lable at ScienceDirect

Biomaterials 32 (2011) 2233e2240

Contents lists avai

Biomaterials

journal homepage: www.elsevier .com/locate/biomater ia ls

Recombinant spider silk particles as drug delivery vehicles

Andreas Lammel a,1, Martin Schwab b,1, Markus Hofer b, Gerhard Winter b, Thomas Scheibel a,*a Lehrstuhl Biomaterialien, Universitätsstraße 30, Universität Bayreuth, D-95440 Bayreuth, GermanybDepartment of Pharmacy, Pharmaceutical Technology and Biopharmaceutics, Ludwig-Maximilians-Universität, D-81377 Munich, Germany

a r t i c l e i n f o

Article history:Received 17 September 2010Accepted 26 November 2010Available online 24 December 2010

Keywords:BiomimeticsBiodegradationControlled releaseDrug loadingProtein

* Corresponding author. Fax: þ49 (0) 921 55 7346.E-mail address: thomas.scheibel@bm.uni-bayreuth

1 These authors contributed equally to this work.

0142-9612/$ e see front matter � 2010 Elsevier Ltd.doi:10.1016/j.biomaterials.2010.11.060

a b s t r a c t

Spider silk has been in the focus of research mainly due to the superior mechanical characteristics of silkfibers. However, it has been previously shown that spider silk proteins can also adopt other morphologiessuch as submicroparticles. This study examines the applicability of such particles as drug carriers. Particlecharacterization revealed that particles made of the engineered spider silk protein eADF4(C16)are colloidally stable in solution. Here, it is shown that small molecules with positive net-charge candiffuse into the negatively charged spider silk protein matrix driven by electrostatic interactions. Theloading efficiencies correlate with the distribution coefficient (logD) of small molecules of weak alkalinenature. Interestingly, constant release rates can be realized for a period of two weeks at physiologicalconditions in vitro, with accelerated release rates within acidic environments. Enzymatic degradationstudies of eADF4(C16) particles indicated that the silk proteins degrade slowly and the particles decreasein size. Along with their all-aqueous and easy preparation, drug loaded eADF4(C16) particles providea high potential for diverse applications in which controlled release from biodegradable carriers isdesired.

� 2010 Elsevier Ltd. All rights reserved.

1. Introduction

In order to achieve constant drug levels during therapy,sophisticated drug depot systems have been developed previously.With these systems toxic side effects are reduced and the numberof drug administration is decreased, while cellular uptake andbioavailability are improved [1e4]. Especially colloidal micro- andnanoparticulate carriers have been extensively investigated asa platform for controlled drug delivery [5e7]. In general, thematerial employed as carrier for pharmaceutical agents shouldoffer control of structure, morphology and function, while alsoexhibiting good mechanical stability [8]. Therefore, biodegradableand non-cytotoxic polymers are preferred since they retain theirproperties for a limited period of time before they graduallydecompose into soluble nontoxic degradation products that can beexcreted from the body. Many synthetic (aliphatic polyesters, poly-glycolic acid (PGA), polylactid acid (PLA), etc.) and natural (poly-saccharides and proteins) polymers have been employed toproduce degradable vehicles for encapsulation, incorporation orbinding of active compounds [9e13].

.de (T. Scheibel).

All rights reserved.

While synthetic polymers offer the potential of sustaining therelease of the encapsulated therapeutic agent over a period of daysup to several months, they typically demand organic solvents orrelatively harsh formulation conditions during processing withoften limited biocompatibility, due to remaining toxic solvents andacidic degradation products [14].Many natural polymers in contrastcan be processed at ambientmild conditions. In spite of the possibleadvantages of natural polymers concerning biocompatibility andprocessability, most biopolymers present a main drawback of rapidre-solubilization in aqueous environments since they are oftenhydrophilic, thus resulting in fast drug release profiles [15]. In orderto resolve this problem, chemical cross-linking procedures (e.g.using glutaraldehyde and formaldehyde treatment) have beenconsidered [16e18]. Unfortunately, the presence of residual cross-linking agents could lead to toxic side effects. In addition, unwantedreactions between the drug and the cross-linker could result in theformation of toxic or even inactivated derivatives [19,20].

Thedisadvantages of a systembasedonahydrophilic biopolymercan be diminished upon using a hydrophobic biopolymer capable ofsustained drug release [12]. Silk proteins are amphiphilic biopoly-mers and unify all aforementioned properties necessary for anoptimized drug delivery system [21e27]. Silk proteins from spidersand insects, especially Bombyx mori fibroin, have been investigatedfor their use in drug delivery devices [28e30]. In the area ofparticulate drug carriers, fibroin microspheres with diameters of

A. Lammel et al. / Biomaterials 32 (2011) 2233e22402234

several microns and loading efficiencies of up to 21% have beenreported [31]. Potentially, encapsulation efficiencies can beincreased up to 100% with the trade-off of having large sphere sizeswith diameters of 100 to 440 mm [28]. Further, the applied prepa-ration techniques are highly sophisticated and lack scalability.

Here, we employed a previously established recombinantlyproduced engineered spider silk protein eADF4(C16)mimicking theknown amino acid sequence of the natural spider silk protein ADF4from the European garden spider Araneus diadematus [32,33].Detailed studies of the thermodynamic assembly process of eADF4(C16) revealed that stable microspheres are formed upon additionof high concentrations (>400 mM) of lyotropic salts such aspotassium phosphate [34e36]. Analysis of morphology and struc-ture of the obtained micro- and sub-microspheres showed thatthey have a smooth surface and are solid with high b-sheet content(>60%) and no apparent sub-microstructure [34]. Further studies ofthe process parameters of eADF4(C16) sphere formation revealedthat sphere size (250 nme3 mm) and size distribution can becontrolled by mixing intensity and concentration of potassiumphosphate and concentration of eADF4(C16) [36].

The aim of this studywas to investigate the applicability of thesespider silk protein particles as drug carriers. Since the solventconditions only involve water and potassium phosphate at roomtemperature and neutral pH, and since the processing conditionscan be easily scaled-up, protein spheres made of engineered spidersilk are per se suitable for the encapsulation of sensitive drugs. Here,several model drugs were applied, and loading efficiencies weredetermined in dependence of drug molecule properties such aspartition coefficient and dissociation constant. Further, loadingefficiencies, release kinetics and biodegradation of the silk proteincarriers have been investigated.

2. Materials and methods

2.1. Materials

2.1.1. Engineering of recombinant spider silk protein eADF4(C16)The amino acid sequence of eADF4 (C16) was adapted from the natural sequence

of ADF4 from Araneus diadematus [33]. Our previously established engineeringapproach allowed the combination andmultimerization of singlemotifs, resulting inthe eADF4(C16) protein comprising 16 repeats of the sequence GSSAAAAAAAA-SGPGGYGPENQGPSGPGGYGPGGP resulting in a molecular mass of 48 kDa [33]. Theproteinwas purifiedwith purity higher than 98% as described previously [33]. Due toits amino acid composition, eADF4(C16) has a theoretical isoelectric point of 3.48indicating a net negative charge at a physiological pH of 7.4.

2.1.2. Small molecular model drugsAll drugs were dissolved in water at a concentration of 0.21 mmol/ml. Drug

substances and their featured properties are depicted in Table 1. The main selectioncriteria were solubility in aqueous media (expressed by the octanol/water partitioncoefficient (logP)), the acidic dissociation constant (pKa for protonated bases (BHþ)

Table 1List of small molecular weight model drugs used for eADF4(C16) sphere loading. Values fotaken from literature [37e40]. The partition coefficient (logP) accounts for the individual ufor each substance. All substances were purchased from SigmaeAldrich (Deisenhofen, G

Model drug Molecularweight [Da]

lAbs [nm] Dissociation constantof BHþ (pKa)

Chlorthalidone 338 276 e

Nipagin 152 254 e

Acetaminophen 151 242Phenol red 354 510 e

Tetracaine*HCl 301 310 8.20Procaine*HCl 272 290 8.05Papaverine*HCl 376 248 8.07Ephedrine*HCl 202 256 9.60Propranolol*HCl 295 290 9.10Biperiden*HCl 347 210 9.60Ethacridine lactate 343 365 11.00Methyl violet 407 590 e

or for acids (HA)) and the resulting net-charge in aqueous media (predominant orpermanent charge).

2.2. Preparation of eADF4(C16) particles

Lyophilized protein eADF4(C16) was dissolved in 6 M guanidiniumthiocyanate.Dialysis was performed against 10 mM tris(hydroxymethyl)aminomethane(Tris)/HCl, pH 8, at 4 �C using membranes having a molecular weight cut-off at6000e8000 Da (Spectrum Laboratories, Rancho Dominguez, USA). The concen-tration of eADF4(C16) solution was determined by UV-Vis-spectroscopy at 20 �Cusing a Cary100 spectrophotometer (Varian Medical Systems, Palo Alto, USA) andthe molar extinction coefficient of eADF4(C16) at 276 nm (e ¼ 46,400 M�1 cm�1).eADF4(C16) particles were prepared using a phase separation procedure asdescribed previously [34,36]. Briefly, an aqueous eADF4(C16) (c ¼ 1.0 mg/ml)solution was mixed with potassium phosphate (2 M, pH 8) in volumetric ratios of1:10 using a pipette. The resulting particles were centrifuged for 10 min at 10.000 gand washed three times with purified water. Particles were re-dispersed in water,and particle concentrations (particles in mg/ml) were determined gravimetrically.A stock dispersion of known protein particle concentration was used for allexperiments.

2.3. Colloidal stability of eADF4(C16) particles

The colloidal stability of eADF4(C16) particles in suspension was studied byadding 1.0 mg of particles to 1.0 ml of (NH4)2SO4 solutions (pH 6.5) of varyingconcentration (0e2.0 M) and measuring the intensity of scattered light at a wave-length of 400 nm after 15 min. (NH4)2SO4 was employed due to its kosmotropicnature enhancing the hydrophobic effect yielding particle agglomeration in aqueoussolutions. Based on Mie theory the intensity of scattered light in forward directionincreases with increasing particle size [37]. Therefore, the onset of electrolyte-induced flocculation in dilute dispersions can be detected by an increase in intensityof scattered light in forward direction [38].

2.4. Drug loading of eADF4(C16) particles

Drug loading of spider silk particles was conducted as follows: 100 ml of spidersilk particle suspension containing 21 nmol silk protein were mixed with 1.0 ml ofmodel drug solution containing 210 nmol model drug. After 10 min of incubation atroom temperature samples were centrifuged for 10 min at 10.000 g, and thesupernatant was analyzed for residual drug concentration using UV-Vis-spectros-copy. Standard calibration curves for model drugs were used for drug quantification.A control group of samples containing only 100-ml water mixed with 1.0 ml of modeldrug solutionwas prepared for each experiment. Drug concentrations of control andsample supernatants were used to calculate the amount of drug incorporated in thespider silk particles. All experiments were performed in triplicate at pH 6.5.Encapsulation efficiency and loading were determined using Eqs. (1) and (2),respectively:

encapsulation efficiency�w=w%

� ¼ amount of model drug in particlesmodel drug initially added

(1)

loading�w=w%

� ¼ amount model drug in particlesamount of particles

(2)

2.5. In vitro release studies

Drug loaded eADF4(C16) particles were washed with distilled water and sus-pended in 1 ml PBS (pH 7.4) before incubation at 37 �C with constant shaking. Eachvial contained 2mg of drug-loaded particles comprising 42-nmol spider silk protein.

r molecular weight, dissociation constants (pKa) and partition coefficients (logP) arenprotonated forms. The absorptionwavelength lAbs was determined experimentallyermany).

Dissociation constantof HA (pKa)

logP Predominantcharge at pH7

Permanentlycharged

8.9 �0.03 Uncharged No8.3 1.86 Uncharged No9.3 0.38 Uncharged No1.7; 7.7 3.00 Negative Yese 4.00 Positive Noe 2.40 Positive Noe 3.50 Positive Noe 1.30 Positive Noe 3.18 Positive Noe 3.50 Positive Noe 2.50 Positive Noe 3.20 Positive Yes

Fig. 1. eADF4(C16) particle characterization. a) Size distribution of eADF4(C16) parti-cles analyzed using laser diffraction spectrometry. The inset shows a scanning electronmicrograph of corresponding eADF4(C16) particles. The average diameter of theparticle ensemble was davg ¼ 332 � 95 nm. b) Investigation of colloidal stabilityassessed by intensity of scattered light at 400 nm. R2 is the correlation coefficient of thelinear fit.

A. Lammel et al. / Biomaterials 32 (2011) 2233e2240 2235

The solvent was periodically removed from each sample and replacedwith fresh PBS(pH 7.4). Then, the drug content in the medium was analyzed using UV-Vis-spec-trometry. The percentage of cumulative model drug release (% w/w) was investi-gated as a function of incubation time. Each experiment was performed in triplicate.To study the effect of different pH values on the release behavior of drug loadedeADF4(C16) particles, 1 mg of drug loaded silk particles were incubated in 1.0 ml PBSat 5 different pH values (pH 2, 4, 6, 7.4 and 8.8) for 5 days. The solvent was with-drawn daily and the particles were re-dispersed in fresh media. Supernatants ofdrawn samples were analyzed for drug content with UV-Vis-spectrometry.

2.6. In vitro degradation of eADF4(C16) particles

In order to analyze the degradability of eADF4(C16) silk particles, we employeda mixture of elastase and trypsin, two naturally occurring proteases in vertebrates.Elastase is a single polypeptide chain of 240 amino acid residues and contains fourdisulfide bridges. It is synthesized as an inactive zymogen, which requires limitedproteolysis at the N-terminal domain by trypsin in order to produce the activeenzyme [39]. Elastase is a serine protease with a broad specificity, as it will cleaveproteins at the carboxyl side of small hydrophobic amino acids such as Ile, Gly, Ala,Ser, Val, and Leu [38]. According to the product information of SigmaeAldrichelastase and trypsin have an optimum activity at 25 �C. To assess the degradability,1.0 mg of silk particles with an average diameter of 700 nmwere incubated at 25 �Cin 1.0 ml PBS in the presence of 0.8-mg elastase and 12.5-mg trypsin. Over twoweeks,samples were drawn on a daily basis and centrifuged. The pellets, containing eADF4(C16) particles, were re-dispersed in distilled water and washed three times forfurther analysis of size and morphology using laser diffraction spectrometry andscanning electron microscopy. Elastase and trypsin from hog pancreas weresupplied by SigmaeAldrich (St. Louis, USA).

2.7. In vitro release upon degradation

The release upon degradation was exemplarily studied with methyl violetloaded eADF(C16) particles. Therefore, 2 mg of loaded particles were suspended in1 ml PBS (pH 7.4) in presence of a mixture of elastase and trypsin at five differentconcentrations [control: no enzyme; C1: (0.8 mg elastase þ 12.5 mg trypsin)/1 mgeADF4(C16); C2: (1.6 mg elastase þ 25 mg trypsin)/1 mg eADF4(C16); C3: (2.4 mgelastase þ 37.5 mg trypsin)/1 mg eADF4(C16); C4: (4.0 mg elastase þ 62.5 mg trypsin)/1 mg eADF4(C16); C5: (8.0 mg elastase þ 125 mg trypsin)/1 mg eADF4(C16)].According to the product information of SigmaeAldrich, elastase and trypsin havean optimal activity at 25 �C. Therefore, the release upon degradationwas conductedby incubation in different release media at 25 �C with constant shaking for 7 days.The solvent was periodically removed from each sample and replaced with freshrelease media. Then, the drug content in the medium was analyzed using UV-Vis-spectrometry. The percentage of cumulative methyl violet release (% w/w) wasinvestigated as a function of incubation time. Each experiment was performed intriplicate.

2.8. Characterization of eADF4(C16) particles

2.8.1. Scanning electron microscopyThe eADF4(C16) particles were immobilized on Thermanox plastic cover slips

(Nagle Nunc, USA), dried at room temperature, gold sputtered under vacuum, andanalyzedusinga JSM5900LVscanningelectronmicroscope (JEOLLtd., Japan, at20kV).

2.8.2. Laser diffraction spectrometryParticle sizes and size distributions were determined in triplicate using laser

diffraction spectrometry (Horiba, Partica LA-950, Japan). Refractive indices of 1.33for water and 1.60 for protein were taken for computation of particle sizes. In orderto eliminate concentration effects, all samples were measured at equal concentra-tions resulting in a transmittance of 82%. In addition, a dry specimen of eachpreparation was analyzed by scanning electron microscopy (SEM) to confirmspherical shape and sphere size.

2.8.3. Fourier transform infrared spectroscopy (FTIR)Fourier transform infrared (FTIR) spectra were collected using a Bruker Equinox

55 FTIR spectrometer. The samples were prepared by placing a droplet of eADF4(C16)particle suspension on CaF2 disks and subsequent air-drying. Absorbance spectrawere recorded between 400 and 4000 cm�1 with non-polarized light at a resolutionof 4 cm�1. The measurements were carried out at 25 �C and 30% relative humidity,and each spectrum was accumulated 32 times. The secondary structure of eADF4(C16) particles was analyzed using the amide I band (1600e1700 cm�1). Peaks at1648e1660 cm�1, 1625e1640 cm�1 and 1660e1668 cm�1 can be assigned toa-helical, b-sheet and b-turn structures, respectively [40]. For data comparison andelucidation of concentration effects, all spectrawere rescaled to themaximum of thecontrol sample.

2.8.4. UV-Vis-spectrometryUltravioletevisible spectroscopy, using a Cary100 spectrophotometer (Varian

Medical Systems, Palo Alto, USA), was employed for determination of the drug

concentration in supernatants as a basis for the calculation of loading efficienciesand release behavior. Calibration curves for all model drugs were obtained by usingfive different concentrations of all stock solutions.

2.8.5. Zeta-potential analysisIn order to elucidate and characterize the loading mechanism of eADF4(C16)

particles with model drugs, zeta-potential measurements were conducted as a func-tion of amount of model drug added. The zeta-potential was determined usingaNanoseriesMalvernZetasizer (Malvern,Worcestershire,UK).Automatic titrationwasconductedwith aMalvernMultipurposeTitratorMPT-2. Experimentswereperformedin distilled water (pH 7) at 25 �C. Each measurement was performed in triplicate.

3. Results and discussion

3.1. eADF4(C16) particle characterization

To characterize the morphology and determine the size ofobtained eADF4(C16) particles, the prepared stock dispersion wasexamined by SEM and laser diffraction spectrometry. As shown inFig. 1a) particles of spherical shape with diameters from 170 nm to700 nm were obtained. The average diameter of particles wasdavg ¼ 332 � 95 nm. The yield of particle formation by salting outwas higher than 99% with remaining soluble protein belowthe detection limit. We determined that eADF4(C16) particles arecolloidally stable within the complete studied concentration range

A. Lammel et al. / Biomaterials 32 (2011) 2233e22402236

from 0 to 2.0 M (NH4)2SO4 (Fig. 1b). The slight linear decrease ofintensity with increasing concentration of (NH4)2SO4 can beexplained by the linear increase in ion concentration yieldinga decrease of number of particles per volume.

3.2. Loading efficiencies and loading procedure

Due to its negative charge at pH 7, eADF4(C16) can formcomplexeswith positively chargedmolecules based on electrostaticinteractions. In order to elucidate if small molecules attach to theparticle surface or are able to permeate into the interior, wecompared loading efficiencies of glass beads with that of eADF4(C16) particles assuming that permeation processes of drug mole-cules into the dense glass matrix cannot occur. Due to the highnegative zeta-potential (z�50 mV) of glass beads, the loadingefficiency of glass beads should be higher than that of spider silkparticles (zeta-potentialz�22mV) in case no diffusion occurs intothe proteinmatrix, even under the consideration that the surface ofthe silk beadsmight be slightly rougher than that of the glass beads.

For this experiment methyl violet (MV) was employed withloading efficiencies above 95% at molar ratios of MV: eADF4(C16) of10:1. Online zeta-potential measurements during methyl violetloading revealed that the change of zeta-potential during eADF4(C16) particle loading is a tri-phasic process (Fig. 2a). First, the zeta-potential changes gradually after addition of the methyl violetsolution. After an initial constant slope, the zeta-potential curveexhibits a plateau phase indicating no further change of surfaceloading upon increasing methyl violet concentration. Finally thezeta-potential decreases further. The reduction of the zeta-poten-tial, as seen in the titration curve, is a direct consequence of the

Fig. 2. Characterization of loading procedure. a) Zeta-potential of eADF4(C16) particlesas a function of added methyl violet. For comparison the inlay shows the Zeta-potentialof glass beads with methyl violet. b) Loading and loading efficiency of eADF4(C16) withmethyl violet particles as a function of molar ratio.

interaction of the silk particles with molecules of opposite charge.The initial lowering of surface charge can be explained by thecharge compensation due to the addition of opposite chargedmethyl violet molecules. The plateau region indicates an equilib-rium state of drug adsorbed at the solid-liquid particle interfaceand a diffusion of molecules into the hydrophobic core of theprotein sphere. After the core matrix is saturated, the influx ofmethyl violet molecules is reduced and eventually terminated. Atthat point the zeta-potential starts to decrease again, as can be seenby the second slope in Fig. 2a, due to further loading of the particlesurface. Interestingly, this decrease occurs at a molar ratio of MV:eADF4(C16) of 10:1 which was identified to be the molar ratio atwhich the loading efficiency decreases. Fig. 2b shows the obtainedloading and loading efficiencies of eADF4(C16) particles as a func-tion of molar ratio. Up to a molar ratio of MV: eADF4(C16)z 10 theloading increases linearly with the amount of methyl violet added.Above a molar ratio of 10 the loading reaches a plateau leading toa decrease of loading efficiency.

In contrast, the zeta-potential of glass microparticles duringmethyl violet addition showed no distinctive changes (insetFig. 2a). The initial assumption that methyl violet cannot permeateinto the glass particle matrix was confirmed by analyzingthe supernatant after completing the titration experiment. Whilethe surface charge of glass particles is approximately two timeshigher compared to silk particles, the determined loading efficiencywas only 0.03%. Furthermore the loaded methyl violet could beeasily washed off the surface of glass particles by three washingsteps using Millipore water.

In order to investigate the influence of molecular parameters onthe loading efficiency, twelve different lowmolecular weight drugswere used in this study (Table 1). Since an individual eADF4(C16)molecule is amphiphilic with a dominating hydrophobic character(hydropathicity index ¼ �0.46), exhibits 17 negative charges (oneat each C module and one at the carboxyl terminus), one positivecharge at the amino terminus, and one positive charge in the T7-TAG, we concluded that loading of eADF4(C16) particles with drugsis mainly driven by three parameters: the charge of the drugmolecule determined by its proton dissociation constant Ka

(accounted for BHþ or HA); the octanol water partition coefficient(logPo/w) as an indicator of solubility of the model drug; themolecular weight (MW) which plays an important role in diffusiondriven mass transport processes. The distribution betweena hydrophobic and a hydrophilic phase of two different species ofa specific drug, i.e. the native and the protonated form, can bedescribed by its apparent distribution coefficient (logD), which canbe calculated using Eqs. (3) and (4), respectively [46].For acids:

log D ¼ log P � log�1þ 10pH�pKa� (3)

For bases:

log D ¼ log P � log�1þ 10pKa�pH� (4)

The logP and pKa values of individual species used for calculationof logD are listed in Table 1. Table 2 summarizes the determinedloading efficiencies, maximal (calculated by employing loadingefficiencies of 100%) and experimental amount of entrapped drug,as well as the calculated distribution coefficient (logD) at pH 7.

Protonated weak organic bases could be loaded onto eADF4(C16) particles with efficiencies ranging between 20.7% and 53.0%.For this class of low molecular weight drugs the ratio of calculatedlogD divided by the molecular weight of the individual moleculecorrelates linearly with the obtained loading efficiencies (Fig. 3).This linear relationship clearly indicates that the combination ofcharge and solubility (expressed by the apparent distribution

Table 2List of employedmodel drugs classified according to their chemical nature. The table provides an overview of theoretical and experimental model drug content of loaded spidersilk particles (expressed as percentage of wt drug/wt spider silk protein particles), corresponding encapsulation efficiencies and calculated distribution coefficients (logD).

Model drug Chemicalnature

Maximal drug content [w/w%] Experimental drug content [w/w%] Encapsulation efficiency [%] logD

Ephedrin*HCl Base 4.23 0.88 20.7 �1.321Procain*HCl Base 5.71 2.16 38.0 0.396Biperiden*HCl Base 7.28 3.05 42.0 0.899Propranolol*HCl Base 6.19 2.78 45.0 1.197Papaverine*HCl Base 7.89 3.71 47.0 2.395Tetracaine*HCl Base 6.30 3.34 53.0 2.773Ethacridine lactat Strong base 7.20 7.07 98.2 2.899Acetaminophen Weak acid 3.16 0.06 0.2 0.378Chlortalidon Weak acid 7.09 0.68 9.7 �0.315Nipagin Weak acid 3.19 0.55 17.3 0.544Phenol red Strong acid 7.12 0.00 0.0 e

Methyl violet e 8.54 8.37 98.1 e

A. Lammel et al. / Biomaterials 32 (2011) 2233e2240 2237

coefficient logD) and diffusion coefficient (expressed by the inverseproportionality of molecular weight) are the dominating factorsresponsible for effective loading of eADF4(C16) particles with smallweakly alkaline molecules.

Investigation of molecules with permanent charge revealed thatpositively charged molecules such as methyl violet were mostsuccessfully incorporated, whereas negatively charged moleculessuch as phenol red could not be incorporated in eADF4(C16) parti-cles, and slightly acidic molecules yielded relatively low loadingefficiencies from 0.2 to 17.3%. Strongly alkaline molecules likeethacridine lactate showed a loading efficiency of more than 98%.Despite the relatively high logD of acetaminophen (comparable toNipagin), a loading efficiency of only 0.2% was determined, whichmight be explained by its structure acting as weak phenolic baseleading to the appearance of negatively charged acetaminophenspecies. We suspect that due to electrostatic repulsion of the nega-tive charges these species cannot diffuse into the sphere matrix.

According to our results, the distribution coefficient logD can beused to estimate the obtainable loading efficiency for weak organicbases without permanent charge. Molecules with permanentcharge cannot be included in this model since it is not possible tocalculate the corresponding logD value. Due to the physico-chemicalproperties of eADF4(C16) and particles made thereof, for efficientloading drug molecules should ideally feature a positive chargecombined with a certain degree of hydrophobicity and a molecularweight enabling diffusion into the spider silk protein matrix.

Fig. 3. Loading efficiencies of eADF4(C16) particles for model drugs of weak alkalinenature such as Ephedrin (Eph), Procain (Prc), Biperiden (Bip), Propranolol (Prp),Papaverine (Pap) and Tetracaine (Tet) plotted over logD MW�1.

3.3. In vitro release studies

The in vitro release behavior of model drugs from eADF4(C16)particles was exemplarily studied with methyl violet and ethacri-dine lactate. Cumulative release profiles showed that both mole-cules were released over a period of 30 days (Fig. 4a). Mostinterestingly, only a very small drug burst could be detected, i.e. an

Fig. 4. Release studies of ethacridine lactate and methyl violet. a) Experimental andtheoretical release kinetics of both model drugs over a period of 35 days. b) Release ofethacridine lactate as a function of pH. (Buffer capacity PBS: pH 5.8 e pH 8; nonbuffered conditions for pH 2.0, pH 4.0 and pH 8.8).

A. Lammel et al. / Biomaterials 32 (2011) 2233e22402238

initial higher drug release within the first 24 h of incubation. Therelease of ethacridine lactate and methyl violet within the first 24 hwas 11% of the total amount encapsulated. Subsequently, eADF4(C16) particles released approximately 5% of the entrapped mole-cules per day within the first week (Fig. 4a).

Next, the influence of pH on drug release was evaluated. Releaseexperiments with ethacridine lactate loaded eADF4(C16) particlesincubated in PBS at 37 �C, 40 rpm and different pH values showed

Fig. 5. Characterization of eADF4(C16) particles upon enzymatic degradation. a) Size distribu(C16) particles distribution over time. c) Percentage of particles and agglomerations of eADF4d) Second derivative of FTIR spectra of eADF4(C16) particles upon degradation. e) Cumulaconcentrations as indicated*. f) Averaged release rate between two successive measuremetrations as indicated*. *[control: no enzymes; C1: (0.8 mg elastase þ 12.5 mg trypsin)/1 melastase þ 37.5 mg trypsin)/1 mg eADF4(C16); C4: (4.0 mg elastase þ 62.5 mg trypsin)/1 mg

a strong pH influence on the release rates (Fig. 4b) with an acidicenvironment accelerating drug release. In order to investigate therelease mechanism, also non-physiological pH values (pH 4 and pH2) were tested. Almost 80% of the loaded drug was released after24 h from silk spheres incubated at pH 2 (non buffered conditions).For silk particles incubated at pH 4 (non buffered conditions) aninitial release of almost 40% was obtained after the first day ofincubation. Particles incubated at pH 6 showed even less release,

tion of eADF4(C16) particles upon enzymatic degradation. b) Mean and mode of eADF4(C16) particles after degradationwith elastase (c ¼ 4 mg/ml) and trypsin (c ¼ 50 mg/ml).tive release of methyl violet upon degradation with elastase and trypsin at differentnts of methyl violet upon degradation with elastase and trypsin at different concen-g eADF4(C16); C2: (1.6 mg elastase þ 25 mg trypsin)/1 mg eADF4(C16); C3: (2.4 mgeADF4(C16); C5: (8.0 mg elastase þ 125 mg trypsin)/1 mg eADF4(C16)].

Fig. 6. Loading and release mechanism for spider silk particles. 1) Drug molecules are attracted by the particle through electrostatic forces. 2) After particle surface saturation lowmolecular weight drugs start to diffuse into the sphere matrix. 3) Drug molecules are bonded by attractive hydrophobic and electrostatic interactions. 4) After complete loading andincubation in release media drug molecules are transported to the particle surface due to concentration gradient driven transport processes. 5) Upon time drug molecules are slowlyreleased from the surface allowing constant release rates.

A. Lammel et al. / Biomaterials 32 (2011) 2233e2240 2239

and the lowest release was seen at pH 7.4 and 8.8. The observedresults confirm the predicted importance of electrostatic interac-tions between eADF4(C16) and drug molecules. Presumably, aninflux of protons into the biopolymer sphere leads to a displace-ment of drug molecules from the matrix. In addition, the decreasedpH influences the distribution of charged drug species by shiftingthe equilibrium towards the charged species. As these species aredriven towards the negatively charged surface of the protein, theycan easily be washed away by the solvent.

3.4. In vitro degradation

Degradability of drug depot systems is a highly desirableproperty, since the risk of inflammation and intoxication isdramatically lower than for non-degradable systems [41,42]. Asmost biopolymers feature the ability of enzymatic degradation [43],degradation studies were conducted using proteases (trypsin andelastase) [44,45] naturally occurring in vertebrates simulatingbiodegradation of eADF4(C16) drug carriers. Elastase and trypsin,i.e. serine proteases attack the carbonyl carbon of peptide bonds ofsmall, hydrophobic amino acids such as glycine, alanine, and valine,leading to efficient peptide bond cleavage [46]. Due to the relativelyhigh content of glycine and alanine in eADF4(C16) (z50 % of thetotal amino acid composition), such proteases may cleave severalpeptide bonds of the primary structure of eADF4(C16).

Size and morphology analysis of particle ensembles taken fromdegradation experiments using LDS and SEM showed that after twodays of degradation particles form clusters (Fig. 5a). Comparing themode value, which represents the particle size most commonlyfound in the distribution, with the mean size of the particles leadsto the conclusion that bigger particles of the ensemble aredegraded preferentially (Fig. 5b). At t ¼ 0 the mean is larger thanthemode indicating an asymmetric size distribution towards largerparticles. Upon enzymatic degradation for two days, mean andmode approach each other indicating that larger particles disap-pear and the particle distribution becomes symmetric. The particledistribution remains symmetrical up to day 8 on which the meanfalls below the mode, indicating an asymmetric size distributiontowards smaller particles. Analysis of the relative relation of singleparticles to agglomerations indicates an oscillatory agglomerativebehavior (Fig. 5a,c). The oscillatory behavior is presumably due to

changes in surface structure upon degradation. Seemingly, uponenzymatic degradation there is a dynamic equilibrium between anintermediate surface structure state which cannot be recognized bythe chosen enzymes (elastase & trypsin). Presumably, in a first stepenzymatic degradation renders the initial hydrophilic surface morehydrophobic. Therefore, particle agglomeration is fostered in orderto minimize the particles’ surface free energy. Upon time, theparticle surface structure changes due to protein rearrangementuntil a new surface exhibits an enzymatic degradation site, andagglomerated particles can be further degraded.

Secondary structure of eADF(C16) particles during degradationexperiments was detected by analyzing the 2nd derivative changesof FTIR spectra at the wavenumbers 1648e1660 cm�1 and1625e1640 cm�1. Our results indicate thatonlyminor changes in theratio of b-sheet and a-helical content occur, and, thereby,the structural features of eADF4(C16) particles are apparentlyconserved. This is an important result regarding the long-termstability and release behavior of eADF4(C16) particles at physiolo-gical conditions since structural changes would significantly alterthe release properties.

The release study upon degradation employing methyl violetloaded eADF(C16) particles revealed that during thefirst day the totalrelease is quite equal independent of the amount of enzyme added(Fig. 5e). From day two on, a clear trend can be seen that increasingenzymeconcentrationyields toanaccelerated releasebehavior. Fig. 5fillustrates the average release rate between two successive releasemeasurements. The results show that for higher enzyme concentra-tions (C3, C4, and C5) the release rate increases upon incubation time(Fig. 5f). Furthermore, the higher the enzyme concentration theearlier the maximum release rate is reached (Fig. 5f).

4. Conclusion

Our results show the potential to use engineered spider silkprotein eADF4(C16) particles for sustained controlled delivery ofpositively charged and sufficiently hydrophobic drug molecules.According to the presented findings, spider silk particle loading andrelease follows the following mechanism: In a first step drugmolecules are attracted to a silk particle by electrostatic forces.After particle surface saturation low molecular weight drugs startto diffuse into the biopolymer matrix. Drug molecules are bonded

A. Lammel et al. / Biomaterials 32 (2011) 2233e22402240

by attractive hydrophobic and electrostatic interactions. Aftercomplete loading and incubation in release media, drug moleculesare transported to the particle surface due to concentrationgradient driven transport processes. Upon time, drug molecules areslowly released from the surface leading to constant release rates(Fig. 6). One major advantage of the system is that eADF4(C16)particles can be produced and loadedwithin an all-aqueous processunder ambient conditions, which is extremely important consid-ering encapsulation of labile compounds and the biocompatibilityof the product. We conclude that engineered spider silk particleshave the potential for diverse applications where controlled releasefrom mechanically tough and slowly biodegradable carriers isdesirable.

Acknowledgements

This work was supported by the International Graduate Schoolof Science and Engineering (IGSSE to AL) within the Elite Networkof Bavaria and DFG(SCHE 603/9e1) to TS.

Appendix

Figures with essential colour discrimination. Certain figures inthis article, particularly Fig. 5, are difficult to interpret in black andwhite. The full colour images can be found in the online version, atdoi:10.1016/j.biomaterials.2010.11.060.

References

[1] Yasutomo N, Yoshiharu N, Masahisa W, Shigenori K, Jun M. Mechanical prop-erties of silk fibroin-microcrystalline cellulose composite films. J Appl Polym Sci2002;86:3425e9.

[2] Vasir JK, Tambwekar K, Garg S. Bioadhesive microspheres as a controlled drugdelivery system. Int J Pharm 2003;255:13e32.

[3] Roser M, Fischer D, Kissel T. Surface-modified biodegradable albumin nano-and microspheres II: effect of surface charges on in vitro phagocytosis andbiodistribution in rats. Europ J Pharm Biopharm 1998;46:255e63.

[4] Uhrich KE, Cannizzaro SM, Langer RS, Shakesheff M. Polymeric systems forcontrolled drug release. Chem Rev 1999;99:3181e98.

[5] Allen TM, Cullis PR. Drug delivery systems: entering the mainstream. Science2004;303:1818e22.

[6] Edlund U, Albertsson AC. Degradable polymer microspheres for controlleddrug delivery. In: Albertsson AC, editor. Degradable aliphatic polyesters.Berlin: Springer; 2002. p. 67e112.

[7] Arshady R. Biodegradable microcapsular drug delivery systems:manufacturing methodology, release control and targeting prospects. J BioactCompat Pol 1990;5:315e42.

[8] Altman GH, Diaz F, Jakuba C, Calabro T, Horan RL, Chen J, et al. Silk-basedbiomaterials. Biomaterials 2003;24:401e16.

[9] Freiberg S, Zhu XX. Polymer microspheres for controlled drug release. IntJ Pharm 2004;282:1e18.

[10] Langer R, Peppas NA. Advances in biomaterials, drug delivery, and bionano-technology. AIChE J 2003;49:2990e3006.

[11] George M, Abraham TE. Polyionic hydrocolloids for the intestinal delivery ofproteindrugs: alginateandchitosan- a review. JControlRelease2006;114:1e14.

[12] Liu X, Sun Q, Wang H, Zhang L, Wang JY. Microspheres of corn protein, zein,for an ivermectin drug delivery system. Biomaterials 2005;26:109e15.

[13] Won YW, Kim YH. Recombinant human gelatin nanoparticles as a proteindrug carrier. J Control Release 2008;127:154e61.

[14] Morlock M, Koll H, Winter G, Kissel T. Microencapsulation of rh-erythropoi-etin, using biodegradable poly(,-lactide-co-glycolide): protein stability andthe effects of stabilizing excipients. Europ J Pharm Biopharm 1997;43:29e36.

[15] Soppimath KS, Aminabhavi TM, Kulkarni AR, Rudzinski WE. Biodegradable poly-meric nanoparticles as drug delivery devices. J Control Release 2001;70:1e20.

[16] Vandelli MA, Rivasi F, Guerra P, Forni F, Arletti R. Gelatin microspheres cross-linked with ,-glyceraldehyde as a potential drug delivery system: preparation,characterisation, in vitro and in vivo studies. Int J Pharm 2001;215:175e84.

[17] Sahin S, Selek H, Ponchel G, Ercan MT, Sargon M, Hincal AA, et al. Preparation,characterization and in vivo distribution of terbutaline sulfate loaded albuminmicrospheres. J Control Release 2002;82:345e58.

[18] Latha MS, Rathinam K, Mohanan PV, Jayakrishnan A. Bioavailability of theoph-ylline from glutaraldehyde cross-linked casein microspheres in rabbitsfollowing oral administration. J Control Release 1995;34:1e7.

[19] Speer DP, Chvapil M, Eskelson CD, Ulreich J. Biological effects of residualglutaraldehyde in glutaraldehyde-tanned collagen biomaterials. J BiomedMate Res 1980;14:753e64.

[20] Digenis GA, Gold TB, Shah VP. Cross-linking of gelatin capsules and its rele-vance to their in vitro-in vivo performance. J Pharm Sci 1994;83:915e21.

[21] Vollrath F, Barth P, Basedow A, Engström W, List H. Local tolerance to spidersilks and protein polymers in vivo. In Vivo 2002;16:229e34.

[22] Allmeling C, Jokuszies A, Reimers K, Kall S, Choi CY, Brandes G, et al. Spider silkfibres in artificial nerve constructs promote peripheral nerve regeneration.Cell Prolif 2008;41:408e20.

[23] Lammel A, Keerl D, Römer L, Scheibel T. Proteins: polymers of natural origin. In:Hu J, editor. Recent advances in biomaterials research. Trivandrum: TransworldResearch Network; 2008. p. 1e22.

[24] Hardy JG, Römer LM, Scheibel TR. Polymeric materials based on silk proteins.Polymer 2008;49:4309e27.

[25] Römer L, Scheibel T. The elaborate structure of spider silk: structure andfunction of a natural high performance fiber. Prion 2008;2:1e8.

[26] Heim M, Keerl D, Scheibel T. Spider silk: from soluble protein to extraordinaryfiber. Angew Chem Int Ed Engl 2009;48:3584e96.

[27] Spieß K, Lammel A, Scheibel T. Recombinant spider silk proteins for applica-tions in biomaterials. Macromol Biosci 2010;10:998e1007.

[28] Wenk E, Wandrey AJ, Merkle HP, Meinel L. Silk fibroin spheres as a platformfor controlled drug delivery. J Control Release 2008;132:26e34.

[29] Liebmann B, Hümmerich D, Scheibel T, Fehr M. Formulation of poorly water-soluble substances using self-assembling spider silk protein. Colloid Surf A2008;331:126e32.

[30] Lammel A, Hu X, Park S-H, Kaplan D, Scheibel T. Controlling silk fibroinparticle features for drug delivery. Biomaterials 2010;31:4583e91.

[31] Wang X, Wenk E, Matsumoto A, Meinel L, Li C, Kaplan DL. Silk micro-spheres for encapsulation and controlled release. J Control Release 2007;117:360e70.

[32] Scheibel T. Spider silks. recombinant synthesis, assembly, spinning, andengineering of synthetic proteins. Microb Cell Fact 2004;3:14.

[33] Huemmerich D, Helsen CW, Quedzuweit S, Oschmann J, Rudolph R, Scheibel T.Primary structure elements of spider dragline silks and their contribution toprotein solubility. Biochemistry 2004;43:13604e12.

[34] Slotta UK, Rammensee S, Gorb S, Scheibel T. An engineered spider silk proteinforms microspheres. Angew Chem Int Ed Engl 2008;47:4592e4.

[35] Rammensee S, Slotta U, Scheibel T, Bausch AR. Assembly mechanism ofrecombinant spider silk proteins. P Natl Acad Sci USA 2008;105:6590e5.

[36] Lammel A, Schwab M, Slotta U, Winter G, Scheibel T. Processing conditionsfor the formation of spider silk microspheres. Chem Sus Chem 2008;1:413e6.

[37] Morrison ID, Ross S. Sizing by light scattering. In: Morrison ID, Ross S, editors.Colloidal dispersions e suspensions, emulsions and foams. New York: JohnWiley & Sons; 2002. p. 87e94.

[38] Riley T, Govender T, Stolnik S, Xiong CD, Garnett MC, Illum L, et al. Colloidalstability and drug incorporation aspects of micellar-like PLA-PEG nano-particles. Colloid Surf B 1999;16:147e59.

[39] Boyer. The enzymes e volume III. New York: Academic Press; 1971.p. 323e373.

[40] Tamm LK, Tatulian SA. Infrared spectroscopy of proteins and peptides in lipidbilayers. Q Rev Biophys 1997;30:365e429.

[41] Shard AG, Tomlins PE. Biocompatibility and the efficacy of medical implants.Regen Med 2006;1:789e800.

[42] Schlosser M, Wilhelm L, Urban G, Ziegler B, Ziegler M, Zippler R. Immuno-genicity of polymeric implants: long-term antibody response against poly-ester (Dacron) following the implantation of vascular prostheses into LEW.1Arats. J Biomed Mater Res 2002;61:450e7.

[43] Nair LS, Laurencin CT. Biodegradable polymers as biomaterials. Prog Polym Sci2007;32:762e98.

[44] Shimizu T, Yamashiro Y, Igarashi J, Fujita H, Ishimoto K. Increased serumtrypsin and elastase-1 levels in patients undergoing l-asparaginase therapy.Europ J Pediatrics 1998;157:561e3.

[45] Del Favero G, Fabris C, Plebani M, Panucci A, Piccoli A, Perobelli L, et al. Serumelastase 1 in chronic pancreatic disease. J Mol Med 1985;63:603e6.

[46] Page MJ, DiCera E. Serine proteases and serine protease inhibitors. In:Begley TP, editor. Encyclopedia of chemical biology. New York: John Wiley &Sons; 2008.