmasonic Apparatusfrom Elma GmbH (Singen, Germany). At the end of the cycles, theformed emulsion was quickly poured into 0.3 wt % PVA aqueoussolution (25 mL) under vigorous stirring for 3−4 h to allow fororganic solvent evaporation. At the end of reaction, drug loadednanoparticles were collected by centrifugation at 8000 × g for 15 minand washed three times with deionized water. The supernatant wasanalyzed both by dynamic light scattering (DLS) and by Nanosight toestablish if all drug-loaded nanoparticles were precipitated and we didnot find any traces of particles within the sensitivity of these methods.Nanoparticulate pellets were resuspended in water and dried bylyophilization using a Christ Alpha 1−2 D freeze-dryer from Martin

Christ Gefriertrocknungsanlagen GmbH (Osterode an Harz, Ger-many). In this way, a fine yellow powder was obtained (38 mg).Control void nanoparticles (VPNPs) were synthesized by the methoddescribed above with the exception of the drug molecule. Nanoparticlecharacterization, drug content evaluation, morphology, critical micelleconcentration (CMC), DLS distribution, ζ-potential, Nanosightdistribution, fluorescence analyses, and kinetics of drug release wereevaluated following nanoparticles synthesis.

Drug Content and Encapsulation Efficiency (EE%). UV−visand fluorescence spectroscopies were used to determine the amount ofdrug encapsulated in PCNPs. A known amount of dry PCNPs (250μg) was dispersed in water and chloroform was added to the waterphase. After 3 h under stirring, residual curcumin released fromnanoparticles was determined by UV−vis and fluorescence emission.We checked that pure PLGA and PVA did not interfere in the analysisof curcumin. The encapsulation efficiency was calculated (eq 1) as

μ

μ=

·

·

−

−EE(%)drug ( g mg )

drug ( g mg ).100

loaded drug1

NPs

initial drug1

NPs (1)

UV−vis quantification at λmax = 420 nm was recorded by using aNanodrop 2000C spectrophotometer from Thermo Fisher Scientific(Wilmington, Germany). Fluorescence spectra were recorded using aFluoromax-4P spectrofluorometer from Horiba Scientific (New Jersey,U.S.A.). Samples were excited at a fixed wavelength (λex = 450 nm)and spectra were recorded in a wavelength range between 480 and 700nm. The fluorescence emission of curcumin was detected at 500 nm;the slit widths (for controlling magnitude and resolution oftransmitted light) were standardized at 3 and 3 nm for excitationand emission wavelength. The results of the experiment are expressedas average of three different analyses.

Morphology. TEM and SEM images of nanoparticles wereobtained by a Zeiss EM-109 microscope (Oberkhochen, Germany)operating at 80 kV. For analyses, nanoparticles were dispersed undersonication in water (50 μg mL−1) and a drop of the resulting solutionwas placed on a Formvar/carbon-coated copper grid and air-dried.

Particle Size and ζ-Potential Analyses. DLS measurementswere performed at 90° with a 90Plus Particle Size Analyzer fromBrookhaven Instrument Corporation (Holtsville, NY) working at 15mW of a solid-state laser (λ = 661 nm). Zeta-potential measurementswere elaborated on the same instrument equipped with AQ-809electrode and data were processed by ZetaPlus software. Viscosity andrefractive index of water were used to characterize the solvent. PCNPsand VPNPs were dispersed under mild sonication in the solvent inorder to avoid the formation of large aggregates. The final sampleconcentration used for measurements was typically 20 μg mL−1. Allmeasurements (accepted PDI below 0.3) were performed in triplicateand the average values were taken. Sight distribution spectra werecollected by NanoSight LM10 from NanoSight Limited (Amesbury,U.K.) and analyzed with Nanoparticle Tracking Analysis (NTA)software, version 2.2 Build 0363; samples were in a range ofconcentration around 10.0 × 108 nanoparticles mL−1 working at 23°C.

Evaluation of Nanoparticles Stability in Cell CultureMedium. We evaluated nanoparticles aggregation under physiologicalconditions during time with DLS measurements. Viscosity andrefractive index of DMEM were used to characterize the solvent.PCNPs and VPNPs were dispersed in DMEM at a concentration of 50and 100 μg mL−1 and their hydrodynamic diameter was estimated at37 °C during time from 1 h up to 72 h of incubation. Allmeasurements were performed in triplicate and the average valueswere taken.

Critical Micelle Concentration (CMC). CMC of VPNPs wasestimated by fluorescence spectroscopy using pyrene as a fluorescenceprobe.31−34 A solution of pyrene (6.0 mM) in acetone was preparedand added to water to give a pyrene concentration of 1.20 μM andthen acetone was evaporated (acetone-free pyrene solution). Solutionsof VPNPs in a range from 5.0 × 10−5 to 1.0 mg mL−1 were preparedand mixed together with pyrene solutions (Vtot = 2 mL); the finalconcentration of pyrene in each sample solution was 6.0 μM. The

Biomacromolecules Article

dx.doi.org/10.1021/bm3017324 | Biomacromolecules 2013, 14, 672−682673

fluorescence spectra of pyrene were recorded monitoring theexcitation fluorescence behavior at 334 and 338 nm. The CMC wasestimated as the cross-point when extrapolating the intensity ratio I338/I334 at low and high concentration regions.Fluorescence Stability Analysis of Encapsulated Drug.

Fluorescence emission of curcumin in different biological mediumwas analyzed to show the efficacy of nanoparticle protection of thedrug from external biological environment and degradation. Acetatebuffer, 20 mM (pH = 4.75), phosphate buffer, 20 mM (pH = 7.40),ammonium buffer, 20 mM (pH = 9.25), and cell culture Dulbecco’sModified Eagle’s medium (DMEM; pH = 7.40) were prepared. Freecurcumin (10 mM in DMSO) was diluted in all the buffers above-described at a concentration of 5.43 μM; however, PCNPs weredispersed directly in the medium at a concentration of 25 μg mL−1

(drug content equal to 5.43 μM) and VPNPs at the sameconcentration as a reference. The fluorescence emission spectra wererecorded using with the same instrument described above in awavelength from 490 to 700 nm (λex = 480 nm).In Vitro Drug Release Profiles. In vitro drug release of the drug

from PCNPs was evaluated by the dissolution technique.35,36 Thestudies of curcumin release were performed at 37 °C in phosphatebuffer solution PBS (150 mM, pH = 7.40) with 2 wt % albumin bovineserum to mimic a biological environment. A nanoparticulate dispersionof PCNPs (1 mg mL−1) in PBS was maintained at 37 °C in a closedglass tube, to prevent evaporation of the solvent, under mild stirring;samples were withdrawn at regular time intervals and replaced withfresh medium every 24 h to create a sink condition. After nanoparticlecentrifugation, solubilized curcumin in each sample was analyzedspectrophotometrically at 420 nm. All measurements were performedin triplicate and the average values were taken. The kinetic analysis ofthe release data was done using Peppas−Korsmeyer’s model (eq 2):37

= ·

= + ·

f k t

f k n tlog log log( )

nt

t (2)

where f t is the fractional amount of drug release, k is the releaseconstant, n is the release exponent, and t is the time of release.

Cell Cultures and In Vitro Experiments. The human HER2-positive MCF7 breast carcinoma cell line was maintained in DMEM-F12 medium, supplemented with 10% FBS and 2 mM L-glutamine andgrown at 37 °C in a 5% CO2 atmosphere. For each test performed,dried PCNPs or VPNPs were resuspended prior to use in completeculture medium to the final concentration (50 and 100 μg mL−1) andadded to cells. Stock solution of curcumin (100 mM) were prepared inDMSO and diluted with the culture medium to get the same amountof curcumin carried by PCNPs.

Cell Toxicity. To test PLGA nanoparticles toxicity, MTT assay wasperformed. Cells (1.5 × 103) were seeded in 96-well plate, eight wellsper type, and concentration of nanoparticle (i.e., VPNPs and PCNPs,50 and 100 μg mL−1). 24 h after plating, the culture media wasremoved and the toxicity was monitored after 24, 48, and 72 h ofexposure. According to the manufacturer’s instructions (CellTiter 96Non-Radioactive Cell Proliferation Assay, Promega), at the end ofeach exposure time, tetrazolium salt was added and formazan productwas detected after 4 h incubation, by means of 96-well plate reader at595 nm. Results are normalized with untreated control and areexpressed as the mean absorbance of ± SEM of three independentbiological replicates.

Apoptosis Measurement. Apoptosis has been detected by meansof Annexin V/7-AAD staining (BD Pharmingen) and analyzed by flowcytometry. MCF7 cells were seeded in 6-well plates (1.0 × 105 cellsper well) and 24 h after plating were incubated with PCNPs or VPNPs100 μg mL−1 for 24, 48, and 72 h or left untreated as control. Next,cells were washed twice with cold PBS, carefully trypsinized to avoidmechanical damage of membrane and resuspended in Annexin binding

Figure 1. Size and shape characterization of VPNPs and PCNPs. SEM images (inset: TEM magnification) of (A) as-synthesized VPNPs and (B)PCNPs. Nanosight distributions in deionized water of (C) VPNPs and (D) PCNPs.

Biomacromolecules Article

dx.doi.org/10.1021/bm3017324 | Biomacromolecules 2013, 14, 672−682674

buffer in the presence PE labeled Annexin V and 7-AAD. Sampleacquisition was performed using FACS SCAN (Becton Dickinson, SanJose, CA, U.S.A.) and analyzed with CellQuest Pro Software (BD).Cell Cycle Analysis. MCF7 cells were seeded in six-well plates and

incubated with PCNPs or VPNPs 100 μg mL−1 or with free curcuminat 21.7 μM for 24, 48, and 72 h. Then, cells were collected and fixed in70% ethanol. For DNA staining, after ethanol removal, cells wereincubated overnight at 4 °C with propidium iodide 10 μg mL−1 andRNase A 18 μg mL−1. Sample acquisition was performed using a flowcytometer equipped with doublet-discriminator module (FACSCalibur; Becton Dickinson, San Jose, CA, U.S.A.), and the DNAcontent analyzed by Flowjo software (TreeStar Inc. OR, U.S.A.).Determination of Curcumin Release by FACS Analysis.

MCF7 cells were seeded in six-well plates (1.0 × 105 cells per well)and 24 h after plating were incubated with 100 μg mL−1 PCNPs orwith 21.7 μM free curcumin or left untreated as control. After 24 hincubation, the cell culture media containing nanoparticles or freecurcumin was removed, cells washed three times with PBS and newculture media without PCNPs or curcumin was added. Total cellfluorescence was monitored sampling cells at 24, 48, or 72 h afterincubation. For fluorescence analysis, cells were washed with PBS,trypsinized, and collected in FACS tubes in complete culture medium.Sample acquisition has been performed using FACS SCAN equippedwith 488 nm argon laser and analyzed with CellQuest Pro Software(BD).

Confocal Laser Scanning Microscopy (CLSM). Cells weregrown on a coverslip, seeded in six-well plates (1.0 × 105 cells), andthe same experimental procedure for FACS analysis was followed.After incubation, cells were washed with PBS and fixed for 10 min at 4°C with 4% paraformaldehyde−PBS solution. Images were obtainedwith Leica DM IRE2 confocal microscope equipped with an argon/krypton laser to excite curcumin at 458 nm. At least five fields per eachsample of two biological replicates were taken.

■ RESULTS AND DISCUSSIONPreparation and Characterization of PLGA-Curcumin

Nanoformulation. After preliminary studies, the “single-emulsion” technique was selected to prepare water-solublePCNPs and VPNPs (Scheme S1). This method afforded at thesame time the maximum encapsulation efficiency and a narrowparticle size distribution. PVA (5.0 and 0.3 wt %) was used assurfactant to stabilize the nanoparticle emulsion during itsformation, leading to an increase in the solubility of the drugwithout nanoparticle agglomeration. As expected, TEM andSEM analyses revealed no evidence of PCNPs and VPNPsalteration in their morphological surface and structure: imagesfrom Figure 1 showed that VPNPs and PCNPs were sphericalin shape with diameters of 116.9 ± 3.8 and 128.37 ± 6.7 nm,respectively (Figure 1A,B). Mean size distribution by Nanosight

Figure 2. Drug structure conservation from solvent effect in different biological buffers. Fluorescence emission spectra of free curcumin (orangeline), PCNPs (blue line), VPNPs (green line), and solvent (gray line). Excitation wavelength was fixed at 480 nm. The concentration of free drugwas calculated equal to the same amount of drug encapsulated in polymeric nanoparticles. (A) Nanoparticles were dispersed in 20 mM acetatebuffer, pH 4.75; (B) 20 mM phosphate buffer, pH 7.40; (C) 20 mM ammonium buffer, pH 9.25; and (D) DMEM cell culture medium, pH 7.40.

Biomacromolecules Article

dx.doi.org/10.1021/bm3017324 | Biomacromolecules 2013, 14, 672−682675

analysis (Figure 1C,D) confirmed nanoparticle diameters with amaximum in intensity of 93 ± 23 nm for VPNPs and 104 ± 28nm for PCNPs.To follow their behavior in aqueous solution, we analyzed

the size distribution and surface charge of both kinds ofnanoparticles in deionized water by DLS and ζ-potential. Wechecked these parameters before and after the freeze-dryingprocess. This method could be the critical synthetic stepbecause it is known that very low temperatures could affectnegatively the colloidal stability during powder resuspen-sion.38,39 In this case, lyophilization was a necessary stepbecause PLGA nanoparticles are biodegradable, and thepolymer structure has a tendency to collapse after long-termexposures to water. From DLS analyses in water, initialhydrodynamic diameters of 99.6 ± 10.5 and 116.8 ± 18.2 nmwere measured for VPNPs and PCNPs, respectively, beforefreeze-drying. After lyophilization, we found values of 110.8 ±12.4 and 133.2 ± 4.3 nm, respectively. The mean surface chargewas about −21.0 ± 1.9 mV for VPNPs and −30.2 ± 3.7 mV forPCNPs without observing alteration following the cryodesicca-tion process.To predict the nanoparticle kinetic of aggregation in

biological fluids, we analyzed the evolution of hydrodynamic

changes of VPNPs and PCNPs in DMEM cell culture medium,performing a kinetic experiment from 1 h up to 72 h ofincubation at 37 °C. Aliquots of nanoparticles at fixed intervalsof time were withdrawn and analyzed by DLS (see Figure S1).Even in this case, we checked nanoparticle changes before (A)

Figure 3. In vitro kinetic of drug release. (A) Drug diffusion profile ofcurcumin from PCNPs using the dissolution technique in 150 mMPBS (pH 7.4, 2.0 wt % BSA) to mimic biological environmentconditions. Values represent mean ± SD of three batches (some errorbars are too small to be shown). (B) Kinetic calculation of the drugrelease rate keeping into consideration the Peppas−Korsmeyes’smodel (inset: full equation used to build up the kinetic curve).

Figure 4. Effect of VPNPs, PCNPs, and curcumin on cell proliferation.Proliferation assays have been performed on MCF7 breast cancer cellsin the presence of (A) VPNPs and (B) PCNPs (50 and 100 μg mL−1)and of (C) free curcumin (20 and 40 μM) at different times, asindicated. Untreated cells were used as control and represent 100% ofthe proliferation. Values are mean ± SD of three independent sets ofexperiments. Analyzed by one-way ANOVA, samples show statisticaldifference at *p ≤ 0.05 and **p ≤ 0.01.

Biomacromolecules Article

dx.doi.org/10.1021/bm3017324 | Biomacromolecules 2013, 14, 672−682676

and after (B) lyophilization. Interestingly, we observed thatbefore freeze-drying, VPNPs and PCNPs had an averagediameter after 1 h of 83.6 ± 2.2 nm (100 μg mL−1) and 85.1 ±4.3 nm (100 μg mL−1), respectively. These values were almostunchanged after the first 24 h of incubation, but later on, bothnanoparticle sizes raised up slowly to 100.1 ± 9.2 nm (100 μgmL−1 VPNPs, after 72 h) and 118.5 ± 3.9 nm (100 μg mL−1

PCNPs). However, after lyophilization, VPNPs increased theirsize from 95.8 ± 5.2 nm (100 μg mL−1) after 1 h to 130.3 ± 9.9nm (100 μg mL−1) after 72 h. PCNPs followed the samebehavior from 107.2 ± 12.3 nm (100 μg mL−1) up to 123.2 ±10.8 nm (100 μg mL−1) after 72 h of incubation. This kineticexperiment was tested also at a lower nanoparticle concen-tration (50 μg mL−1) to simulate the dilution tested in cellculture experiments. However, we did not observe anyalteration of hydrodynamic diameters up to 72 h of incubationfor both nanoparticle formulations (details are summarized inTable S1, Supporting Information).Assessments of Nanoparticulate Formulation and

Drug Protection from Solvent Effect. First, to find outthe optimal concentration of nanoparticles and their potentialuse as drug delivery system, we evaluated the critical micelleconcentration (CMC) of VPNPs using pyrene as a fluorescentprobe (Figure S3). This technique was very useful and gave usthe opportunity to calculate the lower concentration limit ofnanoparticles in aqueous solution. From the fluorescence graph,the mean intensity ratios (I338/I334) of pyrene excitation spectra

increased by increasing polymer concentration. Because theincrement in the intensity ratio indicates the aggregation ofpyrene into the hydrophobic reservoirs of the polymers, CMCwas determined from the crossover point of the secondderivative sigmoidal curve at low concentration ranges. In thisway, we estimated that the lower limit concentration of VPNPsin solution was approximately 9.07 ± 1.21 mg L−1.Hence, to achieve an ideal nanoparticulate formulation for

drug delivery, our first objective was to assess that the syntheticprocess did not change the chemical structure of the drug, fromwhich it derives its pharmacological properties. A secondimportant issue concerns the maintenance of the polymer shell,which should protect the drug from the solvent until thenanocomplex reaches the desired site of action. For instance,with the aim to demonstrate that curcumin was reallyencapsulated inside the polymer matrix and, at the sametime, was able to maintain its chemical properties, we furtherset up a series of fluorescence analyses in different physiologicalbuffers commonly used both in cellular and clinical treatments(Figure 2).We performed this experiment in 20 mM sodium acetate

buffer, pH 4.75 (panel A); 20 mM phosphate buffer, pH 7.40(panel B); 20 mM ammonium buffer, pH 9.25 (panel C); and acomplete cell culture growth medium DMEM, pH 7.40 (panelD). At first impression, a high fluorescence intensity of PCNPs(25 μg mL−1, 5.43 μM curcumin encapsulated) was detected incomparison with free drug showing a broad fluorescence peak,

Figure 5. Apoptosis assay. Cell death has been assayed by means of 7-AAD/annexin-V double staining on cells exposed to 100 μg mL−1 of PCNPsup to 72 h, or left untreated as control. The assay allows to distinguish viable cells (both annexin-V- and 7-AAD-negative) from early apoptotic (7-AAD−/annexin-V+), late apoptotic (7-AAD+/annexin-V+) and necrotic cells (7-AAD+/annexin-V−). Due to their high fluorescence intensity, PCNPscause a nonspecific signal (visible as a smear that crosses the dot plot quadrants) that increases the signal of late apoptotic cells of 1.5%. To theshown values has been already subtracted this percentage of false positive signal and represent the mean ± SD of three biological independent sets ofexperiments.

Biomacromolecules Article

dx.doi.org/10.1021/bm3017324 | Biomacromolecules 2013, 14, 672−682677

which became increasingly lower following the increment inpH. As expected, no fluorescence signal was found in blankbuffer solutions and in samples containing VPNPs at the sameconcentration of PCNPs. However, the analysis in DMEMrevealed a small peak nearby the fluorescence signal ofcurcumin in PCNPs and we observed this behavior, althoughwith minor intensity, in the sample of VPNPs and in theuntreated buffer. This behavior was probably attributable to thecolored dye contained in the medium. Despite this fact, tomimic exactly the cell growth conditions, we preferred toperform the analyses in the complete medium without changingany parameter. The emission spectrum of free curcuminshowed a strong shift of maximal intensity toward 610 nmprobably due to conformational changes of its molecular

structure caused by pH environment and by protein chelationeffect from the fetal bovine serum contained in themedium.40−42 It was worth noting that the strong fluorescencesignal of PCNPS was maintained throughout the overall pHrange. According to this result, we could demonstrate thatPCNPs were able to maintain their stability in a broad range ofphysiological conditions without showing any collapse ofnanoparticles from the solution and, concomitantly, that thesynthetic process to gain our formulation did not affect thechemical structure of curcumin, which should be the idealcondition to exploit its pharmacological properties for in vivoexperiments (Figure S5).

In Vitro Drug Release Kinetics. Based on the abovegeneral considerations, in vitro release kinetics helped us tohave access to fundamental implications on the porosity of ournanoformulation on a molecular level, possible interactionsbetween drug and the external environment or the polymermatrix and their influence on the mechanism and rate of drug

Figure 6. Cell cycle analysis. Cells were treated for different timeintervals with 100 μg mL−1 of VPNPs, PCNPs, or 21.7 μM ofcurcumin. At the end of each time, cells were ethanol-fixed, stainedwith PI and analyzed in the DNA content by flow cytometry. Relativedistribution of cell population in the cell cycle phases is presented.Exposure to nanoparticles or curcumin occurred (A) for 24, (B) 48,and (C) 72 h. Values are mean ± SE of three independent sets ofexperiments.

Figure 7. FACS analysis of curcumin release by PCNPs. MCF7 cellswere incubated with 100 μg mL−1 of PNCPs or free curcumin at 21.7μM. After 24 h, cells were washed, the medium was replaced, andcurcumin-associated fluorescence was monitored for the following 72h. (A) T = 0 is the starting time after 24 h of incubation; (B−D) timeintervals at 24, 48, and 72 h postincubation. The values shown are thewhole fluorescence of total cell population gated and are the mean ±SD of three biological replicates.

Biomacromolecules Article

dx.doi.org/10.1021/bm3017324 | Biomacromolecules 2013, 14, 672−682678

release. Such information would assist us in a predictiveapproach to the design and development of future sustaineddelivery in in vivo biological systems.The amount of curcumin into the nanoparticle system was

previously calculated by both UV−vis and fluorescencespectroscopy. For this purpose, a calibration curve of thepure compound in a mixture of 9:1 chloroform/ethanol waselaborated for the analyses (Figure S2). A known amount ofPCNPs was dispersed in aqueous solution under sonication andvigorous stirring. The organic phase was poured for 3 h to forma biphasic system, just the time necessary to dissolve the druginto the organic solvent, which was later detected according to

the two techniques previously described. We decided to expressthe encapsulation efficiency by the wt% difference of the finalencapsulated drug from the initial amount of curcumin, dividedby the weight of the PLGA polymer. In this way, we calculatedthat 80 ± 7 μg of drug were encapsulated for each mg of PLGAnanoparticles (8.0 wt %). Because at the beginning of thereaction the amount of curcumin was 10.0 wt %, we concludedthat an 80% drug encapsulation into the polymer matrix couldbe achieved by a single-emulsion process.Our release study was performed in 150 mM phosphate

buffer solution containing 2.0 wt % BSA with the aim toenhance curcumin dissolution because it exhibited limited

Figure 8. Confocal microscopy analysis of curcumin release by PCNPs. Cells were grown on coverslips, and the same experimental procedure asdescribed in Figure 7 has been followed. At least five fields for a sample of two biological replicates were taken. Scale bar = 40 μm.

Biomacromolecules Article

dx.doi.org/10.1021/bm3017324 | Biomacromolecules 2013, 14, 672−682679

solubility in pure buffer.42 PCNP suspension was maintained at37 °C with gentle stirring and small amounts of sample werewithdrawn at fixed times, while the same volume was replacedby fresh medium to mimic a physiological sink condition. Asexpected, we observed an initial burst in drug release in the first24 h of incubation (50.3 ± 6.5%) and then this values increasedup to 73.3 ± 8.0% within the following 72 h (overall 4 days).Next, the release of residual cargo became very slow and thecurve slope decreased to reach a plateau corresponding to amaximum after 10 days (81.3 ± 2.3%), as shown in Figure 3.Thus, it was clear that the incorporation of curcumin in

PLGA nanoparticles could significantly sustain its continuousand prolonged release. The data obtained from kinetic studieswas fitted according to the Korsmeyer−Peppas model:43 wecalculated that the regression coefficient of log f t (fractionalamount of drug release) against log t was found to be 0.972,with values of release exponent (n) and release constant (k) of0.456 and 0.674, respectively. By this calculation, we coulddemonstrate that the release of curcumin from PCNPs followeda Fickian-law diffusion with a release exponent of 0.45. For thedetermination of the exponent (n), we decided to investigatethe linear portion of the kinetic curve until the fractional releaseamount reached 60%.Assessment of Nanoparticle Toxicity in Cells. To study

the interactions of PLGA nanoformulation with MCF7 as acellular model, the toxicity of nanoparticles was firstinvestigated by measuring the cell proliferation as an index ofcell viability by means of the MTT specific assay (Figure 4).Following incubation of MCF7 cells for 24, 48, and 72 h with

VPNPs at concentrations up to 100 μg mL−1, we observed thatnanoparticles do not alter cell proliferation, indicating that thePLGA vector is nontoxic even after prolonged exposure (Figure4A). Once the intrinsic nanocarrier safety was established, weturned our attention to the properties of PCNPs to inhibit cellproliferation in our tumor cell model. In contrast to VPNPs,MTT assay revealed that, ready after 24 h, 50 μg mL−1 ofPCNPs were sufficient to efficiently trigger inhibition of tumorproliferation. The registered effect was dose- and time-dependent, leading to a decrease of proliferation by over 50%after incubating the cells for 72 h with 100 μg mL−1 ofnanoparticles (Figure 4B). These data were compared withdirect biological effect of curcumin assessed with MTT assay(Figure 4C). As expected, by incubating MCF7 breast cancercells with different concentrations of curcumin free drug, asignificant dose- and time-dependent growth inhibition wasnoticeable.Further evidence supporting the carrier safety derived from

the study of apoptotic parameter in cancer cells, by using theAnnexin-V specific assay to find possible signs of cell death thatthe MTT assay was unable to detect. To this end, flowcytometry was performed to monitor apoptosis-inducedchanges in the plasma membrane. It is well documented,indeed, that phosphatidylserine, which is asymmetricallylocalized on the cytosolic side of cell membrane under normalconditions, is reversed and exposed on the outer monolayer ofthe cell membrane during the early stages of apoptosis.44

Exploiting high-affinity of Annexin-V for phosphatidylserine,apoptotic cells could be quantitatively determined by usingfluorescently labeled Annexin-V in combination with a vital dyesuch as 7-amino-actinomycin (7-AAD).45 This assay allows todistinguish viable cells from early and late stages of apoptosisand from necrotic cells, which could be quantitativelydetermined. Viable cells with intact membranes exclude both

7-AAD and Annexin V, whereas the membranes of dead anddamaged cells are selectively permeable to these markers. Forexample, viable cells are Annexin-V- and 7-AAD-negative; cellsthat are in early apoptosis are Annexin-V-positive and 7-AAD-negative, and cells that are in late apoptosis or already dead areboth Annexin-V- and 7-AAD-positive. The incubation of MCF7even with high concentrations of PLGA nanocarrier resulted inneither apoptosis nor early stages of cell death (Figure S6),demonstrating a good safety profile of these nanoparticles.Next, we turned our attention to PCNPs to assess whether

the observed effect of reduced proliferation was actuallyattributable to apoptotic effects (Figure 5). Surprisingly,although the signal in this assay resulted disturbed by theintense fluorescence of curcumin (visible as a smear that crossesthe dot plot quadrants) by incubating cells with PCNPs,nanoparticle-treated samples showed the same physiologicalbasal apoptotic rate recovered with untreated samples, thus,excluding apoptotic events at the basis of the observed reducedproliferation.Previous studies have shown that curcumin inhibits the

proliferation of cancer cells by blocking the cell cycleprogression.7,46,47 To test whether the reduced proliferationobserved was indeed due to cell cycle arrest, an analysis ofDNA content in the samples treated with VPNPs and PCNPs,respectively, has been performed (Figure 6).Monoparametric DNA analysis with propidium iodide

staining showed that VPNPs treatment did not affect the cellcycle phases in comparison with control cells. In contrast, asignificant alteration of cell cycle has been monitored insamples treated with curcumin, leading to a reduction of Sphase and to a G2/M block. Moreover, a similar change incellular phases is also present in samples incubated withPCNPs, indicating a specific biological effect owned by thesenanoparticles. Taken together, these experiments suggested acytotoxic activity of curcumin-loaded PLGA nanoparticlesessentially attributable to the cell cycle arrest induced bycurcumin release. Further investigations were necessary toassess the extent of curcumin release into the cytoplasm.

Sustained Drug Release in Tumor Cells. To demon-strate that the biological effects of PCNPs was actuallycorrelated with curcumin release from nanoparticles, weexploited the fluorescent properties of the drug by consideringthe whole increase of cell fluorescence due to curcumin uponits release into the cytoplasm. After 24 h incubation of MCF7cells with PCNPs, the cell culture containing nanoparticles wasremoved, a fresh medium without PCNPs was then added, andthe whole cell fluorescence was monitored for the next threedays (Figure 7).FACS analysis demonstrated that the total fluorescence of

PCNP-treated cells underwent a strong increase compared tountreated cells, particularly noticeable immediately after the 24h period of incubation, and the levels of whole fluorescenceremained higher as compared with samples treated with thecurcumin free drug for the following two days. Of particularinterest is the fluorescence profile that greatly differs betweencells incubated with free drug and samples treated with PCNPs.In fact, the fluorescence curve of curcumin-treated cellsexhibited a narrow shape, indicating a homogeneous distribu-tion of the molecule into the cell population, while cells treatedwith PCNPs had a broader fluorescence profile, suggesting amarked heterogeneity of the population. Indeed, in nano-particle-treated samples, it is possible to distinguish two maincell populations. The first one is characterized by a larger

Biomacromolecules Article

dx.doi.org/10.1021/bm3017324 | Biomacromolecules 2013, 14, 672−682680

number of cells, which exhibit an average low level offluorescence intensity represented by a homogeneous curvewith shape and trend similar to those of samples treated withfree drug. The second population instead is characterized by asmaller number of cells having very high fluorescence intensity.Furthermore, it is worth noting that the fluorescence intensityof the main population remains unchanged over time while thesecond population progressively disappears. A possibleinterpretation of these observations is that the main populationconsists of cells with free curcumin derived from degradation ofinternalized nanoparticles, while the minor population isrepresented by cells harboring large nanoparticles not yetdissolved and responsible for the high fluorescence signal.According to this model, large nanoparticles gradually dissolveover time, allowing for the release of the drug inside the cell,thus sustaining the fluorescence of the main population. Thishypothesis was corroborated by direct observation of the cellsby laser scanning confocal microscopy (Figures 8 and S7).By using the above-mentioned experimental procedure, cells

were grown on a coverslip and, after incubation of MCF7 cellswith the PCNPs for 24 h, the cell culture containingnanoparticles was removed, fresh medium without PCNPswas added, and the whole cell fluorescence analyzed byconfocal microscopy for the next 72 h. As expected, accordingto FACS analysis, images indicated that the majority of the cellpopulation had a widespread cytoplasmic fluorescence, similarto curcumin-treated cells. In addition, in a small fraction of cellstogether with a diffuse cytoplasmic pattern of fluorescence,highly fluorescent spots were clearly visible, representing large-sized nanoparticles still loaded with curcumin. However, sizeand number of these nanoparticle-associated spots progres-sively decreased, suggesting the occurrence of a continuativedissolution of PCNPs and, thus, supporting the hypothesis of asustained release of curcumin in tumor cell cytoplasm.

■ CONCLUSIONPLGA is a polymer approved by the FDA and EMA for clinicaltrials in various drug delivery systems. PLGA-based nano-particles present well-documented advantages for drug deliveryin vivo, as they can protect drugs from degradation, enhancetheir stability, and by exploiting their nanoscale size, penetratespecific cancer and inflamed tissues via “enhanced permeationand retention” (EPR) effect. Moreover, the sustained release ofthe therapeutic agent from stable PLGA nanoparticles canincrease the efficacy of treatments improving pharmacokineticand pharmacodynamic profiles of drugs.25

Although several studies have been carried out on thepotential utilization of these nanoparticles both in vitro and invivo, poor direct evidence is available so far on the intracellularfate of PLGA nanocomplexes and, in particular, on thesustained release of drugs in cellular microenvironment. Inthis paper, we have encapsulated the naturally availablenutraceutic agent, curcumin, in PLGA nanoparticles, inves-tigated their kinetics of degradation in vitro and studied thedrug release in HER2-positive MCF7 breast cancer cells. Ourstudy demonstrates that PLGA nanoparticles are able to releasecurcumin intracellularly inducing time- and dose-dependentinhibition of proliferation via a G2/M block even at lowconcentrations of drug. The cytotoxic behavior was actuallytriggered by specific cell cycle arrest caused by free curcuminreleased in the cytoplasm, while PLGA nanoparticles proved tobe completely innocuous toward cells in absence of drug. Asthe therapeutic dosages commonly required for curcumin-based

treatments of inflammatory and cancer diseases are extremelyhigh due to a poor bioavailability of this molecule in vivo, ourresults suggest a great potential of PLGA-based nano-formulation of curcumin for the treatment of malignant breastcancer.

■ ASSOCIATED CONTENT*S Supporting InformationDLS analyses and stability tests, calibration curves for thecalculation of drug loading, CMC assay, fluorescence analyses,and apoptosis assay. This material is available free of charge viathe Internet at http://pubs.acs.org.

■ AUTHOR INFORMATIONCorresponding Author*Phone: +39-02-64483302. Fax: +39-02-64483565. E-mail:davide.prosperi@unimib.it.

Author Contributions‡These authors contributed equally.

NotesThe authors declare no competing financial interest.

■ ACKNOWLEDGMENTSM.C. and L.P. acknowledge the research fellowships ofCMENA (University of Milan). This work was supported byNanoMeDia Project (Regione Lombardia) and FondazioneRegionale per la Ricerca Biomedica (FRRB).

■ REFERENCES(1) Shah, S.; Prasad, S.; Knudsen, K. E. Cancer Res. 2012, 72, 1248−1259.(2) Nagaraju, G. P.; Aliya, S.; Zafar, S. F.; Basha, R.; Diaz, R.; El-Rayes, B. F. Integr. Biol. 2012, 4, 996−1007.(3) Sandu, S. K.; Pandey, M. K.; Sung, B.; Ahn, K. S.; Murakami, A.;Sethi, G.; Limtrakul, P.; Madmaev, V.; Aggarwal, B. B. Carcinogenesis2007, 28, 1765−1773.(4) Aggarwal, B. B.; Sundaram, C.; Malani, N.; Ichikawa, H. Adv. Exp.Med. Biol. 2007, 595, 1−75.(5) Manikandam, P.; Sumitra, M.; Aishwarya, S.; Manohar, B. M.;Lokanadam, B.; Puvanakrishnan, R. Int. J. Biochem. Cell Biol. 2004, 36,1967−1980.(6) Ak, T.; Gulcin, I. Chem. Biol. Interact. 2008, 174, 22−37.(7) Sa, G.; Das, T. Cell Div. 2008, 3, 14.(8) Campbell, F. C.; Collet, P. G. Future Oncol. 2005, 1, 405−414.(9) Bhattacharyya, S.; Mandal, D.; Sen, G. S.; Pal, S.; Banerjee, S.;Lahiry, L.; Finke, J. H.; Tannebaum, C. S.; Das, T.; Sa, G. Cancer Res.2007, 67, 362−370.(10) Lai, C. S.; Wu, J. C.; Yu, S. F.; Badmaev, V.; Nagabhushanam,K.; Ho, C. T.; Pan, M. H. Mol. Nutr. Food Res. 2011, 55, 1819−1828.(11) Rodwell, C. Nat. Rev. Cancer 2012, 12, 376.(12) Shishodia, S.; Amin, H. M.; Lai, R.; Aggarwal, B. B. Biochem.Pharmacol. 2005, 70, 700−713.(13) Mackenzie, G. G.; Queisser, N.; Wolfson, M. L.; Fraga, C. G.;Adamo, A. N.; Oteiza, P. I. Int. J. Cancer 2008, 123, 56−65.(14) Fang, J.; Lu, J.; Holmgren, A. J. Biol. Chem. 2005, 280, 25284−25290.(15) Junj, Y.; Xu, W.; Kim, H.; Ha, H.; Neckers, L. Biochim. Biophys.Acta 2007, 1773, 383−390.(16) Banerjee, M.; Singh, P.; Panda, D. FEBS J. 2010, 277, 3437−3448.(17) Kunwar, A.; Barik, A.; Mishra, B.; Rathinasamy, K.; Pandey, R.;Priyadarsini, K. I. Biochim. Biophys. Acta 2008, 1780, 673−679.(18) Shao, Z. M.; Shen, Z. Z.; Liu, C. H.; Sartippour, M. R.; Go, V.L.; Heber, D.; Nguyen, M. Int. J. Cancer 2002, 98, 234−240.

Biomacromolecules Article

dx.doi.org/10.1021/bm3017324 | Biomacromolecules 2013, 14, 672−682681

(19) Balasubramanian, S.; Eckert, R. L. J. Biol. Chem. 2007, 282,6707−6715.(20) Goel, A.; Kunnumakkara, A. B.; Aggarwal, B. B. Biochem.Pharmacol. 2008, 75, 787−809.(21) Sun, S. H.; Huang, H. C.; Huang, C.; Lin, J. K. Eur. J. Pharmacol.2012, 690, 22−30.(22) Syng, A. C.; Kumar, A. L.; Khar, A. Mol. Cancer Ther. 2004, 3,1101−1108.(23) Anand, P.; Kunnumakkara, A. B.; Newman, R. A.; Aggarwal, B.B. Mol. Pharmaceutics 2007, 4, 807−818.(24) Aggarwal, B. B.; Sung, B. Trends Pharmacol. Sci. 2009, 30, 85−94.(25) Danhier, F.; Ansorena, E.; Silva, J. M.; Coco, R.; Le Breton, A.;Preat, V. J. Controlled Release 2012, 161, 505−522.(26) Anand, P.; Thomas, S. G.; Kunnumakkara, A. B.; Sundaram, C.;Harikumar, K. B.; Sung, B.; Tharakan, S. T.; Misra, K.; Pryadarsini, I.K.; Rajasekharan, K. N.; Aggarwal, B. B. Biochem. Pharmacol. 2008, 76,1590−1611.(27) Shaikh, J.; Ankola, D. D.; Beniwal, V.; Singh, D.; Kumar, M. N.Eur. J. Pharm. Sci. 2009, 37, 223−230.(28) Song, Z.; Feng, R.; Sun, M.; Guo, C.; Gao, Y.; Li, L.; Zhai, G. J.Colloid Interface Sci. 2011, 354, 116−123.(29) Mathew, A.; Fukuda, T.; Nagaoka, T.; Hasumura, T.; Morimoto,H.; Yoshida, Y.; Maekawa, T.; Venugopal, K.; Kumar, D. S. PloS One2012, 7, e32616.(30) Cartiera, M. S.; Ferreira, E. C.; Caputo, C.; Egan, M. E.; Caplan,M. J.; Saltzman, W. M. Mol. Pharmaceutics 2010, 7, 86−93.(31) Ananthapadmanabhan, K. P.; Goddard, E. D.; Turro, N. J.;Kuot, P. L. Langmuir 1985, 1, 352−355.(32) Saito, K.; Ingalls, L. R.; Lee, J.; Warner, J. C. Chem. Commun.2007, 24, 2503−2505.(33) Peng, K.-T.; Chen, C.-F.; Chu, I.-M.; Li, Y-M; Hsu, W.-H.; Hsu,R. W-W.; Chang, P.-J. Biomaterials. 2010, 31, 5227−5236.(34) Lui, C. W.; Lin, W. J. Int. J. Nanomed. 2012, 7, 4749−4767.(35) D’Souza, S. S.; DeLuca, P. P. Pharm. Res. 2006, 23, 460−474.(36) Hickey, T.; Kreutzer, D.; Burgess, D. J.; Moussy, F. Biomaterials2002, 23, 1649−1656.(37) Seju, U.; Kumar, A.; Sawant, K. K. Acta Biomater. 2011, 7,4169−4176.(38) Abdelwahed, W.; Degobert, G.; Stainmesse, S.; Fessi, H. Adv.Drug Delivery Rev. 2006, 58, 1668−171.(39) Lemoine, D.; Francois, C.; Kedzierewicz, F.; Preat, V.; Hoffman,M.; Maincent, P. Biomaterials 1996, 17, 2191−2197.(40) Esatbeyouglu, T.; Huebbe, P.; Ernst, I. M. A.; Chin, D.; Wagner,A. E.; Rimbach, G. Angew. Chem., Int. Ed. 2012, 51, 2−27.(41) Wang, Y. J.; Pan, M. H.; Cheng, A. L.; Lin, L. I.; Ho, Y. S.;Hsieh, C. Y.; Lin, J. K. J. Pharm. Biomed. Anal. 1997, 15, 1867−1876.(42) Mitra, S. P. J. Surface Sci. Technol. 2007, 23, 91−110.(43) Costa, P.; Lobo, J. M. S. Eur. J. Pharm. Sci. 2001, 13, 123−133.(44) Koopman, G.; Reutelingsperger, C. P.; Kuijten, G. A.; Keehnen,R. M.; Pals, S. T.; van Oers, M. H. Blood 1994, 84, 1415−1420.(45) van Tilborg, G. A. F.; Mulder, W. J. M.; Chin, P. T. K.; Storm,G.; Reutelingsperger, C. P.; Nicolay, K.; Strijkers, G. J. BioconjugateChem. 2006, 17, 865−868.(46) Simon, A.; Allais, D. P.; Duroux, J. L.; Basly, J. P.; Durand-Fontanier, S.; Delage, C. Cancer Lett. 1998, 129, 111−116.(47) Choudhori, T.; Pal, S.; Das, T.; Sa, G. J. Biol. Chem. 2005, 280,20059−20068.

Biomacromolecules Article

dx.doi.org/10.1021/bm3017324 | Biomacromolecules 2013, 14, 672−682682